|Home | About | Journals | Submit | Contact Us | Français|
Prostatic acid phosphatase (PAP) is expressed in nociceptive dorsal root ganglia (DRG) neurons, functions as an ectonucleotidase and generates adenosine extracellularly. Here, we found that PAP inhibits noxious thermal sensitivity and sensitization that is associated with chronic pain through sustained activation of the adenosine A1 receptor (A1R) and phospholipase C-mediated depletion of phosphatidylinositol 4,5-bisphosphate (PIP2). In mice, intrathecal injection of PAP reduced PIP2 levels in DRG, inhibited thermosensation through TRPV1 and enduringly reduced thermal hyperalgesia and mechanical allodynia caused by inflammation, nerve injury and pronociceptive receptor activation. This included inhibitory effects on lysophosphatidic acid (LPA), purinergic (ATP), bradykinin and protease activated (thrombin) receptors. Conversely, PIP2 levels were significantly elevated in DRG from Pap−/− mice and this correlated with enhanced thermal hyperalgesia and mechanical allodynia in Pap−/− mice. To directly test the importance of PIP2 in nociception, we intrathecally injected PIP2 into mice. This transiently (2 hr) elevated PIP2 levels in lumbar DRG and transiently (2 hr) enhanced thermosensation. Additionally, thermal hyperalgesia and mechanical allodynia were enduringly enhanced when PIP2 levels were elevated coincident with injury/pronociceptive receptor stimulation. Nociceptive sensitization was not affected if PIP2 levels were elevated in the absence of ongoing pronociceptive receptor stimulation. Taken together, our data suggest that PIP2 levels in DRG directly influence thermosensation and the magnitude of nociceptive sensitization. Moreover, our data suggest there is an underlying “phosphoinositide tone” that can be manipulated by an adenosine-generating ectonucleotidase. This tone regulates how effectively acute nociceptive insults promote the transition to chronic pain.
Nociceptive neurons in the DRG sense noxious thermal and mechanical stimuli and can be sensitized by diverse pronociceptive chemicals (Hucho and Levine, 2007; Basbaum et al., 2009). Once sensitized, animals often display long-lasting thermal hyperalgesia and mechanical allodynia—two common symptoms of chronic pain.
Recently, we found that nociceptive neurons express two molecularly distinct ectonucleotidases that generate adenosine extracellularly by dephosphorylating adenosine 5’-monophosphate (AMP). These ectonucleotidases include the transmembrane (TM) isoform of PAP (also known as ACPP) and ecto-5’-nucleotidase (NT5E, also known as CD73) (Zylka et al., 2008; Sowa et al., 2010). Interestingly, PAP knockout (Pap−/−) mice, Nt5e−/− mice and adenosine A1 receptor knockout (A1R−/−) mice all display enhanced nociceptive responses following inflammation or nerve injury (Wu et al., 2005; Zylka et al., 2008; Sowa et al., 2010). These observations suggest that deficiencies in adenosine production or A1R signaling enhance nociceptive sensitization; however, the mechanism underlying these enhanced nociceptive responses is currently unknown.
The secretory isoform of PAP (S-PAP) also generates adenosine by dephosphorylating AMP (Vihko, 1978; Zylka et al., 2008; Sowa et al., 2009). When injected intrathecally (i.t.), S-PAP has long-lasting (3 day) thermal antinociceptive effects in naïve mice as well as long-lasting antihyperalgesic and antiallodynic effects in sensitized animals (Zylka et al., 2008; Sowa et al., 2009). These antinociceptive effects were transiently blocked by an A1R antagonist and were eliminated in A1R−/− mice, indicating that S-PAP exclusively activates A1R over a sustained time period.
Adenosine and A1R agonists have antinociceptive effects when administered acutely to rodents and humans (Eisenach et al., 2003; Sawynok, 2007). Additionally, acute A1R activation inhibits neurotransmitter release from nociceptive neurons, voltage-gated calcium channels, and postsynaptic neurons in spinal cord (Dolphin et al., 1986; Li and Perl, 1994; Lao et al., 2001). While these inhibitory mechanisms may account for some aspects of A1R-mediated antinociception, inhibition of neurotransmission cannot readily explain why sustained A1R activation by PAP selectively inhibits noxious thermal sensitivity in naïve mice without affecting mechanical sensitivity (Zylka et al., 2008; Sowa et al., 2009). This selectivity suggests that PAP might regulate thermal nociception by acting through a specific thermosensory channel.
Indeed, in our present study, we found that PAP acts through A1R to reduce the levels of PIP2 in cultured cells and in vivo. In turn, this reduction in PIP2 inhibits thermosensation, in part through the capsaicin and noxious heat receptor TRPV1 (Supplemental Figure S1)(Caterina et al., 1997; Caterina et al., 2000; Davis et al., 2000). This mechanism is consistent with several studies showing that PIP2 is required for TRPV1 to function optimally in vitro (reviewed in (Rohacs et al., 2008)). Pronociceptive G protein-coupled receptors (GPCRs) also require PIP2 to signal, suggesting alterations in PIP2 might regulate sensitization through these receptors. Indeed, we found that PAP inhibited signaling and sensitization through diverse pronociceptive GPCRs by depleting PIP2 (Supplemental Figure S1), with PIP2 levels at the time of stimulation/injury enduringly influencing the level of nociceptive sensitization. Collectively, our experiments suggest PIP2 plays a central role in nociceptive mechanisms.
All procedures and behavioral experiments involving vertebrate animals were approved by the Institutional Animal Care and Use Committee at the University of North Carolina at Chapel Hill.
Full-length expression constructs for mouse TM-PAP (nt 64- 1314 from GenBank accession # NM_207668) and human TM-PAP (nt 51-1304 from GenBank accession # BC007460) were generated by RT-PCR amplification using C57BL/6 mouse trigeminal cDNA or human placental cDNA (Clontech) as template and Phusion polymerase. The red fluorescent protein mCherry was then fused in-frame to the C-terminus of all TM-PAP constructs. Mouse TM-PAP(H12A) was generated by PCR-based mutagenesis using mouse TM-PAP as template (His12 corresponds to His43 of the mPAP preprotein). This active site mutant was previously described and lacks catalytic activity (Schneider et al., 1993; Ostanin et al., 1994). All constructs have a Kosak consensus sequence, were cloned into pcDNA3.1 and were sequence verified. We obtained additional constructs from others (see Acknowledgments). We confirmed that adenosine receptors were expressed in Rat1 fibroblasts by RT-PCR (A1R primers: 5’ CATTGGGCCACAGACCTACT and 5’ GGCAGAAGAGGGTGATACA).
Cell lines were grown on glass bottom culture dishes (MatTek Corp, P35G-0-10-C) in DMEM containing 10% Fetal Bovine Serum, 100 U/mL penicillin and 100 µg/mL streptomycin and transfected with Lipofectamine Plus (Invitrogen) according to manufacturer’s protocol. The total amount of DNA per transfection was adjusted to 1 µg by adding pcDNA 3.1. Following transfection (18–24 hours), cells were loaded for one hour at room temperature with 2 µM Fura-2 AM (Invitrogen, F-14185) in Hank’s Buffered Salt Solution (HBSS + Calcium and Magnesium) assay buffer (HBSS + 9 mM HEPES + 11 mM D-Glucose + 0.1% fatty-acid free BSA, pH 7.4). Cells were then washed 3 times with HBSS assay buffer and sat for at least 30 min prior to imaging. A Nikon TE2000U microscope and Sutter DG4 light source were used to image calcium responses (excitation 340 nm / 380 nm; emission 510 nm). We manually pipetted and aspirated solutions for all calcium imaging experiments. Cells were stimulated with 1 µM capsaicin, 100 nM LPA, 1 U/mL thrombin, 10 µM ATP or 1 µM bradykinin for 1–5 min, washed in HBSS assay buffer for 1 min, then stimulated with 0.006% SDS to evoke maximal calcium responses for normalization. We did not use ionomycin to normalize responses because this calcium ionophore activates calcium-dependent PLC enzymes. As a result, the magnitude of the ionomycin-induced calcium response is also proportional to PIP2 levels in cells.
Calcium responses were normalized by calculating the area under the curve (AUC) during ligand stimulation for each cell, and then dividing by the maximum SDS-evoked calcium response in each cell. These values were averaged over all cells for a given condition and then normalized relative to untransfected cells in the same field of view or relative to control cells (with the untransfected or control cell response set to 100%).
For experiments with capsaicin, Rat1 fibroblasts were transfected with TRPV1-GFP alone or TRPV1-GFP plus various constructs. The same amount of TRPV1-GFP was used for each transfection and the total amount of DNA per transfection was adjusted to 1 µg by adding pcDNA 3.1. Cells were stimulated with 1 µM capsaicin (from 100 mM stock in 100% DMSO, dissolved to final concentration in HBSS assay buffer) for 1 min, followed by a 5 min wash in HBSS assay buffer, then stimulation with 0.006% SDS.
For experiments with the PIP2 shuttle, Rat1 fibroblasts were stimulated with 100 nM LPA for 1 min, followed by a 15 min wash with HBSS assay buffer + 3 nM PIP2 + 3 nM Carrier 2 (PIP2 Shuttle Kit, Echelon, P-9045) or HBSS assay buffer + 3 nM Carrier 2 alone. Cells were then stimulated with 100 nM LPA again, washed for 1 min with HBSS assay buffer, and stimulated with 0.006% SDS.
For thapsigargin experiments, HBSS assay buffer lacking calcium and containing 1 mM EGTA was used to eliminate extracellular calcium. 10 µM thapsigargin was added for 5 min, the cells were then washed for 2 min with HBSS assay buffer, and stimulated with 0.006% SDS. For PTX experiments, Rat1 fibroblasts were incubated for 18 hours with 500 ng/mL PTX prior to loading with Fura-2 AM and stimulation with 100 nM LPA. For experiments with adenosine receptor antagonists, PLC inhibitor (U73122), PKC inhibitor (staurosporine), or PKA inhibitor (KT5720), cells were incubated with antagonists for 3–4 hours, loaded in the presence of antagonists/inhibitors with Fura-2 AM for one hour, and then stimulated with pronociceptive ligands.
A HEK293-TRPV1 stable cell line (Kim et al., 2008) was transfected with TM-PAP-mCherry or TM-PAP(H12A)-mCherry. Patch clamp recordings were made from mCherry-expressing cells using a Multiclamp 700B amplifier and pClamp 9.2 software as described (Campagnola et al., 2008). Heat ramps were generated by exchanging bath solution with a pre-heated solution via a 2-to-1 port. Solution was preheated with an in-line heater controlled by a TC-324B temperature controller modified for high temperature (Warner Instruments). Only one current recording was made per cover slip. The bath solution consisted of (in mM) NaCl 140, KCl 4, CaCl2 2, MgCl2 2, NaHEPES 10, Glucose 5, (pH 7.4, mOsm 295–310) and was perfused at a rate of 2–3 mL/min by gravity flow. Electrodes were pulled from borosilicate glass on a Sutter Instruments P-2000 and filled with intracellular solution that contained (in mM), KCl 135, MgATP 3, HEPES 10, Na2ATP 0.5, CaCl2 1.1, EGTA 2, Glucose 5, with pH adjusted to 7.5 with HCl and osmolarity adjusted to 300 mOsm with sucrose. Tip resistances ranged from 2.5 to 5 MΩ. Series resistance was not compensated; however, recordings with series resistances greater than 15 MΩ were discarded.
For quantification of PIP2 in vitro, HEK293 cells or Rat1 fibroblasts were plated onto glass coverslips and transfected with the construct PLCδ-PH-GFP along with indicated constructs using Lipofectamine Plus (Invitrogen), according to the manufacturer’s protocol. 18–24 hours later, the cells were fixed with 4% paraformaldehyde (PFA)-PBS. Cells were imaged on a confocal microscope. GFP fluorescence on the plasma membrane of cells compared to the cytoplasm was quantified using ImageJ by taking cross-sectional averages of pixel intensity at the plasma membrane and dividing by the average of pixel intensity in the cytoplasm.
For quantification of PIP2 in DRG, age-matched, adult male mice were injected i.t. with 5 µL of 15% lidocaine + 50 U/mL hPAP (250 mU total), 15% lidocaine + 3 nM Carrier 2 (Echelon, P-9C2), 15% lidocaine + 3 nM PIP2 (Echelon, P-9045) + 3 nM Carrier 2 or 15% lidocaine alone. Prior to injection, an equimolar mixture of PIP2 + Carrier 2 was incubated for 15 min. at room temperature. Lidocaine causes transient (5–20 min.) paralysis of both hindlimbs, permitting us to visually determine if each mouse received a successful i.t. injection (we only quantified PIP2 levels in mice that showed transient bilateral paralysis). One day later, mice were sacrificed and L3-L6 DRGs were dissected bilaterally (n=8 ganglia / sample) and placed in PBS on ice. For each sample, DRG wet weight was determined then lipids were extracted and quantified using the PI(4,5)P2 Mass ELISA Kit from Echelon (K-4500) following the manufacturer’s protocol. PIP2 levels were normalized by dividing by the wet weight of DRG tissue.
Pap−/−, A1R−/− and Trpv1−/− (B6.129X1-Trpv1tm1Jul/J) mice were backcrossed to C57BL/6 mice for at least 10 generations. For all other experiments male, C57BL/6 mice were purchased from Jackson Laboratories. Male, 2–4 month-old mice were used for all behavioral studies. All mice were acclimated to testing room, equipment, and experimenter for 1–3 days before behavioral testing. The experimenter was blind to genotype and drug treatment during behavioral testing. Thermal and mechanical sensitivity were measured as described previously (Zylka et al., 2008). For intrathecal drug delivery, 5 µL was injected into unanesthetized mice using the direct lumbar puncture method (Fairbanks, 2003). The CFA model of inflammatory pain and the SNI model of neuropathic pain were performed as described previously (Shields et al., 2003; Zylka et al., 2008).
S-hPAP (Sigma, P1774) and heat-inactivated S-hPAP were prepared as described previously (Zylka et al., 2008). 18:1 Lysophosphatidic acid (Avanti Polar Lipids, 857130) was dissolved in 0.9% ethanol and then diluted to final concentrations in either HBSS assay buffer (calcium imaging) or 0.9% saline (injections). Adenosine 5’-triphosphate (ATP, Sigma, A26209) was dissolved in either HBSS assay buffer (calcium imaging) or 0.9% saline (injections). Capsaicin (Sigma, 2028 – 1 mg) was dissolved in 0.9% saline/10% ethanol/0.5% Tween 80 and 5 µL was injected for intrathecal delivery, while 20 µL was injected for intraplantar delivery. U73122 (Tocris, 1268) was first dissolved into DMSO, then further diluted in 0.9% saline for i.t. injection. The PI(4,5)P2 Shuttle PIP Kit (Echelon, P-9045) was used to increase PIP2 levels in vivo. PtdIns(4,5)P2 di-C16 was first dissolved into 10% DMSO in 0.9% saline. Carrier 2 (Histone H1) was dissolved into 0.9% saline. Prior to injection, PIP2 and Carrier 2 were mixed in a 1:1 molar ratio and incubated at room temperature for 15 min. Thrombin (Sigma, T4648) was first dissolved to 100 Units/mL in 0.1% BSA and further diluted in HBSS assay buffer to final concentrations. BK was dissolved to 1 mM in DMSO and further diluted in HBSS assay buffer to final concentrations. PTX (Sigma, P7208) and caffeine (Sigma, C0750) were dissolved in water. 8-Cyclopentyl-1,3-dimethylxanthine (CPT) (Sigma, C102), 8-Cyclopentyl-1,3-dipropylxanthine (CPX) (Sigma, C101), SCH58261 (Sigma, S4568), MRS1754 (Sigma, M6316), MRS 1523 (Sigma, M1809), staurosporine (Sigma, S4400), KT5720 (Tocris, 1288), and U73122 (Tocris, 1268) were dissolved in DMSO and further diluted in HBSS assay buffer to final concentrations.
We previously found that S-PAP reduced noxious thermal sensitivity by activating A1R for a sustained three day time period (Zylka et al., 2008; Sowa et al., 2009). Since TRPV1 functions as a noxious heat and capsaicin receptor (Caterina et al., 1997; Caterina et al., 2000), we hypothesized that PAP might reduce thermal sensitivity by inhibiting TRPV1. To test this hypothesis, we transfected Rat1 fibroblasts with TRPV1 and mouse transmembrane PAP (TM-PAP) then measured capsaicin-evoked responses with the calcium indicator Fura-2 AM. We observed that both the amplitude and duration of the capsaicin-evoked calcium response was reduced by 25% in TM-PAP transfected cells relative to cells expressing TRPV1 alone (Figure 1A). Inhibition was dependent on TM-PAP catalytic activity because capsaicin-evoked calcium responses were not inhibited in cells transfected with TM-PAP(H12A), a phosphatase-dead mutant of TM-PAP (Figure 1B). Moreover, the A1R-selective antagonist 8-cyclopentyl-1,3-dipropylxanthine (CPX) blocked the effect of TM-PAP on capsaicin-evoked signaling (Figure 1B; Rat1 cells express A1R, data not shown). Lastly, heat-evoked current density through TRPV1 was reduced in cells expressing TM-PAP relative to cells expressing catalytically inactive TM-PAP(H12A) (Figure 1C). All TM-PAP constructs were fused to the red fluorescent protein mCherry and were expressed at similar levels, but only the H12A mutant lacked catalytic activity (as assessed using enzyme histochemistry, data not shown). Collectively, our data suggest that TM-PAP acts through A1R to reduce capsaicin- and heat-evoked activation of TRPV1.
Next, to determine if PAP reduced noxious thermal sensitivity by acting through TRPV1 in vivo, we measured thermal withdrawal latencies in wild-type (WT) and Trpv1−/− mice before and after i.t. injection of secreted human PAP (S-hPAP). As previously observed (Caterina et al., 2000), there were no significant differences at baseline between WT and Trpv1−/− mice when stimulating the hindpaw with radiant heat (Figure 1D). Following S-hPAP injection, paw withdrawal latency significantly increased (relative to baseline) at the 30 min time point and remained elevated for three days in WT mice (Figure 1D), reproducing previous results (Zylka et al., 2008). In contrast, the thermal antinociceptive effect of S-hPAP was blunted in magnitude (p < 0.001 by two-way ANOVA; relative to WT) and duration (2 days; relative to baseline) in Trpv1−/− mice (Figure 1D; this difference between genotypes was independently reproduced in Figure 1F-control paw). S-hPAP was equally effective at reducing mechanical allodynia in WT and Trpv1−/− mice following CFA-induced inflammation (Figure 1E; black dashed line verses red dashed line), ruling out the trivial possibility that Trpv1−/− mice were less sensitive to all antinociceptive effects of S-hPAP. Although S-hPAP had long lasting thermal antinociceptive effects in WT mice (Figure 1F), we were unable to study the thermal antinociceptive effects of S-hPAP in Trpv1−/− mice because Trpv1−/− mice did not develop thermal hyperalgesia following inflammation (Figure 1F), as previously found by others (Caterina et al., 2000; Davis et al., 2000). Taken together, our cell-based and in vivo data suggest PAP reduces thermal sensitivity, in part, by inhibiting TRPV1.
Our findings raised the question of how sustained, PAP-dependent activation of A1R inhibited TRPV1 at a mechanistic level. A1R stimulation inhibits protein kinase A (PKA) via pertussis toxin (PTX)-sensitive Gαi/o-proteins. In addition, A1R stimulation activates phospholipase C (PLC, including PLCβ3) via PTX-sensitive Gβγ subunits (Murthy and Makhlouf, 1995; Dickenson and Hill, 1998). PLC enzymes then hydrolyze PIP2 in the membrane to diacylglycerol (DAG) and inositol triphosphate (IP3). These facts suggested that sustained activation of A1R might inhibit TRPV1 activity by inhibiting PKA, activating protein kinase C (PKC; via DAG), depleting intracellular calcium stores (via IP3) or depleting PIP2 (via PLC activation). Although TRPV1 can be modulated by PKC and PKA (Bhave et al., 2002; Bhave et al., 2003; Huang et al., 2006), TM-PAP did not inhibit TRPV1 through PKC or PKA pathways (Supplemental Figure S2A, S2B). Nor did TM-PAP deplete intracellular calcium stores (Supplemental Figure S2C).
TRPV1 is also modulated by PIP2 (Prescott and Julius, 2003; Liu et al., 2005; Stein et al., 2006; Lishko et al., 2007; Lukacs et al., 2007; Klein et al., 2008; Rohacs et al., 2008; Yao and Qin, 2009). At high capsaicin concentrations (≥ 1 µM) and in the presence of extracellular calcium, PIP2 is required for TRPV1 channel activity while depletion of PIP2 desensitizes the channel. This requirement for PIP2 suggested TM-PAP might inhibit TRPV1 by reducing cellular levels of PIP2 through sustained A1R activation.
To test this possibility, we quantified the levels of PIP2 in cells using the biosensor PLCδ-PH-GFP (Varnai and Balla, 1998). When PIP2 levels are high, PLCδ-PH-GFP is primarily localized to the plasma membrane (PM). When PIP2 levels are reduced, PLCδ-PH-GFP translocates from the membrane to the cytosol. This translocation can be quantified by measuring the GFP signal intensity on the PM relative to the cytosol (expressed as the ratio PM/Cytosol). We used HEK293 cells for these experiments because this biosensor was difficult to visualize in Rat1 fibroblasts (although we reproduced our key finding in Rat1 cells; Supplemental Figure S3).
In HEK293 cells expressing only PLCδ-PH-GFP, the majority of the GFP signal was in the PM, giving a PM/cytosol ratio of 3.43 ± 0.35 (Figure 2A, 2E). In contrast, PLCδ-PH-GFP was redistributed to the cytosol in cells co-transfected with TM-PAP or PLCβ3 (PM/Cytosol ratio of 1.60 ± 0.06 and 1.70 ± 0.9, respectively) (Figure 2B, 2E). This finding suggested TM-PAP and PLCβ3 depleted PIP2 to a similar extent when expressed for an extended (~24 h) time period. Importantly, the A1R antagonist CPX and the PLC inhibitor U73122 blocked the TM-PAP mediated redistribution of PLCδ-PH-GFP to the cytosol (Figure 2C, 2E), suggesting that TM-PAP depleted PIP2 by acting through A1R and PLC. Additionally, the TM-PAP- and PLCβ3-mediated redistribution of PLCδ-PH-GFP was blocked by overexpressing phosphatidylinositol-4-phosphate-5-kinase (PIPK; Figure 2D, 2E). PIPK increases PIP2 levels in transfected cells (Lin et al., 2005; Milosevic et al., 2005), suggesting TM-PAP- and PLCβ3 altered PLCδ-PH-GFP membrane localization by depleting PIP2.
Next, we genetically manipulated PIP2 levels to determine if increasing or decreasing PIP2 affected capsaicin-evoked calcium responses. Both TM-PAP and PLCβ3 deplete PIP2 to a similar extent (Figure 2E) but only PLCβ3 hydrolyzes PIP2 directly. Likewise, both TM-PAP and PLCβ3 reduced capsaicin-evoked calcium responses to a similar extent (Figure 2F), suggesting indirect or direct depletion of PIP2 was sufficient to reduce TRPV1 activity. Conversely, increasing PIP2 levels by overexpressing PIPK (which regenerates PIP2) blocked TM-PAP- and PLCβ3 from inhibiting capsaicin-evoked calcium responses (Figure 2F). These findings suggested signaling through TRPV1 was reduced as a direct result of PIP2 depletion, consistent with the findings of others using cultured cells (Liu et al., 2005; Stein et al., 2006; Lishko et al., 2007; Lukacs et al., 2007; Klein et al., 2008; Rohacs et al., 2008; Yao and Qin, 2009). In addition, TM-PAP did not affect the capsaicin-evoked calcium response in cells expressing TRPV1Δ42(777–820) (data not shown), a TRPV1 mutant that is missing a putative PIP2 binding domain (Prescott and Julius, 2003; Kwon et al., 2007; Kim et al., 2008). Taken together, these data show that TM-PAP reduces TRPV1 activity in vitro through sustained activation of A1R and subsequent depletion of PIP2.
Since PAP regulated PIP2 levels in cultured cells, we hypothesized PAP might also regulate PIP2 levels in vivo. To test this hypothesis, we measured PIP2 levels in lumbar (L)3-L6 DRG from WT mice that were injected (i.t.) with vehicle or S-hPAP. We collected DRG one day after injections because S-hPAP has maximal A1R-dependent antinociceptive effects at this time (Zylka et al., 2008). Injection of S-hPAP significantly reduced PIP2 levels by 40 ± 9 % (Figure 3A) and this reduction was dependent on A1R activation [as evidenced by the observation that PIP2 levels in L3-L6 DRG were not significantly different in A1R−/− mice injected (i.t.) one day earlier with 250 mU S-hPAP (250 mU) relative to heat inactivated S-hPAP (0 mU); 159.6 ± 23.5 pmol/mg DRG verses 174.4 ± 28.0 pmol/mg DRG, respectively. n=4 mice per condition]. Conversely, PIP2 levels were elevated by 89 ± 23 % in DRG from Pap−/− mice (Figure 3A). Taken together, these data suggest PIP2 levels are inversely related to the amount of PAP ectonucleotidase activity and A1R stimulation.
Since our cell based data suggested that PAP depleted PIP2 by activating PLC, we next evaluated whether the thermal antinociceptive effect of S-hPAP could be blocked using the PLC inhibitor U73122. This inhibitor was previously injected i.t. (at 5.4 nmol) to block PLC activation by a δ-opioid receptor ligand (Narita et al., 2000). Indeed, i.t. injection of U73122 at the same dose transiently and completely blocked the thermal antinociceptive effect of S-hPAP, providing evidence that S-hPAP acted through PLC to reduce thermal sensitivity in vivo (Figure 3B, 3C). Importantly, the 5.4 nmol dose had no effect on thermal sensitivity when injected alone (Fig. 3B, C) (Narita et al., 2000). However, a higher U73122 dose (12.5 nmol) transiently enhanced thermal sensitivity (Supplemental Figure S4). Since strong PLC inhibition would be predicted to elevate PIP2 levels, this latter result suggests that the PLC/PIP2 pathway tonically modulates thermal thresholds.
To directly assess whether S-hPAP reduced thermal sensitivity by depleting PIP2, we transiently replenished PIP2 in lumbar DRG using a PIP2 shuttle (Ozaki et al., 2000). We found that injection (i.t.) of PIP2 (complexed with carrier) transiently (2 hr) elevated PIP2 levels well above normal levels in lumbar DRG (Figure 3D) whereas injection of carrier alone (the control) had no effect (based on the observation that PIP2 levels in L3-L6 DRG were not significantly different in WT mice relative to WT mice injected i.t. with Car; 112.0 ± 14.6 pmol/mg DRG verses 102.5 ± 10.5 pmol/mg DRG, respectively. N=4 mice per condition). Strikingly, i.t. injection of PIP2 (complexed with carrier) also transiently (2 hr) reversed S-hPAP-mediated thermal antinociception whereas carrier alone had no effect (Figure 3E, 3F). The fact that behavior was significantly altered only during the time at which PIP2 was significantly elevated makes it unlikely these effects on thermal sensitivity were a coincidence. Instead, these results strongly suggest that PAP inhibits thermal sensitivity as a direct result of PIP2 depletion. Moreover, control animals injected with PIP2 displayed transient thermal hyperalgesia, suggesting thermal sensitivity can be enhanced when PIP2 levels are elevated above normal levels. The magnitude of this effect on thermal sensitivity in control animals was smaller (2.6 s) than in animals that were injected with S-hPAP and PIP2 (4.0 s). This argues that PIP2 replenishment was sufficient to block the thermal antinociceptive effect of S-hPAP independent of how PIP2 affects thermal sensitivity in control animals.
Taken together, our data strongly support a mechanism (Figure 3G) where (1) TM- and S-PAP function as ectonucleotidases that generate adenosine. Adenosine then stimulates (2) A1R in a sustained fashion, followed by (3) PLC activation and (4) PIP2 hydrolysis. (5) Sustained reductions in PIP2 levels decreased TRPV1 activity and decreased noxious thermal sensitivity. In addition, our data suggest PIP2 levels are regulated by tonic ectonucleotidase-dependent adenosine production and A1R activation.
Diverse chemicals are released upon injury and inflammation and sensitize nociceptive neurons, in many cases, by activating pronociceptive GPCRs (Hucho and Levine, 2007; Basbaum et al., 2009). Since many pronociceptive receptors signal through PLC (and hence require PIP2 for signaling), we hypothesized that PAP might reduce signaling through pronociceptive receptors via sustained activation of A1R and PIP2 depletion. Such a mechanism could have important physiological implications because reduced signaling through pronociceptive receptors would be predicted to reduce nociceptive sensitization, a key symptom of chronic pain. LPA is a pronociceptive ligand that sensitizes nociceptive neurons, causes long-lasting (>7 day) sensitization in vivo (including thermal hyperalgesia and mechanical allodynia) and is implicated in neuropathic pain mechanisms (Elmes et al., 2004; Inoue et al., 2004; Park and Vasko, 2005). In addition, LPA receptors are coupled to Gαq/11 proteins and signal through PLC in many cell-types, including Rat1 fibroblasts (which endogenously express LPA receptors) (Mills and Moolenaar, 2003; Kelley et al., 2006).
Using Rat1 fibroblasts, we found that the amplitude and duration of LPA-evoked calcium responses were significantly reduced in cells expressing TM-PAP relative to untransfected cells in the same field of view (Figure 4A, 4D). This “PAP effect” was species-conserved as cells transfected with human TM-PAP (TM-hPAP) were also less responsive to LPA stimulation (Figure 4B, 4D). In contrast, the LPA-evoked calcium response was not reduced in cells transfected with the catalytically inactive TM-PAP(H12A) mutant (Figure 4C, 4D).
To determine if TM-PAP inhibited LPA-evoked signaling by generating adenosine and activating A1R, we assessed whether PTX (an inhibitor of Gαi/o-coupled receptors) or adenosine receptor antagonists could block the effect of TM-PAP on LPA-evoked signaling. We found that PTX completely blocked the PAP effect, as evidenced by no significant differences between untransfected cells and TM-PAP + PTX-treated cells (Figure 4D). Additionally, the PAP effect was blocked by the A1/A2B adenosine receptor antagonist caffeine (Caff) and by two different A1R-selective antagonists: CPT (8-cyclopentyl-1,3-dimethylxanthine) and CPX (Figure 4D). In contrast, selective antagonists of all other adenosine receptors (A2AR: SCH 58261; A2BR: MRS 1754; A3R: MRS 1523) did not block the PAP effect (data not shown).
Next, we evaluated whether increasing or decreasing PIP2 affected LPA-evoked calcium responses in our cell-based assay. We found that both TM-PAP and PLCβ3 reduced LPA-evoked calcium responses to a similar extent (Figure 4D), suggesting indirect or direct depletion of PIP2 was sufficient to reduce signaling. This is consistent with a previous study showing that LPA-evoked Ca2+ responses were reduced to baseline levels when PIP2 was depleted using an inducible PIP2 phosphatase (Varnai et al., 2006). Conversely, increasing PIP2 levels by overexpressing PIPK blocked the TM-PAP- and PLCβ3-mediated reduction in LPA-evoked calcium responses (Figure 4D). Moreover, restoring PIP2 levels with the PIP2 shuttle blocked the inhibitory effect of TM-PAP (Supplemental Figure S5). These experiments provide complimentary support that TM-PAP inhibits LPA receptor signaling as a direct result of PIP2 depletion. The inhibitory effect of TM-PAP was also blocked with the PLC inhibitor U73122 (Figure 4D), further indicating that TM-PAP acts through PLC to deplete PIP2. As expected, the magnitude of the LPA-evoked Ca2+ influx was smaller in all cells when PLC was inhibited. Lastly, TM-PAP did not reduce LPA signaling by acting through other pathways that are downstream of A1R, including Gαi/o-mediated inhibition of PKA or DAG-mediated PKC activation (data not shown).
Since many pronociceptive receptors are Gαq/11-coupled and signal via PLC, we hypothesized that TM-PAP might reduce signaling through additional pronociceptive receptors, including purinergic receptors (using the ligand adenosine 5’-triphosphate, ATP), protease activated receptors (using the ligand Thrombin, Thr) and bradykinin (BK) receptors. Importantly, activation of these receptors evokes transient calcium responses in Rat1 cells and causes nociceptive sensitization in vivo (Kelley et al., 2006; Wang et al., 2006; Burnstock, 2007; Sawynok, 2007; Dale and Vergnolle, 2008). Strikingly, calcium responses induced by all three ligands were reduced in TM-PAP transfected cells relative to untransfected cells, and these reductions were blocked by the A1R antagonist CPX and by overexpressing PIPK (Figure 4E).
Collectively our data further support a mechanism (Figure 4F) where (1) PAP functions as an ectonucleotidase to generate adenosine over a sustained time period. (2) Adenosine then stimulates A1R followed by (3) PLC activation and (4) PIP2 hydrolysis. With less PIP2 available for (5) Gαq/11/PLC-mediated receptor signaling, less IP3 (and DAG) is generated following receptor stimulation, resulting in smaller pronociceptive ligand-evoked calcium responses.
Since PAP reduced pronociceptive receptor signaling in Rat1 cells by depleting PIP2, we hypothesized that PAP might also reduce signaling through pronociceptive receptors in vivo. To test this possibility, we took advantage of the fact that both LPA and ATP produce long lasting (>7 day) thermal hyperalgesia and mechanical allodynia when injected i.t. (Inoue et al., 2004; Nakagawa et al., 2007). This is longer than the three day antihyperalgesic and antiallodynic effects of S-hPAP (Zylka et al., 2008). Thus, by injecting S-hPAP one day before ATP or LPA, we could ascertain whether S-hPAP reduced initiation of LPA- or ATP-evoked signaling by quantifying hyperalgesia and allodynia on days four and eight (i.e. after the three day antinociceptive effects of PAP wore off).
First, we measured baseline (BL) thermal and mechanical sensitivity in two groups of WT mice. Next, we injected S-hPAP (i.t.) into one of the groups and heat inactivated S-hPAP (control; catalytically dead) into the other group (Figure 5; a third group of S-hPAP injected mice from a different experiment is shown, to provide a visual reference for how S-hPAP typically affects naïve mice). One day later, S-hPAP increased paw withdrawal latency to the noxious thermal stimulus but had no effect on mechanical sensitivity, whereas inactive S-hPAP had no effects on thermal or mechanical sensitivity (Figure 5). These results were expected (Zylka et al., 2008) and confirmed that mice received an active or inactive dose of S-hPAP. After taking these measurements, we injected (i.t.) either 5 nmol LPA (Figure 5A, 5B) or 100 nmol ATP (Figure 5C, 5D). These doses produce maximal sensitization in animals (Inoue et al., 2004; Nakagawa et al., 2007). Both pronociceptive compounds produced long-lasting thermal hyperalgesia and mechanical allodynia in the control (inactive S-hPAP-injected) mice. In contrast, mice injected with S-hPAP and then LPA/ATP were significantly different from controls at all times post LPA/ATP injections and did not develop thermal hyperalgesia or mechanical allodynia, as evidenced by latencies and thresholds that were at or near baseline levels on days four and eight. These data suggest that PAP, an enzyme that reduces PIP2 levels in vivo (Figure 3A), blocked physiologically-relevant signaling and sensitization through two distinct pronociceptive receptors. Conversely, thermal hyperalgesia and mechanical allodynia were significantly and enduringly enhanced following LPA or ATP injection in Pap−/− mice (Figure 6)—that is, mice which have elevated levels of PIP2 in lumbar DRG (Figure 3A).
Although LPA and ATP produce long-lasting sensitization in vivo, injection of these chemicals may not fully model the sensitization and pathology that is associated with chronic pain conditions. To determine if reducing PIP2 levels with S-hPAP had a more generalized effect on the signals that initiate pain sensitization, we tested S-hPAP in the CFA model of inflammatory pain and in the spared nerve injury (SNI) model of neuropathic pain. Strikingly, i.t. injection of S-hPAP prior to CFA-induced inflammation nearly eliminated thermal hyperalgesia and significantly reduced mechanical allodynia for the duration of the experiment compared to controls injected with inactive S-hPAP (Figure 7A, 7B). These preemptive antinociceptive effects of S-hPAP were dependent on A1R activation (Supplemental Figure S6). In addition, i.t. injection of S-hPAP prior to nerve injury eliminated thermal hyperalgesia and significantly reduced mechanical allodynia for the duration of the experiment compared to mice injected with inactive S-hPAP (Figure 7C, 7D). Collectively, these findings indicate that S-hPAP enduringly blocks sensitization in two chronic pain models when injected before inflammation/injury.
Since PIP2 levels in DRG were indirectly elevated and sensitization was enhanced in Pap−/− mice (data above and in (Zylka et al., 2008)), we next sought to determine if direct elevation of PIP2 in DRG could enhance sensitization. To accomplish this, we co-injected (i.t.) WT mice with LPA + Car (the control) or LPA + PIP2 + Car then measured noxious thermal and mechanical sensitivity for several days. Importantly, there is a critical window of 3 hr over which LPA (injected i.t.) signals to produce nociceptive sensitization in mice (Ma et al., 2009). Since PIP2 levels are only elevated in DRG for 2 hr post injection (Figure 3D), these co-injection experiments allowed us to elevate PIP2 levels only when LPA receptors were active in vivo. Strikingly, when PIP2 levels were elevated coincident with LPA receptor activation, thermal hyperalgesia and mechanical allodynia were significantly and reproducibly enhanced for the duration of the experiment (Figure 8A, 8B). In contrast, injection of Car alone or PIP2 + Car in the absence of a pronociceptive stimulus had no long-term effects on thermal or mechanical sensitivity (Figure 8C, 8D). And, injection of Car alone or PIP2 + Car did not sensitize mice when injected three days after LPA was injected (i.e. well past the 3 hr critical window for LPA receptor signaling; Figure 8E, 8F). In this same experiment, PIP2 + Car (but not Car alone) caused a transient (2 hr) enhancement in thermal sensitivity (data not shown), thus reproducing our findings above (Figure 3E, 3F) and confirming that these PIP2 injections were successful and had the capacity to affect behavior. Taken together, these experiments strongly argue that PIP2 levels must be elevated in DRG at the time of pronociceptive receptor activation to enhance thermal and mechanical sensitization.
Lastly, we sought to determine if direct elevation of PIP2 could enhance thermal and mechanical sensitization caused by CFA. For this experiment, all mice were injected i.t. with Car or PIP2 + Car immediately before injecting CFA into one hindpaw. CFA sensitizes nociceptors through the release of an “inflammatory soup” composed of diverse pronociceptive ligands (Basbaum et al., 2009). Since this soup could activate pronociceptive receptors for an extended period of time (and since PIP2 is only elevated for 2 hr after a single injection), we re-injected (i.t.) all mice 2 hr later with PIP2 + Car (or Car alone). This ensured that PIP2 levels remained elevated while CFA “initiated” sensitization. Strikingly, CFA-induced thermal hyperalgesia and mechanical allodynia were significantly enhanced for the duration of the experiment when PIP2 levels were transiently elevated (Figure 8G, 8H; compare the inflamed paw of PIP2 + Car injected mice to the inflamed paw of Car alone (control) mice). In contrast, thermal and mechanical responses were not altered in the contralateral (uninflamed) paws of mice injected with PIP2 + Car (Figure 8G, H), further demonstrating that acute elevation of PIP2 does not sensitize mice in the absence of a pronociceptive stimulus.
In our effort to determine how PAP regulated nociception at a mechanistic level, we found that sustained A1R activation reduced the levels of PIP2 in cells and in DRG. This reduction in PIP2 reduced noxious thermal sensitivity, in part through inhibition of TRPV1. And, this reduction in PIP2 enduringly reduced nociceptive sensitization to thermal and mechanical stimuli.
Numerous studies found that TRPV1 can be modulated by PIP2 in vitro (reviewed in (Rohacs et al., 2008)). At low capsaicin concentrations and in the absence of extracellular calcium, PIP2 partially inhibits TRPV1 (Prescott and Julius, 2003). In contrast, at high capsaicin concentrations and in the presence of extracellular calcium, PIP2 is required for TRPV1 activity (Liu et al., 2005; Stein et al., 2006; Lishko et al., 2007; Lukacs et al., 2007; Klein et al., 2008; Yao and Qin, 2009). These contrasting in vitro results made it difficult to predict how alterations in the levels of PIP2 might affect thermosensation in animals. In our present study, we found that PAP inhibited TRPV1 in cultured cells through sustained A1R activation and PLC-mediated PIP2 depletion. Likewise, PAP inhibited thermosensation in mice through A1R activation (Zylka et al., 2008), PLC activation (Figure 3B, 3C) and A1R-dependent PIP2 depletion (Figure 3A, and data above). Moreover, the inhibitory effect of PAP on thermosensation was partially dependent on TRPV1 activation (Figure 1D). Conversely, thermosensation was modestly enhanced when PIP2 levels were elevated in vivo (Figure 3E, 3F). Taken together, our findings suggest that PIP2 facilitates thermosensation in vivo through TRPV1-dependent and independent mechanisms. Our findings are consistent with a study showing that TRPV1-dependent thermosensation was enhanced through interactions with PIRT, a PIP2-binding protein (Kim et al., 2008).
Pronociceptive GPCR activation can sensitize TRPV1 via PKC/PLC (Bhave et al., 2003; Huang et al., 2006). Since A1R is also coupled to PLC (Murthy and Makhlouf, 1995; Jacobson and Gao, 2006), this raises the question of why PAP inhibits TRPV1 upon A1R activation, as we observed, instead of sensitizing TRPV1? This likely reflects differences in how pronociceptive GPCRs and A1R couple to downstream signaling pathways. Pronociceptive receptors, including LPA receptors, are coupled to PLC isoforms via Gαq/11 and Gβγ proteins. Stimulation of these receptors evokes transient PKC activation and large PLC-dependent calcium responses that desensitize rapidly (Mills and Moolenaar, 2003; Kelley et al., 2006). In contrast, A1R is coupled to PLC isoforms exclusively via Gβγ proteins and A1R does not desensitize, at least when activated by ectonucleotidase-generated adenosine. Specifically, we found that A1R does not desensitize when activated for a sustained 3 day time period by PAP (Zylka et al., 2008; Sowa et al., 2009) or upon repeated injection of S-hPAP (data not shown). And importantly, the PLC inhibitor U73122 transiently inhibited PAP antinociception (Figure 3B, 3C) arguing that PLC is active over this 3 day period. These data suggest that ectonucleotidase-dependent activation of A1R is sufficient to deplete PIP2 and inhibit TRPV1 but is not sufficient to activate PKC, sensitize TRPV1 or detectably elevate calcium levels. Indeed, we found that PAP did not inhibit TRPV1 through PKC (Supplemental Figure S2A) nor did PAP reduce intracellular (IP3-sensitive) calcium stores (Supplemental Figure S2C). And, acute stimulation with the A1R agonist N6-cyclopentyladenosine (300 nM) did not evoke Ca2+ influx in cultured small diameter DRG neurons from adult mice (Eric McCoy and Zylka, data not shown).
We found that most of the thermal antinociceptive effects of PAP were lost in Trpv1−/− mice while the mechanical antinociceptive effects of PAP were preserved (Figure 1D–F). This dissociation suggests that most of the thermal antinociceptive effects of PAP were mediated through TRPV1, while the remaining thermal and mechanical antinociceptive effects of PAP were likely mediated by other PIP2 sensitive channels or proteins. Depletion of PIP2 generally reduces ion channel activity, including KCNQ, P2X4, and N-type calcium channels (Gamper et al., 2004; Suh and Hille, 2005; Bernier et al., 2008), all of which could affect nociception (Basbaum et al., 2009). In addition, PIP2 depletion inhibits synaptic vesicle exocytosis (Di Paolo and De Camilli, 2006). Further studies will be needed to determine if additional antinociceptive effects of PAP are due to inhibition or modulation of other PIP2 sensitive channels, proteins or mechanisms.
We unexpectedly found that nociceptive sensitization could be enduringly altered by manipulating the levels of PIP2 at the time of stimulation/injury. This suggests that any manipulation, be it genetic or environmental, which alters PIP2 levels could have a lasting impact on nociceptive sensitization. Indeed, we found that thermal hyperalgesia and mechanical allodynia were enduringly enhanced only when PIP2 levels were elevated coincident with LPA receptor activation. PIP2 injection alone or PIP2 injection several days after LPA injection did not produce, enhance or prevent sensitization. Direct elevation of PIP2 in DRG also enhanced sensitization caused by peripheral injection of CFA, further supporting the physiological importance of PIP2 in nociceptive sensitization. Intriguingly, these findings also hint that PIP2 levels in DRG regulate sensitization regardless of whether pain-producing stimuli are administered centrally (i.e. LPA) or peripherally (i.e. CFA).
Likewise, we found that sensitization could be enduringly altered by increasing or decreasing PIP2 levels through manipulation of PAP activity. Pap−/− mice have elevated PIP2 levels in DRG and enhanced LPA-, ATP-, CFA- and nerve injury-induced nociceptive sensitization (data above and in (Zylka et al., 2008)). Conversely, S-hPAP injections reduced PIP2 levels in DRG and enduringly inhibited LPA-, ATP-, CFA- and nerve injury-induced nociceptive sensitization. S-hPAP acted exclusively through A1R to mediate these enduring effects (Supplemental Figure S6), ruling out the possibility that PAP acted through other adenosine receptor subtypes, including A2A (Loram et al., 2009). Importantly, these alterations in nociceptive sensitization outlasted the three-day antinociceptive effects of S-hPAP and the acute (2 hr) elevation of PIP2, arguing that neither maintained activity of PAP nor long-term elevations of PIP2 contributed to these enduring effects. Instead, these long-term changes in nociceptive sensitization could be due to reduced or enhanced engagement of transcriptional and non-transcriptional mechanisms that are downstream of pronociceptive receptor/PLC stimulation (Ji et al., 2009).
Intrathecal injections target DRG and spinal cord (Luo et al., 2005). This raises the question of precisely where S-hPAP and PIP2 act to regulate nociception. Since S-hPAP (i.t.) regulates PIP2 levels and nociception through A1R (data above)(Zylka et al., 2008), S-hPAP has the potential to act on any cell that expresses A1R. This includes peptidergic and nonpeptidergic nociceptive neurons as well as postsynaptic neurons in the spinal cord (Reppert et al., 1991; Li and Perl, 1994; Lao et al., 2001; Schulte et al., 2003) but excludes microglial cells because they do not express A1R (Orr et al., 2009). Despite repeated attempts, we were unable to determine at a cellular level which DRG neurons incorporated PIP2 (by studying the distribution of PIP2 conjugated to fluorochromes, data not shown). However, nociception was only affected over the time period in which PIP2 levels were elevated in DRG, arguing that nociceptive neurons in DRG were targeted. Furthermore, the elevated levels of PIP2 in DRG from Pap−/− mice are likely to be restricted to the subset of peptidergic and nonpeptidergic nociceptive neurons that normally express PAP (Zylka et al., 2008). We cannot further pinpoint at the cellular level where S-hPAP and PIP2 act to regulate nociception with existing technologies.
Lastly, PIP2 levels can be increased or decreased relative to an intermediate level in DRG (Figure 3A, 3D), suggesting nociception may be influenced by an underlying “phosphoinositide tone”. This tone may be coordinated with the “adenosine tone” that is present in diverse tissues, including the nervous system (Boison, 2008). Our data indicate that adenosine-generating ectonucleotidases like PAP contribute to this underlying phosphoinositide tone both positively and negatively. It may thus be possible to harness any molecule or mechanism that causes a sustained reduction in this tone to enduringly prevent or treat symptoms associated with chronic pain.
We thank Sang-Kyou Han for PLCβ3 (EGFP- and FLAG-tagged versions), Tamas Balla for PLCδ-PH-GFP, Ira Milosevic for PIPK (mRFP-PIPK1-γ), David Julius for TRPV1-GFP and Xinzhong Dong for TRPV1Δ42 and HEK293-TRPV1 stable cells, Yvette Chuang, Bonnie Taylor-Blake and Eric McCoy for expert technical assistance and Paul Farel, Ben Philpot and Anthony LaMantia for comments on the manuscript. This work was supported by grants to M.J.Z. from The Sloan Foundation, The Searle Scholars Program, The Klingenstein Foundation, The Whitehall Foundation, Rita Allen Foundation and NINDS (R01NS060725) and grants to P.V. from The Sigrid Juselius Foundation, The Finnish Cancer Foundation and The Research Council for Medicine of the Academy of Finland. N.A.S. was supported by NINDS (F30NS063507) and a MSTP grant (T32GM008719). Confocal imaging was performed at the UNC-CH Confocal Imaging Facility, which is co-funded by NINDS and NICHD (P30NS045892). Calcium imaging was performed at the Michael Hooker Microscopy Facility, which is funded by an anonymous private donor. M.J.Z. is a Rita Allen Foundation Milton E. Cassel Scholar.