Secondary Logo

Journal Logo

The Role of Heme Oxygenase in Neuropathic and Incisional Pain

Li, Xiangqi MD; Clark, J. David MD, PhD

doi: 10.1097/00000539-200003000-00031

Heme oxygenase (HO) catalyzes the formation of free iron, biliverdin, and the second messenger molecule carbon monoxide from heme. We document a role for HO in both neuropathic and incisional pain models. For our neuropathic model, the L5 and L6 nerve roots of rats were ligated unilaterally resulting in mechanical allodynia and thermal hyperalgesia in the ipsilateral hind paws. Both changes were dose-dependently reversed by systemic administration of the HO inhibitor tin protoporphyrin (Sn-P). Likewise, a 1-cm incision made in one hind paw resulted in mechanical allodynia and thermal hyperalgesia, again reversible by using Sn-P. The 50% effective doses for Sn-P ranged from 4.0 to 6.8 μmol/kg depending on the model and nociceptive stimulus. We also observed that the blood-brain barrier impermeable HO inhibitor zinc protoporphyrin had little analgesic activity in these models when injected systemically. Using an enzymatic assay, we observed increased HO activity in lumbar spinal cord tissue from either nerve root ligated or incised animals as compared with tissue from sham-operated animals. Taken together, we interpret our results to indicate that an increase in spinal cord HO activity at least partially underlies the allodynia and hyperalgesia seen in rat models of neuropathic and incisional pain.

Implications: Central nervous system heme oxygenase likely plays a role in nociceptive signaling in both neuropathic and incisional models of pain. Therefore, inhibitors of heme oxygenase activity may be viable analgesics in these settings.

Department of Anesthesiology, Stanford University and Veterans’ Affairs Palo Alto Health Care Services, Palo Alto, California

November 30, 1999.

Address correspondence and reprint requests to J. David Clark, MD, PhD, Stanford University and VAPAHCS, Department of Anesthesiology, MC: 112A, 3801 Miranda Ave., Palo Alto, CA 94304.

The gaseous second messenger molecule nitric oxide (NO), produced by NO synthase, has well established roles in pain of several etiologies, including neuropathic and inflammatory (1–3). Carbon monoxide (CO), a gaseous second messenger molecule produced by heme oxygenase (HO), has not been well investigated with respect to its possible involvement in pain or nociceptive signaling despite its many similarities to NO. These similarities include its pattern of distribution within the central nervous system (CNS) and peripheral nervous system (4) and the ability of these enzyme systems to produce small, diffusible, gaseous second messenger molecules (NO and CO), both of which are capable of activating guanylate cyclase (5).

There are two reports linking the enzyme HO to pain. The first (6) examined the ability of intrathecal injections of zinc protoporphyrin (Zn-P) to reduce mechanical allodynia in the hind paws of rats after inflammation was induced by the intraplantar injection of zymosan or the intrathecal injection of glutamate receptor agonists. Interestingly, no reversal of thermal hyperalgesia was noted nor were there any effects of Zn-P on mechanical or thermal thresholds in unperturbed tissue. Later, Yamamoto et al. (7) showed that intrathecal administration of Zn-P reduced flinching behavior in the formalin model of inflammatory pain. In neither of the studies was blood-brain-barrier-permeable HO inhibitors injected peripherally, leaving open the question of whether protoporphyrin inhibitors of HO have significant analgesic efficacy when injected peripherally. Also, there is no information addressing the question of possible analgesic efficacy of HO inhibitors in other models of pain. Two types of pain with great clinical relevance are neuropathic pain and incisional pain. Neuropathic pain is both common and difficult to treat. Although a number of different classes of drugs delivered by a number of different modalities are effective in reducing incisional pain, e.g., systemic opioids, epidural local anesthetics, systemic nonsteroidal antiinflammatory drugs, etc., no single medication or modality is ideal. Thus, there is a need for new treatments which might reduce pain by novel mechanisms. The data presented below address the possibility that systemically injected HO inhibitors have analgesic efficacy in models of neuropathic and incisional pain.

Back to Top | Article Outline


All animal protocols used in these experiments were approved by our local animal care committee. Male Sprague-Dawley rats, approximately 200 g, were obtained from B&K (Freemont, CA). These animals were kept under standard husbandry conditions with a 12:12 h light:dark cycle and were fed a soy-based diet.

Rats used in the neuropathic model were anesthetized with pentobarbital. Segmental nerve ligations of L5 and L6 were performed as described elsewhere (8). Briefly, after shaving and skin incision, the paraspinous musculature of the low back was divided, and the transverse processes of the lower lumbar vertebra was identified by using the sacral ala as a landmark. The appropriate segmental nerves were identified, and a tight 4–0 silk suture was placed around each one. The muscle was closed with additional silk suture, and the skin was stapled before the application of antibiotic ointment. Sham-operated animals underwent anesthesia and skin incision. Preliminary experiments showed that a finite percentage of animals undergoing only nerve root dissection would display hyperalgesia at some point postoperatively. Starting several days after surgery, animals were tested for thermal hyperalgesia and mechanical allodynia in the L3-4 dermatome on the plantar aspect of the hind paw by using the methods of Hargreaves et al. (9) and Chaplan et al. (10), respectively. We generally found that animals that would ever become hyperalgesic would do so within 14–21 days. More than 90% of our animals demonstrated thermal hyperalgesia (withdrawal latency < 75% of baseline) with mechanical allodynia (<30% baseline) and could be used in experiments. Both types of changes persisted to approximately 80 days from the time of surgery. Baseline thermal and mechanical thresholds were determined on each experimental day for all animals used in the studies. Animals were given no more than one dose of test drug per week, thus allowing for full dissipation of effects. Each animal was used in testing four or fewer times.

Rats used in the incisional model were treated as described by Brennan et al. (11). In this model, a 1-cm incision was made on the plantar surface of one of the rat’s hind paws with the animal under isoflurane anesthesia by using a No. 15 scalpel blade. The incision was started immediately distal to the heel and extended to a point just proximal to the first set of tori (foot pads). Care was taken to identify and divide the plantaris muscle. After hemostasis was obtained, two 4–0 silk sutures were placed in mattress fashion along the wound. Antibiotic ointment was then applied. Animals in the sham group underwent anesthesia without incision. These animals were always used 16–19 h after incision. New animals were used for every dose and type of drug tested in these experiments. Again, thermal hyperalgesia and mechanical allodynia were measured by using methods as described above. For mechanical experiments, the von Frey fiber was applied to a point 2–3 mm distal to the distal end of the incision. When using the Hargreaves-type assay, the focused beam of light was directed to this same spot. Animals that appeared to have uneven wound closures or that were applying their hind paws to the glass surface of the Hargreaves instrument unevenly, were not used.

The metalloporphyrins tin protoporphyrin (sn-P) and Zn-P were obtained from Porphyrin Products (Ogden, UT). These were dissolved using NaOH 0.05M/0.9% NaCl, and pH wascorrected to 7.5–8.0 with HCl. These substances were injected subcutaneously in a 100–250 μL volume 30–45 min before behavioral assays as preliminary experiments indicated that maximal effects were present by that time. Rats were weighed on the day of experimentation to calculate the appropriate dose of test substance. Control animals received saline injections of the same volume.

HO enzymatic activity was measured by using the method of Tenhunen et al. (12). In these experiments, segments of spinal cord ipsilateral to either nerve root ligation or hind paw incision from the L2-3 to L6-S1 region, approximately 60 mg of tissue, were homogenized with three 5-s bursts of a sonicator (Fisher Scientific, St. Louis, MO), in 1.2 mL 0.1M KPO4 buffer with pH = 7.40. Samples were then centrifuged for 10 min at 14,000 g, 4°C, and the supernatants were collected. This supernatant was divided between two reaction tubes, and hemin was added for a final concentration of 17 μM. To these tubes either buffer (blanks) or nicotinamide adenine dinucleotide phosphate (180 μM final concentration) was added. Tubes were incubated for 30 min at 37°C in a rocking waterbath, a time course over which bilirubin accumulated linearly. Bilirubin formation was followed by measuring absorbance at 468 nm by using a spectrophotometer (Perkin-Elmer, Wellesley, MA). The molar extinction coefficient for bilirubin has been determined to be 27.7 in this system (12). Under these assay conditions biliverdin, a primary reaction product, is converted to bilirubin in rapid and stoichiometric fashion (13). Homogenate protein determinations were made by using commercially available “Bio-Rad DC” reagents (Bio-Rad, Hercules, CA).

Estimation of 50% effective doses (ED50) and confidence intervals were accomplished with nonlinear regression analysis by using Prism 2.0 software (Graphpad, San Diego, CA). Specifically, sigmoidal dose-response curves of variable slope were constructed allowing estimation of ED50 values and 95% confidence intervals (95% CI). Comparisons of HO activity were made by using t-tests for paired (ipsilateral versus contralateral rat spinal cord tissue) or independent data (sham operated versus incisional rat spinal cord tissue) as appropriate. Our cutoff for significance was P < 0.05.

Back to Top | Article Outline


Animals that underwent L5-6 nerve root ligation became maximally hyperalgesic in the hind paw ipsilateral to the nerve root ligations within 21 days of surgery, although significant changes were often observed within the first week of nerve root ligation. For mechanical stimuli as assessed using von Frey filaments (10), the threshold for paw withdrawal was decreased from 20.5 to 2.1 g (Figure 1A). The mechanical thresholds for sham-operated animals were, however, unchanged. The thresholds for hind paws contralateral to the nerve root ligations were also unchanged (data not shown). Likewise, hyperalgesia to thermal stimuli was observed in the nerve root ligation model with the latency for withdrawal from the focused beam of light decreasing from 10.9 to 7.8 s (Figure 1B). Again, there was no hyperalgesia measurable in the hind paws contralateral to nerve root ligation as compared with sham-operated animals.

Figure 1

Figure 1

With mechanical allodynia and thermal hyperalgesia in our neuropathic pain model established, we turned to the question of whether inhibitors of HO would reduce the hyperalgesia and allodynia. As shown in Figure 1A, the blood-brain-barrier-permeable protoporphyrin sn-P reversed mechanical allodynia in a dose-dependent manner. The ED50 for this effect was 4.0 μmol/kg (95% CI 3.4–4.5 μmol/kg). At no dose did Sn-P reduce sensitivity to mechanical stimuli to a point less than baseline sensitivity. Data in Figure 1A also demonstrate that sn-P has no effect on baseline (sham operated animals) mechanical thresholds across the dose range used here. In parallel experiments, we found that Sn-P reversed thermal hyperalgesia in the same animals to a level not statistically different from sham-operated animals with an ED50 of 5.7 μmol/kg (95% CI 4.5–7.1 μmol/kg), not statistically different from the ED50 value for mechanical stimuli. Again, no dose in the range we used affected the thermal paw flick threshold of sham-operated animals consistent with previously reported data for the intrathecal administration of very small amounts of other HO inhibitors (6). These observations were unlikely to be caused by nonspecific sedating effects of Sn-P, as the latencies of the sham-operated animals did not change within the dose range used.

We present in Figure 2 data demonstrating that unilateral hind paw incision renders rats hyperalgesic to thermal and mechanical stimuli as previously reported (11). The average reduction in mechanical threshold was from 18.0 to 3.5 g (Figure 2A). This apparent allodynia was reversed toward baseline in a dose-dependent manner by Sn-P with an ED50 of 6.8 μmol/kg (95% CI 6.4–7.2 μmol/kg). Similar to the observation made with neuropathic data, the thermal hyperalgesia was also reversed with an ED50 of 4.9 μmol/kg (95% CI 3.7–6.5 μmol/kg) (Figure 2B). Again, Sn-P had no measurable effect on the mechanical or thermal thresholds in sham-operated animals.

Figure 2

Figure 2

Next, we attempted to address the question of whether HO inhibitors were acting centrally or peripherally to reduce responses to noxious stimuli using the potent but blood-brain-barrier-impermeable compound Zn-P (14–16). Zn-P is approximately equipotent to Sn-P at inhibiting HO in broken cell systems (14). Figure 3 displays data showing that 50 μmol/kg Zn-P failed to significantly reduce thermal hyperalgesia in either model of pain. In other experiments, we observed that Zn-P failed to reverse mechanical allodynia in these models (data not shown).

Figure 3

Figure 3

We next attempted to evaluate the hypothesis that an increased level of HO enzymatic activity is present in spinal cord tissue from rats subjected to either nerve root ligation or hind paw incision. Although no assay for the determination of HO activity in histological sections of tissue exists, we were able to measure HO activity in tissue from the lumbar segment of the rat’s spinal cord. Figure 4 presents data supporting our hypothesis. When HO enzymatic activity was measured by using the well established method of Tenhunen et al. (12), we observed that the overall HO enzymatic activity was increased approximately 3-fold in tissue from the lumbar region of the spinal cord ipsilateral to either nerve root ligation or hind paw incision as compared with tissue from sham-operated animals. HO enzymatic activity in lumbar segments contralateral to incision or nerve root ligation tended to be higher than in sham-operated animals, although the differences were not statistically significant.

Figure 4

Figure 4

Back to Top | Article Outline


Because of a small but promising study documenting that intrathecal administration of HO inhibitors could reduce mechanical allodynia in a model of inflammatory pain (6) and a separate study showing a reduction in formalin-induced licking behavior in rats (7), we set out to determine if systemic administration of HO inhibitors would prove analgesic in other pain models. The two models chosen were a nerve root constriction model of neuropathic pain (8), and a model of incisional pain (11). Whereas the latter model probably has inflammatory, neuropathic, and perhaps other components and is therefore not a model of any specific nociceptive stimulus, it does have the virtue of reflecting a common, clinically relevant situation. We were able to demonstrate in both models that the potent HO inhibitor Sn-P dose-dependently reduced mechanical allodynia and thermal hyperalgesia. Importantly, the blood-brain-barrier-impermeable protoporphyrin Zn-P had little effect on allodynia and hyperalgesia in either model. This suggests that the HO activity involved in the nociceptive signaling processes is located in the CNS rather than the peripheral nervous system. Lastly, overall HO activity was significantly increased in the spinal cords of animals having nerve root ligation or hind paw incision. This observation implicates HO in nociceptive signaling independently of the data using HO inhibitors. Taken together, our results seem to indicate that HO inhibitors can reduce allodynia and hyperalgesia from various etiologies by inhibiting CNS HO activity. Furthermore, it seems plausible that increased spinal cord HO activity levels could be partially responsible for the observed allodynia and hyperalgesia in these two models of pain.

Not only do these results suggest that systemically administered HO inhibitors could be analgesics for pain of various etiologies, they also imply that spinal cord HO activity increases in various pain states. Activation of CNS HO has not been demonstrated in situations in which pain was involved. It has been observed, however, that CNS HO activity increases after oxidative stress (17). This process probably involves the rapid increased expression of HO Type 1 (HO-1). More recently it was demonstrated in CNS cultures and in a cell line that oxidative stress can increase HO activity through the activation of HO Type 2 (HO-2) (18). This process apparently involves protein kinase C (PKC). Because the HO inhibitors used here are nonselective and the HO enzymatic assay measures total HO activity, we cannot be certain which isotype is involved in our observed responses. It is clear, however, that HO-2 is the predominant form of HO in the CNS, including the spinal cord (19). Also, PKC activation and/or translocation has been documented in spinal cord tissue in neuropathic and inflammatory models of pain (20,21). Because HO-2 seems to be activated by PKC, it may be that in our models of pain HO-2 is the HO isotype primarily responsible for the observed increase in HO activity and possibly partially responsible for the observed allodynia and hyperalgesia. Furthermore, chronic exposure to morphine can cause hyperalgesia in rats (22), and we have recently demonstrated that HO-2 expression and activity are greatly increased in spinal cord tissue in morphine tolerant rats (23). Although not specifically addressed, an increase in HO-1 or HO-2 expression as opposed to enzymatic activity could be responsible for the observed increase in spinal cord HO activity in either model. The availability of an HO-2 knockout mouse model may assist in further investigations directed at understanding the role of HO-2 in allodynia and hyperalgesia of various etiologies (24).

On the clinical level, our results suggest that HO inhibitors might have use in treating neuropathic or incisional pain. New drugs acting through novel mechanisms might be welcomed for the treatment of neuropathic pain in particular, where treatment with currently available drugs is often unsatisfactory. Because an HO inhibitor would presumably act through a novel mechanism, one might hope for effects at least additive if not synergistic with those of opioids, nonsteroidal antiinflammatory drugs, or other anlgesics. The same agent used in our studies, Sn-P, has been used in clinical studies for the treatment of hyperbilirubinemia and was generally well tolerated (25). However, total doses of 1–2 μmol/kg are most common in clinical studies in humans, well below the ED50 values for the analgesic effects seen in these rat studies. Therefore, opportunities may still exist for finding compounds with sufficient selectivity, potency and bioavailability to be considered potential pharmaceuticals.

Back to Top | Article Outline


1. Hao JX, Xu XJ. Treatment of a chronic allodynia-like response in spinally injured rats: effects of systemically administered nitric oxide synthase inhibitors. Pain 1996; 66:313–9.
2. Roche AK, Cook M, Wilcox GL, et al. A nitric oxide synthesis inhibitor (L-NAME) reduces licking behavior and Fos-labeling in the spinal cord of rats during formalin-induced inflammation. Pain 1996; 66:331–41.
3. Yoon YW, Sung B, Chung JM. Nitric oxide mediates behavioral signs of neuropathic pain in an experimental rat model. Neuroreport 1998; 9:367–72.
4. Snyder SH, Jaffrey SR, Zakhary R. Nitric oxide and carbon monoxide: parallel roles as neural messengers. Brain Res Brain Res Rev 1998; 26:167–75.
5. Brune B, Ullrich V. Inhibition of platelet aggregation by carbon monoxide is mediated by activation of guanylate cyclase. Mol Pharmacol 1987; 32:497–504.
6. Meller ST, Dykstra CL, Gebhart GF. Investigations of the possible role for carbon monoxide (CO) in thermal and mechanical hyperalgesia in the rat. Neuroreport 1994; 5:2337–41.
7. Yamamoto T, Nozaki-Taguchi N. Zinc protoporphyrin IX, an inhibitor of the enzyme that produces carbon monoxide, blocks spinal nociceptive transmission evoked by formalin injection in the rat. Brain Res 1995; 704:256–62.
8. Kim SH, Chung JM. An experimental model for peripheral neuropathy produced by segmental spinal nerve ligation in the rat. Pain 1992; 50:355–63.
9. Hargreaves K, Dubner R, Brown F, et al. A new and sensitive method for measuring thermal nociception in cutaneous hyperalgesia. Pain 1988; 32:77–88.
10. Chaplan SR, Bach FW, Pogrel JW, et al. Quantitative assessment of tactile allodynia in the rat paw. J Neurosci Methods 1994; 53:55–63.
11. Brennan TJ, Vandermeulen EP, Gebhart GF. Characterization of a rat model of incisional pain. Pain 1996; 64:493–501.
12. Tenhunen R, Marver HS, Schmid R. Microsomal heme oxygenase: characterization of the enzyme. J Biol Chem 1969; 244:6388–94.
13. Tenhunen R, Ross ME, Marver HS, et al. Reduced nicotinamide-adenine dinucleotide phosphate dependent biliverdin reductase: partial purification and characterization. Biochemistry 1970; 9:298–303.
14. Zakhary R, Gaine SP, Dinerman JL, et al. Heme oxygenase 2: endothelial and neuronal localization and role in endothelium-dependent relaxation. Proc Natl Acad Sci USA 1996; 93:795–8.
15. Mark JA, Maines MD. Tin-protoporphyrin-mediated disruption in vivo of heme oxygenase-2 protein integrity and activity in rat brain. Pediatr Res 1992; 32:324–9.
16. Bing O, Grundemar L, Ny L, et al. Modulation of carbon monoxide production and enhanced spatial learning by tin protoporphyrin. Neuroreport 1995; 6:1369–72.
17. Ewing JF, Maines MD. Glutathione depletion induces heme oxygenase-1 (HSP32) mRNA and protein in rat brain. J Neurochem 1993; 60:1512–9.
18. Dore S, Takahashi M, Ferris CD, et al. Bilirubin, formed by activation of heme oxygenase-2, protects neurons against oxidative stress injury. Proc Natl Acad Sci U S A 1999; 96:2445–50.
19. Dwyer BE, Nishimura RN, Lu SY. Differential localization of heme oxygenase and NADPH-diaphorase in spinal cord neurons. Neuroreport 1995; 6:973–6.
20. Mao J, Price DD, Mayer DJ, et al. Pain-related increases in spinal cord membrane-bound protein kinase C following peripheral nerve injury. Brain Res 1992; 588:144–9.
21. Malmberg AB, Chen C, Tonegawa S, et al. Preserved acute pain and reduced neuropathic pain in mice lacking PKCgamma. Science 1997; 278:279–83.
22. Mao J, Price DD, Mayer DJ. Thermal hyperalgesia in association with the development of morphine tolerance in rats: roles of excitatory amino acid receptors and protein kinase C. J Neurosci 1994; 14:2301–12.
23. Li X, Clark JD. Chronic exposure to morphine increases expression of heme oxygenase type 2. Brain Research. In press.
24. Poss KD, Thomas MJ, Ebralidze AK, et al. Hippocampal long-term potentiation is normal in heme oxygenase-2 mutant mice. Neuron 1995; 15:867–73.
25. Kappas A, Drummond GS, Manola T, et al. Sn-protoporphyrin use in the management of hyperbilirubinemia in term newborns with direct Coombs-positive ABO incompatibility. Pediatrics 1988; 81:485–97.
© 2000 International Anesthesia Research Society