Involvement of CX3CL1/CX3CR1 Signaling in Spinal Long Term Potentiation

The long-term potentiation (LTP) of spinal C-fiber-evoked field potentials is considered as a fundamental mechanism of central sensitization in the spinal cord. Accumulating evidence has showed the contribution of spinal microglia to spinal LTP and pathological pain. As a key signaling of neurons-microglia interactions, the involvement of CX3CL1/CX3CR1 signaling in pathological pain has also been investigated extensively. The present study examined whether CX3CL1/CX3CR1 signaling plays a role in spinal LTP. The results showed that 10-trains tetanic stimulation (100 Hz, 2s) of the sciatic nerve (TSS) produced a significant LTP of C-fiber-evoked field potentials lasting for over 3 h in the rat spinal dorsal horn. Blockade of CX3CL1/CX3CR1 signaling with an anti-CX3CR1 neutralizing antibody (CX3CR1 AB) markedly suppressed TSS-induced LTP. Exogenous CX3CL1 significantly potentiated 3-trains TSS-induced LTP in rats. Consistently, spinal LTP of C-fiber-evoked field potentials was also induced by TSS (100 Hz, 1s, 4 trains) in all C57BL/6 wild type (WT) mice. However, in CX3CR1-/- mice, TSS failed to induce LTP and behavioral hypersensitivity, confirming an essential role of CX3CR1 in spinal LTP induction. Furthermore, blockade of IL-18 or IL-23, the potential downstream factors of CX3CL1/CX3CR1 signaling, with IL-18 BP or anti-IL-23 neutralizing antibody (IL-23 AB), obviously suppressed spinal LTP in rats. These results suggest that CX3CL1/CX3CR1 signaling is involved in LTP of C-fiber-evoked field potentials in the rodent spinal dorsal horn.

CX3CL1, a chemokine, has two functional forms: membrane-anchored CX3CL1 and soluble CX3CL1 [16], which is released from membrane by lysosomal cysteine protease Cathepsin S (Cat S) [17] or disintegrin and metalloproteinase (ADAM) 10/17 [18,19]. In the central nerve system, CX3CL1 is produced mostly in neurons, and its sole receptor CX3CR1, a G proteincoupled receptor, is mainly expressed in microglia [20][21][22]. Therefore, interaction between neurons and microglia may be mediated via CX3CL1/CX3CR1 signaling [20]. Increasing evidence suggests that spinal CX3CL1/CX3CR1 signaling plays a key role in the development and maintenance of pathological pain [23][24][25][26][27]. To address whether CX3CL1/CX3CR1 signaling is involved in central sensitization in the spinal cord, the present study was designed to illustrate the influence of CX3CL1/CX3CR1 signaling on spinal LTP.

Materials and Methods Animal
Male adult Sprague Dawley rats (200-300 g, n = 128) were supplied by Shanghai Experimental Animal Center of the Chinese Academy of Sciences. C57BL/6NTac-[KO] CX3CR1 mice were purchased from Taconic Farms Inc. [28], and C57BL/6 background wild type (WT) control mice (male, 8-10 weeks) were purchased from The Jackson Laboratory and bred in the Animal Center of Fudan University. All animals were housed in a 12 h light/dark cycle with a room temperature of 22±1°C, and received food and water ad libitum. All experimental protocols and animal handling procedures were permitted by the Shanghai Animal Care and Use Committee, and were in line with the policies issued by the International Association for the Study of Pain.

Cerebrospinal fluid and tissue collection
After defined survival times, rats were sacrificed by overdose of urethane and the L4-L6 spinal dorsal horn was rapidly removed. The dorsal horn tissues were homogenized with ultrasonic cell processor in an SDS sample buffer that contained a mixture of proteinase inhibitors and PMSF. To collect cerebrospinal fluid (CSF), a catheter (PE-10 tube) was inserted through the gap between the L4 and L5 vertebrae and extended to the subarachnoid space under sodium pentobarbital anesthesia (80 mg/kg, i.p.) and sterilizing. The CSF flowed out spontaneously through the catheter, when the rat body was erected.

Western blots
Equal amount of protein or CSF sample was loaded and separated in 10% Tris-Tricine SDS-PAGE gel and transferred to PVDF membrane (Millipore). The membranes were blocked with 5% nonfat milk in Tris-buffered saline (pH 7.5) with 0.1% Tween-20 for 2 h at room temperature (RT), and incubated overnight at 4°C with goat anti-CX3CL1 antibody (1:500, R&D Systerms, USA), rabbit anti-CX3CR1 antibody (1:2000, Torrey Pines Biolabs, USA) or goat anti-Cathepsin S antibody (1:500, ABcam, Hong Kong). The blots were then incubated with HRP-conjugated secondary antibodies (1:1000, Pierce) for 2 h at RT. Signals were finally detected using enhanced chemiluminescence (ECL, Thermo, USA), and the bands were visualized with the ChemiDoc XRS system (Bio-Rad, USA). All Western blot analysis was performed at least three times, and consistent results were obtained. Experiment 1: To test the effect of anti-CX3CR1 antibody (CX3CR1 AB) on the expression of CX3CR1, the spinal dorsal horn tissues were removed from rats of sham group (n = 4) in Experiment 1 of Electrophysiological recording, CX3CR1 AB group (n = 4) and IgG group (n = 4) in Experiment 2 of Electrophysiological recording, at the end of electrophysiological recording (3 h after TSS). Experiment 2: To examine the expression of CX3CL1 after TSS, the spinal dorsal horn tissues and CSF were removed from rats applied with 10-trains TSS (0.5 h after TSS, n = 4) or sham (n = 4).

ELISA Assay
To determine soluble CX3CL1 expression in CSF after TSS, we collected the spinal CSF (n = 6) from rats applied with 10-trains TSS (0.5 h after TSS) or sham. A rat CX3CL1 "Sandwich" ELISA kit (RayBiotech, USA) was used to examine CX3CL1 content the CSF. Rat recombinant FKN standards and samples in 100 μl were run in duplicate according to the manufacturer's instructions. The optical density of each well was read at 450 nm. Data are expressed as percentage of FKN content in the basal fractions.

Behavioral tests
The mechanical threshold was measured by probing von Frey filaments (Stoelting, USA). Each mouse was placed in a chamber (10cm×10cm×20cm) with customized platform that contains 1.5 mm diameter holes in a 5 mm grid of perpendicular rows throughout the entire area of the platform. Mice were allowed to acclimate for approximately 30min. A series of von Frey filament stimuli (0.16, 0.4, 0.6, 1.0, 1.4, 2.0g) were delivered to the central region of the plantar surface of the hindpaw with increasing bending force until the mouse withdrew the foot. Each filament was applied five times and each time maintained for 2s with 15s intervals. When the hindpaw withdrew from a filament at least three of the five applications, the value of the filament in grams was considered to be the"paw withdrawal threshold" (PWT).
The thermal threshold was measured by Hargreavestest. Mice were placed individually in transparent plastic chambers on an elevated glass surface. After acclimation to the test chambers for about 30min, a radiant heat source (IITC/Life Science Instruments) was focused on the hindpaw. The heat source was turned off when the mouse lifted the foot. The time from the onset of radiant heat application to withdrawal of the hindpaw was defined as the hindpaw withdrawl latency (PWL). To prevent tissue damage, the cut-off latency was set at 15s. The average of three trials was determined and the interval between trials is 10 min.
Hargreaves'test and von Frey test were performed before and 4 days after TSS in the same groups of CX3CR1 -/-(n = 8) and C57BL/6 WT (n = 8) mcie, started with von Frey test followed by Hargreaves' test with an interval of 2 hours.

Data analysis
All the data were expressed as means ± SEM. Student's t-test (for comparisons of two groups) or One-(or two-) way ANOVA (for multiple group comparisons) followed by post hoc Student-Newmann-Keuls test was used to identify significant differences. In all cases, p < 0.05 was considered as being statistically significant.

Expression of CX3CL1 and CX3CR1 in the spinal dorsal horn
Double immunostaining was performed on sections of the L4-6 spinal cord in rats. The distributions of CX3CL1 and CX3CR1 were examined in the spinal dorsal horn of naïve rats. Consistent with previous reports [21,30], CX3CL1 was mostly expressed in NeuN (neuron marker)-labeled neuron and slightly in GFAP (astrocytic marker)-labeled cells in the spinal cord (Fig. 1A). Its receptor, CX3CR1 was mainly colocalized with Iba1 (microglia marker) (Fig. 1B).
Blockade or knockout of CX3CR1 impairs spinal LTP As described in our previous studies [5,15,31], tetanic stimulation of the sciatic nerve (TSS) produced a significant long-term potentiation (LTP) of C-fiber-evoked field potentials lasting for over 3 h in the rat spinal dorsal horn. The representative LTP was illustrated in Fig. 2. After 10-trains TSS, the C-fiber-evoked field potential was amplified about 3 folds than that before TSS. In contrast, in the sham group without TSS, no obvious change in C-response was observed (Two-way ANOVA, treatments: F 1, 9 = 138.261, p < 0.01) (Fig. 2).
To examine whether CX3CL1/CX3CR1 signaling is involved in the spinal LTP, an anti-CX3CR1 neutralizing antibody (CX3CR1 AB)was applied to block CX3CL1/CX3CR1 signaling. As shown in Fig. 3A, the induction of spinal LTP was remarkably blocked by administration of CX3CR1 AB (30 μg/30 μl) 2h before 10-trains TSS, compared with control IgG (Two-way ANOVA, treatments: F 1, 12 = 11.981, p<0.01). In addition, at the end of electrophysiological recording (3 h after TSS), the spinal dorsal horns were removed and the expression of CX3CR1 was examined by Western blots. Although no striking upregulation of CX3CR1 was observed after TSS, the expression of CX3CR1 was significantly decreased by delivering CX3CR1 AB, as compared with IgG (One-way ANOVA, F 2, 9 = 5.399, p < 0.01) (Fig. 3B).
To further confirm CX3CL1/CX3CR1 signaling contributes to spinal LTP, TSS was delivered to the sciatic nerve for induced LTP of C-fiber-evoked field potentials in C57BL/6 WT and CX3CR1 knock-out mice. As described in the previous studies [32], spinal LTP of C-fiberevoked field potentials was induced by a 4-trains TSS (100 Hz, 1s) in all five C57BL/6 WT mice, lasting for >2 h, with an amplitude increase of 92% at 1 h (Fig. 3C). Of note, LTP failed   to be induced in CX3CR1 knock-out mice (Fig. 3C). Two-way ANOVA analysis revealed significant difference between groups (F 1, 16 = 100.208, p<0.01). Combination with the results from rats and mice indicate an essential role of CX3CR1 in the induction of rodent spinal LTP.
Our previous studies showed that following TSS, a robust mechanical allodynia was observed in rats from day 1 after TSS and lasted at least for 7 days [4]. Blockade of CX3CR1 by anti-CX3CR1 antibody significantly suppressed TSS-induced mechanical allodynia [43]. In the present study, we further demonstrated that TSS, which conventionally induces LTP of C-fiber-evoked field potential in the WT mouse spinal dorsal horn, also produced a longlasting mechanical allodynia and thermal hyperalgisa (Fig. 3D and 3E). Consistent with the electrophysiological results, behavioral tests showed that TSS-induced mechanical allodynia and thermal hyperalgesia did not occurred in CX3CR1 knock-out mice (Student's t-test, p<0.01) (Fig. 3D-3F).

CX3CL1 facilitates spinal LTP
To further verify the contribution of CX3CL1/CX3CR1 signaling to spinal LTP, exogenous CX3CL1 was applied to test whether LTP was facilitated. Considering spinal LTP could be saturated by strong stimulation of sciatic nerve [33], 3-trains TSS was used to induce spinal LTP to avoid the potential ceiling effect of 10-trains TSS on LTP in rats. The results showed that 3-trains TSS induced LTP with smaller potentiated extent than that of 10-trains TSS-induced LTP, and 3-trains TSS-induced LTP was robustly potentiated by spinal application of CX3CL1 (0.75 μg/30 μl) 30 min before 3-trains TSS (Two-way ANOVA, treatments: F 2, 18 = 6.618, p < 0.01) (Fig. 4A).
Several studies indicate that membrane-bound CX3CL1 is cleaved by the protease Cathepsin S (CatS), which is expressed and released by activated microglia [17,34,35]. Therefore, in the current work, whether soluble CX3CL1 was cleaved from neuronal membranes after TSS was examined. As shown in Fig. 4D, there were two bands detected by an anti-CX3CL1 antibody, a 95 kDa band predominantly expressed in the rat spinal dorsal horn (SDH) tissues and a 72 kDa band strongly expressed in the CSF (Fig. 4D inset), corresponding to membranebound CX3CL1 and soluble CX3CL1, respectively [25,36,37]. Following 10-trains TSS, membrane-bound CX3CL1 in the SDH was markedly reduced (Student t-test, t = 3.022, p < 0.05), whereas soluble CX3CL1 in the CSF was obviously increased at 30 min (Student t-test, t = 4.036, p < 0.05) (Fig. 4D). The soluble CX3CL1 in the CSF was further confirmed by ELISA assay (Fig. 4E). In addition, an increased protease Cathepsin S (Cat S) was also detected in the CSF 30 min after TSS (Student t-test, t = 2.720, p < 0.05) (Fig. 4F).

Discussion
Unmyelinated C-fibers predominantly terminate in the superficial laminae of the spinal dorsal horn and mainly transfer nociceptive information. It is proved that the sensitization of unmyelinated C-fibers is the peripheral substrate of pathological pain [38][39][40][41]. C-fiber-evoked field potentials reflect the activation of pain-sensitive neurons in the superficial spinal dorsal horn. Long-term potentiation (LTP) of C-fiber-evoked field potentials is a phenomenon of central sensitization in the spinal cord, contributing to the development of pathological pain [2,42]. It is showed that acute nerve injury can evoke both pathological pain and spinal LTP of C-fiberevoked field potentials [31]. Compelling evidence has confirmed that tetanic stimulation of the sciatic nerve (TSS) not only evoked LTP of C-fiber-evoked field potentials, but also induced a long-lasting allodynia and hyperalgesia, the common symptom of neuropathic pain [4,5,43]. Accordingly, the investigation of spinal LTP of C-fiber-evoked field potentials will help us to understand the central mechanism underlying pathological pain.

CX3CL1/CX3CR1 in Spinal LTP
Over the past decades, lots of neuronal factors were demonstrated to be involved in spinal LTP [6]. In recent years, the contribution of spinal glia to spinal LTP has also been focused on, and several glial factors have been considered to participate in spinal LTP, such as P2X4 receptors and p38 mitogen-activated protein kinase (p38 MAPK) [14], interleuk-1beta (IL-1beta) [15], tumor necrosis factor alpha (TNF-alpha) and P2X7 receptors [12,13]. In the present study, another spinal microglial factor, CX3CL1/CX3CR1 signaling, was also involved in longterm potentiation (LTP) of C-fiber-evoked field potentials in the spinal dorsal horn.
Increasing evidence suggests that the activation of spinal glia plays an essential role in the development and maintenance of pathological pain [44,45]. As a molecular model of central sensitization in the spinal cord [1][2][3], spinal long-term potentiation (LTP) has also been showed to be related with the activation of spinal glia [11,12,14,15]. CX3CR1, a G proteincoupled receptor and the sole receptor of CX3CL1, is mainly expressed in spinal microglia [20,21]. Binding with CX3CL1, microglia can be activated through p38MAPK signaling [25,27], ERK1/2 signaling [46] and ERK5 signaling [47]. In addition, it has been demonstrated that CX3CL1/CX3CR1 signaling activity in spinal microglia is an essential process for development and maintenance of inflammatory pain [48,49], neuropathic pain [25,47] and cancer pain [27]. In line with such reports, the present findings of contribution of CX3CL1/CX3CR1 signaling to spinal LTP presents new evidence that CX3CL1/CX3CR1 signaling is involved in the potentiation of nociceptive transmission under the pathological pain condition.
CX3CL1 exists two functional forms: either membrane-bound or as a soluble glycoprotein [16]. The soluble form CX3CL1 performs chemoattractant activity for T cells and monocytes whilst membrane-bound CX3CL1 acts as an adhesion molecule contributing to leukocyte capture [16,50]. The studies from Clark et al. showed that the levels of soluble CX3CL1 in CSF increased significantly after peripheral nerve injury, and lysosomal cysteine protease Cathepsin S played a key role in the release of soluble CX3CL1 from neuron membrane to CSF [17]. On the other hand, exogenous Cathepsin S-induced hyperalgesia and allodynia were attenuated by a neutralizing antibody against CX3CL1 [35]. Therefore, under pathological pain conditions, soluble CX3CL1 may be the main functional form, which is cleaved from neuronal membranes to activate the microglia via CX3CR1 and then contributes to amplification and maintenance of pathological pain. Although we did not observe the upregulation of CX3CR1 in the spinal dorsal horn at 3 h after TSS, another work from our laboratory showed that significant upregulation of CX3CR1 in the spinal cord occurred at 24 hours after TSS [51]. It is suggested that TSS-induced de novo synthesis of CX3CR1 may take more than 3 h. Interestingly, TSS induced an increased soluble CX3CL1 release, which may play an essential role in the enhanced CX3CL1/CX3CR1 signaling during spinal LTP.
In the current study, we also found the contribution of IL-18 and IL-23 to spinal LTP. In the spinal cord, IL-18 was considered to be a key modulator in pathological pain [52][53][54] and mediated microglia/astrocyte interaction [53]. Miyoshi et al. reported that the production of IL-18 in the spinal cord was regulated by p38MAPK [53]. On the other hand, exposing to exogenous CX3CL1, the p38MAPK signaling was activated in spinal microglia [25]. Consequently, it is reasonable to infer that CX3CR1 may be the upstream regulator of IL-18 in microglia.
As to IL-23, its role in the pathogenesis of multiple sclerosis (MS) has been studied [55][56][57]. However, the acquaintance with involvement of IL-23 in pathological pain remains limited. In the injured sciatic nerve of a mouse chronic constriction injury (CCI) model, the upregulation of IL-23 mRNA was observed [58]. In the current study, the finding of the involvement of IL-23 in spinal LTP provided direct evidence that spinal IL-23 may contribute to the potentiation of nociceptive transmission. The previous studies manifested that there are NF-kappa-B binding sites in p19 subunit gene promoter of IL-23, by binding with which NF-kappa-B could regulate IL-23 expression [59][60][61]. It was also found that NF-kappa-B could be activated in spinal IL-18R-expressing astrocytes after nerve injury, and the IL-18-induced allodynia was dosedependently alleviated by intrathecal injection of an NF-kappa-B inhibitor, SN50, suggesting that nerve injury induces NF-kappa-B activation in the spinal astrocytes via the IL-18 signaling [53]. Accordingly, IL-23 may be regulated through IL-18/NF-kappa-B signaling. Therefore, it is conceivable that there may be a CX3CL1/IL-18/IL-23 signaling pathway contributing to spinal LTP.
Contrary to our finding of the facilitated effect of CX3CL1 on spinal LTP, the inhibitory influence of CX3CL1 on neuron excitability and central sensitization was reported. In the in vitro studies of cultured microglia, it was observed that CX3CL1 suppressed the releases of proinflammatory cytokines from activated microglia, such as TNF-alpha, IL-1beta, nitric oxide (NO) and IL-6 [62][63][64]. Some studies on hippocampus showed that CX3CL1 reduced excitatory postsynaptic response [65][66][67] and impaired the induction of LTP [68]. With regard to the contradictory effect of CX3CL1 in the central nervous system, one possibility may be attributed to the different concentrations to be used. In the work of Mizutani et al, 0.03 nM CX3CL1 significantly reduced LPS (lipopolysaccharide)-induced phosphorylation of ERK1/2 and secretion of TNF-alpha and IL-6 in macrophages, however, 3nM CX3CL1 elevated the expression of IL-23, which subsequently upregulated the production of TNF-alpha and abolished suppressive effect of low concentration of CX3CL1 [69]. This phenomenon suggests that different doses of CX3CL1 may induce different intracellular signaling and then perform inverse effects. In addition, given that two novel functional isoforms of CX3CR1 have identified [68], it is possible that the different isoforms of CXCR1 exert contradictory functions via diverse signaling pathways [46].
In conclusion, the present study showed that CX3CL1/CX3CR1 signaling was involved in long-term potentiation (LTP) of C-fiber-evoked field potentials in the spinal dorsal horn.