BRIEF COMMUNICATION

Hyperalgesia in Experimental Neuropathy Is Dependenton the TNF Receptor 1

Claudia Sommer, Christine Schmidt,1 and Annette GeorgeNeurologische Klinik, Universitat Wurzburg, Josef-Schneider-Strasse 11, 97080 Wurzburg, Germany

Received September 15, 1997; accepted January 28, 1998

Recent evidence points to a role of cytokines liketumor necrosis factor-a (TNF) in the generation ofhyperalgesia not only in inflammatory, but also inneuropathic pain. We used the model of chronic con-strictive injury (CCI) of one sciatic nerve in the mouseto investigate which of the two known TNF receptorsis involved in the process that leads to hyperalgesiaafter nerve injury. Neutralizing antibodies to TNF, tothe TNF receptor 1 (TNFR1), and to the TNF receptor 2(TNFR2) were administered by epineurial injectiononce daily to mice with CCI. Testing of the animals’hind paws with thermal and innocuous mechanicalstimuli revealed a reduction in thermal hyperalgesiaand mechanical allodynia in mice treated with neutral-izing antibodies to TNF and to TNFR1. Neutralizingantibodies to TNFR2 had no effect. We conclude thatTNFR1, but not TNFR2, is mediating thermal hyperal-gesia and mechanical allodynia after nerve injury. r 1998

Academic Press

Key Words: neuropathic pain; tumor necrosis factor;TNF-receptor; neutralizing antibodies.

Hyperalgesia and allodynia are symptoms frequentlyoccurring in painful neuropathies (9). Animal models ofpartial nerve injury allow quantification of withdrawalthresholds to thermal and mechanical stimuli, thusproviding an analogy to the phenomena of hyperalgesiaand allodynia in humans. Recent studies provide evi-dence for a role of proinflammatory cytokines not onlyin inflammatory pain (6) but also in neuropathic pain(2, 3). Particularly tumor necrosis factor-a (TNF) hasbeen shown to induce axonal degeneration, demyelin-ation, and thermal hyperalgesia when injected into theendoneurium and to evoke ectopic activity in isolatednerve fibers when applied topically (8, 18, 22). Sub-stances inhibiting TNF-production or -shedding reducethermal hyperalgesia and mechanical allodynia in ani-

mals with a chronic constriction injury of the sciaticnerve (CCI) (15, 16). TNF exerts its action mainlythrough the two TNF-receptors, the TNF receptor 1(TNFR1) and the TNF receptor 2 (TNFR2) (19). Differ-ential actions have been ascribed to each receptor, withthe majority of the TNF effects being transmittedthrough TNFR1 (for review see 21). The inducibleTNFR2 is supposed to function by ligand passing toTNFR1 and thereby to enhance and regulate TNFeffects. Thus, TNF-induced cytotoxicity and TNF-enhanced expression of adhesion molecules are par-tially inhibited by blockade of either receptor. In addi-tion, membrane bound 26-kDa TNF as opposed tosoluble 17-kDa TNF seems to be the main ligand forTNFR2, which has been suggested to be of importancein local inflammatory responses (7). Given that nerveinjury first induces a local reaction in the endoneurium,we speculated that TNF-expression on Schwann cellsand fibroblasts (17, 23) might be involved in the induc-tion of hyperalgesia by signaling through TNFR2. Incontrast, this study presents evidence for a major roleof TNFR1 in hyperalgesia associated with nerve injury.

Three experimental groups of 10 female C57BL/6-mice (16–21 g, Harlan–Winkelmann, Borchen, Ger-many) were used. The animals were housed on a 14:10light:dark cycle with standard rodent chow and waterad libitum. Of each group, 8 animals were anesthetizedby intraperitoneal injection of a barbiturate solutionand the sciatic nerve was exposed unilaterally at themid-thigh level. Three ligatures (10-0 prolene) wereplaced around the nerve with 1-mm spacing. Theligatures were tied until they elicited a brief twitch inthe hindlimb. In four mice, an antibody solution (seebelow) was instilled into the wound; four mice receivednormal hamster serum. The wound was closed inlayers. Behavioral testing was performed at the samehour every time (7 AM to 12 noon). The thermalnociceptive threshold was measured in each hindpawon two consecutive days before surgery (baseline) and1 This study forms part of the doctoral thesis of Christine Schmidt.

EXPERIMENTAL NEUROLOGY 151, 138–142 (1998)ARTICLE NO. EN986797

1380014-4886/98 $25.00Copyright r 1998 by Academic PressAll rights of reproduction in any form reserved.

daily from day 3 to day 7 after surgery as describedbefore (14). Paw withdrawal latencies on the controlside were subtracted from those on the experimentalside, the mean of five such assessments resulted in a‘‘difference score’’ giving a measure of thermal hyperal-gesia. Mechanical allodynia was assessed with vonFrey hairs using the up-and-down method of Dixon (1,4) from day 3 to 7 after surgery. The 50% withdrawalthreshold (force of the particular von Frey hair to whichan animal reacts in 50% of the presentations) wasrecorded. In each of the experimental groups, 4 of the 8operated animals were treated with neutralizing ham-ster anti-mouse monoclonal antibodies (AB): Group 1was treated with anti-TNF-AB (125 µg daily in avolume of 62,5 µl, 2 µg/µl), group 2 with anti-TNFR1-AB (50 µg daily in a volume of 50 µl, 1 µg/µl),and group 3 with TNFR2-AB (50 µg daily in a volume of50 µl, 1 µg/µl). Antibodies were purchased from Gen-zyme (Russelsheim, Germany) and have been exten-sively characterized (11, 12). In particular, the antibod-ies are neutralizing and not antigenic, thus there wasno need for immunosuppression of the animals. Theother four operated animals in each group received therespective volume of normal hamster serum (CharlesRiver, Schweinfurt, Germany), i.e., 62.5 µl in group 1and 50 µl in groups 2 and 3, as a sham treatment.Antibodies and serum were administered by local epi-neurial administration intraoperatively (see above) andthen daily until the end of the experiment. Postopera-tive injections were performed at 6 AM daily. The micewere briefly anesthetized in ether. A 26 G needle wasinserted just cranially to the middle of the operationscar. The needle was inserted in a cranio-caudal direc-tion for approximately 6 mm. This is the distance thatin preliminary tests was shown to reach the area of theligatures in the operated nerve. The antibody solutionwas then slowly injected. Two animals in each groupserved as unoperated controls for the behavioral tests.To control for the effects of the injection itself, in apreliminary experiment, three mice received CCI withno additional treatment, and three mice received CCIand daily injections of 50 µl of normal saline using themethod described above. Behavioral testing for 1 weekrevealed no differences in pain-related behavior be-tween these groups.

Sciatic nerve tissue of all animals was harvested onday 8 post-OP. Nerve sections from under the ligatures(3 mm) and distal to the ligatures (4 mm) were snapfrozen in methylbutane precooled in liquid nitrogen.Penetrance into the nerve was controlled for by react-ing frozen sections with anti-hamster secondary anti-bodies using an ABC-system (Vector, Burlingame,U.S.A.) and DAB as the chromogen. Comparison ofbehavioral data for thermal hyperalgesia was per-formed by repeated measures—ANOVA for comparison

of groups and subsequent Scheffe test for comparison ofindividual means. Friedman’s test was used for statisti-cal analysis of 50% thresholds to mechanical stimuliwith Newman–Keul’s test for post hoc analysis. Signifi-cance was assumed below P values of 0.05. All valuesare given as mean 6 standard deviation.

Thermal hyperalgesia was present from day 3 in thethree mice with CCI only and in the three micereceiving daily saline injections under ether anesthesiato the same degree (difference scores between22.29 6 0.28 and 23.06 6 0.39 for CCI only and be-tween 23.02 6 0.47 and 22.99 6 0.28 for CCI withsaline injections). Thermal hyperalgesia after CCI waspresent in all mice receiving sham-treatment withhamster serum from day 3 until the end of the experi-ment on day 7 (difference score between 22.47 6 0.09and 23.13 6 0.27). Thermal hyperalgesia was signifi-cantly reduced in mice treated with anti-TNF-AB dur-ing the observation period (difference score between21.26 6 0.20 and 22.02 6 0.08; Fig. 1a). In mice treatedwith anti-TNFR1-AB, thermal hyperalgesia was alsoreduced. Differences to sham-treated mice were signifi-cant on days 5 to 7 with difference scores between21.43 6 0.20 and 22 6 0.24 (Fig. 1b). Treatment withanti-TNFR2 did not affect thermal withdrawal thresh-olds (Fig. 1c). Reduced withdrawal thresholds to me-chanical stimuli were present in mice receiving sham-treatment from day 3 until the end of the experiment(50% thresholds between 0.005 6 0.006 and 0.019 60.009 mN). Mice treated with anti-TNF-AB had a lesserreduction of mechanical withdrawal thresholds with asignificant difference to sham-treated animals on days4, 6, and 7 (0.036 6 0.04 mN to 0.075 6 0.03 mN, Fig.1d). Mice treated with anti-TNFR1-AB also had adiminished reduction of mechanical withdrawal thresh-olds, which became more prominent toward the end ofthe experiment with significant difference to sham-treated mice on days 6 and 7 and 50% thresholdsbetween 0.02 6 0.01 and 0.09 6 0.001 mN (Fig. 1e).Treatment with anti-TNFR2-AB had no effect on me-chanical thresholds (Fig. 1f ).

Staining with anti-hamster secondary antibodies re-vealed the epineurial and also endoneurial presence ofthe monoclonal antibodies in the area under the liga-tures of lesioned nerves of the respective animals(Fig. 2).

The present data indicate that hyperalgesia afternerve injury can be attenuated by neutralizing antibod-ies against TNF and against TNFR1.Antibodies againstTNFR2 did not affect the animals’ pain-related behav-ior. Thus, a prominent role of TNFR1 in the signalingthat eventually entails hyperalgesia, can be assumed.Control staining confirmed penetrance of the epineuri-ally administrated antibodies into the endoneurium, sothat this route of admission can be used to enhance the

139BRIEF COMMUNICATION

local concentration, reduce possible systemic side ef-fects, and avoid nerve damage as it occurs with endoneu-rial injection (8).

Both thermal hyperalgesia and mechanical allodyniawere attenuated to similar degrees by the antibodiesagainst TNF and against TNFR1. Withdrawal thresh-

olds were not completely reversed to normal. This couldbe due to either an insufficient dose of the antibody ordue to the fact that other mediators than TNF areinvolved in the generation of hyperalgesia. Other inves-tigators used doses of 250 and 200 µg of the sameanti-TNF-AB, respectively, per mouse for intraperito-

FIG. 1. Thermal hyperalgesia after CCI is reduced compared to sham-treated mice in mice treated with (a) anti-TNF-antibodies (AB) and(b) anti-TNFR1-AB, but not (c) anti-TNFR2-AB. Equally, reduction of mechanical withdrawal thresholds is diminished by (d) anti-TNF-ABand (e) anti-TNFR1-AB, but not (f ) anti-TNFR2-AB. (*P , 0.05 vs sham-treated mice).

140 BRIEF COMMUNICATION

neal application (5, 20) to achieve neutralization, sothat the dose we administered, 125 µg given locally,should be sufficient. Similarly, the anti-receptor-ABhave been previously used in mice systemically in dosesof 40 µg (12). However, although we could prove thatthe antibodies do penetrate the blood–nerve barrier, weare not sure of the final concentration at their endoneu-rial binding sites.

Contrary to our hypothesis, TNFR2 does not seem toplay a major role in the signaling involved in thedevelopment of hyperalgesia after nerve injury. Wecould not even find an indication of ligand passing, i.e.,TNFR2-mediated augmentation of the TNFR1 re-sponse, since the anti-TNFR2-had no effect at all on thewithdrawal thresholds. Most biological activities ofTNF known so far can be exerted via TNFR1, and lowerreceptor densities are needed than for TNFR2 to achievesimilar effects (21). Nevertheless, antagonistic antibod-ies to either of the receptors may inhibit the biologicaleffects at least in part (10, 13). Given that TNFR2 is theinducible receptor, and that it preferentially reactswith membrane bound TNF (7), an enhanced actionthrough TNFR2 might have been expected. However,from the present study we conclude that TNF, via theTNFR1, is involved in the generation and maintenanceof thermal hyperalgesia and mechanical hypersensitiv-ity in the CCI-model of neuropathic pain.

ACKNOWLEDGMENTS

Supported by Deutsche Forschungsgemeinschaft So 328/2-1. Criti-cal reading of the manuscript by K. V. Toyka is gratefully acknowl-edged.

REFERENCES

1. Chaplan, S. R., F. W. Bach, J. W. Pogrel, J. M. Chung, and T. L.Yaksh. 1994. Quantitative assessment of tactile allodynia in therat paw. J. Neurosci. Methods 53: 55–63.

2. Deleo, J. A., R. W. Colburn, M. Nichols, and A. Malhorta. 1996.Interleukin-6-mediated hyperalgesia/allodynia and increasedspinal IL-6 expression in a rat mononeuropathy model. J.Interferon Cytokine Res. 16: 695–700.

3. Deleo, J. A., R. W. Colburn, and A. J. Rickman. 1997. Cytokineand growth factor immunohistochemical spinal profiles in twoanimal models of mononeuropathy. Brain Res. 759: 50–57.

4. Dixon, W. 1965. The up-and-down method for small samples. J.Am. Statist. Assoc. 60: 967–978.

5. Ferran, C., F. Dautry, S. Merite, K. Sheehan, R. Schreiber, G.Grau, J. F. Bach, and L. Chatenoud. 1994. Anti-tumor necrosisfactor modulates anti-CD3-triggered T cell cytokine gene expres-sion in vivo. J. Clin. Invest. 93: 2189–2196.

6. Ferreira, S. 1993. Inflammatory pain: The role of cytokines andits control by drugs which release nitric oxide. Ann. Ist. SuperSanita 29: 367–373.

7. Grell, M., E. Douni, H. Wajant, M. Lohden, M. Clauss, B.Maxeiner, S. Georgopoulos, W. Lesslauer, G. Kollias, K. Pfizen-maier, and P. Scheurich. 1995. The transmembrane form oftumor necrosis factor is the prime activating ligand of the 80kDa tumor necrosis factor receptor. Cell 83: 793–802.

FIG. 2. Frozen sections from the sciatic nerve of a mouse with CCI and treatment with anti-TNF-antibody, stained with anti-hamstersecondary antibodies using an ABC-system and DAB as the chromogen. (a) Positive immunostaining in the perineurium (arrow) and in thesubperineurial endoneurium (small arrows), (b) positive immunostaining (small arrows) indicates penetrance of the anti-TNF-antibody intothe endoneurial space. Bar, 10 µm.

141BRIEF COMMUNICATION

8. Redford, E. J., S. M. Hall, and K. J. Smith. 1995. Vascularchanges and demyelination induced by the intraneural injectionof tumour necrosis factor. Brain 118: 869–878.

9. Scadding, J. W. 1992. Neuropathic Pain. In Diseases of theNervous System. Clinical Neurobiology (A. K. Asbury, G. M.McKhann, and W. I. McDonald, Eds.), Vol. 2, pp. 858–872. Saun-ders, Philadelphia.

10. Shalaby, M. R., A. Sundan, H. Loetscher, M. Brockhaus, W.Lesslauer, and T. Espevik. 1990. Binding and regulation ofcellular functions by monoclonal antibodies against humantumor necrosis factor receptors. J. Exp. Med. 172: 1517–1520.

11. Sheehan, K. C., N. H. Ruddle, and R. D. Schreiber. 1989.Generation and characterization of hamster monoclonal antibod-ies that neutralize murine tumor necrosis factors. J. Immunol.142: 3884–3893.

12. Sheehan, K. C., J. K. Pinckard, C. D. Arthur, L. P. Dehner, D. V.Goeddel, and R. D. Schreiber. 1995. Monoclonal antibodiesspecific for murine p55 and p75 tumor necrosis factor receptors:Identification of a novel in vivo role for p75. J. Exp. Med. 181:607–617.

13. Slowik, M. R., L. G. De Luca, W. Fiers, and J. S. Pober. 1993.Tumor necrosis factor activates human endothelial cells throughthe p55 tumor necrosis factor receptor but the p75 receptorcontributes to activation at low tumor necrosis factor concentra-tion. Am. J. Pathol. 143: 1724–1730.

14. Sommer, C., J. A. Galbraith, H. M. Heckman, and R. R. Myers.1993. Pathology of experimental compression neuropathy pro-ducing hyperesthesia. J. Neuropathol. Exp. Neurol. 52: 223–233.

15. Sommer, C., C. Schmidt, A. George, and K. V. Toyka. 1997. Ametalloprotease-inhibitor reduces pain associated behavior inmice with experimental neuropathy. Neurosci. Lett. 237: 45–48.

16. Sommer, C., and M. Marziniak. 1996. Experimental painfulmononeuropathy: Inhibitors of TNF-a-production induce a de-crease in hyperalgesia and an increase of spinal met-enkepha-lin. Soc. Neurosci. Abs. 22: 511.

17. Sommer, C., and M. Schafers. 1997. Painful mononeuropathy inC57BL/Wld mice with delayed Wallerian degeneration: differen-tial effects of cytokine production and nerve regeneration onthermal and mechanical hypersensitivity. Brain Res. in press.

18. Sorkin, L., W. Xiao, R. Wagner, and R. Myers. 1997. Tumournecrosis factor-a induces ectopic activity in nociceptive primaryafferent fibres. Neuroscience 81: 255–262.

19. Tartaglia, L. A., and D. V. Goeddel. 1992. Two TNF receptors.Immunol. Today 13: 151–153.

20. Tripp, C. S., M. K. Gately, J. Hakimi, P. Ling. and E. R. Unanue.1994. Neutralization of IL-12 decreases resistance to Listeria inSCID and C.B-17 mice. Reversal by IFN-gamma. J. Immunol.152: 1883–1887.

21. Vandenabeele, P., W. Declerq, R. Beyart, and W. Fiers. 1995. Twotumour necrosis factor receptors: Structure and function. TrendsCell. Biol. 5: 392–399.

22. Wagner, R., and R. R. Myers. 1996. Endoneurial injection ofTNF-alpha produces neuropathic pain behaviors. Neuroreport7: 2897–2901.

23. Wagner, R., and R. R. Myers. 1996. Schwann cells producetumor necrosis factor alpha: expression in injured and non-injured nerves. Neuroscience 73: 625–629.

142 BRIEF COMMUNICATION