Received 6 January 2006; Accepted 15 September 2006; Published online: 22 October 2006.

GTP cyclohydrolase and tetrahydrobiopterin regulate pain sensitivity and persistence

Irmgard Tegeder1, 2, 10, Michael Costigan1, 10, Robert S Griffin1, Andrea Abele2, Inna Belfer3, 4, Helmut Schmidt2, Corina Ehnert2, Jemiel Nejim4, 9, Claudiu Marian2, Joachim Scholz1, Tianxia Wu4, Andrew Allchorne1, Luda Diatchenko5, Alexander M Binshtok1, David Goldman3, Jan Adolph2, Swetha Sama5, Steven J Atlas7, William A Carlezon8, Aram Parsegian8, J"orn L"otsch2, Roger B Fillingim6, William Maixner5, Gerd Geisslinger2, Mitchell B Max4 & Clifford J Woolf1

1 Neural Plasticity Research Group, Department of Anesthesia and Critical Care, Massachusetts General Hospital & Harvard Medical School, 149 13th Street, Room 4309, Charlestown, Massachusetts 02129, USA.

2 Pharmazentrum Frankfurt, Institut f邦r Klinische Pharmakologie / Zentrum f邦r Arzneimittelforschung, Entwicklung und Sicherheit, Klinikum der Johann Wolfgang Goethe-Universit"at, Theodor Stern Kai 7, Frankfurt am Main, 60590, Germany.

3 Laboratory of Neurogenetics, National Institute on Alcohol Abuse and Alcoholism, National Institutes of Health, Department of Health and Human Services, 5625 Fishers Lane, Room 3S-32, Rockville, Maryland 20852, USA.

4 National Institute of Dental and Craniofacial Research, National Institutes of Health, Department of Health and Human Services, 10 Center Drive, Building 10, Room 3C-405, Bethesda, Maryland 20892, USA.

5 Center for Neurosensory Disorders, School of Dentistry, 2110 Old Dental Building, CB# 7455 University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599-7455, USA.

6 University of Florida College of Dentistry, Community Dentistry and Behavioral Science, 1329 SW 16th Street, Gainesville, Florida 32608, USA.

7 General Medicine Division and the Clinical Epidemiology Unit, Massachusetts General Hospital & Harvard Medical School, 15 Parkman Street, WAC 615, Boston, Massachusetts 02114, USA.

8 Department of Psychiatry, Harvard Medical School, McLean Hospital, 115 Mill Street, Belmont, Massachusetts 02478, USA.

9 Howard Hughes Medical Institute每National Institutes of Health Research Scholars Program, 1 Cloister Court, Building 60, Bethesda, Maryland 20892-1460, USA.

10 These authors contributed equally to this work.

Discussion

Tetrahydrobiopterin synthesis increased in rat sensory neurons in response to both axonal injury and peripheral inflammation. Blocking the increased BH4 synthesis by independently inhibiting two successive enzymes in the synthesis cascade reduced neuropathic and inflammatory pain, and in contrast, BH4 administration produced pain in naive animals and enhanced inflammatory and neuropathic pain sensitivity. Furthermore, a haplotype of GCH1 that reduces its upregulation in response to a forskolin challenge was protective against persistent neuropathic pain and was associated with reduced sensitivity to experimental pain in humans. We conclude that we have identified both a previously unknown pathway involved in the production and modulation of pain and a genetic marker of pain sensitivity.

Regulation of BH4 concentrations is normally achieved by a feedback inhibition mediated by BH4 acting in concert with the GTP cyclohydrolase feedback regulatory protein (GFRP). GFRP, however, unlike GTP cyclohydrolase, is not upregulated after nerve injury (QRT-PCR data not shown). This altered stoichiometry may disrupt efficient feedback inhibition, resulting in accumulation of excess BH4 in DRG neurons, and thereby increasing nitric oxide and possibly biogenic amine neurotransmitter synthesis. Seven days after SNI, nitric oxide concentrations were elevated in the DRG, suggesting that overproduction of NO contributes to the pain evoked by BH4. Pain-producing effects of nitric oxide probably involve direct nitrosylation of target proteins34, modulation of NMDA receptor activity35 and activation of the guanylyl cyclase每cyclic GMP每PKG pathway36, 37, resulting in increased glutamatergic transmission38. Supporting this, inhibition of GTP cyclohydrolase prevented the increases in both BH4 and NO, and NOS inhibition reduced mechanical and cold allodynia after SNI. BH4 may act in a paracrine as well as an autocrine fashion, as it is released from neurons39, and may both increase enzymatic activity and produced cofactor-independent effects40, 41. Considering the latter, we found that BH4 produced a short-latency calcium influx in cultured adult DRG neurons that is partly mediated through nitric oxide synthesis. Although neuronal tryptophan hydroxylase mRNA was upregulated in DRG neurons after SNI, serotonin concentrations remained below detection limits in this tissue. In the spinal cord serotonin is expressed in descending inhibitory and excitatory fibers. DAHP treatment did not, however, significantly reduce serotonin concentrations in the spinal cord and brain stem (data not shown) or alter performance in the forced water swim test. This model of anxiety and depressive behavior is sensitive to changes in serotonin abundance26. We suggest, therefore, that changes in serotonin production do not contribute to BH4-mediated increases in pain sensitivity. Because BH4 produces pain rapidly, it is likely that these immediate effects do not involve changes in transcription, activation of microglia42 or induction of neuronal cell death9. Moreover, the efficacy of DAHP in the formalin test, peripheral inflammation and multiple models of neuropathic pain points to a common BH4-dependent mechanism in diverse pain conditions.

To evaluate the potential role of BH4 in human pain, we analyzed whether polymorphisms in the rate-limiting BH4-synthesizing enzyme GCH1 are associated with specific pain phenotypes. If BH4 is absent or very substantially reduced in humans due to rare missense, nonsense, deletion or insertion mutations in the coding regions of GTP cyclohydrolase43 or sepiapterin reductase genes, DOPA-responsive dystonia and other severe neurological problems occur19, 20. These neurological diseases are caused by dopamine or serotonin neurotransmitter deficiencies that result from the lack of BH4 as a cofactor for the enzymes that synthesize these transmitters. The homozygotes for the pain-protective haplotype in our study did not have any neurological diseases. We therefore speculated that the pain-protective haplotype embodies a variation in a regulatory site that causes a more modest impairment in GTP cyclohydrolase production or function. In support of this, we found that constitutive expression of GTP cyclohydrolase and BH4 production was equivalent in cells of carriers and noncarriers of the pain-protective haplotype. However, forskolin-evoked upregulation was significantly smaller in carriers of the pain-protective haplotype. Although the precise locations mediating the regulation of GCH1 transcription have still to be determined, they likely involve elements in the region 5' to exon 1 and within the large 20 kb intron 1, because the SNPs found exclusively in the pain-protective haplotype are located in the putative promoter region of GCH1 (C.-9610G>A) and in intron 1 (C.343+8900A>T), respectively. These SNPs might modify the efficiency of transcriptional modulation by signals mediated by cyclic AMP每dependent transcription factors. Although hundreds of transcripts are regulated in DRGs by nerve injury or sustained nociceptor stimulation, and although many chemical agents and biological molecules affect pain behavior in experimental settings, only a few genes have been identified so far that modulate pain sensitivity in humans11, 44, 45. The current finding for GCH1 is one of the first to be replicated across three independent human study populations.

Our results demonstrate that alterations in the concentration of the essential enzyme cofactor BH4 modify the sensitivity of the pain system, and that SNPs in the gene for the rate-limiting BH4-producing enzyme GTP cyclohydrolase alter both responses in healthy humans to noxious stimuli and the susceptibility of patients to the development of persistent neuropathic pain. Because the pain-protective haplotype in GCH1 is associated with a reduction in the risk of developing persistent pain without signs of dystonia, a treatment strategy that could reduce excess de novo synthesis in the DRG of BH4, but not constitutive concentrations of BH4, by targeting only induction of GTP cyclohydrolase or by leaving the recycling pathway intact, might prevent the establishment or maintenance of chronic pain. In addition, the identification of a predictor of the intensity and chronicity of pain will be a useful tool to assess an individual's risk for developing chronic pain. The effect of the pain-protective haplotype on both experimental and persistent pain, and the involvement of BH4 in both inflammatory and neuropathic pain, may explain why sensitivity to acute experimental pain is a predictor of postsurgical and eventually chronic pain46, 47.

Methods

Nociceptive models

For the SNI model two branches of the rat sciatic nerve, the common peroneal and the tibial, were ligated and sectioned distally. In the CCI model the sciatic nerve was constricted with three ligatures; in the SNL model the L5 spinal nerve was tightly ligated. For the formalin test, 50米l of 5% formaldehyde solution were injected into a hindpaw and flinches were counted per minute up to 60 min. Paw inflammation was induced with 50米l complete Freund's adjuvant (CFA) injected into a hindpaw. Nociceptive analysis was conducted blinded and animals were fully habituated to the room and test cages. Mechanical allodynia was assessed with graded-strength monofilament von Frey hairs (0.0174每20.9 g, log scaled), cold allodynia with the acetone test and heat hyperalgesia with the Hargreaves test. Drugs (Sigma) were injected i.p. or i.t. through a spinal catheter; osmotic pumps were used for infusion. Control animals received vehicle. L4-5 DRG and spinal cord tissue was processed for QRT-PCR, western blotting, in situ hybridization and immunofluorescence studies (Supplementary Methods online). Animal procedures were approved by the Committee on Research Animal Care of the Massachusetts General Hospital.

Concentrations of DAHP, neopterin and biopterin

Concentrations were determined by liquid chromatography coupled to tandem mass spectrometry on a tandem quadrupole mass spectrometer (PE Sciex API 4000; Applied Biosystems). Biopterin and neopterin analysis is described at Nature Protocols online (DOI: 10.1038/nprot.2006.298); DAHP analysis is described in Supplementary Methods.

Electrophysiology and calcium imaging

Miniature excitatory postsynaptic currents were recorded at -70 mV by whole-cell patch clamp in adult rat transverse spinal cord slices48. [Ca2+]i was measured fluorometrically as the ratio of the absorbances at 340 and 380 nm (忖F 340/380) in cultured adult DRG neurons loaded with the Ca2+-sensitive fluorescent dye Fura-2. BH4 (0.3每10米M), DEA-NONOate (50米M) and L-NAME (10每100米M) were applied using a multibarrel fast drug-delivery system.

Immortalization of leukocytes and forskolin stimulation

Peripheral blood lymphocytes were immortalized with EBV transfection. WBCs were stimulated with PHA in RPMI medium; EBV was then added and cells were incubated at 37 ~C, 4.5% CO2, 90% relative humidity. Immortalized cells were stimulated with 10米M forskolin for 12 h.

Data analysis.

Data are means ㊣ s.e.m. The number of animals per group was 9每12. Areas under the effect-versus-time curves were calculated using the linear trapezoidal rule and compared with Student's t-test or univariate analysis of variance (ANOVA) with subsequent t-tests employing a Bonferroni 汐-correction for multiple comparisons. All other data were analyzed with univariate ANOVA or ANOVA for repeated measurements. P at 0.05 was considered significant for all tests.

Human genetic studies

We genotyped 15 single nucleotide polymorphisms (SNPs) spaced evenly through GCH1 using the 5' exonuclease method (primer sets and probes in Supplementary Table 3 online). GCH1 haplotypes were identified in silico using the SAS/genetics software package (SAS Institute, Inc.), which implements a modified expectation-maximization algorithm to reconstruct haplotypes from population genotype data. Linkage disequilibrium between SNPs was used to describe the nonindependence of alleles (Supplementary Fig. 4).

Chronic lumbar root pain

We collected DNA from 168 Caucasian adults who participated in a prospective observational study of surgical diskectomy for persistent lumbar root pain29. The primary endpoint was persistent leg pain over the first postoperative year, using four 'leg pain' variables (details in Supplementary Methods) normalized to z-scores with mean 0 and standard deviation 1. The primary pain outcome variable for each individual was the mean of these four z-scores. There were 147 subjects who completed the 1-year questionnaire. Genotype-phenotype associations for each SNP were sought by regression analysis. The covariates were a number of demographic, psychological and environmental factors, including sex, age, worker's compensation status, delay in surgery after enrollment and Short-Form 36 general health scale. Stepwise regression31 was applied to assess the association between pain scores and haplotypes with frequencies >1%, obtained from the Ensemble database v.38, April 2006. These haplotypes accounted for 94% of chromosomes studied. If a haplotype was identified to be significantly (P < 0.05) associated with pain scores, phenotype-diplotype association analysis was performed by regression analysis. The collection of DNA and genetic analyses were carried out with the approval of the National Institute of Dental and Craniofacial Research institutional review board and informed consent was obtained from all subjects.

Experimental pain sensitivity in healthy subjects

In two separate cohorts of healthy volunteers we analyzed the association of heat, ischemic and mechanical pain with GCH1 haplotypes. One cohort was examined at the University of North Carolina at Chapel Hill and the second cohort was examined at the University of Florida. Experimental procedures used to assess pain perception are described in refs. 11,49. In order to combine the data across the two cohorts, each subject's value for a given pain measure was standardized to z-scores. Differences between the diplotype groups were determined using one way ANOVA followed by Bonferroni post hoc testing (P at 0.05). Results of the individual cohorts are in Supplementary Table 4. The studies were carried out with the approval of the institutional review boards of the University of North Carolina and University of Florida. Informed consent was obtained from all subjects.

Supplementary Methods

Real time RT-PCR

Quantitative real-time PCR was performed using the Sybr green detection system with primer sets designed on Primer Express. Specific PCR product amplification was confirmed with gel electrophoresis. Transcript regulation was determined using the relative standard curve method per manufacture's instructions (Applied Biosystems). For each time point four samples of pooled tissue of two rats was analyzed.

In situ hybridization

Fresh frozen DRGs were cut at 14 um, postfixed and acetylated. Riboprobes were obtained by in vitro transcription of cDNA and labeled with digoxigenin (Dig-labeling kit, Roche). Sections were hybridized with 200 ng/ml of sense or antisense probes in a prehybridization mix (5 x SSC, 50% formamide, 2 x Denhardt's, 500ug/ml herring sperm DNA, 250 ug/ml yeast tRNA), hybridized with 200 ng/ml of sense or antisense probes in prehybridization mix, incubated with anti-Dig-alkaline phosphatase in 0.1 M maleic acid buffer, developed with NBT/BCIP/levamisole (Boehringer Mannheim), embedded in glycerol/gelatin or subjected to post in situ immunostaining. Primary antibodies: sheep Dig-AP 1:1000(Roche), rabbit ATF-3 1:300 (SantaCruz). Blocking and antibody incubations in 1% blocking reagent (Roche) in maleic acid buffer.

Drug treatment

DAHP was dissolved in 1:1 polyethylene glycol (PEG400) and 1 x PBS, pH 7.4 (15 mg/ml) and administered i.p. or intrathecally (250 ug/kg/h; 5ul/h). For all i.t. injections/infusions a spinal catheter (Recathco) was used. Infusions were done with an osmotic pump (Alzet). 6R-BH4 and neopterin in artificial cerebrospinal fluid were injected i.t. (10 ug, single 10ul injection). N-acetyl-serotonin in 1 x PBS pH 7.4 containing 3% ethanol was delivered by i.t. infusion (100 ug/kg/h; 5 ul/h) or single i.p. injection (50 mg/kg). L-NAME (25 mg/kg, dissolved in 0.9% NaCl) was injected i.p. Control animals received the appropriate vehicle. All drugs were from Sigma-Aldrich.

Plasma and CSF concentrations of DAHP

Concentrations of DAHP were determined by LC/MS-MS on a tandem quadrupole mass spectrometer (PE Sciex API 3000; Applied Biosystems). Extraction by acetonitrile precipitation; chromatographic separation on a Nucleosil C18 Nautilus column (125 x 4 mm I.D., 5 um particle size, 100 A pore size). Mobile phase was acetonitrile:water (80:20%, v/v), and formic acid (0.1%, v/v). Flow rate 0.2 ml/min, injection volume 5 ul. DAHP eluted at 4.7 min. Mass spectrometer in positive ion mode, 5200 V, 400 ~C, auxiliary gas flow 6 l/min. The mass transition for the MRM was m/z 127↙60. Quantification with Analyst software V1.1 (Applied Biosystems). The coefficient of variation over the calibration range of 10每4000 ng/ml was < 5%.