|Home | About | Journals | Submit | Contact Us | Français|
Small unmyelinated sensory neurons classified as nociceptors are divided into two subpopulations based on phenotypic differences including expression of neurotrophic factor receptors. Approximately half of unmyelinated nociceptors express the NGF receptor TrkA and half express the GDNF Family Ligand (GFL) receptor Ret. The function of NGF/TrkA signaling in the TrkA population of nociceptors has been extensively studied and NGF/TrkA signaling is a well established mediator of pain. The GFLs are analgesic in models of neuropathic pain emphasizing the importance of understanding the physiological function of GFL/Ret signaling in nociceptors. However, perinatal lethality of Ret-null mice has precluded the study of the physiological role of GFL/Ret signaling in the survival, maintenance and function of nociceptors in viable mice. We deleted Ret exclusively in nociceptors by crossing nociceptor-specific Nav1.8 Cre and Ret conditional mice to produce Ret-Nav1.8 conditional knock out (CKO) mice. Loss of Ret exclusively in nociceptors results in a reduction in nociceptor number and size indicating Ret signaling is important for the survival and trophic support of these cells. Ret-Nav1.8 CKO mice exhibit reduced epidermal innervation, but normal central projections. In addition, Ret-Nav1.8 CKO mice have increased sensitivity to cold and increased formalin-induced pain, demonstrating that Ret signaling modulates the function of nociceptors in vivo. Enhanced inflammation-induced pain may be mediated by decreased Prostatic Acid Phosphatase (PAP) as PAP levels are markedly reduced in Ret-Nav1.8 CKO mice. The results of this study identify the physiological role of endogenous Ret signaling in the survival and function of nociceptors.
During embryonic development small-diameter sensory neurons in the dorsal root ganglion (DRG) which are classified as nociceptors express the NGF receptor TrkA and depend on NGF for survival (Johnson et al., 1980; Crowley et al., 1994; Silos-Santiago et al., 1995). During late embryonic and early postnatal life small-diameter unmyelinated nociceptors (C-fiber nociceptors) diverge into two subpopulations (Molliver et al., 1997). One population continues to express TrkA. In the other population, TrkA expression is down-regulated and Ret, the signaling receptor for the GDNF Family Ligands (GFLs) is up-regulated. Nociceptors that express TrkA express neuropeptides, do not bind the lectin isolectin B4 (IB4), do not express thiamine monophosphatase (TMP), and project to lamina I and lamina II outer (IIo) of the dorsal horn while nociceptors that express Ret do not typically express neuropeptides, do bind IB4, do express TMP, and project to lamina II inner (IIi) of the dorsal horn (Molliver et al., 1997; Braz et al., 2002; Zylka et al., 2005). While much is known about the role of NGF/TrkA signaling in the TrkA population of C-fiber nociceptors, notably the well established role of NGF/TrkA signaling in pain (Pezet and McMahon, 2006), little is known of the physiological role of GFL/Ret signaling in the Ret population.
The GFL family includes: glial cell line-derived neurotrophic factor (GDNF), neurturin (NRTN), artemin (ARTN), and persephin (PSPN) (Airaksinen and Saarma, 2002). The GFL receptor is a complex consisting of a signaling component, the receptor tyrosine kinase Ret, and a ligand-binding component, a cell surface receptor called GDNF Family Receptor α (GFRα 1–4) (Airaksinen and Saarma, 2002). Each GFRα co-receptor binds with high-affinity to one of the four ligands and confers ligand specificity to the receptor complex: GDNF-GFRα1, NRTN-GFRα2, ARTN-GFRα3, and PSPN-GFRα4. Ret is the common signaling component of all GFL receptor complexes and is required in partnership with a GFRα co-receptor to form a functional receptor complex.
GFL/Ret signaling influences sensory neuron survival as well as sensory neuron structure and function in many experimental paradigms (Bennett et al., 1998b; Bennett et al., 2000; Airaksinen and Saarma, 2002; Ernsberger, 2008; Malin and Davis, 2008). Perinatal death of Ret null mice has precluded study of the physiological function of Ret in mature sensory neurons (Schuchardt et al., 1994). To overcome this limitation we deleted Ret exclusively in the non-peptidergic IB4-binding subpopulation of C-fiber nociceptors by crossing mice harboring a conditional RET allele (Jain et al., 2006) with mice expressing Cre recombinase under the sodium channel α subunit 1.8 promoter (Nav1.8-Cre) (Stirling et al., 2005), which is expressed exclusively in nociceptors. Analysis of these mice reveals that Ret is required for the survival, trophic maintenance and normal function of IB4 nociceptors.
Institution approved protocols were followed for all animal studies. Generation of Ret conditional mice with a floxed human RET9 allele followed by the EGFP reporter for identification of cells with Ret deletion were generated as previously described (Jain et al., 2006). To produce mice with conditional Ret deletion in nociceptors, heterozygous mice with one Ret-null allele (RetTGM or Retlacz) (Enomoto et al., 2001; Gould et al., 2008) knocked-into the Ret locus were crossed with Nav1.8Cre/+ mice (Stirling et al., 2005) to produce double heterozygous mice (Ret+/−:Nav1.8Cre/+). Double heterozygous mice were then crossed with Ret conditional mice (RetRETfloxEGFP/+) to produce Ret-Nav1.8 Conditional Knock Out (CKO) mice (RetRETfloxEGFP/−:Nav1.8Cre/+) and Ret-Nav1.8 Heterozygous Control (Het) mice (RetRETfloxEGFP/+:Nav1.8Cre/+). Note that EGFP reporter expression in these mice indicates that Cre-mediated deletion of the floxed allele has occurred.
Behavior testing was performed using male and female mice 7–10 weeks of age. Litter mates or age-matched control mice of the appropriate genotypes were used. Ret-Nav1.8 CKO mice (RetRETfloxEGFP/−:Nav1.8Cre/+) were compared with control mice that harbor at least one functional Ret allele in non-peptidergic nociceptors. Control mice include Ret-Nav1.8 Het mice (RetRETfloxEGFP/+:Nav1.8Cre/+) or others obtained in crosses to produce Ret-Nav1.8 CKO mice (i.e. Ret+/−:Nav1.8Cre/+, Ret+/+:Nav1.8Cre/+, RetRETfloxEGFP/+:Nav1.8+/+, etc.). We did not observe any significant difference between Ret-Nav1.8 Het mice (RetRETfloxEGFP/+:Nav1.8Cre/+) and these other control mice in behavior tests; therefore, they were combined into a single group and are referred to as control mice. To assess general health, mice were weighed and internal and surface body temperature was recorded. The experimenter was blind to the genotypes of the mice during data acquisition. Prior to testing, mice were placed in open bottom Plexiglas boxes (5×5×10 inches) on an elevated plastic (formalin), wire mesh (von Frey, acetone), or warmed (30°C) glass (Hargreaves) platform for one hour (formalin, acetone) or 2–3 hours (von Frey, Hargreaves).
Mice were given a subcutaneous injection of 10μl of 2.5% formaldehyde into the plantar surface of the right paw and spontaneous pain behavior was recorded in 5 minute bins for a total of 1 hour. Spontaneous pain behavior included shaking flinching, or licking of the paw as well as holding the paw in an elevated position.
Mechanical sensitivity was determined using the von Frey test. Beginning with the smallest filament and continuing from smallest to largest calibrated von Frey filaments were pressed to the plantar surface of the hind paw until the filament just bent. The withdrawal threshold is defined as the force which produces a withdrawal response in 3 of 5 consecutive applications within one trial. The threshold was determined in 3 trials per hind paw with a 15 minute interval between trials. Tail mechanical sensitivity was measured with an Analgesia-Meter (Ugo Basile; Stoelting) using a modification of the Randall-Selitto test as previously described (Morales and Gereau, 2009). The withdrawal threshold was obtained by averaging the values obtained in 3 trials which were separated by 15 minutes.
The thermal threshold was determined by measuring the withdrawal latency to a radiant heat source applied to the plantar surface of the hind paw in 5 separate trials for each hind paw with a 15 minute interval between trials. The withdrawal threshold was determined by averaging the withdrawal latency obtained in each of the trials.
Sensitivity to a cold stimulus was measured using the acetone test. One drop of acetone was applied to the plantar surface of the hind paw using a 1 ml syringe. Mice were observed for 5 minutes after each acetone application. Spontaneous pain behavior (defined as above in the formalin test) which occurred within five minutes after acetone application was counted as a positive response. Spontaneous pain behavior which occurred within the first 15 seconds after acetone application was not counted since most mice had some reaction to the initial application. In addition, the amount of time spent in pain behavior was recorded. Five trials were performed on each hind paw with a 5 minute interval between trials.
Mice were tested in a thermal gradient apparatus with zone temperatures ranging from 9°C to 51°C as described previously (Dhaka et al., 2007). The temperature zone in which mice spend the most time during the 2 hour test is defined as the “preferred temperature”.
An accelerating rotarod (Ugo Basile, Italy) was used to evaluate motor coordination and balance. Five consecutive acceleration trials were performed with a 5 minute interval between trials as described previously (Montana et al., 2009).
Tests in the sensorimotor battery were used as previously described to evaluate balance, strength, coordination, and movement (Jain et al., 2006).
Locomotor activity was measured in an open field using a VersaMax Animal Activity Monitoring System (AccuScan Instruments, Columbus, OH). Mice were habituated to the test room in their cages for 1 hour prior to testing. Locomotor activity was assessed by recording photobeam breaks in a 42L × 42W × 30H cm chamber for 60 minutes. Total distance traveled, time spent moving, and the number of beam breaks (horizontal activity), were calculated for the entire chamber.
Mice were deeply anesthetized and perfused with PBS followed by 4% paraformaldehyde in PBS. L4 DRG were removed, rinsed in PBS and embedded in paraffin. The entire DRG was sectioned at 8μm in the transverse plane, collected in sequence and stained with 0.25% cresyl violet. DRG neurons containing one or more nucleoli were counted in every tenth section. The number of neurons counted was multiplied by 10 to obtain the total number. Counts were corrected for split nuclei (Silos-Santiago et al., 1995).
To take advantage of the EGFP reporter as a marker for neurons in which Ret is deleted Ret-Nav1.8 CKO mice and Ret-Nav1.8 Het control mice, each of which contain one floxed Ret allele followed by the EGFP reporter and one Cre allele in the Nav 1.8 locus, were used in IHC experiments. Foot pad skin was removed from the plantar surface of the hind paw, immersion fixed in Zamboni’s fixative for 2 hours, rinsed in PBS, and cryoprotected in 30% sucrose. Mice were then perfused as described above. DRG and the lumbar enlargement of the spinal cord were removed, rinsed in PBS and cryoprotected in 30% sucrose. Skin was sectioned at 30 μm in a plane perpendicular to the skin surface. Transverse sections of lumbar spinal cord and DRG (30 μm and 18 μm, respectively) were cut on a cryostat and collected on slides. The following antibodies were used: rabbit anti-Ret 1:200 (IBL), rabbit anti-GFP 1:1000 (Invitrogen), rabbit anti TrkA 1:300 (Upstate), rabbit anti-βIII tubulin (Covance), goat anti-GFRα2 and GFRα3 (1:200 and 1:50, respectively R and D Systems), and goat anti-CGRP 1:400 (Serotec). Alexa conjugated to IB4 (1:100) (Invitrogen) was used for IB4 labeling.
To determine the percent of labeled neurons relative to the number of EGFP-positive neurons, labeled profiles and EGFP-positive profiles were counted in at least 3 randomly selected sections (separated by > 50 μm) per DRG. To determine the total number of EGFP-, βIII Tubulin-, or TrkA-positive neurons, 18 μm sections of the entire DRG were collected in series. Neurons were counted in every fourth section and the number of neurons was multiplied by 4 to obtain the total number.
The dermal-epidermal border was traced and the length determined using MetaMorph software. Labeled nerve fibers in the epidermis were counted and the fiber density (# of fibers/unit of dermal-epidermal border) was calculated (Lindfors et al., 2006).
30μm frozen spinal cord sections were incubated for 30 minutes at room temperature in Tris Maleate buffer containing 0.08% lead nitrate and 0.25% thiamine monophosphate chloride (Sigma) and developed in a 1% aqueous solution of sodium sulfide (Akkina et al., 2001).
All images were captured using an upright microscope equipped for epifluorescence microscopy (Nikon 80i; CoolSnapES camera) and were processed using MetaMorph or Adobe Photoshop software using global adjustments in brightness and contrast.
RNA was purified from lumbar DRG from adult Ret-Nav1.8 CKO and control mice. RNA (1μg) was reverse transcribed using M-MLV Reverse Transcriptase (Invitrogen). qRT-PCR was performed using Fast SYBR Green Master Mix (Applied Biosystems) and reactions were run in an Applied Biosystems 7900HT Real Time Thermal Cycler. All reactions were run in triplicate and relative fold changes in RNA amount were calculated by the ΔΔCT method using GAPDH to normalize samples (Livak and Schmittgen, 2001).
Real time PCR (qRT-PCR) primer sequences are as follows (5′→3′):
Data are reported as the Mean±SEM. Statistically significant differences were determined using the Student’s t Test.
Tissue-specific conditional deletion of genes is a powerful strategy to study the function of genes whose embryological loss results in lethality. Ret loss results in perinatal death due to renal agenesis (Schuchardt et al., 1994). We reasoned that generating mice expressing a conditional RET allele (RETfloxEGFP) and the nociceptor-specific Nav1.8 Cre, would allow us to study the function of Ret specifically in these neurons. We first determined if Cre expressed from the Nav1.8 locus could delete the conditional RET allele in an efficient and specific manner. We generated compound heterozygous mice that harbor one RET conditional allele and one Nav1.8 Cre allele (RetRETfloxEGFP/+:Nav1.8Cre/+, “+” denotes the endogenous wild-type mouse allele). This strategy allows simultaneous visualization of the excised allele by direct detection of EGFP signal and fluorescent labeled markers (including Ret) for non-peptidergic nociceptors (Jain et al., 2006). In spinal cord and DRG from RetRETfloxEGFP/+:Nav1.8Cre/+ postnatal day (PND) 1 mice EGFP signal is restricted to DRG neurons and to afferents in the superficial lamina of the dorsal horn where small diameter DRG neurons terminate, suggesting specific deletion of Ret in non-peptidergic nociceptors (Figure 1a, c, d). EGFP is not detected in motor neurons in the ventral horn of the spinal cord or in sympathetic ganglia in these mice, sites where Ret is known to be expressed, as shown by EGFP signal in Ret-EGFP reporter mice which express EGFP in all Ret-positive cells (Figure 1a-f and data not shown). In addition, we observed EGFP expression in spinal dorsal horn neurons in Ret-EGFP reporter mice (Figure 1b, e) which is absent in Ret-Nav1.8 CKO mice (Figure 1a, c). These results suggest specific deletion of the Ret conditional allele in small diameter nociceptors.
We determined the efficiency of conditional Ret deletion in small diameter DRG neurons using IB4 labeling which overlaps virtually completely with Ret in small diameter non-peptidergic DRG neurons (Molliver et al., 1997; Bennett et al., 1998b). IB4 labeling in lumbar DRG from PND 1 Ret-Nav1.8 Het mice (RetRETfloxEGFP/+:Nav1.8Cre/+) reveals specific and efficient deletion of Ret in IB4-labeled neurons (Figure 1g-i). Of 560 neurons expressing EGFP (neurons from which Ret is deleted allowing expression of the EGFP reporter from the Ret locus), 557 were labeled with IB4 indicating greater than 99% specificity of Ret deletion in the IB4 binding population (N=3). Of 603 IB4-positive neurons, 562 express EGFP indicating Ret is excised from 93% of IB4-positive DRG neurons (N=3). These results indicate that Ret is deleted with high efficiency and specificity in the population of small diameter DRG neurons that bind IB4 (non-peptidergic nociceptors).
We next generated Ret-Nav1.8 CKO mice (RetRETfloxEGFP/−:Nav1.8Cre/+). Ret-Nav1.8 CKO mice were viable, grew normally to adulthood and were grossly indistinguishable from control littermates that harbor at least one functional Ret allele in non-peptidergic nociceptors. There was no difference in adult body weight, or in internal or surface body temperature between Ret-Nav1.8 CKO and control mice, consistent with equivalent general health of the two groups (not shown).
In order to confirm that Ret is deleted from small diameter DRG neurons (nociceptors), lumbar DRG from adult Ret-Nav1.8 CKO and WT mice were labeled with antibodies to Ret (Figure 2a, b). In WT DRG Ret is observed in both large and small diameter DRG neurons (Figure 2a). In Ret-Nav1.8 CKO mice Ret labeling is absent in small diameter neurons but remains in large diameter neurons (Figure 2b). Ret labeling in DRG from PND 10 Ret-Nav1.8 CKO mice shows no overlap between EGFP signal (neurons in which Ret is deleted) and Ret immunoreactivity indicating specific deletion of Ret from non-peptidergic nociceptors (Figure 2c–e). As expected, qRT-PCR shows a significant decrease in Ret expression in lumbar DRG from Ret-Nav1.8 CKO mice compared to control mice (Table 1).
While GFLs promote survival and provide trophic support to many types of neurons, the physiological role of Ret in small nociceptive DRG neurons has not been examined in adult animals and remains unclear. Studies in GFL and Gfrα co-receptor null mice have reported variable results (Airaksinen and Saarma, 2002). To determine if Ret is required for survival and trophic support of nociceptors we compared the total number of L4 DRG neurons in adult Ret-Nav1.8 CKO mice and control mice. We found a 33% decrease in the total number of L4 DRG neurons in Ret-Nav1.8 CKO mice compared to control mice (Figure 3a–c). Because small diameter non-peptidergic nociceptors are a subset of total DRG neurons, we investigated the extent of neuron loss in the EGFP-positive sub-population of DRG neurons by counting the total number of EGFP-positive cells. There is a 52% decrease in the number of EGFP-positive L4 DRG neurons in Ret-Nav1.8 CKO mice compared to Ret-Nav1.8 Het control mice (Figure 3d–f).
To determine whether neuron loss in the population of EGFP-positive neurons accounts for the cell loss observed in total neuron counts (33% above), we counted EGFP and βIII tubulin-positive neurons to determine the percentage of total DRG neurons that are EGFP-positive (Table 2). In Ret-Nav1.8 Het mice EGFP-positive neurons account for 46.7±2.5% of total DRG neurons. Loss of 52 % of EGFP-positive neurons would therefore correspond to a ~24% decrease in total neuron number, accounting for most if not all of the DRG neuron loss in Ret-Nav1.8 CKO mice.
The soma size of DRG neurons is influenced by the expression of GFL ligands and GFL co-receptors (Zwick et al., 2002; Elitt et al., 2006; Lindfors et al., 2006). We compared the soma area of EGFP-positive DRG neurons in Ret-Nav1.8 CKO and Ret-Nav1.8 Het control mice to determine if Ret signaling is required for trophic support of non-peptidergic nociceptors. We found a 39% decrease in the soma area of EGFP-positive DRG neurons in adult Ret-Nav1.8 CKO mice compared to Ret-Nav1.8 Het control mice (Figure 3d, e, g). Size frequency analysis of EGFP-positive neurons suggests that size is decreased across the entire population of EGFP-positive neurons (Figure 3h). These results indicate that Ret signaling is required for trophic maintenance of mature non-peptidergic DRG neurons and is in agreement with findings in adult Gfrα2 null and PND 14 Ret-Wnt1 CKO mice (Lindfors et al., 2006; Luo et al., 2007).
In the DRG, Gfrα2 and Gfrα3 expression is largely restricted to small diameter neurons while Gfrα1 is expressed in both large diameter and small diameter DRG neurons (Bennett et al., 1998b; Orozco et al., 2001; Stucky et al., 2002; Lindfors et al., 2006; Luo et al., 2007). Because it has been reported that Gfrα2 deletion alters sensory function (Stucky et al., 2002; Lindfors et al., 2006) and Ret alters Gfrα2 expression (Luo et al., 2007) we examined Gfrα2 expression in Ret-Nav1.8 CKO mice. We also examined expression of Gfrα3, the other Gfrα co-receptor expressed exclusively in small diameter DRG neurons (Orozco et al., 2001; Elitt et al., 2006). In agreement with previous reports (Bennett et al., 1998b; Lindfors et al., 2006; Luo et al., 2007) we found that virtually all Gfrα2-positive neurons in Ret-Nav1.8 Het control mice are also EGFP-positive (express Ret) and the majority of EGFP-positive neurons co-express Gfrα2 (Table 2). Deletion of Ret had no effect on the percentage of EGFP-positive neurons which co-express Gfrα2 suggesting that Gfrα2 expression is not regulated by Ret in non-peptidergic nociceptors (Table 2). The percent of Gfrα2-positive neurons that express EGFP is significantly less in Ret-Nav1.8 CKO mice compared to Ret-Nav1.8 Het control mice, likely reflecting the loss of EGFP neurons in Ret-Nav1.8 CKO mice. Surprisingly, we did not find extensive overlap between EGFP (Ret) and GFRα3 in Ret-Nav1.8 Het control mice (Figure 3i, k; Table 2). This is in contrast to previous reports which have found greater co-localization of Ret and Gfrα3 (Orozco et al., 2001; Luo et al., 2007). We do not believe that our failure to detect significant co-expression of EGFP and Gfrα3 is due to lack of Nav1.8 expression in Gfrα3-positive neurons because Gfrα3 is expressed exclusively in small diameter DRG neurons (Orozco et al., 2001; Elitt et al., 2006) and Nav1.8 is expressed in nearly all small diameter DRG neurons (Stirling et al., 2005) suggesting significant overlap of Nav1.8 and Gfrα3 expression. Deletion of Ret results in an increase in the proportion of Gfrα3-positive neurons which are EGFP-positive (Figure 3j, k; Table 2) suggesting that Ret negatively regulates Gfrα3 expression in small diameter DRG neurons.
To determine whether neuron loss due to Ret deletion is accompanied by a corresponding loss of peripheral innervation we quantified the density of EGFP-positive fibers in the epidermis (Figure 4a, d, g). Labeling with an anti-EGFP antibody was required to visualize EGFP afferents in the skin. The density of EGFP-positive sensory afferents in the epidermis of Ret-Nav1.8 CKO mice is significantly decreased compared with Ret-Nav1.8 Het control mice. This decrease in non-peptidergic afferents in the epidermis likely reflects the loss of non-peptidergic neurons. Surprisingly, the total epidermal fiber density, as determined by the density of βIII tubulin-positive fibers, is not significantly different in Ret-Nav1.8 CKO mice compared with Ret-Nav1.8 Het control mice (Figure 4b, e, g). The majority of nerve fibers in the epidermis are from non-peptidergic DRG neurons which express Ret or from DRG neurons that express the NGF receptor TrkA and co-express neuropeptides (Zylka et al., 2005). To determine whether expansion of the TrkA/peptidergic fiber population in the epidermis accounts for the normal number of βIII tubulin-labeled fibers observed in Ret-Nav1.8 CKO mice we compared the density of CGRP-positive fibers in Ret-Nav1.8 CKO mice and Ret-Nav1.8 Het control mice (Figure 4c, f, g). We found that there is no difference in the density of CGRP-positive fibers in the epidermis of Ret-Nav1.8 CKO and control mice indicating that expansion of peptidergic afferents does not account for the normal density of βIII tubulin-labeled epidermal afferents observed in Ret-Nav1.8 CKO mice.
In the spinal cord dorsal horn, TrkA (peptidergic) and Ret (non-peptidergic) afferents are segregated with TrkA afferents occupying Lamina I and Lamina IIo and Ret afferents occupying Lamina IIi (Molliver et al., 1997). Double labeling of EGFP (Ret) and CGRP (TrkA) afferents in the lumbar dorsal horn shows that the non-overlapping pattern of Ret and TrkA afferents observed in lamina I and lamina II of Ret-Nav1.8 Het control mice is indistinguishable from the pattern observed in Ret-Nav1.8 CKO mice (Figure 5a-f). In addition, the density of EGFP labeling in Ret-Nav1.8 CKO mice is similar to Ret-Nav1.8 Het control mice indicating that, in contrast to peripheral projections, the density of the central projections of non-peptidergic Ret afferents is unchanged in the absence of Ret. Similar results were obtained with higher dilution of the GFP antibody suggesting that oversaturation of the fluorescent signal did not obscure loss of EGFP-positive afferents (not shown). IB4 labeling of non-peptidergic afferents in the lumbar dorsal horn was also similar in Ret-Nav1.8 CKO mice and Ret-Nav1.8 Het control mice (Figure 8f, g). Therefore, although there is a significant loss of non-peptidergic neurons in Ret-Nav1.8 CKO mice, the density and topography of the central projections of these neurons in the dorsal horn appear normal. Ret signaling is therefore dispensable for the development and maintenance of central projections from non-peptidergic DRG neurons since surviving non-peptidergic neurons reach their central targets in the absence of Ret.
To determine whether loss of Ret in nociceptors affects mechanical sensitivity, von Frey testing was performed on Ret-Nav1.8 CKO and control mice. There is no difference in hind paw mechanical sensitivity between Ret-Nav1.8 CKO mice and control mice (Figure 6a, Left). Sensitivity to a noxious mechanical stimulus, determined using a modified Randall Selitto test, is also not different in Ret-Nav1.8 CKO mice compared to control mice (Figure 6a, Right). These results indicate that Ret signaling in non-peptidergic DRG neurons is not required for normal behavioral sensitivity to innocuous or noxious mechanical stimuli. Recent reports demonstrating increased mechanical thresholds in mice in which most/all non-peptidergic DRG neurons are killed using Diphtheria Toxin (DTA) suggest that non-peptidergic nociceptors are required for normal mechanical sensitivity (Abrahamsen et al., 2008; Cavanaugh et al., 2009). In Ret-Nav1.8 CKO mice in which ~50% of non-peptidergic DRG neurons are lost we observed no change in mechanical sensitivity, indicating that 50% or less of the normal number of non-peptidergic DRG neurons is sufficient to maintain normal mechanical sensitivity.
Previous studies suggest that GFL/Ret signaling may modulate noxious heat transduction in IB4 nociceptors (Stucky et al., 2002; Lindfors et al., 2006). We found that noxious heat sensitivity, as measured by the Hargreaves test, is not different between Ret-Nav1.8 CKO and control mice (9.2±0.5s, 8.8±0.3s, respectively; p=0.4; Ret-Nav1.8 CKO, N=25, Control, N=55). Consistent with this finding, qRT-PCR analysis reveals no change in the expression of the noxious heat-activated channel TRPV1 in Ret-Nav1.8 CKO mice compared to control mice (Table 1). Because male and female mice respond differently to noxious thermal stimuli (Mogil et al., 2000) we analyzed data from male and female mice separately. There was no difference between Ret-Nav1.8 CKO and control mice in the male or female group (Figure 6b). As previously reported, male control mice had significantly longer withdrawal latencies than female control mice (Mogil et al., 2000). Thermal preference, measured using a thermal gradient apparatus, was not different in male or female Ret-Nav1.8 CKO compared to control mice (Figure S1).
To determine if Ret loss in non-peptidergic nociceptors influences sensitivity to a cold stimulus we assessed cold sensitivity in Ret-Nav1.8 CKO mice and control mice using the acetone test. Female Ret-Nav1.8 CKO mice respond to a significantly greater percentage of acetone applications than female control mice (Figure 6c). In addition, the total amount of time spent in pain-like behavior is significantly greater for female Ret-Nav1.8 CKO mice compared to female controls (Figure 6d). In contrast, male Ret-Nav1.8 CKO mice are not different from male control mice (Figure 6c, d). This result suggests that disruption of Ret signaling in non-peptidergic DRG neurons in female Ret-Nav1.8 CKO mice results in greater sensitivity to cold and is consistent with the observation that a mixed group of male and female Gfrα2 null mice show increased sensitivity to cold (Lindfors et al., 2006). In addition, our findings suggest that cold sensitivity may be influenced by gender. This possibility is consistent with a recent report of gender based differences in cold sensitivity in TRPA1 null mice (Kwan et al., 2006).
Our results suggest that Ret signaling in non-peptidergic nociceptors regulates cold sensation. Alternatively, loss of non-peptidergic neurons could contribute to increased cold sensitivity in Ret-Nav1.8 CKO mice. However, this possibility is inconsistent with recent reports in which DTA is used to kill specific subpopulations of DRG neurons and cold sensitivity is either abolished or unchanged (Abrahamsen et al., 2008; Cavanaugh et al., 2009). Examination of the expression of the cold-activated channels TRPA1 and TRPM8 in Ret-Nav1.8 CKO by qRT-PCR reveals a significant increase only in TRPM8 expression in Ret-Nav1.8 CKO mice compared to control mice (Table 1). This result suggests that Ret signaling normally suppresses the expression of TRPM8 and that the mechanism underlying increased cold sensitivity in Ret-Nav1.8 CKO mice may involve TRPM8. Previous reports indicating that TRPM8 is not co-localized with Nav1.8 (Abrahamsen et al., 2008) or IB4 (Dhaka et al., 2008) suggest that TRPM8 and Ret are not co-localized and may indicate that modulation of TRPM8 expression by Ret is indirect.
We performed the formalin test in Ret-Nav1.8 CKO and control mice to determine if deletion of Ret in small diameter DRG neurons affects spontaneous pain behavior induced by inflammation. Ret-Nav1.8 CKO mice have an increased response, compared to control mice, in both phases of the formalin test (Figure 7). In the first phase of the formalin test Ret-Nav1.8 CKO mice spent 25% more time engaged in spontaneous pain behavior compared to control mice. In the second phase Ret-Nav1.8 CKO mice spent ~40% more time engaged in spontaneous pain behavior compared to control mice. These data suggest that Ret signaling in nociceptors attenuates formalin-induced inflammatory pain.
Our results are contrary to the decreased formalin-induced pain behavior observed in Gfrα2 null mice. Although decreased cell size suggests Ret signaling is decreased in Gfrα2 null mice (Lindfors et al., 2006), it is not known to what extent Ret signaling is impaired in Gfrα2 null mice since many sensory neurons express more than one Gfrα co-receptor. In addition, Ret-Nav1.8 CKO mice exhibit cell loss and changes in peripheral innervation which are different from anatomical changes observed in Gfrα2 null mice. In addition, Gfrα2 null mice are reported to be severely growth retarded which could influence pain behavior. Therefore, it is not surprising that the phenotypes of these two mutant mice are not identical.
To ensure that behavioral changes observed in Ret-Nav1.8 CKO are sensory specific and are not related to disturbances of general sensorimotor function or to motor deficits we performed sensorimotor testing in Ret-Nav1.8 CKO mice and control mice. The performance of Ret-Nav1.8 CKO mice is not different from Ret control mice in any of these tests (Figures S1–2).
NGF is a well established mediator of inflammation and attenuation of NGF/TrkA signaling decreases inflammation-induced pain behavior in the formalin test (Pezet and McMahon, 2006; Ugolini et al., 2007). Expanded TrkA expression could therefore explain the enhanced response to formalin observed in Ret-Nav1.8 CKO mice. We examined TrkA expression in sensory neurons from which Ret is deleted to determine if TrkA expression is expanded in this population. In control mice we found that TrkA is co-expressed in 9.3±2.2% of DRG neurons that express Ret (EGFP-positive neurons), consistent with previous reports (Table 2) (Molliver et al., 1997). There is an increase in the percentage of EGFP-positive neurons that co-express TrkA in Ret-Nav1.8 CKO mice compared with Ret-Nav1.8 Het control mice (Figure 8a, b, c Left; Table 2). The increase in TrkA co-expression in EGFP-positive neurons could be a result of expansion of TrkA expression in the non-peptidergic population of nociceptors or may reflect the decreased number of EGFP-positive neurons in Ret-Nav1.8 CKO mice. Quantification of the total number of TrkA-positive neurons in lumbar DRG revealed that the total number of TrkA-positive neurons is not different in Ret-Nav1.8 CKO mice compared to Ret-Nav1.8 Het control mice (Figure 8c Right). Although the total number of TrkA-positive DRG neurons is not significantly altered in Ret-Nav1.8 CKO, the proportion of TrkA-positive nociceptors is increased due to loss of non-peptidergic nociceptors in Ret-Nav1.8 CKO mice. An increase in the proportion of “pro-inflammatory” TrkA nociceptors may contribute to increased inflammation-induced pain observed in Ret-Nav1.8 CKO mice.
TMP is a well known marker for non-peptidergic DRG neurons and is known to co-localize extensively with IB4 (Bennett et al., 1998b; Zylka et al., 2008). A recent report has identified TMP as the transmembrane form of prostatic acid phosphatase (PAP) an enzyme which dephosphorylates extracellular adenosine monophosphate (AMP) to produce the analgesic molecule adensosine (Zylka et al., 2008). PAP null mice show normal sensitivity to mechanical and noxious thermal stimuli but are hypersensitive in chronic inflammatory and neuropathic pain models. The expression of PAP specifically in non-peptidergic DRG neurons and the similar behavior phenotypes of PAP null mice and Ret-Nav1.8 CKO mice suggested the possibility that Ret signaling might influence PAP expression or activity. We performed TMP histochemistry and found that TMP staining is greatly reduced in the superficial laminae of the lumbar dorsal horn of Ret-Nav1.8 CKO mice compared with Ret-Nav1.8 Het control mice, suggesting that Ret modulates TMP expression and/or activity (Figure 8d, e). Since the central projections of non-peptidergic DRG neurons are apparently normal in Ret-Nav1.8 CKO mice as shown by EGFP (Figure 5) and IB4 labeling (Figure 8f, g), it is unlikely that the decrease in TMP staining is due to loss of non-peptidergic afferents in the dorsal horn. Using qRT-PCR we found that PAP expression is significantly decreased in lumbar DRG from Ret-Nav1.8 CKO mice compared to control mice (Table 1). These results are consistent with the possibility that Ret positively regulates PAP expression and/or activity and that loss of Ret in non-peptidergic nociceptors results in a decrease in PAP with a consequent increase in formalin-induced pain.
Specific deletion of Ret in non-peptidergic nociceptors of Ret-Nav1.8 CKO mice has, for the first time, provided an opportunity to examine the physiological function of GFL/Ret signaling in non-peptidergic nociceptors in mature mice. Our results demonstrate that Ret is required for the survival, trophic support, and normal function of non-peptidergic nociceptors and provide evidence that decreased PAP (Prostatic Acid Phosphatase) expression or activity may be the molecular mechanism underlying increased formalin-induced pain following Ret deletion in nociceptors.
Ret expression increases from embryonic day 17 to PND 7 in maturing non-peptidergic DRG neurons while TrkA expression is extinguished between PND 0 and PND 21 (Molliver et al., 1997). A recent study in which Ret was conditionally deleted in DRG neurons using Wnt1-Cre reported that Ret is required for TrkA extinction in non-peptidergic neurons (Luo et al., 2007). In contrast, we found that TrkA expression is not maintained in non-peptidergic neurons in adult Ret-Nav1.8 CKO mice. Although the rate of TrkA extinction is decreased in Ret-Wnt1 CKO mice, TrkA expression is significantly decreased during the first two postnatal weeks in these mice (Luo et al., 2007) suggesting that, in agreement with our results, TrkA extinction occurs in non-peptidergic DRG neurons in the absence of Ret. In addition, extinction of TrkA expression in non-peptidergic DRG neurons is reportedly not complete until PND 21 (Molliver et al., 1997). Since Ret-Wnt1 CKO mice die in the third postnatal week it was not possible to determine if TrkA extinction continues after P14 or if TrkA extinction is completed in these mice.
Deletion of Ret in non-peptidergic nociceptors in Ret-Nav1.8 CKO mice results in ~50% loss of these neurons. Although GFL/Ret signaling is reported to influence sensory neuron survival under various experimental conditions (Matheson et al., 1997; Airaksinen and Saarma, 2002; Zwick et al., 2002; Elitt et al., 2006), the role of endogenous GFL/Ret signaling in sensory neuron survival in mature animals is unclear. Analysis of GFL and GFL co-receptor null mice has produced inconsistent results (Airaksinen and Saarma, 2002) due to overlapping expression of Gfrα co-receptors in many DRG neurons (Bennett et al., 1998b). Determination of neuron numbers in mice in which Ret is deleted has overcome this confound since Ret partners with all of the GFL co-receptors.
The incomplete loss of non-peptidergic DRG neurons in Ret-Nav1.8 CKO suggests that a subpopulation of non-peptidergic neurons is not dependent on Ret signaling for survival. DRG neurons become increasingly independent of trophic factors for survival during the first three postnatal weeks, the same period during which the transition from TrkA to Ret expression occurs in non-peptidergic neurons (Molliver et al., 1997; Vogelbaum et al., 1998). During the transition from TrkA to Ret expression, a subpopulation of non-peptidergic neurons co-expresses both receptors raising the possibility that non-peptidergic neurons which down regulate TrkA late in postnatal development depend on NGF/TrkA signaling for survival in the absence of Ret. In support of this possibility, prior to TrkA extinction in non-peptidergic DRG neurons, NGF can support the survival of these neurons in vitro (Molliver et al., 1997), suggesting that NGF/TrkA signaling can substitute for GFL/Ret signaling to maintain survival of non-peptidergic neurons. Neurons that lose TrkA expression early in the period of transition when they are still trophic factor-dependent likely die in Ret-Nav1.8 CKO because they lack both Ret and TrkA.
DRG neuron loss was not observed at PND 14 in Ret-Wnt1 CKO mice suggesting that DRG neuron survival is not dependent on Ret (Luo et al., 2007). However, at P14 TrkA extinction is not yet complete in non-peptidergic neurons (Molliver et al., 1997). Non-peptidergic neurons which continue to express TrkA at PND 14 could survive through NGF/TrkA signaling in the absence of Ret. There may be a sub-population of DRG neurons lost by PND14 in Ret-Wnt1 CKO mice (likely those which down-regulate TrkA prior to PND 14) which is too small to detect.
Surprisingly, the density of central projections of non-peptidergic nociceptors appears normal in adult animals despite significant loss of non-peptidergic DRG neurons in Ret-Nav1.8 CKO mice. One possible explanation for this observation is that surviving non-peptidergic neurons elaborate additional collateral branches in the dorsal horn to compensate for afferents which are lost with dying neurons. Whether or not these branches form appropriate central connections is unknown and could impact functional changes observed in Ret-Nav1.8 CKO mice.
Loss of non-peptidergic afferents in the epidermis is consistent with the loss of non-peptidergic neurons observed in Ret-Nav1.8 CKO mice. Development of epidermal innervation is dependent on NGF/TrkA signaling and independent of GFL/Ret signaling (Molliver et al., 1997; Patel et al., 2000). Whether the maintenance of peripheral innervation by non-peptidergic neurons requires GFL/Ret signaling in mature mice is unknown. Although 50% of non-peptidergic DRG neurons are lost, ~30% of the epidermal innervation density from non-peptidergic fibers is maintained in Ret-Nav1.8 CKO mice suggesting that the majority of surviving non-peptidergic neurons maintain target innervation in the absence of GFL/Ret signaling. This indicates that GFL/Ret signaling is not invariably required for the maintenance of peripheral target innervation although a sub-population of surviving non-peptidergic neurons may depend on Ret for maintenance of peripheral projections. Interestingly, NGF/TrkA signaling is required for the maintenance of peptidergic/TrkA fiber innervation of the epidermis in mature animals (Bennett et al., 1998a) revealing diverse requirements for trophic factors in the maintenance of peripheral projections in peptidergic and non-peptidergic nociceptors in mature animals.
Surprisingly, total epidermal fiber density is not decreased in Ret-Nav1.8 CKO mice. Maintenance of normal epidermal innervation density in Ret-Nav1.8 CKO mice is not due to expansion of TrkA/peptidergic afferents or to increased branching of surviving non-peptidergic afferents. Because innervation density is regulated by the amount of available trophic factor in the target field (Kessler et al., 1983; Zwick et al., 2002; Elitt et al., 2006), other Ret-expressing axons which respond to GFLs in the skin could potentially extend into areas of the epidermis from which non-peptidergic fibers are lost leaving unoccupied target areas of epidermis with available GFLs. Although peptidergic and non-peptidergic fibers from small diameter DRG neurons account for the majority of nerve fibers in the epidermis, other fiber types are present (Rice et al., 1998). Some of these, such as sympathetic fibers, express Ret (Airaksinen and Saarma, 2002) and may expand in the epidermis of Ret-Nav1.8 CKO mice.
Ret-Nav1.8 CKO mice exhibit increased formalin-induced pain behavior suggesting that endogenous Ret signaling in non-peptidergic nociceptors attenuates inflammation-induced pain. TMP (Thiamine Monophosphatase), a well known marker of non-peptidergic nociceptors, was recently identified as the trans-membrane form of PAP (Zylka et al., 2008). PAP expressed on the terminals of non-peptidergic nociceptors in Lamina II of the dorsal horn dephosphorylates extracellular AMP to produce the analgesic molecule adenosine, which activates adenosine A1 receptors on lamina II neurons producing an anti-nociceptive effect in models of neuropathic and inflammatory pain (Zylka et al., 2008).
The Ret ligands GDNF (Boucher et al., 2000), ARTN (Gardell et al., 2003), and NRTN (unpublished data) have potent analgesic activity in nerve-injury models of neuropathic pain. Exogenous GDNF restores TMP activity in non-peptidergic DRG neurons following nerve injury (Bennett et al., 1998b). The identification of the analgesic activity of PAP in nerve injury-induced pain suggests that regulation of PAP expression or activity may be a mechanism by which GFLs produce analgesia after nerve injury. In addition, PAP has analgesic activity in a model of inflammation-induced pain (Zylka et al., 2008). We have found that PAP activity is markedly reduced in the dorsal horn of Ret-Nav1.8 CKO mice (Figure 8). Our findings suggest that regulation of PAP may be a mechanism by which endogenous GFL/Ret signaling in non-peptidergic nociceptors modulates inflammation-induced pain behavior.
Previously published studies provide conflicting information as to whether GFLs are proalgesic or analgesic. A number of studies have found that exogenous GFLs administered to nerve injured animals prevent or reverse neuropathic pain (Boucher et al., 2000; Gardell et al., 2003; Hao et al., 2003; Wang et al., 2003; Pezet et al., 2006). However, there are conflicting reports regarding the effects of exogenous GFLs administered to normal animals. There have been reports that subcutaneous administration of GFLs produces sensitization to noxious stimuli (Malin et al., 2006; Bogen et al., 2008). However, other studies have reported that subcutaneous administration of GFLs does not alter sensory thresholds (Boucher et al., 2000; Gardell et al., 2003; Hao et al., 2003). Here, we focused on the role of endogenous GFL/Ret signaling in nociceptors in inflammation-induced pain. Our findings suggest that endogenous GFL/Ret signaling in nociceptors attenuates inflammation-induced spontaneous pain. We propose that increased inflammation-induced spontaneous pain in Ret-Nav1.8 CKO mice may be due to decreased PAP activity in these mice. This hypothesis is based on the reported analgesic action of PAP in models of both inflammatory and neuropathic pain (Zylka et al., 2008). Our results do not address the possibility that exogenous GFLs do or do not produce hyperalgesia in normal animals. Together with previous studies, our results suggest that modulation of PAP activity may be a mechanism by which GFL/Ret signaling produces analgesia in both inflammation- and nerve injury-induced pain models.
In summary, our results demonstrate that Ret signaling is required for the survival of a sub-population of small diameter non-peptidergic sensory neurons but is dispensable for the development and maintenance of central and peripheral target innervation of surviving non-peptidergic neurons. Importantly, Ret deletion in non-peptidergic DRG neurons results in altered cold sensitivity and increased pain following formalin-induced inflammation indicating that Ret signaling modulates nociceptor function in vivo.
We thank Amanda Knoten, Angela Lluka, Gary Wu and Sherri K. Vogt for providing excellent technical assistance and animal care. We thank Maria Elena Morales for preparation of the figures. This work was supported by NIH grant R21NS059566 and ACS IRG-58-010-47(to J.P.G.); HD047396, DK081644 and DK082531, and Children Discovery Institute grant MDII2009177 (to S.J.); NIH grant AGO13370 (to E.M.J. and J.M.); NIH grant NS042595 (to R.G.W.); and BBSRC and WCU grant R31-2008-000-10103-0 (to J.N.W.). This work was supported by NIH Neuroscience Blueprint Interdisciplinary Center Core Grant P30 NS057105 and George M. O’Brien Center for Kidney Disease Research (P30-DK079333) to Washington University.