Anti-NGF therapy profoundly reduces bone cancer pain and the accompanying increase in markers of peripheral and central sensitization

Pain 115 (2005) 128–141 www.elsevier.com/locate/pain

Anti-NGF therapy profoundly reduces bone cancer pain and the accompanying increase in markers of peripheral and central sensitization Molly A. Sevcika, Joseph R. Ghilardia,b, Christopher M. Petersa, Theodore H. Lindsaya, Kyle G. Halvorsona, Beth M. Jonasa, Kazufumi Kubotaa, Michael A. Kuskowskic, Leila Boustanyd, David L. Sheltond, Patrick W. Mantyha,b,* a

Neurosystems Center and Departments of Preventive Sciences, Psychiatry, Neuroscience, and Cancer Center, University of Minnesota, 515 Delaware Street, Minneapolis, MN 55455, USA b Research Service, VA Medical Center, Minneapolis, MN 55417, USA c GRECC, VA Medical Center, Minneapolis, MN 55417, USA d Rinat Neuroscience Corporation, Palo Alto, CA 94304, USA Received 8 November 2004; received in revised form 1 February 2005; accepted 14 February 2005

Abstract Bone cancer pain can be difficult to control, as it appears to be driven simultaneously by inflammatory, neuropathic and tumorigenic mechanisms. As nerve growth factor (NGF) has been shown to modulate inflammatory and neuropathic pain states, we focused on a novel NGF sequestering antibody and demonstrated that two administrations of this therapy in a mouse model of bone cancer pain produces a profound reduction in both ongoing and movement-evoked bone cancer pain-related behaviors that was greater than that achieved with acute administration of 10 or 30 mg/kg of morphine. This therapy also reduced several neurochemical changes associated with peripheral and central sensitization in the dorsal root ganglion and spinal cord, whereas the therapy did not influence disease progression or markers of sensory or sympathetic innervation in the skin or bone. Mechanistically, the great majority of sensory fibers that innervate the bone are CGRP/TrkA expressing fibers, and if the sensitization and activation of these fibers is blocked by anti-NGF therapy there would not be another population of nociceptors, such as the non-peptidergic IB4/RET-IR nerve fibers, to take their place in signaling nociceptive events. q 2005 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved. Keywords: Tumor; Skeletal malignancies; NGF; Metastasis; Nociception

1. Introduction The most frequent presenting symptom of tumor metastasis to the skeleton is bone pain. Pain originating from skeletal metastases usually increases in intensity with the evolution of the disease and is commonly divided into three categories: ongoing pain, spontaneous breakthrough (incident) pain and movement-evoked breakthrough pain (Mercadante and Arcuri, 1998; Portenoy and Hagen, 1990). * Corresponding author. Address: Neurosystems Center and Departments of Preventive Sciences, Psychiatry, Neuroscience, and Cancer Center, University of Minnesota, 515 Delaware Street, Minneapolis, MN 55455, USA. Tel.: C1 612 626 0180; fax: C1 612 626 2565. E-mail address: [email protected] (P.W. Mantyh).

Ongoing pain, which is the most frequent initial symptom of bone cancer, begins as a dull, constant, throbbing pain that increases in intensity with time and is exacerbated by use of involved portions of the skeleton (Mercadante, 1997; Portenoy and Lesage, 1999). As bone cancer progresses, intermittent episodes of extreme pain can occur spontaneously, or more commonly, after weight-bearing or movement of the affected limb (Mercadante, 1997; Portenoy and Lesage, 1999). Of these types of pain, breakthrough pain is the more difficult to control, as the dose of opioids required to control this pain are usually significantly greater than that needed to control ongoing pain, and the doses of opioids needed to completely control breakthrough pain are often accompanied by significant unwanted side effects such

0304-3959/$20.00 q 2005 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.pain.2005.02.022

M.A. Sevcik et al. / Pain 115 (2005) 128–141

as sedation, somnolence, respiratory depression and constipation (Mercadante, 1997; Portenoy, 1999). To define the mechanisms that drive bone cancer pain, we utilize a model in which mouse osteosarcoma tumor cells are injected into the intramedullary space of the mouse femur (Honore et al., 2000b; Schwei et al., 1999), which produces extensive tumor-induced bone destruction and behaviors indicative of ongoing and movementevoked pain, similar to that observed in patients with metastatic bone cancer pain. Following injection and confinement of sarcoma cells to the mouse femur, painrelated behaviors in mice first appear 6 days following injection and continue through sacrifice at 14 days postinjection. Similar to humans with bone cancer pain, both the ongoing and movement-evoked pain in the mouse sarcoma model are relatively resistant to opioid therapy so that in general 10-fold higher doses of morphine are required to control bone cancer pain as compared to chronic inflammatory pain (Luger et al., 2002). Additionally, neuropathic pain is frequently resistant to conventional analgesic therapy such as opioids (Dellemijn, 1999; Woolf and Mannion, 1999), suggesting a potential neuropathic or tumorigenic component may be involved in driving bone cancer pain. Although there seem to be components similar to classic inflammatory and neuropathic pain states in this model, the neurochemical signature of observed changes in the spinal cord and DRG of tumor-bearing animals is unique and distinguishable from either classical pain state (Honore et al., 2000b; Schwei et al., 1999; Urch et al., 2003). Nerve growth factor (NGF) has been shown to be important in modulating inflammatory (Bennett, 2001; Jaggar et al., 1999; Lamb et al., 2003; Woolf et al., 1994) and neuropathic (Ramer et al., 1998; Ro et al., 1999) pain states. NGF has also been shown to be expressed by several tumor, inflammatory and immune cells (Dolle et al., 2003; Vega et al., 2003). NGF sequestering agents have been shown to attenuate several pain states, such as some neuropathic (Gwak et al., 2003; Ro et al., 1999) and inflammatory pain (Koltzenburg et al., 1999; McMahon et al., 1995). In the present report, we focus on a novel antiNGF sequestering antibody and its potential utility in attenuating bone cancer pain.

2. Methods 2.1. Animals Experiments were performed on a total of 158 adult male C3H/HeJ mice (The Jackson Lab, Bar Harbor, ME), weighing 20–25 g. The mice were housed in accordance with the National Institutes of Health guidelines under specific pathogen free (SPF) conditions in autoclaved cages maintained at 22 8C with a 12-h alternating light and dark cycle and were given autoclaved food and water ad libitum. All procedures were approved by

129

the Institutional Animal Care and Use Committee at the University of Minnesota. 2.2. Culture and injection of tumor cells Osteolytic murine sarcoma cells were obtained (NCTC 2472, ATCC, Rockville, MD), stably transfected with green fluorescent protein (GFP) and maintained as previously described (Sabino et al., 2002). Injection of tumor cells were performed as previously described (Honore et al., 2000a,b; Luger et al., 2001). In brief, following induction of general anesthesia with sodium pentobarbital (50 mg/kg, i.p.), an arthrotomy was performed exposing the condyles of the distal femur. Hank’s buffered sterile saline (HBSS, Sigma Chemical Co., St. Louis, MO; 20 ml; sham, nZ40) or HBSS containing 105 osteolytic murine sarcoma cells (20 ml, NCTC 2472, ATCC, Rockville, MD; sarcoma, nZ90) was injected into the intramedullary space of the mouse femur and the injection site sealed with dental amalgam (Dentsply, Milford, DE), followed by irrigation with sterile filtered water. A day 14 endpoint was used, as this is the time point when the tumor is still confined to the bone. To ensure confinement to the femur, the tumor can be visualized using hematoxylin and eosin (H&E), green fluorescent protein (GFP) and radiographical analysis. At day 14 there was also maximal presentation of cancer-related pain behaviors and maximal changes in expression of neurochemical markers of peripheral and central sensitization. Sham animals were used for control analysis of neurochemical changes and bone histology. They were not significantly different behaviorally, neurochemically or histologically from naı¨ve animals. 2.3. Treatment with anti-NGF antibody To assess the effect of the anti-NGF antibody (mAb 911, Rinat Neuroscience Corp., Palo Alto, CA) on pain-related behaviors, neurochemical changes, tumor growth and bone destruction, the anti-NGF antibody was administered twice (10 mg/kg, i.p. at days 6 and 11 post sham or sarcoma injection) beginning 6 days postsarcoma cell injection when observable bone destruction began and animals were euthanized at 14 days post-sarcoma cell injection, when significant bone destruction and pain behaviors were observed. The doses of anti-NGF antibody used in the current study caused no adverse effects, such as hypoalgesia, in naive mice. To monitor the general health of the mice, weights were recorded at the beginning and end of the experiments and animals were monitored for side effects, such as ataxia, illness, or lethargy. Mice were randomly placed into treatment groups receiving either sterile saline (shamCvehicle: nZ28; sarcomaCvehicle: nZ35; 1.4 ml/g, i.p. at days 6 and 11 post sham injection.) or an anti-NGF antibody, (shamCanti-NGF; nZ4; sarcomaCantiNGF: nZ23, 10 mg/kg, i.p. at days 6 and 11 post sham or sarcoma injection, Rinat Neuroscience Corp.). For behavioral comparison of anti-NGF antibody to morphine sulfate, separate experiments were performed where mice were given an acute dose of morphine 15 min (Hasselstrom et al., 1996) prior to behavioral testing (naı¨ve: nZ6; shamCvehicle: nZ8; sarcomaCvehicle: nZ8; sarcomaC anti-NGF: nZ8; sarcomaCmorphine 10 mg/kg, i.p.: nZ8; sarcomaCmorphine 30 mg/kg, i.p.: nZ8). For thermal and mechanical sensitivity testing and the assessment of hindpaw skin innervation, mice were divided into two treatment groups receiving either sterile saline (naı¨veCvehicle: nZ11) or

130

M.A. Sevcik et al. / Pain 115 (2005) 128–141

an anti-NGF antibody (naı¨veCanti-NGF: nZ11, 10 mg/kg/ at days 0 and 5, i.p., Rinat Neuroscience Corp.) for 2 weeks. 2.4. Characterization of therapy The sequestering antibody, anti-NGF, (mAb 911, Rinat Neuroscience Corp.) is effective in blocking the binding of NGF to the TrkA and p75 NGF receptors and inhibiting Trk A autophosphorylation (Hongo et al., 2000). It binds to both human and rodent NGF, does not bind to other members of the neurotrophin family and has been characterized previously (Hongo et al., 2000). The antibody was administered every five days, and this dose was based on previous work in other rodent models of pain (Shelton et al., submitted and unpublished observations) and preliminary findings that the antibody has a half-life of approximately five days in the mouse (data not shown), thus ensuring a relatively high concentration of antibody from day 6 through the duration of the study. 2.5. Euthanasia and processing of tissue Mice were sacrificed at day 14 post tumor injection and the tissues were processed for immunohistochemical analysis of spinal cord, dorsal root ganglia (DRG), hindpaw skin and femora as previously described (Honore et al., 2000a,b; Luger et al., 2001; McCarthy et al., 1995). Briefly, mice received a normally nonnoxious mechanical stimulation of the injected knee 1.5 h prior to euthanasia for induction of c-Fos expression (Honore et al., 2000b; Hunt et al., 1987). Following this manipulation, mice were euthanized with CO2 and perfused intracardially with 12 ml 0.1 M phosphate buffered saline (PBS) followed by 25 ml 4% formaldehyde/12.5% picric acid solution. Spinal cord segments (L2–L4), DRG (L1–L5) and serial plantar skin cross-sections were removed, post-fixed in the perfusion fixative overnight and cryoprotected in 30% sucrose at 4 8C for 24 h. Serial frozen spinal cord and plantar skin sections, 60 mm thick, were cut on a sliding microtome, collected in PBS, and processed as free floating sections. Serial DRG sections, 15 mm thick, were cut on a cryostat and thaw-mounted on gelatin-coated slides for processing. Following sectioning, DRG, spinal cord and plantar skin sections were briefly rinsed in PBS and then incubated in blocking solution at 22 8C (3% normal donkey serum (NDS) 0.3% Triton X100 in PBS) for 1 h followed by incubation overnight in the primary antibody. Spinal cord sections were immunostained for cFos protein (1:2,000, Oncogene Research, San Diego, CA) and dynorphin (polyclonal guinea pig anti-dynorphin, 1:1,000, Neuromics, Minneapolis, MN). DRG sections were immunostained for activating transcription factor 3 (ATF-3) (polyclonal rabbit antiATF-3, 1:500, Santa Cruz Biotechnologies, Santa Cruz, CA) and CD68 (ED-1; polyclonal rat anti-CD68, 1:5,000, Serotec, Raleigh, NC). Skin sections were immunostained for calcitonin gene related peptide (CGRP, 1:15,000; Sigma, St. Louis, MO), tyrosine hydroxylase (TOH, polyclonal rabbit anti-TOH, 1:2,000, Chemicon, Temecula, CA) and neurofilament H (Clone RT97, polyclonal rabbit anti-RT-97, 1:2,500 Sigma). After incubation in primary antibody, sections were rinsed in PBS three times for 10 min each and then incubated in the secondary antibody solution for 3 h at 22 8C. Secondary antibodies, conjugated to Cy3 or biotin (Jackson ImmunoResearch, West

Grove, PA), were used at 1:600 or 1:500 respectively. In order to detect secondary antibodies conjugated to biotin, following secondary incubation, sections were rinsed in PBS and incubated in Cy3 conjugated streptavidin (1:4000; Jackson ImmunoResearch) for 45 min. To confirm specificity of the primary antibodies, controls included omission of the primary antibody or preabsorption with the corresponding synthetic peptide. Following immunostaining procedures, spinal cord and plantar skin sections were mounted onto gelatin-coated slides. All mounted sections were then dehydrated in alcohol gradients (70, 90, 100%), cleared in xylene and coverslips were mounted with DPX (Fluka, Switzerland). Following radiological examination, the animals were perfused at day 14 as described above and the right (internal control) and left (tumor-bearing) femora were fixed in picric acid and 4% formalin at 4 8C overnight and decalcified in 10% EDTA (Sigma) for no more than 14 days. Bones were then embedded in paraffin. Femoral sections, 5 mm thick, were cut in the lateral plane and stained with tartrate-resistant acid phosphatase (TRAP) and hematoxylin and eosin (H&E) to visualize histological features of the normal bone marrow, tumor, osteoclasts and macrophages. To visualize sarcoma cells using fluorescence microscopy, femoral sections 5 mm thick were stained with an antibody raised against green fluorescent protein (GFP) (rabbit anti-GFP, 1:6,000, Molecular Probes, Eugene, OR). GFP staining was performed using TSA-Plus Cyanine 3 System (Perkin–Elmer Life Sciences, Inc., Boston, MA), as previously described (Sabino et al., 2002; Sevcik et al., 2004). Immunohistochemical analysis of the sham and cancerous femora was performed on decalcified, paraffin embedded 14 mm serial sections. Endogenous peroxidases were quenched by incubating the sections in 2% hydrogen peroxide for 1 h at 22 8C. Sections were then rinsed three times with PBS for 10 min and blocked in TSA blocking buffer for 1 h. Primary antiserum was added upon removal of the blocking buffer and allowed to incubate at room temperature overnight. Primary afferent unmyelinated and thinly myelinated sensory nerve fibers were labeled using an antibody raised against rabbit anti-calcitonin gene related peptide (CGRP, 1:15,000; Sigma). Sections were rinsed three times in TSA wash buffer for 10 min followed by 45 min incubation in streptavidin HRP (1:4,000). Sections were then rinsed three times at 22 8C with TSA wash buffer for 10 min. CY3-conjugated tyramine (1:600) from the TSA-Plus cyanine 3 System (Perkin– Elmer Life Sciences, Inc.) was applied to the femora for 7 min, washed twice with TSA wash buffer and once with PBS. Finally, the sections were air dried, dehydrated through an alcohol gradient (70, 90 and 100%), cleared in xylene and mounted with DPX (Fluka, Switzerland). 2.6. Radiographical analysis of bone Radiographs (Faxitron X-ray Corp., Wheeling, IL) of dissected femora were obtained at the day 14 time point to optimally assess bone destruction. Images were captured on Kodak Min-R 2000 mammography film (Eastman Kodak Co., Rochester, NY; exposure settings: 7 sec, 21 kVp). The extent of tumor-induced femoral bone destruction was radiologically assessed in the lateral plane of whole bone images at 5! magnification using a 0 to 5 scale (0, normal bone with no signs of destruction and 5,

M.A. Sevcik et al. / Pain 115 (2005) 128–141

full-thickness bicortical bone loss) (Honore et al., 2000a,b; Luger et al., 2001). 2.7. Osteoclast and macrophage proliferation analysis of bone Osteoclast proliferation was determined by quantifying the number of TRAPC osteoclasts at the bone/tumor interface for sarcoma-injected mice and at the normal marrow/bone interface for sham controls on TRAP stained femoral sections as previously described (Honore et al., 2000a). In brief, osteoclasts are histologically differentiated cells appearing as TRAPC and which are closely associated with regions of bone resorption. These cells are multinucleate and are found in Howship’s lacunae along the cortical and trabecular bone. Tumor associated macrophage (TAMs) proliferation was determined by quantifying the number of TRAPC cells that were freely dispersed throughout the tumor mass. Macrophages within the bone become activated due to tumor released factors that stimulate the cells (Orr et al., 1995), and the cellular appearance of these activated TAMs is marked by their highly irregular surface, multiple lamellipodia and phagocytic vacuoles (McBride, 1986). Results are expressed as the mean number of osteoclasts per mm or TAMs per mm2, respectively. 2.8. Quantification of tumor growth Femora containing GFP-expressing sarcoma cells were imaged using a yellow 515 nm long pass emission filter on a Nikon E600 fluorescence microscope equipped with a SPOT II digital camera utilizing SPOT image capture software (Diagnostic Instruments, Sterling Heights, MI). The total area of intramedullary space and the percent of intramedullary space occupied by tumor was calculated using Image Pro Plus v3.0 software (Media Cybernetics, Silver Spring, MD) (Sabino et al., 2002; Sevcik et al., 2004). The tumor characteristics of sarcoma cells transfected with GFP, such as growth rates, rate of bone resorption and the ability to induce bone cancer-related pain behaviors, were temporally, behaviorally and physically identical to non-transfected sarcoma cells (Sabino et al., 2002). Results are expressed as the percent of tumor growth per mm2. 2.9. Quantification of sensory fibers in bone The number of sensory nerve fibers was determined as previously described (Mach et al., 2002). Briefly, the number of CGRP-Immunoreactivity (IR) fibers in three bone regions (proximal, distal and diaphyseal) and the three bone tissues (periosteum, mineralized bone and marrow) were quantified. Only nerve fibers greater than 30 mm in length were included in the analysis. Six sections per animal were analyzed, and the fibers counted were expressed as fibers per total bone area. 2.10. Quantification of spinal cord, dorsal root ganglion and hindpaw skin Fluorescently labeled spinal cord, DRG and skin tissue sections were analyzed using either an MRC 1024 confocal microscope imaging system (Bio-Rad, Philadelphia, PA), or a SPOT II digital camera on an Olympus BX-60 fluorescence microscope with SPOT image capture software (Diagnostic Instruments, Inc.).

131

The number of DRG neurons expressing activating transcription factor 3 (ATF-3) were counted under 200! magnification with a 1 cm2 eyepiece grid. The total number of neurons (small, medium and large) was determined by counting both labeled and non-labeled neuronal cells bodies (non-labeled cell bodies exhibit background labeling that could be examined through a rhodamine or FITC filter) and results are expressed as percent of total number of neurons which express ATF-3-IR. To prevent duplicate counting of neuronal cell bodies, counts were conducted on every fourth serial section for each marker. To quantify the activated or infiltrating macrophages in DRG, SPOT camera grayscale images were obtained on a minimum of four ipsilateral and contralateral DRG sections per animal and analyzed using Image Pro Plus v3.0 software (Media Cybernetics). For each image, regions of the DRG containing only sensory neuronal cell bodies (excluding peripheral nerve) were outlined. While viewing the monitor, upper and lower thresholds of gray level density were set such that only specific CD68-IR cellular profiles were discriminated from the background in the outlined DRG. The SPOT camera output had been calibrated such that the actual area of each outlined region within acquired images could be determined. The section values for CD68-IR cellular profiles and outlined areas were summed for each animal and results were expressed as total number of CD68-IR cellular profiles per unit area (mm2). Quantification was carried out in spinal cord sections at lumbar levels L2–L4 as these spinal segments receive significant afferent input from the L1–L5 DRGs, which are the principal ganglia that provide afferent input to the mouse femur (Edoff et al., 2000; Molander and Grant, 1987; Puigdellivol-Sanchez A et al.,1998, 2000). Quantification of spinal cord sections for dynorphin was obtained from 4 randomly selected L2–L4 coronal spinal cord sections per animal. The number of dynorphin-IR neurons in spinal cord laminae III–VI were counted at 100! magnification and expressed as mean number of neurons per 60 mm L2–L4 section per animal. The number of c-Fos-IR neurons was counted in laminae III–VI of the dorsal horn in eight randomly selected L2– L4 coronal spinal cord sections per animal. To be considered cFos-IR, the immunofluorescence threshold of the nuclear profile was set at three times the mean background immunofluorescence level of the tissue section. Results are given as mean number of cFos-IR neurons per spinal cord section. Quantification of epidermal innervation density was performed on four randomly selected plantar hindpaw skin cross-sections per animal. The total number of CGRP, TOH and RT97-IR nerve fibers were counted at 200! magnification. Criteria were established to count only single intra-epidermal fibers and not multiple branches of the same fiber (McCarthy et al., 1995). The total length of epidermis in all sections quantified was measured using a 1 cm2 eyepiece grid. Only nerve fibers that were at least 25 mm in length, and projected into the superficial epidermis were counted. RT97IR nerve fibers were only counted within 500 mm of the apex of the foot pat as the density of myelinated nerve fibers is highest in this region (Navarro et al., 1995). Results are given as the mean number of intra-epidermal nerve fibers per mm length of skin per mouse. 2.11. Behavioral analysis Mice were tested for bone cancer pain-related behaviors 10 and 14 days following sham or tumor injections to assess the efficacy of anti-NGF therapy. The anti-NGF therapy was compared to acute

132

M.A. Sevcik et al. / Pain 115 (2005) 128–141

morphine (Baxter, Deerfield, IL; 10 and 30 mg/kg i.p.) treatment and the morphine was administered 15 min prior to behavioral testing to ensure that animals were tested within the therapeutic window of drug action (Hasselstrom et al., 1996). Mice were also tested 8, 10, 12 and 14 days following tumor or sham injections to assess efficacy of anti-NGF therapy (10 mg/kg, i.p. at days 6 and 11 post sham or sarcoma injection) in attenuating pain-related behaviors throughout the progression of the disease. Animals were observed over a 2-min period and ongoing and palpation-evoked bone cancer pain behaviors were analyzed, as previously described (Luger et al., 2002; Sabino et al., 2002, 2003). Briefly, the number of hindpaw flinches and time spent guarding were recorded as measures of ongoing pain, as these measures mirror patients in a clinical setting with bone cancer who protect or suspend their tumor-bearing limb. In our model, movementevoked pain due to palpation of the injected limb was evaluated using previously validated tests (Luger et al., 2001; Sabino et al., 2003). Palpation-evoked pain behaviors were examined where animals received a normally non-noxious palpation to the tumoror sham-injected limb for 2 min prior to observation (Luger et al., 2001). Mice were monitored over a 2-min period, and the number of flinches and time spent guarding were recorded. Palpationevoked behavior tests were developed to reflect the clinical condition when patients with bone cancer experience pain following normally non-noxious movement of the tumor-bearing limb. Following a 15 min acclimation period, thermal and mechanical sensitivity were measured in naı¨ve and naı¨veCvehicle antiNGF animals to assess whether the normal pain threshold responses were altered with anti-NGF treatment. Thermal sensitivity was measured using a Thermal Paw Stimulator (University of California, San Diego, CA). The intensity of radiant heat was adjusted so that the naı¨ve animals responded to the heat by elevating the hindpaw approximately 9 s after the heat was initiated (Choi et al., 2003). The mice were allowed 5 min to recover between each trial. A single trial consisted of four measurements per hindpaw, the longest latency was eliminated and the remaining three measurements were averaged. Mechanical sensitivity was measured using a previously validated method (Chaplan et al., 1994). Von Frey filaments (Stoelting Co., Wood Dale, IL) were applied to the hindpaw of the animals, and the withdrawal threshold was determined by increasing and decreasing the stimulus intensity between 0.2 and 15.1 g equivalents of force. A positive response was noted if the paw was quickly withdrawn. 2.12. RT PCR analysis of mRNA levels of NGF in the 2472 cell line Total RNA from triplicate mouse tissue samples or 2472 sarcoma cells was prepared according to manufacturer’s instructions using the RNeasy micro kit (Qiagen, Valencia, CA), and the RNA was quantified using Ribogreen reagent (Molecular Probes). Two-step RT-PCR was performed using the TaqMan Gold RT-PCR kit (Applied Biosystems, Foster City, CA). The RNA was reverse transcribed using random hexamers, and the cDNA was amplified using a primer/probe set specific for NGF (muNGF-187F: GGGCTGGATGGCATGCT, muNGF-256R: GCGTCCTTGGCAAAACCTT, muNGF-208T: CCAAGCTCACCTCAGTGTCTGGGCC). The samples were analyzed in duplicate from the RT step and normalized to total RNA input.

2.13. Statistical analysis The SPSS version 11 computer statistics package (SPSS, Chicago, IL) was used to perform statistical analyses. Mixed effects linear regression modeling was used to analyze the repeated measures data and accommodates subjects measured at differing time intervals, include both fixed and time-varying covariates, and estimates individual rates of change. Each dependent outcome variable was also compared across groups at specific timepoints using a non-parametric analysis of variance (Kruskal–Wallis). Significant Kruskal–Wallis analyses were followed by nonparametric post-hoc comparisons between pairs of groups using the Mann–Whitney U test. Results were considered statistically significant at P!0.05. In all cases, the investigator was blind to the experimental status of each animal.

3. Results 3.1. Anti-NGF therapy had no effect on disease progression or macrophage infiltration in the bone The effects of anti-NGF therapy on bone destruction, osteoclast proliferation and tumor growth were examined at day 10 and 14 post tumor injection. Sham-injected mice did not demonstrate significant bone destruction (bone score 0.9G0.4; Fig. 1(A)), osteoclast proliferation throughout the entire intermedullary bone/normal marrow interface including the maintained trabecular regions/marrow interfaces (4.6G0.4 osteoclasts/mm) or tumor growth (Fig. 1(D)), as assessed by radiological, TRAP and H&E/GFP analysis, respectively. In sarcomaCvehicle mice, there was extensive bone destruction as observed and characterized by multifocal radiolucencies and complete loss of trabecular bone regions present in sham-injected animals (bone score 3.5G0.2; Fig. 1(B)), in addition to osteoclasts along the cortical bone/ tumor interface (4.0G0.7 osteoclasts/mm) and the tumor had completely filled the intramedullary space (100G0.0% of intramedullary space; Fig. 1(E)). Treatment of sarcomainjected mice with anti-NGF at days 6–11 post tumor injection resulted in no significant change in bone resorption (3.1G0.6; Fig. 1(C)), no reduction in sarcoma-induced osteoclast proliferation at the bone/tumor interface (3.5G 0.1 osteoclasts/mm) or tumor growth (98.0G0.9% of intramedullary space; Fig. 1(F)) as compared to sarcomaC vehicle animals. Fourteen days following tumor injection, sarcomaC vehicle mice displayed a significant number of TAMs (39.8G12.6 TAMs/mm2) and this number was not significantly different than anti-NGF treated sarcoma-injected mice (29.5G7.3 TAMs/mm2). 3.2. Anti-NGF therapy has no observable effect on sensory or sympathetic innervation in bone or skin Thinly myelinated or unmyelinated peptidergic sensory nerve fibers were labeled with an antibody raised against

M.A. Sevcik et al. / Pain 115 (2005) 128–141

calcitonin gene related peptide (CGRP). CGRP-IR nerve fibers were found throughout the entire bone (periosteum, mineralized bone and bone marrow) of both naı¨veCvehicle and naı¨veCanti-NGF animals. There was no significant difference between the intensity or density of nerve fibers between naı¨veCvehicle and naı¨veCanti-NGF mice (Table 1). In agreement with a recent study (Peters et al., in press), CGRP-IR was not observed in sarcoma-bearing animals and; therefore, no quantification was performed. Thinly myelinated or unmyelinated peptidergic sensory nerve fibers (CGRP-IR), large myelinated sensory fibers (RT97-IR) and noradrenergic sympathetic nerve fibers (TOH-IR) were analyzed in the hindpaw plantar skin by immunohistochemistry using antibodies raised against CGRP, RT97 and TOH, respectively. There was no significant difference between the intensity or density of CGRP-IR fibers in sarcomaCvehicle and sarcomaCantiNGF hindpaw skin samples (Fig. 2(A) and (B); Table 1). Differences in the density and intensity of RT97-IR and TOH-IR fibers were also undetectable in sarcomaCvehicle and sarcomaCanti-NGF treated animals (Table 1). Similarly, there was no significant difference between the intensity or density of CGRP-IR fibers in naı¨veCvehicle and naı¨veCanti-NGF hindpaw skin samples (Fig. 2(C) and (D); Table 1). Differences in the density and intensity of RT97-IR and TOH-IR fibers were also undetectable in naı¨veCvehicle and naı¨veCanti-NGF treated animals (Table 1). There were no significant observable differences between the intensity or density of CGRP, RT97 or TOH-IR fibers in the skin samples of sarcomaCvehicle and sarcomaCanti-NGF versus the naı¨veCvehicle and naı¨veC anti-NGF animals.

133

Fig. 1. Anti-NGF therapy has no significant effect on disease progression in bone at day 14 post tumor injection. Sham animals (nZ8), given vehicle, (A) show no radiographically apparent bone destruction at d14, whereas sarcomaCvehicle animals (nZ13) show a transition from the radio-opaque bone tissue to a (B) radio-lucent appearance by d14. SarcomaCanti-NGF (C) animals (nZ8) present the same pattern and extent of bone destruction as sarcomaCvehicle animals. GFP transfected tumor cells injected into the ipsilateral femurs were immunostained with an antibody raised against GFP. ShamCvehicle animals are not immunoreactive for GFP (D), whereas sarcomaCvehicle and sarcomaCanti-NGF are immunoreactive for GFP (E, F); however, no significant difference in tumor growth was observed. Scale bars: A–C 1 mm; D–F 4 mm. SARC=Sarcoma and VEH=Vehicle.

3.3. Anti-NGF antibody therapy significantly reduces bone cancer pain behaviors Ongoing pain was analyzed by measuring spontaneous guarding and flinching over a 2 min time period. SarcomaC vehicle mice demonstrated a greater time spent guarding as compared to the shamCvehicle controls (Fig. 3(A)). Additionally, sarcomaCvehicle mice exhibited an increased number of flinches as compared to shamCvehicle controls (Fig. 3(B)). Administration of anti-NGF (at day 6 and day 11) in sarcoma-injected mice significantly attenuated spontaneous guarding as compared to sarcomaC vehicle mice (Fig. 3(A)). Anti-NGF treatment also significantly reduced spontaneous flinching in sarcoma-injected mice (Fig. 3(B)) as compared to sarcomaC vehicle. Movement-evoked pain was analyzed by measuring palpation-induced responses. SarcomaCvehicle mice demonstrated a greater time spent guarding after palpation as compared to the shamCvehicle controls (Fig. 3(C)). SarcomaCvehicle mice also exhibited an increased number of flinches after palpation as compared to shamCvehicle controls (Fig. 3(D)). Anti-NGF treatment

in sarcoma-injected mice significantly reduced both palpation-evoked guarding (Fig. 3(C)) and palpationevoked flinching (Fig. 3(D)). In these studies, no significant behavioral differences or side effects, such as ataxia, illness, or lethargy, were observed between animals receiving either vehicle or anti-NGF therapy (results not shown). No significant body weight differences were observed between cancerous or sham animals or animals receiving either vehicle or anti-NGF therapy (Table 2). 3.4. Anti-NGF therapy has no effect on baseline thermal or mechanical thresholds and can be compared to the efficacy of morphine in reducing bone cancer pain There was no significant increase in latency of paw withdrawal to a thermal stimulus or increase in threshold of mechanical stimulation with anti-NGF therapy as compared to normal pain thresholds. Anti-NGF therapy had no effect on either normal thermal response (Fig. 4(A)) as compared

134

M.A. Sevcik et al. / Pain 115 (2005) 128–141

Table 1 Summary of the sensory and sympathetic innervation of the hindpaw of animals treated with anti-NGF therapy NativeCvehicle Innervation markers of the hindpaw CGRP-IR fibers 12.5G0.5 RT97-IR fibers 10.4G0.4 TOH-IR fibers 3.4G0.4

NativeCanti-NGF

SarcomaCvehicle

SarcomaCanti-NGF

11.9G0.7 11.9G0.7 3.0G0.8

12.0G0.8 9.1G1.4 3.1G0.7

12.5G0.6 11.1G1.5 3.6G0.7

Thinly myelinated or unmyelinated peptidergic sensory nerve fibers (CGRP-IR), large myelinated sensory fibers (RT-97-IR) and noradrenergic sympathetic nerve fibers (TOH-IR) were analyzed in the hindpaw plantar skin by immunohistochemistry using antibodies raised against CGRP, RT-97 and TOH, respectively. There was no significant difference between the intensity or density of CGRP-IR, RT-97 or TOH fibers in sarcomaCvehicle and sarcomaCantiNGF hindpaw skin or naı¨veCvehicle and naı¨veCanti-NGF hindpaw skin. There were no significant observable differences between the intensity or density of CGRP, RT97 or TOH-IR fibers in the skin samples of sarcomaCvehicle and sarcomaCanti-NGF versus the naı¨veCvehicle and naı¨veCanti-NGF animals.

to naı¨veCvehicle or normal mechanical stimulation (Fig. 4(B)) as compared to naı¨veCvehicle. Animals were tested to compare the efficacy of anti-NGF antibody to morphine sulfate (MS) in reducing bone cancerrelated behaviors. Behavioral assessment on days 10 and 14 revealed that sarcomaCvehicle animals showed statistically longer time guarding (Fig. 4(C)) and flinching (Fig. 4(E)) and increased time guarding (Fig. 4(D)) and flinching (Fig. 4(F)) in response to palpation of the injected limb as compared to shamCvehicle animals. Treatment with either anti-NGF (10 mg/kg at days 6 and 11 post sham or sarcoma injection, i.p.) or acute morphine sulfate (10 or 30 mg/kg i.p.) significantly reduced both ongoing and movement-evoked guarding and flinching behaviors at days 10 and 14 post tumor injection (Fig. 4(C–F)), as compared to sarcomaC vehicle mice. Anti-NGF therapy significantly attenuated the bone cancer-related pain behaviors more effectively as

compared to 10 mg/kg or 30 mg/kg morphine (P!0.05 vs. SarcomaCanti-NGF). 3.5. Anti-NGF therapy modulates peripheral changes induced by bone cancer in the DRG Activating transcription factor-3 (ATF-3), which is in the ATF/CREB family, has previously been shown to be up-regulated in a model of peripheral nerve injury (Tsujino et al., 2000). This up-regulation is seen in sensory and motor neuron cell bodies and is known to label injured neurons. There was significant increase in the percentage of ATF-3IR neurons in L2 DRG ipsilateral to the sarcoma-injected femur (14.0G5.9% of total neurons in L2 expressed ATF-3; Fig. 5(A)) as compared to shamCvehicle (1.6G0.5% of total neurons in L2 expressed ATF-3). Treatment with antiNGF significantly attenuated the expression of ATF-3

Fig. 2. Anti-NGF therapy has no observable effect on sensory innervation of the skin. Confocal images show representative hindpaw skin samples of both sarcoma-injected (A, B) and naı¨ve (C, D) mice were immunostained for the neuropeptide calcitonin gene-related peptide (CGRP), which labels small diameter peptidergic sensory nerve fibers. Note that there was no difference in intensity or density of CGRP immunostained nerve fibers between sarcomaCvehicle (A, nZ3) mice and sarcomaCanti-NGF (B, nZ8) mice. Similarly, there was no difference in intensity or density of CGRP-IR fibers between naı¨veCvehicle (C, nZ8) mice and naı¨veCanti-NGF (D, nZ8) mice. Also note that there was no change in the number of nerve fibers expressing CGRP in sarcoma-injected and naı¨ve mice (A, B vs. C, D). Scale bar: 50 mm. SARC=Sarcoma and VEH=Vehicle.

M.A. Sevcik et al. / Pain 115 (2005) 128–141

135

Fig. 3. Anti-NGF therapy attenuates bone cancer pain. Anti-NGF treatment (10 mg/kg at days 6 and 11 post sham or sarcoma injection, i.p.) attenuated both ongoing and movement-evoked bone cancer pain behaviors throughout the progression of the disease. The time spent guarding and number of spontaneous flinches of the sarcoma injected limb over a 2-min observation period was used as a measure of ongoing pain 8, 10, 12 and 14 days after injection and confinement of sarcoma cells to the left femur (A, B). Parameters of movement-evoked pain included quantification of time spent guarding and the number of flinches over a 2min observation period following a normally non-noxious palpation of the sarcoma-injected femur (C, D). Note that anti-NGF treatment post tumor injection (triangle) significantly reduced ongoing and palpation-evoked pain behaviors on days 10, 12 and 14 as compared to sarcomaCvehicle (square). At 10, 12 and 14 days post injection, shamCvehicle are significantly different from sarcomaCvehicle. #P!0.05 vs. shamCvehicle; *P!0.05 vs. sarcomaCvehicle. SARC=Sarcoma and VEH=Vehicle.

(2.6G1.0% of total neurons in L2 expressed ATF-3; Fig. 5(B)) 14 days post tumor injection. Macrophage infiltration has been shown to be upregulated due to peripheral nerve damage (Abbadie et al., 2003; Myers et al., 1996; Tofaris et al., 2002). An antibody raised against CD68 (ED-1), a lysosomal protein expressed by activated tissue macrophages, was used to assess macrophage infiltration in sarcoma-injected mice. There was an up regulation in the number of CD68-IR profiles in neurons in the ipsilateral DRG of sarcomaC vehicle mice (119.6G12.1 cellular profiles/L2 ipsilateral DRG; Fig. 5(C)) compared to shamCvehicle (80.6G6.0 cellular profiles/L2 ipsilateral DRG). Anti-NGF treatment significantly reduced the up-regulation of CD68-IR neurons in the ipsilateral DRG (92.0G9.9 cellular profiles/L2 ipsilateral DRG; Fig. 5(D)) in sarcoma-injected mice. 3.6. Anti-NGF therapy modulates central changes induced by bone cancer in the spinal cord Expression of dynorphin has been shown to be involved in the maintenance of chronic pain (Vanderah et al., 1996,

2001). Dynorphin expression has also been shown to be upregulated in the dorsal horn of the spinal cord in several persistent pain states (Iadarola et al., 1988; Noguchi et al., 1991; Schwei et al., 1999). In shamCvehicle mice, a small population of spinal neurons expressed dynorphin in deep spinal laminae (2.3G1.1 dyn-IR neurons/L3/L4 section). In contrast, sarcomaCvehicle mice expressed significantly more dynorphin-IR neurons (6.0G0.5 dyn-IR neurons/L3/L4 section; Fig. 6(A)). Anti-NGF treatment Table 2 Summary of the body weights of animals receiving sham or cancer injections and vehicle or anti-NGF therapy

Animal groups ShamCVEH SarcCVEH SarcCanti-NGF

Body weight (g) before injection

Body weight (g) at day 14

25.0G0.8 25.6G0.5 27.2G0.9

24.8G0.9 24.3G0.6 26.0G1.0

There was no significant difference in body weight between sham and sarcoma-injected animals at the day 14 timepoint. There was also no significant body weight difference between sarcomaCvehicle and sarcomaCanti-NGF-treated animals. SARC=Sarcoma and VEH=Vehicle.

136

M.A. Sevcik et al. / Pain 115 (2005) 128–141

Fig. 4. Anti-NGF therapy has no effect on baseline thermal or mechanical thresholds and is compared to the efficacy of acute morphine (MS) in reducing bone cancer pain. Anti-NGF treatment (10 mg/kg at days 6 and 11 post sham or sarcoma injection, i.p.) in naı¨ve mice had no effect on basal thermal or mechanical responses. Note that there was no significant increase in latency of paw withdrawal to a thermal stimulus or increase in threshold of mechanical stimulation with anti-NGF therapy (A, B). Tumor-induced ongoing pain behaviors were evaluated by measuring spontaneous guarding (C) and flinching (E) over a 2-min observation period. Movement-evoked pain was assessed by measuring the time spent guarding (D) and flinching (F) over a 2-min observation period following normally non-noxious palpation of the distal femur. Note that in mice with bone cancer, there is a significant increase in the duration of guarding and flinching, and treatment with anti-NGF treatment (10 mg/kg, at days 6 and 11 post sham or sarcoma injection, i.p.) administered from 6 days to 14 days post-tumor injection significantly reduced both parameters of ongoing and movement-evoked pain-related behaviors compared with sarcomaCvehicle animals (C–F). Note that antiNGF treatment significantly reduced both ongoing and touch-evoked pain behaviors at day14 more efficiently than morphine at 10 and 30 mg/kg (15 min prior to testing, i.p.) (C–F). Error bars represent S.E.M. #P!0.05 vs. shamCvehicle; *P!0.05 vs. sarcomaCvehicle; CP!0.05 vs. sarcomaCmorphine. SARC=Sarcoma and VEH=Vehicle.

significantly reduced the up-regulation of dynorphin expression (2.0G0.6 dyn-IR neurons/L3/L4 section; Fig. 6(B)) in sarcoma-injected mice. Immediate-early gene activation was attenuated by antiNGF treatment. The expression of c-Fos in the deep dorsal horn (laminae III–VI) has been utilized as a marker of central sensitization in sarcoma-induced bone cancer pain (Honore et al., 2000a,b; Luger et al., 2001; Schwei et al., 1999) states. Normal, non-noxious palpation of shamoperated animals resulted in minimal expression of c-Fos in deep laminae (Sabino et al., 2002). In the bone cancer state, sarcomaCvehicle mice exhibited an increased number of c-Fos-IR neurons (27.7G4.9; cFos-IR neurons/ L3/L4 section; Fig. 6(C)) and treatment with anti-NGF significantly reduced this expression (11.1G1.9; cFos-IR neurons/L3/L4 section; Fig. 6(D)).

3.7. Expression of NGF mRNA by mouse tumor cells as compared to other mouse tissues In order to determine whether the sarcoma tumor cells were a possible source of NGF, 2472 cells grown in culture were assessed for their level of NGF mRNA by RT-PCR. These levels were compared to several normal tissues of the mouse (expressed in arbitrary units), such as brain (1.2G0.8), as well as the level of NGF mRNA from the male mouse salivary gland (1359.1G583.7), a source of aberrantly high exocrine NGF. Sarcoma 2472 cells in vitro do contain readily detectable NGF mRNA (8.0G1.1). This amount is in the range of NGF mRNA levels obtained from normal tissues expressing high levels of NGF mRNA, such as iris (8.8G3.6) (Shelton and Reichardt, 1984). It is several orders of magnitude

M.A. Sevcik et al. / Pain 115 (2005) 128–141

137

Fig. 5. Treatment with anti-NGF sequestering antibody reduced neurochemical changes and macrophage infiltration in dorsal root ganglia (DRG) of tumorbearing animals. Representative confocal images show that fourteen days post tumor implantation, activated transcription factor-3 (ATF-3, blue) increases in the ipsilateral L2 DRG of tumor-bearing animals (A, nZ8). Tumor-induced nerve injury is also accompanied by an increase in the density of CD68KIR macrophages (yellow) around injured sensory neurons within the ipsilateral DRG of sarcomaCvehicle animals (C, nZ7). Treatment with anti-NGF produced a significant reduction in the expression of ATF3 (B, nZ8) and a significant reduction in the number of CD68C macrophages within the ipsilateral L2 DRG of tumor bearing animals (D, nZ7). Scale bars a–dZ5 mm. SARC=Sarcoma and VEH=Vehicle.

below the amount of NGF mRNA present in male mouse salivary gland.

4. Discussion Previous studies have suggested that bone cancer pain has both an inflammatory and tumorigenic component (Honore et al., 2000b; Luger et al., 2001; Sabino et al., 2002). That inflammatory cells may contribute to bone cancer pain comes from the observation that tumorassociated macrophages can comprise a significant portion (2–60%) of the total tumor mass (McBride, 1986; Zhang et al., 2002) and the present sarcoma model is within these estimates with 5–10% of the mass being TRAPC, CD-68C macrophages (Jongen et al., 2002). In both in vitro and in vivo systems, macrophages, other inflammatory cells and tumor cells have been shown to express NGF (Vega et al., 2003) and released NGF would be expected to sensitize and/or activate nearby sensory neurons. As the sarcoma cells proliferate within the mouse femur, the leading edge of the tumor first comes into contact with, surrounds, injures and then ultimately destroys the very distal processes of the sensory fibers that innervate the marrow and mineralized bone (Peters et al., in press). Thus, while the sensory and sympathetic nerve fibers can be observed at and within the leading edge of the tumor, the distal processes of these fibers are ‘trimmed’ at the very distal ends as the tumor grows and destroys the host tissue.

The cell bodies that give rise to the affected sensory fibers begin to express proteins such as ATF-3, which are characteristically expressed by injured neurons (Tsujino et al., 2000). Data from the present mouse model correlates well with data obtained in human tumors where it has been shown that sensory innervation of a wide variety of carcinomas (colorectal, liver and breast) is at best sparse and, when present, is usually associated with the blood vessels that vascularize the tumor (Chamary et al., 2000; Mitchell et al., 1994; Terada and Matsunaga, 2001). In assessing the observed behavioral results it is important to address the potential mechanism(s) by which anti-NGF therapy is exerting its anti-hyperalgesia effects in mice with bone cancer. The most straight forward explanation for the anti-hyperalgesic actions is that anti-NGF therapy induces a reduction in disease progression and/or a loss of peripheral sensory or sympathetic nerve fibers that innervate peripheral tissues. To assess disease progression we examined tumor burden, tumor growth, tumor necrosis, bone destruction and osteoclastogenesis, but did not observe a significant difference in any of these measures between the sarcomaC vehicle vs. sarcomaCanti-NGF treatment groups. NGF is known to be involved in the survival of neurons of the developing animal (Petruska and Mendell, 2004) and to assess the potential loss of sensory or sympathetic innervation, we also examined the sensory (using CGRP as a marker of peptidergic primary afferent fibers that are known to express TrkA) and sympathetic innervation (using TOH which labels nearly all post-ganglionic sympathetic fibers) in

138

M.A. Sevcik et al. / Pain 115 (2005) 128–141

Fig. 6. Neurochemical changes associated with central sensitization are attenuated by administration of anti-NGF therapy. Representative confocal images of dynorphin expression in the dorsal horn of the spinal cord in sarcomaCvehicle (A, nZ9) and sarcomaCanti-NGF (B, nZ4) mice. SarcomaCvehicle mice displayed an increase in dynorphin-IR neurons in deep laminae of the ipsilateral spinal cord (A, nZ9), whereas anti-NGF therapy significantly attenuated the increase in dynorphin expression (B, nZ4). Representative confocal images of c-Fos expressing neurons of the spinal cord in sarcomaCvehicle (C, nZ4) and sarcomaCanti-NGF (D, nZ4) mice. Following a normally non-noxious palpation of tumor-bearing limbs, sarcomaCvehicle mice showed an increased expression of c-Fos protein in neurons within the deep laminae. In sarcoma animals that received administrations of anti-NGF therapy, there was a significant reduction of this increased expression of c-Fos protein. Scale bar: A, B 150 mm; C,D 200 mm. SARC=Sarcoma and VEH=Vehicle.

the plantar surface of the hindpaw skin and the femur in naı¨veCvehicle, naı¨veCanti-NGF, sarcomaCvehicle and sarcomaCanti-NGF treated groups. Similarly, we did not observe any significant differences between the naı¨veC vehicle vs. naı¨veCanti-NGF or sarcomaCvehicle vs. sarcomaCanti-NGF groups, suggesting that a significant denervation did not occur with therapy. Lastly, we examined the baseline thermal and mechanical thresholds where the stimulus was given to the plantar surface of the hindpaw in naı¨veCvehicle vs. naı¨veCanti-NGF treated animals and observed no statistically significant differences between these groups. Thus, while previous studies have shown that high doses of anti-NGF antibody in the adult can reduce the number and function of sensory fibers (Christensen and Hulsebosch, 1997) in the present study the anti-NGF therapy produced no significant effect in terms of disease progression or loss of sensory or sympathetic nerve fibers that innervate the peripheral tissues examined. In the present study, an NGF sequestering therapy was examined in light of the literature demonstrating that in the adult, NGF can play a pivotal role in driving inflammatory (Bennett, 2001; Jaggar et al., 1999; Lamb et al., 2003; Woolf et al., 1994) and neuropathic pain (Ramer et al., 1998; Ro et al., 1999). It has been shown that NGF modulates expression and function of a wide variety of molecules and proteins expressed by sensory neurons including: neurotransmitters (substance P, brain derived neurotrophic factor and CGRP), receptors (bradykinin, P2X3), channels

(TRPV1, ASIC-3 and sodium channels), transcription factors (ATF-3), and structural molecules (neurofilaments and the sodium channel anchoring molecule p11) (Averill et al., 2004; Chuang et al., 2001; Donnerer et al., 1992; Gould et al., 2000; Ji et al., 2002; Mamet et al., 2003; Okuse et al., 2002; Ramer et al., 2001; Rueff et al., 1996; Skoff et al., 2003; Verge et al., 1990; Zhou and Rush, 1996). Additionally, NGF has been shown to modulate the distribution of channels (Nav 1.8) (Gould et al., 2000) and receptors (TRPV1) (Ji et al., 2002) in the sensory neurons, but also modulates the expression profile of supporting cells in the DRG and peripheral nerve, such as non-myelinating Schwann cells and macrophage infiltration (Heumann et al., 1987). Therefore, anti-NGF therapy may be particularly effective in blocking bone cancer pain as NGF appears to be integrally involved in the up-regulation, sensitization and dis-inhibition of multiple neurotransmitters, ion channels and receptors in the primary afferent nerve fibers that synergistically increase the amount of nociceptive signaling in the tumor-bearing bone. While the above data suggests that anti-NGF should have significant anti-hyperalgesic effects, in other pain models, such as the Brennan incision model (Zahn et al., 2004), antiNGF therapy only blocks certain aspects of the pain (thermal but not mechanical hyperalgesia). One rather unique aspect of the sensory innervation of bone, which may partially explain why anti-NGF therapy is so effective in relieving bone cancer pain, is that nearly all nerve fibers that innervate the bone appear to be CGRP-expressing fibers

M.A. Sevcik et al. / Pain 115 (2005) 128–141

and nearly all CGRP-expressing fibers co-express TrkA (Averill et al., 1995; Walsh et al., 1999). Thus, in both human and rodents, fibers from primary afferent sensory neurons innervate bone marrow, mineralized bone and periosteum (Artico et al., 2002; Bjurholm et al., 1988; Chenu, 2002; Freemont et al., 2002; Hill and Elde, 1991; Madsen et al., 2000; Schwab and Funk, 1998), and the great majority of sensory fibers that innervate bone are CGRP expressing fibers with few, if any, of the non-peptidergic IB4/RET-IR nerve fibers being present (Aoki et al., 2004; Mach et al., 2002). Within the marrow and mineralized bone, sensory fibers are generally associated with the blood vessels (Irie et al., 2002; Mach et al., 2002), and nearly all of these fibers express calcitonin gene-related peptide (CGRP). Thus, the great majority of sensory fibers that innervate the bone are CGRP/TrkA expressing fibers and if the sensitization and activation of these fibers is blocked by anti-NGF therapy there would not be another population of nociceptors, such as the non-peptidergic IB4/RET-IR nerve fibers, to take their place in signaling nociceptive events. It is interesting to speculate which NGF receptors are responsible for the hyperalgesic actions revealed in these studies. The anti-NGF antibody used in these studies inhibits the interaction of NGF with both the TrkA and p75 neurotrophin receptor. NGF is capable of causing acute hyperalgesia in p75 knockout mice, showing that trkA is sufficient to mediate the NGF effect in this setting (Bergmann et al., 1998). However, it is known that the p75 receptor is critical for NGF regulation of bradykinin receptors (Petersen et al., 1998). In the complex setting of a growing tumor in the bone, in the presumed presence of multiple inflammatory mediators, it is not clear whether it is the interaction of NGF with TrkA, p75, or both that is important for the behavioral and anatomical effects seen here. Whether anti-NGF therapy can inhibit bone cancer pain in humans remains to be determined. If this anti-NGF therapy were to advance to human trials, it will be important to define the lowest dose that is efficacious in reducing bone cancer pain. The mechanism of tumor-induced bone pain is thought to be similar in mouse and human, as both are mediated by excessive osteoclast activity, release of algogenic inflammatory and tumor products and tumor-induced injury to nerve fibers that innervate the bone (Mantyh et al., 2002). If anti-NGF therapies can block tumor-induced bone pain, retain their analgesic efficacy with disease progression and provide significant opioid-sparing action without significant unwanted side effects, they have the potential to significantly reduce the pain and enhance the quality of life in patients with primary or metastatic bone cancer.

Acknowledgements This work was supported by National Institutes of Health grants (NS23970, NS048021) and a Merit Review from the Veterans Administration.

139

References Abbadie C, Lindia JA, Cumiskey AM, Peterson LB, Mudgett JS, Bayne EK, DeMartino JA, MacIntyre DE, Forrest MJ. Impaired neuropathic pain responses in mice lacking the chemokine receptor CCR2. Proc Natl Acad Sci USA 2003;100:7947–52. Aoki Y, Takahashi Y, Ohtori S, Moriya H, Takahashi K. Distribution and immunocytochemical characterization of dorsal root ganglion neurons innervating the lumbar intervertebral disc in rats: a review. Life Sci 2004;74:2627–42. Artico M, Bosco S, Cavallotti C, Agostinelli E, Giuliani-Piccari G, Sciorio S, Cocco L, Vitale M. Noradrenergic and cholinergic innervation of the bone marrow. Int J Mol Med 2002;10:77–80. Averill S, McMahon SB, Clary DO, Reichardt LF, Priestley JV. Immunocytochemical localization of trkA receptors in chemically identified subgroups of adult rat sensory neurons. Eur J Neurosci 1995; 7:1484–94. Averill S, Michael GJ, Shortland PJ, Leavesley RC, King VR, Bradbury EJ, McMahon SB, Priestley JV. NGF and GDNF ameliorate the increase in ATF3 expression which occurs in dorsal root ganglion cells in response to peripheral nerve injury. Eur J Neurosci 2004;19:1437–45. Bennett DL. Neurotrophic factors: important regulators of nociceptive function. Neuroscientist 2001;7:13–17. Bergmann I, Reiter R, Toyka KV, Koltzenburg M. Nerve growth factor evokes hyperalgesia in mice lacking the low-affinity neurotrophin receptor p75. Neurosci Lett 1998;255:87–90. Bjurholm A, Kreicbergs A, Brodin E, Schultzberg M. Substance P- and CGRP-immunoreactive nerves in bone. Peptides 1988;9:165–71. Chamary VL, Robson T, Loizidou M, Boulos PB, Burnstock G. Progressive loss of perivascular nerves adjacent to colorectal cancer. Eur J Surgical Oncol 2000;26:588–93. Chaplan SR, Bach FW, Pogrel JW, Chung JM, Yaksh TL. Quantitative assessment of tactile allodynia in the rat paw. J Neurosci Methods 1994; 53:55–63. Chenu C. Glutamatergic innervation in bone. Microsc Res Tech 2002;58: 70–6. Choi SS, Han KJ, Lee JK, Lee HK, Han EJ, Kim DH, Suh HW. Antinociceptive mechanisms of orally administered decursinol in the mouse. Life Sci 2003;73:471–85. Christensen MD, Hulsebosch CE. Spinal cord injury and anti-NGF treatment results in changes in CGRP density and distribution in the dorsal horn in the rat. Exp Neurol 1997;147:463–75. Chuang HH, Prescott ED, Kong H, Shields S, Jordt SE, Basbaum AI, Chao MV, Julius D. Bradykinin and nerve growth factor release the capsaicin receptor from PtdIns(4,5)P2-mediated inhibition. Nature 2001;411:957–62. Dellemijn P. Are opioids effective in relieving neuropathic pain? Pain 1999; 80:453–62. Dolle L, El Yazidi-Belkoura I, Adriaenssens E, Nurcombe V, Hondermarck H. Nerve growth factor overexpression and autocrine loop in breast cancer cells. Oncogene 2003;22:5592–601. Donnerer J, Schuligoi R, Stein C. Increased content and transport of substance P and calcitonin gene-related peptide in sensory nerves innervating inflamed tissue: evidence for a regulatory function of nerve growth factor in vivo. Neuroscience 1992;49:693–8. Edoff K, Grenegard M, Hildebrand C. Retrograde tracing and neuropeptide immunohistochemistry of sensory neurones projecting to the cartilaginous distal femoral epiphysis of young rats. Cell Tissue Res 2000;299: 193–200. Freemont AJ, Watkins A, Le Maitre C, Baird P, Jeziorska M, Knight MT, Ross ER, O’Brien JP, Hoyland JA. Nerve growth factor expression and innervation of the painful intervertebral disc. J Pathol 2002;197: 286–92. Gould 3rd HJ, Gould TN, England JD, Paul D, Liu ZP, Levinson SR. A possible role for nerve growth factor in the augmentation of sodium channels in models of chronic pain. Brain Res 2000;854:19–29.

140

M.A. Sevcik et al. / Pain 115 (2005) 128–141

Gwak YS, Nam TS, Paik KS, Hulsebosch CE, Leem JW. Attenuation of mechanical hyperalgesia following spinal cord injury by administration of antibodies to nerve growth factor in the rat. Neurosci Lett 2003;336: 117–20. Hasselstrom J, Svensson JO, Sawe J, Wiesenfeld-Hallin Z, Yue QY, Xu XJ. Disposition and analgesic effects of systemic morphine, morphine-6glucuronide and normorphine in rat. Pharmacol Toxicol 1996;79:40–6. Heumann R, Korsching S, Bandtlow C, Thoenen H. Changes of nerve growth factor synthesis in nonneuronal cells in response to sciatic nerve transection. J Cell Biol 1987;104:1623–31. Hill EL, Elde R. Distribution of CGRP-, VIP-, DbH-, SP-, and NPYimmunoreactive nerves in the periosteum of the rat. Cell Tissue Res 1991;264:469–80. Hongo JS, Laramee GR, Urfer R, Shelton DL, Restivo T, Sadick M, Galloway A, Chu H, Winslow JW. Antibody binding regions on human nerve growth factor identified by homolog- and alanine-scanning mutagenesis. Hybridoma 2000;19:215–27. Honore P, Luger NM, Sabino MAC, Schwei MJ, Rogers SD, Mach DB, O’Keefe PF, Ramnaraine ML, Clohisy DR, Mantyh PW. Osteoprotegerin blocks bone cancer-induced skeletal destruction, skeletal pain and pain-related neurochemical reorganization of the spinal cord. Nat Med 2000a;6:521–8. Honore P, Rogers SD, Schwei MJ, Salak-Johnson JL, Luger NM, Sabino MAC, Clohisy DR, Mantyh PW. Murine models of inflammatory, neuropathic and cancer pain each generates a unique set of neurochemical changes in the spinal cord and sensory neurons. Neuroscience 2000b;98:585–98. Hunt SP, Pini A, Evan G. Induction of c-fos-like protein in spinal cord neurons following sensory stimulation. Nature 1987;328:632–4. Iadarola MJ, Douglass J, Civelli O, Naranjo JR. Differential activation of spinal cord dynorphin and enkephalin neurons during hyperalgesia: evidence using cDNA hybridization. Brain Res 1988;455:205–12. Irie K, Hara-Irie F, Ozawa H, Yajima T. Calcitonin gene-related peptide (CGRP)-containing nerve fibers in bone tissue and their involvement in bone remodeling. Microsc Res Tech 2002;58:85–90. Jaggar SI, Scott HC, Rice AS. Inflammation of the rat urinary bladder is associated with a referred thermal hyperalgesia which is nerve growth factor dependent. Br J Anaesth 1999;83:442–8. Ji RR, Samad TA, Jin SX, Schmoll R, Woolf CJ. p38 MAPK activation by NGF in primary sensory neurons after inflammation increases TRPV1 levels and maintains heat hyperalgesia. Neuron 2002;36:57–68. Jongen JL, Luger NM, Mach DB, Peters CM, Rogers SD, Sabino MAC, Salak-Johnson JL, Mantyh PW. Neurotrophic factors and cancer pain: the expression of NGF, GDNF and BDNF by the murine osteolytic sarcoma cell line 2472 in vitro and in vivo and their potential involvement in bone cancer pain. Soc Neurosci Abstr 2002;52(2). Koltzenburg M, Bennett DL, Shelton DL, McMahon SB. Neutralization of endogenous NGF prevents the sensitization of nociceptors supplying inflamed skin. Eur J Neurosci 1999;11:1698–704. Lamb K, Kang YM, Gebhart GF, Bielefeldt K. Nerve growth factor and gastric hyperalgesia in the rat. Neurogastroenterol Motil 2003;15: 355–61. Luger NM, Honore P, Sabino MAC, Schwei MJ, Rogers SD, Mach DB, Clohisy DR, Mantyh PW. Osteoprotegerin diminishes advanced bone cancer pain. Cancer Res 2001;61:4038–47. Luger NM, Sabino MAC, Schwei MJ, Mach DB, Pomonis JD, Keyser CP, Rathbun M, Clohisy DR, Honore P, Yaksh TL, Mantyh PW. Efficacy of systemic morphine suggests a fundamental difference in the mechanisms that generate bone cancer vs. inflammatory pain. Pain 2002;99: 397–406. Mach DB, Rogers SD, Sabino MAC, Luger NM, Schwei MJ, Pomonis JD, Keyser CP, Clohisy DR, Adams DJ, O’Leary P, Mantyh PW. Origins of skeletal pain: sensory and sympathetic innervation of the mouse femur. Neuroscience 2002;113:155–66. Madsen JE, Hukkanen M, Aspenberg P, Polak J, Nordsletten L. Timedependent sensory nerve ingrowth into a bone conduction chamber. Acta Orthop Scand 2000;71:74–9.

Mamet J, Lazdunski M, Voilley N. How nerve growth factor drives physiological and inflammatory expressions of acid-sensing ion channel 3 in sensory neurons. J Biol Chem 2003;278:48907–13. Mantyh PW, Clohisy DR, Koltzenburg M, Hunt SP. Molecular mechanisms of cancer pain. Nat Rev Cancer 2002;2:201–9. McBride WH. Phenotype and functions of intratumoral macrophages. Biochimica et Biophysica Acta 1986;865:27–41. McCarthy BG, Hsieh ST, Stocks A, Hauer P, Macko C, Cornblath DR, Griffin JW, McArthur JC. Cutaneous innervation in sensory neuropathies: evaluation by skin biopsy. Neurology 1995;45:1848–55. McMahon SB, Bennett DL, Priestley JV, Shelton DL. The biological effects of endogenous nerve growth factor on adult sensory neurons revealed by a trkA-IgG fusion molecule. Nat Med 1995;1:774–80. Mercadante S. Malignant bone pain: pathophysiology and treatment. Pain 1997;69:1–18. Mercadante S, Arcuri E. Breakthrough pain in cancer patients: pathophysiology and treatment. Cancer Treat Rev 1998;24:425–32. Mitchell BS, Schumacher U, Stauber VV, Kaiserling E. Are breast tumours innervated?. Immunohistological investigations using antibodies against the neuronal marker protein gene product 9.5 (PGP 9.5) in benign and malignant breast lesions Eur J Cancer, 30A 1994;1100–3. Molander C, Grant G. Spinal cord projections from hindlimb muscle nerves in the rat studied by transganglionic transport of horseradish peroxidase, wheat germ agglutinin conjugated horseradish peroxidase, or horseradish peroxidase with dimethylsulfoxide. J Comp Neurol 1987;260: 246–55. Myers RR, Heckman HM, Rodriguez M. Reduced hyperalgesia in nerveinjured WLD mice: relationship to nerve fiber phagocytosis, axonal degeneration, and regeneration in normal mice. Exp Neurol 1996;141: 94–101. Navarro X, Verdu E, Wendelscafer-Crabb G, Kennedy WR. Innervation of cutaneous structures in the mouse hind paw: a confocal microscopy immunohistochemical study. J Neurosci Res 1995;41:111–20. Noguchi K, Kowalski K, Traub R, Solodkin A, Iadarola MJ, Ruda MA. Dynorphin expression and Fos-like immunoreactivity following inflammation induced hyperalgesia are colocalized in spinal cord neurons. Mol Brain Res 1991;10:227–33. Okuse K, Malik-Hall M, Baker MD, Poon WY, Kong H, Chao MV, Wood JN. Annexin II light chain regulates sensory neuron-specific sodium channel expression. Nature 2002;417:653–6. Orr FW, Sanchez-Sweatman OH, Kostenuik P, Singh G. Tumor-bone interactions in skeletal metastasis. Clin Orthopaedics Relat Res 1995;19–33. Petersen M, Segond von Banchet G, Heppelmann B, Koltzenburg M. Nerve growth factor regulates the expression of bradykinin binding sites on adult sensory neurons via the neurotrophin receptor p75. Neuroscience 1998;83:161–8. Petruska JC, Mendell LM. The many functions of nerve growth factor: multiple actions on nociceptors. Neurosci Lett 2004;361: 168–71. Portenoy RK. Managing cancer pain poorly responsive to systemic opioid therapy. Oncology 1999;13:25–9. Portenoy RK, Hagen NA. Breakthrough pain: definition, prevalence and characteristics. Pain 1990;41:273–81. Portenoy RK, Lesage P. Management of cancer pain. Lancet 1999;353: 1695–700. Puigdellivol-Sanchez A, Prats-Galino A, Ruano-Gil D, Molander C. Sciatic and femoral nerve sensory neurones occupy different regions of the L4 dorsal root ganglion in the adult rat. Neurosci Lett 1998;251: 169–72. Puigdellivol-Sanchez, A., Forcada-Calvet, P., Prats-Galino, A., Molander, C., Contribution of femoral and proximal sciatic nerve branches to the sensory innervation of hindlimb digits in the rat, the anatomical record. Anat Rec 260 (2000) 180–188 Ramer MS, Bradbury EJ, McMahon SB. Nerve growth factor induces P2X(3) expression in sensory neurons. J Neurochem 2001;77:864–75.

M.A. Sevcik et al. / Pain 115 (2005) 128–141 Ramer MS, Kawaja MD, Henderson JT, Roder JC, Bisby MA. Glial overexpression of NGF enhances neuropathic pain and adrenergic sprouting into DRG following chronic sciatic constriction in mice. Neurosci Lett 1998;251:53–6. Ro LS, Chen ST, Tang LM, Jacobs JM. Effect of NGF and anti-NGF on neuropathic pain in rats following chronic constriction injury of the sciatic nerve. Pain 1999;79:265–74. Rueff A, Dawson AJ, Mendell LM. Characteristics of nerve growth factor induced hyperalgesia in adult rats: dependence on enhanced bradykinin-1 receptor activity but not neurokinin-1 receptor activation. Pain 1996;66:359–72. Sabino MAC, Ghilardi JR, Jongen JL, Keyser CP, Luger NM, Mach DB, Peters CM, Rogers SD, Schwei MJ, de Felipe C, Mantyh PW. Simultaneous reduction in cancer pain, bone destruction, and tumor growth by selective inhibition of cyclooxygenase-2. Cancer Res 2002; 62:7343–9. Sabino MAC, Luger NM, Mach DB, Rogers SD, Schwei MJ, Mantyh PW. Different tumors in bone each give rise to a distinct pattern of skeletal destruction, bone cancer-related pain behaviors and neurochemical changes in the central nervous system. Int J Cancer 2003;104: 550–8. Schwab W, Funk RH. Innervation pattern of different cartilaginous tissues in the rat. Acta Anat (Basel) 1998;163:184–90. Schwei MJ, Honore P, Rogers SD, Johnson JL, Finke MP, Rarmaraine M, Clohisy DR, Mantyh PW. Neurochemical and cellular reorganization of the spinal cord in a murine model of bone cancer pain. J Neurosci 1999; 19:10886–97. Sevcik MA, Luger NM, Mach DB, Sabino MAC, Peters CM, Ghilardi JR, Schwei MJ, Rohrich H, De Felipe C, Kuskowski MA, Mantyh PW. Bone cancer pain: the effects of the bisphosphonate alendronate on pain, skeletal remodeling, tumor growth and tumor necrosis. Pain 2004;111: 169–80. Shelton DL, Reichardt LF. Expression of the beta-nerve growth factor gene correlates with the density of sympathetic innervation in effector organs. Proc Natl Acad Sci USA 1984;81:7951–5. Skoff AM, Resta C, Swamydas M, Adler JE. Nerve growth factor (NGF) and glial cell line-derived neurotrophic factor (GDNF) regulate substance P release in adult spinal sensory neurons. Neurochem Res 2003;28:847–54. Terada T, Matsunaga Y. S-100-positive nerve fibers in hepatocellular carcinoma and intrahepatic cholangiocarcinoma: an immunohistochemical study. Pathol Int 2001;51:89–93.

141

Tofaris GK, Patterson PH, Jessen KR, Mirsky R. Denervated Schwann cells attract macrophages by secretion of leukemia inhibitory factor (LIF) and monocyte chemoattractant protein-1 in a process regulated by interleukin-6 and LIF. J Neurosci 2002;22:6696–703. Tsujino H, Kondo E, Fukuoka T, Dai Y, Tokunaga A, Miki K, Yonenobu K, Ochi T, Noguchi K. Activating transcription factor 3 (ATF3) induction by axotomy in sensory and motoneurons: a novel neuronal marker of nerve injury. Mol Cell Neurosci 2000;15:170–82. Urch CE, Donovan-Rodriguez T, Dickenson AH. Alterations in dorsal horn neurones in a rat model of cancer-induced bone pain. Pain 2003;106: 347–56. Vanderah TW, Laughlin T, Lashbrook JM, Nichols ML, Wilcox GL, Ossipov MH, Malan TP, Porreca F. Single intrathecal injections of dynorphin A or des-Tyr-dynorphins produce long-lasting allodynia in rats; Blockade by MK-801 but not naloxone. Pain 1996;68:275–81. Vanderah TW, Ossipov MH, Lai J, Malan TP, Porreca F. Mechanisms of opioid-induced pain and antinociceptive tolerance: descending facilitation and spinal dynorphin [review]. Pain 2001;92:5–9. Vega JA, Garcia-Suarez O, Hannestad J, Perez-Perez M, Germana A. Neurotrophins and the immune system. J Anat 2003;203:1–19. Verge VM, Tetzlaff W, Bisby MA, Richardson PM. Influence of nerve growth factor on neurofilament gene expression in mature primary sensory neurons. J Neurosci 1990;10:2018–25. Walsh GS, Krol KM, Kawaja MD. Absence of the p75 neurotrophin receptor alters the pattern of sympathosensory sprouting in the trigeminal ganglia of mice overexpressing nerve growth factor. J Neurosci 1999;19:258–73. Woolf CJ, Mannion RJ. Neuropathic pain: aetiology, symptoms, mechanisms, and management. Lancet 1999;353:1959–64. Woolf CJ, Safieh-Garabedian B, Ma QP, Crilly P, Winter J. Nerve growth factor contributes to the generation of inflammatory sensory hypersensitivity. Neuroscience 1994;62:327–31. Zahn PK, Subieta A, Park SS, Brennan TJ. Effect of blockade of nerve growth factor and tumor necrosis factor on pain behaviors after plantar incision. J Pain 2004;5:157–63. Zhang F, Lu W, Dong Z. Tumor-infiltrating macrophages are involved in suppressing growth and metastasis of human prostate cancer cells by INF-beta gene therapy in nude mice. Clin Cancer Res 2002;8: 2942–51. Zhou XF, Rush RA. Endogenous brain-derived neurotrophic factor is anterogradely transported in primary sensory neurons. Neuroscience 1996;74:945–53.