Introduction

Tumor necrosis factor-α (TNFα) plays a central role in initiating inflammatory reactions of the innate immune system, inducing the production of the proinflammatory cytokines interleukin (IL)-1β and IL-6, activating the expression of adhesion molecules and stimulating inflammatory cells. It has been known for some time that peripheral inflammatory conditions resulting in heightened pain sensitivity are characterized by an early release of TNFα at the site of inflammation.1 Direct application of TNFα to peripheral nerve induces spontaneous electrophysiologic activity in neurons of the dorsal root ganglion (DRG),2, 3, 4 and subcutaneous or epineurial application of TNFα produces nocisponsive behaviors.5, 6

A substantial body of evidence suggests that neuroimmune activation of glial cells in the dorsal horn of spinal cord results in the release of proinflammatory cytokines and plays an important role in facilitation of nociceptive neurotransmission and the development of chronic pain.7, 8, 9, 10 Expression of TNFα in the dorsal horn of spinal cord is increased in models of central11 and peripheral12 neuropathic pain. Proinflammatory cytokines are increased in conditions characterized by peripheral inflammation13 and have been shown to enhance nociceptive neurotransmission at the spinal level.13, 14

IL-10, a prototypical antiinflammatory cytokine, was initially characterized as a cytokine synthesis inhibitory factor because of its potent ability to downregulate the synthesis of proinflammatory cytokines. IL-10 inhibits the expression of IL-1β, TNFα and suppresses the activity of proinflammatory Th1 cells. Intrathecal administration of IL-10 protein, or of adenovirus or adeno-associated virus-based vectors that produce IL-10, reduces pain-related behaviors in models of neuropathic pain.15, 16

In previous studies we have shown that vectors constructed from nonreplicating herpes simplex virus (HSV) recombinants can be used effectively to transfer genes into DRG neurons from subcutaneous inoculation, and that release of inhibitory neurotransmitters or antiinflammatory proteins from the central afferent terminals of transduced DRG neurons in vivo can be used to modulate pain-related behaviors.11, 12, 17, 18 In this study, we used a nonreplicating HSV vector constructed to express and release IL-10 in the formalin model of inflammatory pain. We report that prolonged expression and transport of IL-10 by primary afferents to dorsal horn prevents the activation of p38 mitogen-activated protein kinase (MAPK) and decreases expression of mTNFα in microglia with concomitant reduction in pain-related behaviors during the delayed phase of the formalin test. The results implicate microglial mTNFα in spinal neuroimmune activation in inflammatory pain.

Results

We have previously demonstrated that subcutaneous inoculation of an HSV vector can be used to transduce DRG in vivo to deliver transgene products to dorsal horn of spinal cord.11, 12, 17, 18 For the purposes of this study we constructed a non-replicating vector (QHIL10) that contains two copies of a hemagglutinin (HA)-tagged rat IL-10 gene, each under the control of the human cytomegalovirus immediate-early promoter (HCMV IEp; Figure 1a). Control vector QHGFP contains two copies of the green fluorescent protein (GFP) gene under the control of the same promoter (Figure 1b). In a preliminary study we confirmed that transfection of dissociated primary DRG neurons in vitro with vector QHIL10 at multiplicity of infection (MOI) of 1 resulted in robust expression and release of IL-10 (Figures 2a–d). Subcutaneous inoculation in the plantar surface of the hind paw of rats with 30 μl of QHIL10 (1 × 109 PFU per ml) resulted in expression of IL-10 in lumbar L4–L5 DRG (Figure 2e) and transport of vector-derived IL-10 protein to the central terminals of the pseudounipolar DRG axon in the dorsal horn of the lumbar spinal cord (Figure 2f).

Figure 1
figure 1

Schematic representation of the vector constructs QHIL10 (a) and QHGFP (b). Vector QHIL10 contains two copies of the rat interleukin-10 (IL-10) coding sequence in a nonreplicating HSV backbone, defective in expression of the HSV IE genes ICP4, ICP22, ICP27 and ICP47. Control vector QHGFP contains two copies of the green fluorescent protein (GFP) gene in the same HSV backbone.

Full size image
Figure 2
figure 2

(a) Primary DRG neurons 48 h after infection by QHGFP or QHIL10 (MOI of 1 for 2 h) stained for the neuronal marker Tuj1 (green) and interleukin-10 (IL-10, red). Scale bar=10 μm. Western blot of cell lysate (b) and culture medium (c) 48 h after transfection at MOI of 1. (d) ELISA of IL-10 from the medium collected 48 h after transfection. Western blot of protein from DRG (e) and dorsal quadrant of spinal cord (f) 1 week after subcutaneous inoculation of QHIL10 or QHGFP. The identity of IL-10 as vector-derived is confirmed by the anti-HA antibody employed in the western blot.

Full size image

IL-10 attenuates the increase in expression and release of TNFα following lipopolysaccharide

To examine the biological effect of vector-derived IL-10 on microglia in vitro, we used a microglial (HAPI) cell line that has been well characterized and exhibits substantial similarities to primary central nervous system (CNS) microglia.19 HAPI cells, like native microglia, express the cell surface complement receptor 3 detected by OX42 (CD11b) antibody (Figure 3a); we confirmed that the IL-10 receptor was present as well (Figure 3b). To examine the effects of IL-10, we confirmed that HAPI cells transduced in vitro by QHIL10 release substantial amounts of IL-10 into the medium as detected by enzyme-linked immunosorbent assay (ELISA) (Figure 3c). IL-10 expressed in the cell lysate was recognized by both anti-HA and anti-IL-10 antibodies, confirming its identity as a product of the transgene (Figure 3d).

Figure 3
figure 3

(a) HAPI cells immunostained with the microglial cell marker OX42 (green). (b) HAPI cells immunostained with an antibody against the interleukin-10 (IL-10) receptor (red) with Hoechst (blue) counterstaining. (c) ELISA of culture medium collected 48 h after infection of HAPI cells with QHIL10 or QHGFP (MOI of 1 for 2 h). (d) Western blot of HAPI cell lysate 48 h after infection with QHIL10 or QHGFP. Scale bar=10 μm.

Full size image

Lipopolysaccharide (LPS) activates the innate immune response in circulating monocytes/neutrophils, tissue macrophages and CNS microglia through the Toll-like receptor 4. Exposure of HAPI cells for 6 h to LPS resulted in a substantial increase in TNFα mRNA (Figures 4a and b) along with an increase in total TNFα protein (Figures 4c and d). The upregulation in TNFα protein was characterized by an increase in both 26 kDa mTNFα and by the appearance of the 17 kDa cleavage product sTNFα in the cell lysate (Figure 4c), indicating the activation of intracellular processing of 26 kDa mTNFα. In addition, LPS induced a substantial increase in sTNFα released from the cells detected in the culture medium by ELISA (Figure 4e). HAPI cells transduced with QHIL10 showed a significant blunting of the increase in TNFα mRNA (Figures 4a and b) and protein (Figures 4c and d), consistent with inhibition of expression of TNFα and a significant reduction in the amount of 17 kDa sTNFα in the cell lysate (Figure 4c). This was accompanied by a reduction in the amount of sTNFα released from the cells into the medium (Figure 4e), compared to cells transfected with the control vector QHGFP and exposed to LPS.

Figure 4
figure 4

Tumor necrosis factor-α (TNFα) mRNA and protein in HAPI cells transduced with QHGFP or QHIL10 and exposed to lipopolysaccharide (LPS) determined by reverse transcription (RT)–PCR (a) and western blot (c). The amount of mRNA was quantified by relative optical density (b) and the sum of 17 and 26 kDa protein bands quantified by chemiluminescence (d), normalized to β-actin mRNA and protein levels, respectively. The amount of sTNFα released into the culture medium was determined by ELISA (DY510, R&D Systems; e). The data represent the results of three independent experiments. Mean±s.e.m.; **P<0.01.

Full size image

IL-10 blocks the activation of p38 MAPK induced by LPS

To investigate the role of p38 MAPK in LPS-mediated upregulation of TNFα in the microglia, we treated HAPI cells with the p38 MAPK inhibitor SB202190 for 30 min prior to stimulation with LPS. Treatment with SB202190 resulted in a significant reduction in TNFα mRNA (Figures 5a and b), total TNFα in the cell lysate characterized by a reduction in the amount of both 26 and 17 kDa proteins (Figures 5c and d). This was accompanied by a reduction in the amount of sTNFα released from the cells (Figure 5e), compared to LPS-treated cells in the absence of p38 MAPK inhibitor. In addition to the changes in TNFα, exposure to LPS concomitantly resulted in a substantial increase in the phosphorylation of p38 MAPK (p-p38; Figures 5f and g). Transfection of cells with control vector QHGFP had no effect on the amount of p-p38, but IL-10 released from QHIL10-transduced HAPI caused a marked reduction in the level of p-p38 compared to control or QHGFP-transduced cells (Figures 5f and g).

Figure 5
figure 5

Tumor necrosis factor-α (TNFα) mRNA and TNFα protein (mTNFα plus sTNFα) in HAPI cells pretreated with 10 μM SB202190 for 30 min prior to addition of lipopolysaccharide (LPS) determined by RT–PCR (a) and western blot (c). The amount of mRNA was quantified by relative optical density (b) and the sum of the 17 and 26 kDa protein bands quantified by chemiluminescence (d), normalized to β-actin mRNA and protein levels, respectively. sTNFα released in the culture medium by ELISA (e). p-p38 and p38 protein (western blot) in HAPI cells transduced with QHGFP or QHIL10 and exposed the LPS (f). Ratio of p-p38 to p38-determined chemiluminescence (g). Data presented represent the results of three independent experiments. Mean±s.e.m.; *P<0.05; **P<0.01.

Full size image

QHIL10 in vivo reduces pain behavior in the formalin model of inflammatory (nociceptive) pain

In the formalin model of inflammatory pain, administration of formalin subcutaneously in the dorsum of the hind paw induced acute (phase 1) and delayed (phase 2) phases of spontaneous pain behavior as shown by the increased number of flinches (Figures 6a–c). Animals inoculated with QHIL10 ten days prior to formalin injection showed a substantial and significant reduction in the number of flinches in both the acute and delayed phases of the behavioral evaluation (Figures 6a–c).

Figure 6
figure 6

(a) Spontaneous flinching observed after injection of 50 μl of 5% formalin in animals 10 days after subcutaneous inoculation of QHIL10, QHGFP or phosphate-buffered saline (PBS). P<0.05 comparing QHIL10 to QHGFP or PBS by repeated measures analysis. Means±s.e.m., n=8 animals per group. (b) Total number of flinches in phase 1 (1–10 min). (c) Total number of flinches in phase 2 (10–60 min). Mean±s.e.m., n=8 animals per group, **P<0.01.

Full size image

Transgene-mediated expression of IL-10 attenuates the increase of TNFα expression and phosphorylation of p38 MAPK in dorsal horn of spinal cord

The amount of TNFα mRNA (Figures 7a and b) and mTNFα protein (Figures 7c and d) in the dorsal quadrant of the lumbar spinal cord at 45 min after injection of formalin, a time point corresponding to the peak of nocisponsive behavior, were both significantly increased in animals with formalin-induced pain compared to control, and correlated with an increase in p-p38 (Figures 7e and f). There was no sTNF detected by western blot in these samples. TNFα immunoreactivity in the dorsal horn was found in restricted zones in activated microglia, identified by OX42 immunostaining (Figure 7g). Release of IL-10 in dorsal horn achieved by inoculation of vector QHIL10 blocked the increase in mTNFα expression (Figures 7a–d) and the p-p38 MAPK (Figures 7e and f).

Figure 7
figure 7

Tumor necrosis factor-α (TNFα) mRNA and protein in the dorsal quadrant of lumbar spinal cord determined by RT–PCR (a) and western blot (c), quantitated by relative optical density (b) and chemiluminescence (d), normalized to β-actin levels. Levels of p-p38 in the dorsal quadrant of lumbar spinal cord determined by western blot (e) and quantitated by relative optical density (f) normalized to total p38 levels. The data presented represent mean±s.e.m., n=5 animals per group; *P<0.05, **P<0.01. (g) Double-label immunostaining of TNFα (red) and OX42 (green) in laminae I–II of dorsal horn (left, bar=40 μm). Higher power magnification of individual microglial cells (left two panels, bar=40 μm).

Full size image

Discussion

These studies demonstrate that vector-mediated expression of IL-10 in DRG resulting in the delivery of IL-10 to the dorsal horn reduces inflammatory pain and prevents the activation of p38 MAPK and expression of mTNFα in dorsal horn of spinal cord. The results suggest that the 26 kDa transmembrane form of mTNFα is the predominant form of that protein expressed in dorsal horn in the setting of a neuroimmune response induced by inflammatory pain originating at a distant site.

TNFα plays a critical role in the development of both inflammatory and neuropathic pain.20 Peripherally, endoneurial or intramuscular injection of TNFα results in pain-related behaviors,21, 22 and nerve injury results in increased expression of TNFα in DRG.23, 24, 25 TNFα expression is increased in the lumbar spinal cord after thoracic spinal cord injury that results in central neuropathic pain.11 While the early phase of flinching after formalin administration results from C-fiber activation caused by the peripheral stimulus, the late phase of flinching represents the windup phenomenon.26 Spinal cord glia are activated in the nociceptive response to formalin injection,27 and our results indicate that spinal mTNFα plays a critical role in the development of pain.

TNFα is initially expressed as a 26 kDa type II transmembrane protein (mTNFα) that can be cleaved by a TNFα convertase both intracellularly28 and on the cell surface29 to release the 17 kDa sTNFα.28 These conversion steps are critical for the release of the bioactive sTNFα, but mTNFα can mediate an inflammatory response independent of cleavage or the production of sTNFα.30, 31 In our in vivo study of inflammatory pain, we found a marked increase in full-length mTNFα in the spinal dorsal horn 45 min after formalin injection, but sTNFα was not detected. mTNFα colocalized with OX42 at restricted sites in microglia in the spinal cord. In contrast, direct exposure of microglia to a potent inflammatory stimulus (LPS) resulted in increased expression of mTNFα and a concomitant increase in cleavage, beginning in the intracellular compartment, and the release of sTNFα. This suggests that neuroimmune activation in the spinal cord in vivo—occurring distant from the site of injury—differs in important ways from the direct inflammatory response to tissue injury, and that mTNFα plays a critical role in that response.

p38 MAPK plays an important role in the cellular response to environmental stress and inflammatory triggers including LPS. In models of inflammatory pain, p-p38 in microglia is a critical early marker of central neuroimmune activation.32 In mucosal cells IL-10 downregulates the synthesis of TNFα by reducing RNA translation through inhibition of the activating p38 MAPK pathway.33 In microglial cells in vitro, we demonstrated that inhibition of p-p38 by a specific inhibitor blocks the increase in TNFα in response to LPS, and that IL-10 blocks the p-p38 and consequent synthesis and release of TNFα in response to the same stimulus. In vivo, IL-10 similarly blocked the p-p38 and synthesis of mTNFα in microglia of the spinal cord, demonstrating a link between a peripheral inflammatory stimulus, p-p38 and mTNFα expression in dorsal horn, and pain-related behavior. The observation that IL-10 appeared to significantly block p-p38 and the increase in mTNFα expression while incompletely reducing flinching behavior indicates that the p38-TNFα pathway plays a role in part but not all of the nocisponsive behaviors in the formalin model.

IL-10 is a prototypical antiinflammatory cytokine.34 Systemic administration of IL-10 protein produces an analgesic effect in models of inflammatory pain assessed by writhing in response to intraperitoneal administration of acetic acid or zymosan,35 or by knee incapacitation following intra-articular injection of zymosan.35 IL-10 protein also reduces pain-related behaviors in a model of spinal pain caused by intraspinal injection of quisqualic acid.36 There is also extensive evidence that using viral vectors or naked plasmids intrathecally results in expression of IL-10 by meningeal cells, effectively reducing pain-related behaviors in several models of peripheral neuropathic pain.15, 16 The current study extends these observations regarding the pain-relieving effects of IL-10 to a model of inflammatory pain, in which vector-mediated delivery of the gene product is achieved through gene transfer to DRG neurons, and implicate expression of spinal mTNFα in the phenomenon of chronic inflammatory pain.

HSV has unique advantages as a gene transfer vector to peripheral sensory neurons located in the DRG. Transgene products expressed from DRG neurons can provide local effects to modulate neuroinflammatory responses at the level of the central terminations of afferent axons in the spinal cord, and in the peripheral distribution of those same axons without requiring amounts of the transgene product that would alter the systemic immune response. This may prove to be of potential benefit in the adaptation of these strategies to the treatment of chronic pain in patients.

Materials and methods

Vectors

Two viral vector constructs were employed in this study. QHIL10 contains IL-10-HA under the control of the HCMV IEp. Control vector QHGFP is identical to QHIL10, but contains the GFP gene product in place of IL-10-HA (Figure 1). Full-length IL-10 was amplified from a cDNA library prepared from total RNA extracted from rat brain and cloned into BamHI-cut HCMV-polyA/SASB3-16. The HCMV-polyA/SASB3-16 plasmid was cotransfected with the nonreplicating HSV-recombinant UL41E1G6 (kindly provided by Dr Joseph Glorioso, University of Pittsburgh) into 7b cells. Three runs of extraction were performed to select clear plaques and the identity of the insert confirmed by PCR followed by DNA sequencing.

Cell culture

Microglia cells derived from neonatal rat brain19 (HAPI cells, provided by JR Connor Pennsylvania State University College of Medicine, Hershey, PA, USA) were grown in Dulbecco's modified Eagle's medium supplemented with 5% fetal bovine serum. In the LPS experiments, HAPI cells were treated with LPS (1 μg ml−1, Sigma-Aldrich, St Louis, MO, USA) for 6 h. In some experiments 10 μM of p38 MAPK inhibitor SB202190 (Calbiochem, La Jolla, CA, USA) was added 30 min prior to LPS treatment. DRG from 17-day-old rat embryos were cultured as described previously.11

ELISA

The amount of IL-10 released from transduced HAPI cells or DRG neurons was determined using a commercial IL-10 ELISA kit (R&D Systems, Minneapolis, MN, USA). TNFα released from HAPI cells after induction by LPS was determined using TNFα ELISA kits (R&D Systems).

Western blot

The dorsal quadrant was dissected from L4–L6 spinal cord and homogenized in lysis buffer containing 50 mM Tris, 10 mM NaCl, 1% NP40, 0.02% sodium azide and protease inhibitor cocktail (Sigma-Aldrich) at pH 7.4. Cultured DRG cells and HAPI cells were dislodged from culture plates with a cell scraper and homogenized in the same lyses buffer. Cell lysate and tissue homogenates were sonicated then centrifuged at 15 000 g for 10 min at 4 °C. Medium from DRG cell cultures was centrifuged at 1000 g for 10 min at 4 °C. Ammonium sulfate was added to the supernatant to 35% saturation and centrifuged at 10 000 g for 15 min at 4 °C, and the pellet was collected for analysis. Aliquots containing 20 μg of protein were dissolved in Laemmli buffer and boiled at 95 °C for 5 min, and the proteins separated by 12% Tris-glycine sodium dodecyl sulfate-polyacrylamide gel (Bio-Rad Laboratories, Hercules, CA, USA), transferred to nitrocellulose membranes, blocked and then incubated with primary antibodies at 4 °C. Primary antibodies included an antibody against HA (anti-HA, 1:500; Sigma-Aldrich), anti-IL-10 (1:1000; R&D Systems), anti-TNFα (1:500; Chemicon International, Temecula, CA, USA or 1:500; R&D Systems) and anti-phospho-p38 and anti-p38 (1:500; Santa Cruz Biotechnology, Santa Cruz, CA, USA). Peroxidase-coupled secondary antibodies (Calbiochem) were used for amplification and protein bands visualized using X-OMAT AR film (Kodak, Rochester, NY, USA) after chemiluminescence (DuPont NEN, Boston, MA, USA). The membranes were stripped and reprobed with mouse anti-β-actin (1:2000; Sigma-Aldrich) as a loading control. The intensity of each band was determined by quantitative chemiluminescence using a PC-based image analysis system (ChemiDoc XRS System, Bio-Rad Laboratories).

Immunocytochemistry

Cells in culture were fixed, blocked and probed with anti-HA (1:500; Sigma-Aldrich) or anti-Tuj1 (1:8000; Covance Research Products, Berkeley, CA, USA) overnight. HAPI cells were stained with OX42 (1:200; Chemicon International) and IL-10 receptor (1:200; Santa Cruz Biotechnology). The secondary antibodies utilized were fluorescent anti-rabbit IgG Alexa Fluor 594 or anti-mouse IgG Alexa Fluor 488 (1:2000; Molecular Probes, Eugene, OR, USA). HAPI cell nuclei were detected by Hoechst staining. Rats were perfused with 4% paraformaldehyde, the L4–L6 segment of spinal cord was postfixed and cryoprotected, and 20 μm cryostat sections were incubated with goat anti-TNFα (1:25, Santa Cruz Biotechnology) and the microglial marker OX42 (1:200, Chemicon International) followed by fluorescent anti-goat IgG Alexa Fluor 594, and anti-mouse IgG Alexa Fluor 488 (1:2000; Molecular Probes). Images were captured using a Zeiss LSM 510 Meta confocal microscope.

Semi-quantitative RT–PCR

cDNA prepared from RNA isolated from HAPI cells or rat spinal cord was amplified using following primer sets: β-actin-F (5′-CAGTTCGCCATGGATGACGATATC-3′) and β-actin-R (5′-CACGCTCGGTCAGGATCTTCATG-3′) for β-actin, and TNFα-F (5′-TCCGAGATGTGGAACTGGCAGAG-3′) and TNFα-R (5′-GAGCAATGACTCCAAAGTAGACCTGC-3′) for TNFα. All reactions involved initial denaturation at 94 °C for 5 min followed by 28 cycles (for β-actin and TNFα) at 94 °C for 30 s, 68 °C for 3 min, followed by 1 cycle at 68 °C for 8 min using a GeneAmp PCR 2700 (Applied Biosystems, Foster City, CA, USA).

Data analysis

Statistical significance of the difference between vector-treated and control animals was determined by one-way analysis of variance with post hoc comparisons where appropriate. Parametric statistics, using the general linear model for repeated measures, were used to identify significant effects of treatment condition on the behavioral measure of neuroinflammatory pain. The results were examined using the software package SPSS 12.0 for Windows (SPSS, Chicago, IL, USA). Data are expressed as mean±standard error of mean (s.e.m.), with P-value less than 0.05 considered significant.

Experimental animals and behavioral analysis

Male Sprague–Dawley rats weighing 200–250 g were used in all experiments. Housing conditions and experimental procedures were approved by the University of Michigan Committee on Use and Care of Animals. Vector QHIL10 or control vector QHGFP was injected subcutaneously (30 μl of vector at a concentration of 1 × 109 PFU per ml) into one hind foot. After 10 days of vector inoculation, 50 μl of 5% formalin was injected subcutaneously into the dorsal surface of the right hind paw. Spontaneous flinching behavior during both acute phase and long-lasting tonic phase was observed. The number of flinches, characterized as a rapid and brief withdrawal or flexion of the injected paw, was counted for 1 min periods at 1–2, 5–6 and 9–10 min, and then at 5-min intervals during the period 10–60 min after injection.

Disclosure/conflict of interest

The authors declare no conflicts of interest.