Reduced inflammatory hyperalgesia with preservation of acute thermal nociception in mice lacking cGMP-dependent protein kinase I

Abstract

cGMP-dependent protein kinase I (PKG-I) has been suggested to contribute to the facilitation of nociceptive transmission in the spinal cord presumably by acting as a downstream target of nitric oxide. However, PKG-I activators caused conflicting effects on nociceptive behavior. In the present study we used PKG-I^-/- mice to further assess the role of PKG-I in nociception. PKG-I deficiency was associated with reduced nociceptive behavior in the formalin assay and zymosan-induced paw inflammation. However, acute thermal nociception in the hot-plate test was unaltered. After spinal delivery of the PKG inhibitor, Rp-8-Br-cGMPS, nociceptive behavior of PKG-I^+/+ mice was indistinguishable from that of PKG-I^-/- mice. On the other hand, the PKG activator, 8-Br-cGMP (250 nmol intrathecally) caused mechanical allodynia only in PKG-I^+/+ mice, indicating that the presence of PKG-I was essential for this effect. Immunofluorescence studies of the spinal cord revealed additional morphological differences. In the dorsal horn of 3- to 4-week-old PKG-I^-/- mice laminae I-III were smaller and contained fewer neurons than controls. Furthermore, the density of substance P-positive neurons and fibers was significantly reduced. The paucity of substance P in laminae I-III may contribute to the reduction of nociception in PKG-I^-/- mice and suggests a role of PKG-I in substance P synthesis.

The second messenger cGMP is formed by activation of soluble and particulate guanylyl cyclases and has several targets, including cGMP-dependent protein kinase I (PKG-I) and PKG-II, of which PKG-I is expressed in the spinal cord (1, 2). Spinally delivered PKG inhibitors reduce formalin-induced nociceptive behavior in rats (3, 4), suggesting that PKG-I plays an important role in spinal nociceptive processing. It has been speculated that PKG-I mediates hyperalgesic effects of nitric oxide (NO) (5). This idea is supported by the observation that PKG-I inhibition causes a reduction of thermal hyperalgesia induced by injection of the NO donor, NOC-12 (6). Endogenous NO is produced by NO synthases, of which neuronal nitric oxide synthase (nNOS) is activated and up-regulated after N-methyl-d-aspartate receptor stimulation (7-10). NO probably acts as a retrograde messenger (11, 12) at nociceptive synapses, i.e., it is released from the postsynaptic neuron, diffuses back to the presynaptic neuron, and stimulates guanylyl cyclases. The latter step links NO to cGMP production and PKG-I activation. Because inhibition of NOS activity reduces nociception (13, 14), the release of NO is thought to contribute to the development of hyperexcitability of nociceptive neurons under certain circumstances. Under the premise that PKG-I is a mediator of NO at nociceptive synapses, one would expect that PKG-I activation also causes hyperalgesia. However, effects of the spinally delivered PKG activator 8-Br-cGMP have been conflicting (15, 16). (See Supporting Text, which is published as supporting information on the PNAS web site.) Hence, the exact role of PKG-I in nociception remains elusive.

In the present study we used PKG-I^-/- mice to clearly assess the role of PKG-I in nociception. Because these mice have a defective regulation of smooth muscle contraction with vascular and intestinal dysfunctions (17, 18), the overall constitution of homozygous PKG-I^-/- mice deteriorates between 5 and 6 weeks of age (17). We therefore used 3- to 4-week-old animals for nociceptive experiments in the present study.

Materials and Methods

Animals. The generation of the PKG-I null allele and genotyping was done as described (17). Mice were bred and maintained in the animal facility of the Institut für Pharmakologie und Toxikologie der Technischen Universität Munich. For nociceptive testing PKG-I^-/- and litter-matched PKG-I^+/+ mice were shipped to the Institut für Klinische Pharmakologie, Klinikum der Johann Wolfgang Goethe-Universität Frankfurt. Experiments were performed the day after the arrival. Mice had free access to food and water before and during the experiments. The mean ± SD total body mass of wild-type and knockout mice was 16.2 ± 2.7 g and 14.2 ± 2.8 g, respectively (12.5% difference). The lean body mass differs by ≈8%. All experiments were approved by the local Ethics Committee for Animal Research and conformed to IASP ethical guidelines.

Nociceptive Testing. Formalin assay. Fifteen microliters of a 5% formaldehyde solution was injected into the s.c. space at the dorsal side of the right hind paw. The time spent licking the formalin-injected paw was recorded in 5-min intervals up to 45 min, starting right after formalin injection. The PKG-I inhibitor Rp-8-Br-cGMPS (Biolog Life Sciences Institute, Bremen, Germany) was delivered onto the lumbar spinal cord by intrathecal injection as has been described (19). The drug was dissolved in artificial cerebrospinal fluid (141.7 mM Na⁺/2.6 mM K⁺/0.9 mM Mg²⁺/1.3 mM Ca²⁺/122.7 mM Cl^-/21.0 mM mM mM dextrose, bubbled with 5% CO₂ in 95% O₂ to adjust the pH to 7.2) and injected in a volume of 5 μl. The dose (50 nmol) was 1/10th of the dose previously found to reduce flinching behavior in rats (4, 15). Drug injection was performed in short isoflurane anesthesia 10 min before the injection of formalin.

Mechanical hyperalgesia in zymosan-induced paw inflammation. Mice were adapted to the test perspex chamber with a grid bottom for at least 30 min before baseline testing. Fifteen microliters of a 10 mg/ml zymosan (Sigma) suspension in PBS (0.1 M PBS, pH 7.4) was then injected into the plantar side of the right hind paw. Mechanical hyperalgesia was assessed before zymosan injection and then hourly up to 7 h after zymosan injection. The threshold to mechanical nociceptive stimuli was assessed by means of a punctuated stimulation by using von Frey hairs of different strengths (0.008, 0.02, 0.04, 0.07, 0.16, 0.4, 0.6, 1, 1.4, 2, 4, and 6 g; Stoelting). They were placed perpendicularly onto the plantar surface of the right or left paw and bent slightly to apply punctuated pressure. The stimuli were applied at five repetitions each and at increasing order until the paw was withdrawn and then at decreasing order until paw withdrawal stopped. This up-and-down testing was repeated after a short rest. The geometric mean of uppermost (increasing testing) and lowest (decreasing testing) test results was taken as the mechanical paw-withdrawal threshold (MPWT). These data were log-transformed, and the percent decrease of the withdrawal threshold was then calculated in relation to the baseline withdrawal threshold as: % decrease of MPWT = MPWT_baseline - MPWT_zymosan/MPWT_baseline·100.

Mechanical allodynia induced by 8-Br-cGMP. After two baseline measurements 250 nmol of 8-Br-cGMP was injected into the subarachnoid space of the lumbar spinal cord in 5 μl of artificial cerebrospinal fluid. The dose was 1/10th of the dose previously found to cause hyperalgesia in rats (15), i.e., it was equivalent to the rat dose on a per kilogram basis. Based on these previous results, 250 nmol intrathecally can be considered as a high dose. (See Supporting Text and Fig. 4, which is published as supporting information on the PNAS web site, concerning a low dose of 8-Br-cGMP.) The mechanical nociceptive threshold was assessed at 5, 7.5, 10, 20, 30, 40, 50, and 60 min after 8-Br-cGMP injection as described above. Reactions of the right and left paw were identical. The percentage decrease of the MPWT was obtained after log transformation as described above.

Acute thermal nociception. A hot-plate test (temperature, 52°C; cut-off latency, 40 s; Hot Plate FMI, Föhr Medical Instruments, Seeheim/Ober-Beerbach, Germany; time resolution, 0.1 s) was performed to assess acute thermal nociception. The test was repeated three times for each mouse with a rest of 15 min in between, and the mean latency was used for statistical comparison.

Statistics. To compare the nociceptive behavior between groups the total licking time (formalin assay), paw-withdrawal latency (hot plate), or the area under the MPWT versus time course (zymosan and high dose 8-Br-cGMP) was subjected to univariate ANOVA in case of more than two groups or Student's t test in case of two groups. After ANOVA testing groups were mutually compared with t tests with a Bonferroni α-correction for multiple comparisons (α at 0.05). We used spss 11.0 for statistical evaluation.

Immunofluorescence. Mice were intracardially perfused with 0.9% saline followed by 4% paraformaldehyde in 0.1 M PBS (pH 7.4) under deep pentobarbital anesthesia (300 mg/kg). The spinal cord (lumbar enlargement) was removed and postfixed in the same fixative overnight (4°C), and 30-μm-thick transversal sections were cut on a vibratome.

Free-floating sections were incubated in blocking buffer (5% normal goat serum/0.3% Triton X-100 in 0.1 M PBS) for 1 h at room temperature and incubated overnight at 4°C with primary antibodies directed against PKG-I (1:500; ref. 20), substance P (1:500; Santa Cruz Biotechnology), NeuN (1:1,000; Chemicon), N200 (1:1,000; Chemicon), and nNOS (1:500; Alexis Biochemicals, Grünberg, Germany). Binding sites were visualized with Alexa488- or Alexa568-conjugated species-specific secondary antibodies (Molecular Probes). Confocal images (optical sections, 1 μm) were obtained by using a Zeiss LSM 510 confocal laser-scanning microscope.

Image Analysis. For quantitative analysis of cell and fiber densities, midlumbar spinal cord sections (L4/5) from PKG-I^-/- mice (n = 3) and PKG-I^+/+ mice (n = 3) were used. Quantification was performed as described (21), with modifications. Sections stained for substance P and NeuN and sections stained for nNOS and NeuN were analyzed (n = 5 per animal per staining). Confocal images were captured by using identical settings for all sections. Images were analyzed with imagej 1.31 (http://rsb.info.nih.gov/ij). NeuN labeling of neurons was used to identify lamina I-III of the dorsal horn. Cell numbers, mean pixel values, and areas of laminae I-III were determined and compared by using Student's t test (spss 11.0). Values of P < 0.05 were considered as statistically significant.

Results

Formalin Assay. PKG-I^-/- mice spent significantly less time licking the formalin-injected hind paw than PKG-I^+/+ mice (Fig. 1). These differences occurred in the first and second phase of the formalin assay. Injection of the PKG inhibitor Rp-8-Br-cGMPS significantly reduced the licking behavior in PKG-I^+/+ mice but had no effect in PKG-I^-/- mice. The licking behavior in PKG-I^+/+ mice treated with the PKG inhibitor was equivalent to that of untreated PKG-I^-/- mice, suggesting that the dose of Rp-8-Br-cGMPS was sufficient to completely inhibit PKG-I activity in the spinal cord.

Fig. 1.

(A) Time course of the formalin-induced licking behavior in PKG-I^+/+ (•, n = 8) and PKG-I^-/- (○, n = 8) mice. Fifteen microliters of 5% formaldehyde solution was injected s.c. into the right hind paw at time zero, and the time spent licking the injected paw was measured in 5-min intervals for 45 min. Eight mice were used in each group. (B) The bar chart shows the total licking time (0-45 min) after formalin injection in PKG-I^+/+ and PKG-I^-/- mice. Where indicated, 50 nmol of Rp-8-Br-cGMPS (PKG inhibitor) were delivered onto the lumbar spinal cord by injection into the subarachnoid space at the level of L4/L5 (n = 4 in each group). ^*, Statistically significant difference, P < 0.05.

Mechanical Allodynia After Injection of 8-Br-cGMP. 8-Br-cGMP (250 nmol intrathecally) caused mechanical allodynia exclusively in PKG-I^+/+ mice (Fig. 2A). In PKG-I^-/- mice the MPWT remained constant after injection of this dose of 8-Br-cGMP. The t test comparing the area under the “percent decrease of MPWT” versus “time” curves (AUC_MPWT) revealed significant differences between PKG-I^-/- and PKG-I^+/+ mice (P = 0.007).

Fig. 2.

(A) Time course of the percentage change of the MPWT after spinal delivery of the PKG activator 8-Br-cGMP (250 nmol) in PKG-I^+/+ (•, n = 8) and PKG-I^-/- (○, n = 8) mice. The paw-withdrawal threshold was assessed with von Frey hairs. 8-Br-cGMP was injected at time zero. Comparison of the area under the curves revealed a statistically significant difference between PKG-I^+/+ and PKG-I^-/- mice (P < 0.05). (B) Time course of the percentage change of the MPWT after s.c. injection of zymosan at the plantar side of the right hind paw in PKG-I^+/+ (•, n = 8) and PKG-I^-/- (○, n = 8) mice. The paw-withdrawal threshold was assessed with von Frey hairs. Zymosan was injected at time zero. (C) Paw-withdrawal latency (PWL) as assessed in the hot-plate test (52°C, 40-s cut-off latency) in PKG-I^+/+ (n = 4) and PKG-I^-/- (n = 4) mice. The PWL was measured three times in each animal with a rest of 15 min between measurements.

Mechanical Hyperalgesia in Zymosan-Induced Paw Inflammation. At baseline, the MPWT did not differ between PKG-I^-/- and PKG-I^+/+ mice (baseline threshold, 0.26 ± 0.06 g and 0.24 ± 0.06 g in PKG-I^-/- and PKG-I^+/+ mice, respectively; Fig. 2B). After injection of zymosan into the right hind paw, the MPWT decreased in both groups, indicating inflammatory hyperalgesia. In the first 2 h after zymosan injection, no difference occurred between knockout and wild-type mice. However, at >2 h, hyperalgesia decreased in PKG-I^-/- mice, whereas it remained constant up to the end of the observation period at 7 h in PKG-I^+/+ mice (Fig. 2B). The statistical comparison of the area under the MPWT-versus-time curves from 3 to 7 h after zymosan injection revealed significant differences between PKG-I^-/- and PKG-I^+/+ mice (P = 0.04).

Acute Thermal Nociception in the Hot-Plate Test. No difference occurred in the paw-withdrawal latency between PKG-I^-/- and PKG-I^+/+ mice (Fig. 2C; P = 0.88).

Immunofluorescence Studies. Macroscopically, the spinal cord of PKG-I^-/- mice was thinner than that of PKG-I^+/+ mice with similar body weight (±0.5 g). Brightfield micrographs of lumbar sections revealed a reduced ventral-to-dorsal diameter in PKG-I^-/- mice (Fig. 3 A and B). Labeling of neuronal cell bodies with an antibody against the neuron-specific nuclear protein (NeuN) showed the more compact structure of the dorsal horn in PKG-I^-/- mice (Fig. 3 C and D). In particular, the superficial laminae I-III were significantly reduced in size (Table 1). The changes in size were also observed by immunostaining with the neurofilament antibody N200 (Fig. 3 E and F). In control animals laminae II and III showed only weak axonal labeling and, thus, these laminae could readily be recognized as black areas. In PKG-I^-/- mice the width of these laminae was clearly smaller than in PKG-I^+/+ mice. This reduction in size was accompanied by a significant reduction in the number of neurons (Table 1). To study the distribution of functionally important subpopulations of neurons located within laminae II and III, substance P-containing neurons and nNOS-positive neurons were also studied. Compared with control animals (Fig. 3G), substance P-immunoreactive neurons and fibers were scarce in superficial laminae in PKG-I^-/- mice (Fig. 3H). Quantitative analysis revealed a significant reduction of substance P labeling in PKG-I^-/- mice (Table 1). In contrast, no obvious differences in the distribution of nNOS labeling was found (Fig. 3 I and J and Table 1). To demonstrate the colocalization of nNOS (Fig. 3I) and PKG-I (Fig. 3K) in PKG-I^+/+ mice, double-immunofluorescence was used (Fig. 3L). Confocal microscopic imaging revealed numerous double-labeled fibers and cell bodies within the superficial layers (Fig. 3 I, K, and L Insets).

Fig. 3.

Brightfield micrographs of lumbar sections of the spinal cord in PKG-I^+/+ (A) and PKG-I^-/- (B) mice. Note the difference in dorsoventral diameter. Dorsal horn of the lumbar spinal cord in PKG-I^+/+ (C) and PKG-I^-/- (D) mice immunostained for NeuN. Note the more compact structure of the dorsal horn in PKG-I^-/- mice (arrow) compared with PKG-I^+/+ mice. Dorsal horn of the lumbar spinal cord in PKG-I^+/+ (E) and PKG-I^-/- (F) mice immunostained for N200. The fiber-poor superficial layers, which contain mostly cell bodies, are broader in PKG-I^+/+ mice than in PKG-I^-/- mice (arrow). Dorsal horn of the lumbar spinal cord in PKG-I^+/+ (G) and PKG-I^-/- (H) mice immunostained for substance P. Note the numerous substance P-positive neurons in the superficial laminae of PKG-I^+/+ mice (arrow). In contrast, substance P-positive neurons are scarce in these layers in the PKG-I^-/- mouse. Dorsal horn of the lumbar spinal cord in PKG-I^+/+ (I) and PKG-I^-/- (J) mice immunostained for nNOS. Although the nNOS-immunoreactive neurons and fibers appear to be more densely packed in the superficial layers, the overall staining pattern of PKG-I^+/+ and PKG-I^-/- mice is similar. (Inset) High-magnification confocal image of nNOS-positive neurons (arrowhead) and fibers. Same section is shown in K and L. (K) Dorsal horn of the lumbar spinal cord in a PKG-I^+/+ mouse immunostained for PKG-I (same sections as shown in I). PKG-I immunoreactivity is distributed in a pattern that is similar to the one observed for nNOS immunoreactivity. (Inset) High-magnification confocal image of PKG-I-positive neurons (arrowhead) and fibers. (L) Dorsal horn of the lumbar spinal cord in a PKG-I^+/+ mouse immunostained for nNOS (green) and PKG-I (red). Double-labeled fibers appear yellow. (Inset) High-magnification confocal image of a double-labeled neuron (arrowhead) and numerous double-labeled fibers. (Scale bars: A and B, 200 μm; C-L, 100 μm; I, K, and L Insets, 10 μm.)

Table 1. Quantitative analysis of immunofluorescence studies for laminae I-III.

	Lamina I-III area, μm2	NeuN IR neurons	SP (IF)	nNOS (IF)
PKG-I−/−	107,332.4 ± 5,441.4*	223.1 ± 19.8*	52.3 ± 2.5*	67.1 ± 9.2
PKG-I+/+	133,282.3 ± 5,922.6	289.3 ± 28.8	65.0 ± 4.4	73.0 ± 3.3

Data represent the mean ± 2 SEM. SP, substance P; nNOS, neuronal nitric oxide synthase; IF, immunofluorescence intensity of SP- or nNOS-positive structures (neurons and fibers). ^*, P < 0.05, statistically significant difference.

Discussion

We show that cGMP kinase-I deficient mice have reduced formalin-evoked nociception and inflammatory hyperalgesia, whereas reactions to tactile and thermal stimuli to noninflamed paws are normal. This result suggests that the synaptic transmission of innocuous tactile and heat stimuli is PKG-I-independent, whereas the development and maintenance of hyperexcitability of nociceptive neurons involves PKG-I. This finding is in line with previous results obtained with Aplysia. Noxious stimuli to the mollusc caused long-term hyperexcitability of nociceptive sensory neurons that required NO/cGMP/PKG activation and transcription (5). In addition, the late phase of hippocampal long-term potentiation (late-LTP) is reduced in adult PKG-I knockout mice with a hippocampus-specific disruption of the PKG-I gene (22). Because LTP is a memory-like mechanism that resembles the activity-dependent “windup” of nociceptive neurons (11, 12, 23), the function of PKG-I in the sensitization of nociceptive neurons is probably similar to its role in late-LTP.

PKG-I is thought to act as a downstream mediator of NO. Inhibition of NO production reduces hyperalgesia, suggesting that NO facilitates nociceptive transmission (6, 13, 24). PKG inhibitors reduced NO-donor or N-methyl-d-aspartate-evoked hyperalgesia (6), supporting the suggested link between NO and PKG-I. PKG inhibitors also blocked hippocampal LTP (25, 26), again indicating similarities between both mechanisms. The immunofluorescence data of the present study, which show a colocalization of PKG-I and nNOS in spinal cord neurons and axons, provide additional support for the suggested cooperation between NO and PKG-I.

PKG-I additionally regulates the proliferation of sensory neurons during embryogenesis (27), and its deficiency caused axon guidance defects of nociceptive neurons within the developing dorsal root entry zone (2). The present results show that, in terms of formalin-induced nociception, PKG-I inhibition is as effective as PKG-I deficiency. This finding suggests that the previously observed axon guidance defects of nociceptive neurons in PKG-I^-/- mice do not essentially contribute to the observed alterations of nociception in 3- to 4-week-old PKG-I^-/- mice in this test. The defects in axon growth and connectivity and the resulting paucity of fibers in the dorsal funiculus (2) also apparently do not impair acute heat or mechanical nociception, because paw-withdrawal latency in the hot-plate test and baseline von Frey responses were equal in PKG-I^-/- and PKG-I^+/+ mice. However, the defects during embryogenesis may explain the observed size differences of the spinal cord with a remarkable reduction of the dorsoventral diameter in PKG-I^-/- mice. Although PKG-I^-/- mice were somewhat smaller and, in general, have less adipose tissue than their littermates, it seems unlikely that this dorsoventral size reduction is completely accounted for by leaner bodies.

We found that the density of substance P-immunoreactive neurons and fibers was significantly reduced in PKG-I^-/- mice. The potential link between PKG-I and substance P or its precursors has not been directly addressed until now. However, the NO-donor sodium nitroprusside evokes substance P release from spinal cord slices (28). Guanylyl cyclase inhibition prevents this NO-evoked substance P release, suggesting that this effect is mediated by cGMP (28) and therefore possibly by PKG. PKG primarily regulates target protein activity by phosphorylation (29) but may also regulate protein expression (30). The latter appears to be involved in PKG effects on LTP, because the translation inhibitor anisomycin abolished differences in late hippocampal LTP between control and hippocampus-specific PKG-I knockout mice (22). It is therefore conceivable that PKG-I regulates the expression of substance P precursors what would explain the reduced number of substance P-immunoreactive neurons and fibers in PKG-I^-/- mice. The paucity of substance P in laminae I-III may contribute to the reduced nociceptive response in PKG-I^-/- mice.

In summary, our results suggest that spinal PKG-I is involved in the facilitation of synaptic transmission of nociceptive stimuli in the spinal cord in an ongoing activation, whereas acute heat-induced nociception does not require PKG-I activity.