Upregulation of casein kinase 1ε in dorsal root ganglia and spinal cord after mouse spinal nerve injury contributes to neuropathic pain

Using cDNA microarray technique, we previously reported that the expression of ~900 genes out of ~10,000 genes in the Ca _v 2.2^+/+ spinal cord was increased more than 1.20-fold by the lumbar L5 and L6 spinal nerve ligation (L5/L6 SNL) injury as compared with sham-operated Ca _v 2.2^+/+ mice. Among these genes, we recently suggested that glucocorticoid receptor [10] and peripheral-type benzodiazepine receptor [11] are useful targets in the management of neuropathic pain. On the other hand, we also observed that the expression of ~1,300 genes out of ~10,000 genes was reduced more than 1.20-fold by L5/L6 SNL injury in the Ca _v 2.2^-/- spinal cord when compared with sham-operated Ca _v 2.2^-/- mice. Because Ca _v 2.2^-/- mice showed markedly reduced symptoms of the SNL-induced neuropathic pain [3], we speculated that these down-regulated genes have some contribution to the reduced neuropathic pain symptoms. Among these down-regulated genes, we focused on CK1ε in this study, because there has been no information about the expression and function of this molecule in sensory pathway up to this moment, in spite of the fact that CK1 is one of the first serine/threonine protein kinases that were isolated and characterized [5]. The expression of CK1ε mRNA was not changed in Ca _v 2.2^+/+ mice (1.01-fold) but decreased by 2.81-fold in Ca _v 2.2^-/- mice 2 weeks after SNL injury. Quantitative real-time PCR analysis of C57BL/6J mice 2-3 weeks after sham or SNL operation also showed no difference of the CK1ε mRNA expression in the spinal cord (n = 4 for sham and SNL, data not shown).

We next examined the expression of CK1ε protein in the spinal cord by immunoblot analyses (Figure 1A and 1B). The CK1ε expression was significantly enhanced in the Ca _v 2.2^+/+ spinal cord 2 weeks after SNL injury. In contrast, CK1ε expression in the SNL-operated Ca _v 2.2^-/- spinal cord was significantly reduced, which is consistent with the cDNA microarray results.

We further characterized the distribution of CK1ε-positive cells in the spinal cord of sham and neuropathic mice by immunofluorescence techniques. As shown in Figure 1C-E, the CK1ε protein was found to be broadly expressed in the spinal cord.

Consistent with the immunoblot data, the intensity of CK1ε-immunoreactivity (CK1ε-IR) in the dorsal horn of the spinal cord was increased on the ipsilateral side to the nerve injury.

To further characterize the localization profile of the CK1ε, we performed double staining of CK1ε with cell-type specific markers. CK1ε-IR was colocalized with a neuronal marker, neuronal specific nuclear protein (NeuN) (Figure 2Ai-Ci), but not with an astorocytic marker, glial fibrillary acid protein (GFAP) (Figure 2Aii-Cii) or a microglial marker, ionized calcium binding adaptor molecule 1 (Iba 1) (Figure 2Aiii-Ciii). Both GFAP and Iba 1 immunostainings were significantly enhanced on the side ipsilateral to SNL injury (Figure 2Bii, Biii, D and 2E), a finding consistent with the literature [12, 13]. Intriguingly, SNL-induced enhancements of GFAP- and Iba 1-IR were markedly suppressed in Ca _v 2.2^-/- mice (Figure 2Cii, 2Ciii, 2D and 2E). On the other hand, we found SNL injury did not affect the number of NeuN-positive cells in spinal dorsal horn both in wild-type and Ca _v 2.2^-/- mice.

Double staining experiments also revealed that the CK1ε was expressed in 71% (superficial layer) and 66% (middle layer) of the neurons in the L5 dorsal horn of sham-operated spinal cord (Figure 3A and 3B). Two weeks after SNL injury, the percentage was increased to 87% and 75% (superficial and middle layers, respectively) in the L5 dorsal horn ipsilateral to the injury (Figure 3A and 3B) without significant changes in the contralateral side (data not shown). In Ca _v 2.2^-/- spinal cord, SNL injury, in contrast, significantly reduced the percentage of the CK1ε-positive neurons in both ipsilateral superficial and middle layers (Figure 3A and 3B). There were no overt differences in the numbers between right and left sides in naive control and between ipsilateral and contralateral sides in sham-operated animals (data not shown).

Although the increase of CK1ε-positive neurons after SNL injury would explain the upregulation of CK1ε protein revealed by the immunoblot experiments, we also tested another possibility that SNL injury increased the expression level of CK1ε protein in the dorsal horn neurons. Confocal microscopic analyses indicated that the expression of CK1ε protein was indeed up-regulated in the ipsilateral superficial dorsal horn neurons of C57BL/6J mice 2 weeks after SNL injury, whereas the same injury slightly reduced the CK1ε expression in Ca _v 2.2^-/- spinal cord (Figure 3C-F). CK1ε-IR was detected mostly in the cytoplasm, with only faint staining in the nucleus.

To further address the feature of CK1ε expression at neuropathic pain state, we have investigated whether CK1ε is expressed in primary sensory neurons and whether the expression pattern of CK1ε is altered by the SNL injury by immunofluorescence analyses. In agreement with the previous reports [14‐16], we observed marked loss (50%) of ipsilateral L5 DRG neurons 2 weeks after the nerve injury when compared with the ipsilateral side of sham operated mice (Table 1). Most dramatic decrease (87%) was observed in large-sized group, and 53% and 38% losses were observed in small- and medium-sized groups, respectively (Table 1). Similar level of cell loss was also observed in SNL-operated Ca _v 2.2^-/- mice (Table 1).

In L5 DRG, the CK1ε-IR was mainly observed in small- and medium-sized neurons with IR appearing largely in cytoplasm with weak staining in the nucleus (Figure 4). The percentages of CK1ε-positive neurons ipsilateral to sham operation were 68.4%, 29.1% and 15.1% in small-, medium- and large-sized neurons, respectively (Figure 4D). The percentages of CK1ε-positive neurons in the DRG ipsilateral to sham operation were similar to those in contralateral to sham operation and SNL injury, and in naive DRG (data not shown). On the other hand, significant increases in the percentages of CK1ε-positive neurons in the ipsilateral L5 DRG compared with sham group were observed in all size groups 2 weeks after SNL injury (Figure 4B and 4D). Similar changes of CK1ε-positive cell population were also observed in SNL-operated Ca _v 2.2^-/- mice (Figure 4C and 4D). Expression level of CK1ε protein was also analyzed by a computerized image analysis system (Figure 4E). When compared with sham group, significant increases in the intensity of CK1ε-IR were observed in the small- and medium-sized neurons of ipsilateral L5 DRG 2 weeks after SNL injury.

Similar extent of increase was also observed in SNL-operated Ca _v 2.2^-/- mice (Figure 4E). In contrast, the intensity of CK1ε-IR in large-sized neurons was not significantly changed following SNL injury. To clarify the expression profiles of CK1ε protein in sham and neuropathic mice L5 DRGs, we carried out the double immunofluorescence analysis using CK1ε antibodies together with antibodies against two DRG neuron markers. Small and medium-sized DRG neurons were generally classified into two subgroups, which have been designated as peptidergic and non-pepetidergic [17]. The former expresses two major peptidergic neuromodulators, substance P and calcitonin gene-related peptide (CGRP), and the latter expresses isolectin B4 (IB4)-binding glycoprotein. CK1ε protein was found in both CGRP- and IB4-positive populations (Figure 5). The CK1ε expression was detected in 66.4% of CGRP-positive neurons and 44.9% of CK1ε-positive neurons were CGRP-positive in sham operated DRG neurons (Figure 5). Similarly, the CK1ε expression was detected in 82.3% of IB4-positive neurons and 38.3% of CK1ε-positive neurons were IB4-positive. Similar results were also obtained from naive DRG neurons (data not shown). Two weeks after SNL, a marked decrease of IB4 binding (both the cell number and the intensity) was observed in ipsilateral L5 DRGs (Figure 5D), which is similar to that shown by a previous study [16]. Relatively lesser extent of decrease occurred in the number and the intensity of CGRP-positive small- and medium sized neurons (Figure 5B). In injured DRG neurons, CK1ε-IR was detected in 94% and 91% of CGRP- and IB4-positive neurons, respectively. CGRP-IR was detected in 65% of CK1ε-positive neurons, however, IB4 binding was observed in only 3% of CK1ε-positive neurons. The proportions of CK1ε-expressing cells within CGRP- and IB4-positive neurons in the contralateral DRGs were not different from those of naive and sham-operated DRGs (data not shown). The colocalization of CK1ε with CGRP observed in the superficial dorsal horn area of spinal cord slice preparations (Figure 5E and 5F) suggest that part of the CK1ε protein detected by immunoblot analyses (Figure 1A and 1B) and CK1ε-IR in the spinal cord (Figure 1D) originated from the CK1ε present at the primary afferent fibers and terminals, because CGRP is generally accepted as a pure primary afferent marker [18]. It seems clear that the intensity of CK1ε protein in the primary afferent fibers and terminals is enhanced in CGRP-positive DRG neurons after SNL injury (Figure 5F). Because enhanced protein expression of CK1ε was observed in DRG neurons, we have analyzed mRNA level by quantitative real-time PCR method and found that the 2.03 fold increase of CK1ε mRNA was observed 2 weeks after SNL operation (n = 4 for sham and SNL, p < 0.001, data not shown).

To investigate whether the up-regulation of CK1ε protein is involved in the maintenance of neuropathic pain, we next examined the effects of a CK1 inhibitor IC261 [19]. I.t. injections of IC261 (0.1-1 nmol) dose-dependently increased both withdrawal threshold and withdrawal latency of the hind paw ipsilateral to SNL (Figure 6). Another CK1 inhibitor CKI-7 [20] also showed similar effects (data not shown). These results suggest that blocking the CK1ε activity at the spinal level is effective in reducing mechanical allodynia and thermal hyperalgesia. The maximum effects were observed 0.5-1 h after the injection of IC261 and significant analgesic effects on both mechanical allodynia and thermal hyperalgesia were still observed 3-4 h after the injection of the highest doses used in this study (Figure 6A and 6B). IC261 had no significant effects on the contralateral hind paw (Figure 6A and 6B).

Intrathecal injection of saline or dimethyl sulfoxide (DMSO; 3% in saline) used as a solvent for the drugs did not show any effects on mechanical allodynia and thermal hyperalgesia (Figure 6C and 6D).

To explore the mechanism of CK1 inhibitor-induced analgesia at the spinal level, we made L5 dorsal root attached-spinal cord slice preparation from adult mice (9-12 weeks old) and carried out spatio-temporal analyses of the primary afferent fiber-evoked excitatory responses in the dorsal horn by means of imaging techniques using a voltage-sensitive dye 4-(2-(6-(dibutylamino)-2-naphthalenyl)ethenyl)-1-(3-sulfopropyl)-pyridinium hydroxide (di-4-ANEPPS). Repetitive stimulation (10 pulses at 20 Hz) of L5 dorsal root produced prolonged (lasting for 3-4 s) and widely propagating excitatory responses extending from superficial layer to deeper laminae within the ipsilateral L5 spinal dorsal horn (Figure 7). In sham and SNL animals, the magnitude of integrated area of the optical response recorded in the superficial layer of each animal group were significantly larger than those recorded in the corresponding middle layer (Figure 7). Remarkably, the optical responses were not reduced by SNL injury (Figure 7Ci and 7Di), in spite of the fact that 50% of the neurons were lost in DRG. Excitatory synaptic transmission evoked by primary afferents is known to be mainly mediated by glutamate [18, 21], and glutamate receptors, especially NMDA receptor, are considered to play important roles in development and maintenance of neuropathic pain [22, 23]. Application of an NMDA-receptor antagonist, D-APV (50 μM), suppressed the optical responses in the superficial and middle layers elicited by the nerve stimulation in sham and SNL animals. A perfusion of a solution containing both D-APV (50 μM) and a non-NMDA-receptor antagonist, CNQX (20 μM), further reduced the optical responses in both layers in sham and SNL animals (Figure 7). These results suggest that the activation of glutamate receptors is largely responsible for the induction of the excitatory responses evoked by the repetitive electrical stimulation of the primary afferents. To identify a role of CK1ε in neuropathic pain-related spinal nociceptive transmission, we investigated the effects of the CK1 inhibitor on the dorsal root-evoke optical responses (Figure 8). IC261 (1 and 2 μM) showed significant inhibitory effects on the optical responses in the SNL dorsal horn (Figure 8B and 8D). Similar results were obtained when using another CK1 inhibitor CKI-7 (data not shown).

Interestingly, IC261 did not reduce the dorsal root-evoked optical responses elicited in the dorsal horn of sham mice (Figure 8A and 8C). Vehicle control (0.01 and 0.03% DMSO in ACSF) did not change the optical responses evoked in both superficial and middle layers of naive dorsal horn (data not shown). Thus CK1 inhibitor was found to be effective in reducing spinal excitatory response elicited by the presynaptic electrical stimulation only in neuropathic mice.

To test whether the observed effects of CK1 inhibitor originated from the direct effects on the postsynaptic glutamate receptors, we further examined the effect of IC261 on the optical responses induced by NMDA and glutamate in the SNL spinal cord in the presence of tetrodotoxin (TTX; 0.3 μM).

In both superficial and middle layer of the SNL spinal cord, excitatory optical responses evoked by the bath-application of NMDA (300 μM) for 30 s was not inhibited by 1 μM of IC261 but significantly inhibited by 3 μM of IC261 (Figure 9A). These results suggest that blockade of primary afferent fiber-evoked excitatory responses in the dorsal horn by IC261 (Figure 8B and 8D) was partly caused by the blockade of NMDA evoked responses, because only higher concentration of IC261 was effective. On the other hand, 3 μM of IC261 did not show any effect on the excitatory optical responses evoked by the bath-application of glutamate (3 mM) for 1 min (Figure 9B).

Male C57BL/6J mice (7-8 weeks old at the time of operation) were purchased from Clea Japan, Inc. (Tokyo, Japan) and housed under controlled temperature (24 ± 1°C) and humidity (55 ± 10%) with a 12-h light-dark cycle with food and water freely available. Experiments were conducted with the approval of the Animal Care Committee of Tokyo Medical and Dental University (approval No. 0090173), and according to the ethical guidelines for the study of experimental pain in conscious animals published by the International Association of the Study of Pain [29].

L5/L6 SNL was carried out as described previously [3, 10].

I.t. injection was given in a volume of 5 μl by percutaneous puncture through an intervertebral space at the level of the 5th or 6th lumbar vertebra, according to a previously reported procedure [30, 31].

Behavioral studies were conducted in a sound proof room during the light cycle (8:00 a.m.-8:00 p.m.) 2-4 weeks after spinal nerve ligation as described [32]. An investigator, who was unaware of the drug treatment, performed all of the behavioral experiments.

Ca _v 2.2^-/- mice were generated and housed as previously reported [3]. Seven Ca _v 2.2^+/+ and Ca _v 2.2^-/- mice were used for L5/L6 SNL surgery and five Ca _v 2.2^+/+ and Ca _v 2.2^-/- mice were used for sham surgery. cDNA microarray analysis was performed using the CodeLink™ UniSet Mouse 10K I (GE Healthcare, Piscataway, NJ) following the protocol provided by the manufacturer. Four data sets (Ca _v 2.2^+/+ sham, Ca _v 2.2^+/+ SNL, Ca _v 2.2^-/- sham, Ca _v 2.2^-/- SNL) were compared using the CodeLink™ System Software.

Proteins were separated by SDS-PAGE (7.5% gel) and then transferred to a polyvinylidene difluoride membrane (Millipore, Billerica, MA). Anti-CK1ε antibody was used (rabbit; 1: 1000; Santa Cruz Biotechnology, Santa Cruz, CA). We have also tested other anti-CK1ε antibody (mouse; 1: 500; BD Transduction Laboratories, Franklin Lakes, NJ) and found that they showed similar results (single band with same size in immunoblot analysis, data not shown). Immunoreactivity was detected by using the ECL system (GE Healthcare, Buckinghamshire, UK). An anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH) antibody (mouse, 1:20,000; Chemicon, Temecula, CA) was used to normalize protein loading. Relative intensities of the bands were quantified by using an image analysis system with Image J software, version 1.40 g (National Institutes of Health, Bethesda, MD).

Transverse spinal and DRG sections (10 μm) were used. The antibodies used are as follows: CK1ε (rabbit, 1:50; Santa Cruz Biotechnology; mouse, 1:50; BD Transduction Laboratories), CGRP (rabbit, 1:1000; Sigma), NeuN (mouse, 1:100; Chemicon), GFAP (mouse, 1:400; Chemicon) and Iba 1 (rabbit, 1:500, Wako Pure Chemical Industries, Ltd., Osaka, Japan). The sections were then incubated for 2 h at room temperature with Alexa Fluor 488-labeled donkey anti-rabbit IgG (1:1000; Invitrogen, Paisley, UK) or Cy3-conjugated donkey anti-mouse IgG (1:1000; Jackson ImmunoResearch, West Grove, PA). For the detection of IB4 binding, biotinylated IB4 (1:400; Vector Laboratories, Burlingame, CA) and FITC-conjugated extravidin (1:500; Sigma) were used. Sections of a set of control and experimental tissues were concurrently immunostained and images were captured under the same conditions. Control tissue sections, in which the primary antibody was omitted, showed no specific staining. The experiments were carried out at least three times.

Immunofluorescent preparations were examined with a fluorescence microscope (BIOREVO BZ-9000; Keyence Corp, Osaka, Japan).

The numbers of CK1ε-positive neurons were counted in L5 DRGs. On average, five to seven non-adjacent sections of each DRG, where the CK1ε-positive neurons with visible nuclei were equally distributed throughout the rostrocaudal length of the DRG, were randomly selected from 3 to 4 animals in naive control, sham- and SNL-operated groups and numbers of neurons that showed distinctive CK1ε-labeling compared with background labeling were counted as CK1ε-positive by two investigators blinded to the surgical treatment and averaged. The fluorescence intensity was quantified using a 255-level gray scale [33]. To determine the percentage of immunoreactive neurons in each DRG, a threshold of average fluorescence intensity level (for example, 30 in 255-level gray scale for CK1ε) was set by observing several images of normal DRGs. The fluorescence intensity threshold was then applied to all other sections of ipsilateral and contralateral DRGs. The fluorescence intensity and cross-sectional area of CK1ε-positive neurons were quantified using BZ-II analyzer (BZ-H1C software; Keyence Corp, Osaka, Japan). To distinguish cell size-specific changes, we divided the DRG neurons into small-sized (< 600 μm²), medium-sized (600-1200 μm²), and large-sized (> 1200 μm²) groups based on their cross-sectional areas [34, 35]. For the co-localization analyses, double stained DRG sections were similarly selected from 3-4 mice, and CK1ε-, CGRP- and IB4-positive cells were counted.

For counting the dorsal horn cells, five to seven sections from the L5 spinal cord segment were randomly selected from each mouse and numbers of distinctive CK1ε-, Neu N-, GFAP- and Iba 1-positive cells in the superficial and middle layers of the dorsal horn were counted by two investigators blinded to the surgical treatment and averaged. The border between superficial and middle layers was delineated according to a representative immunostaining images of protein kinase C type γ (PKCγ) prepared in our laboratory. PKCγ is present in a subpopulation of neurons in the inner part of lamina II and allows a good localization of the border between laminae II and III. The middle layer was defined as dorsal half of deep dorsal horn (laminae III-VI). The number of cells per one section was averaged in each naive control, sham- and SNL-operated animal and the overall means were calculated.

Proportions of CK1ε-positive cells and intensities of CK1ε-IR measured in single- and double-staining were not significantly different in any of the experimental groups. Therefore, data from both single- and double-labeled cells were combined.

Images of spinal cords stained with the rabbit anti-CK1ε antibody or double labeled with CK1ε (mouse) and CGRP (rabbit) antibodies were collected using a Zeiss LSM 5 Pascal confocal microscope with argon and helium neon lasers (Carl Zeiss Microscopy, Jena, Germany). A × 63, 1.2 NA water-immersion C-apochromatic objective and 2 × zoom value were used for high magnification. The digital images of 20 consecutive z-scan sections (step size approximately 0.5 μm) were analyzed on a computer equipped with an image analysis system (Image J version 1.40 g). CK1ε-IR was quantified using a 255-level gray scale [33]. To quantify CK1ε-IR of the superficial dorsal horn neurons, the immunofluorecence intensities of five randomly selected neurons were quantified using a 255-level gray scale. For each spinal cord section, the ratio of the immunofluorescence intensities of the ipsilateral to the contralateral side was calculated. The ratios for 3-4 non-adjacent sections were averaged in sham- and SNL-operated groups.

Spinal cord slices were prepared according to the method described previously [21, 36]. Transverse slices (thickness, 600-750 μm) of the L5 spinal segments with the L5 dorsal root attached were prepared and stained with a fluorescent voltage-sensitive dye di-4-ANEPPS (Invitrogen).

Changes in voltage-sensitive dye fluorescence in the spinal cord were detected using an optical recording system MiCAM02 (Brainvision Inc., Tsukuba, Japan) [37, 38]. Repetitive electrical stimulation (current: 1.0 mA, duration: 1 ms) comprised of 10 pulses at 20 Hz were applied through a suction electrode attached to the dorsal root, as described previously [39, 40] to induce a long-lasting excitatory component including NMDA receptor transmission. Both A- and C-fibers were thought to be activated by the stimulation mode [41]. For the recording of glutamate receptor agonist-evoked responses in the presence of TTX (0.3 μM), short exposure mode of MiCAM02 recoding and analyzing software (BV analyzer; Brainvision Inc., Tsukuba, Japan) was used to minimize dye bleaching according to the manufacturer's guide. The effect of a pharmacological agent on the nerve- or glutamate receptor agonist-induced responses was evaluated by comparing the averaged magnitude of two or three control responses with the magnitude of response after 25-30 min equilibration for each drug. The concentrations of glutamate receptor antagonists and TTX used were determined according to the previous studies [40, 42, 43] and our preliminary study.

D-APV, CNQX and IC261 were purchased from Tocris Bioscience, Bristol, UK. Glutamate and NMDA were from Sigma, St. Louis, MO, USA. TTX was from Sankyo Co., Ltd., Tokyo, Japan.

Experimental data are expressed as mean ± SEM. Single comparisons were made using Student's two-tailed paired or unpaired t-test. One-way ANOVA followed by the Dunnett's or Tukey's test was used for multiple comparisons. p < 0.05 was considered statistically significant.