Functional characterization of prostaglandin F receptor in the spinal cord for tactile pain (allodynia)

Authors


Address correspondence and reprint requests to Seiji Ito, Department of Medical Chemistry, Kansai Medical University, Moriguchi 570–8506, Osaka, Japan. E-mail: ito@takii.kmu.ac.jp

Abstract

Prostaglandin F (PGF) binds to its receptor (FP) to increase the intracellular-free calcium concentration ([Ca2+]i) by coupling of FP with Gq protein. Spinal intrathecal administration of PGF to mouse induces touch-evoked pain (mechanical allodynia), in which capsaicin-insensitive primary afferent Aβ-fibres and N-methyl-d-aspartate receptor ɛ4 subunit are involved. FP in the spinal cord, however, was not well characterized. Here, we showed constitutive expression of FP mRNA in mouse spinal cord, and functionally characterized spinal FP-expressing cells which were involved in PGF-induced mechanical allodynia. The method for repetitive administration of oligodeoxyribonucleotides through tubing to conscious mice was established for mechanical allodynia evaluation. We identified an antisense oligodeoxyribonucleotide targeting FP mRNA, causing both disappearance of PGF-induced mechanical allodynia and decrease of FP mRNA. With saline-administered mice, PGF rapidly increased [Ca2+]i of the cells in the deeper layer of the dorsal horn. In contrast, when the FP antisense oligodeoxyribonucleotide was repeatedly administered, the population of PGF-responsive cells in the slices reduced, and PGF-induced [Ca2+]i increase of these cells diminished. These data strongly suggested that, in the dorsal horn of the spinal cord, there are the FP-expressing cells which are involved in PGF-induced mechanical allodynia.

Abbreviations used
[Ca2+]i

intracellular-free calcium concentration

FP

prostaglandin F receptor

GAPDH

glyceraldehyde-3-phosphate dehydrogenase

i.t.

intrathecal

NMDA

N-methyl-d-aspartate

PG

prostaglandin

RT-PCR

reverse transcription-polymerase chain reaction

TBOA

DL-threo-β-benzyloxyaspartate

The seven-transmembrane receptor of the physiologically important prostaglandin F (PGF) has an amino-acid sequence and a gene organization conserved among various species, such as human (Abramovitz et al. 1994), mouse (Sugimoto et al. 1994; Hasumoto et al. 1997) and bovine (Sakamoto et al. 1994; Ezashi et al. 1997). PGF receptor (FP) transduces the PGF signal by coupling with Gq protein to elevate the intracellular-free calcium concentration ([Ca2+]i) (Abramovitz et al. 1994; Ito et al. 1994). The reproductive roles of PGF and FP, such as ovulation, luteolysis and parturition, have been emphasized (Sakamoto et al. 1995; Sugatani et al. 1996; Unezaki et al. 1996; Sugimoto et al. 1997).

PGs are also involved in inflammation, including pain, by sensitizing peripheral terminals of primary afferent nociceptors (Vane 1971). Serendipitously, we found that touch-evoked pain, mechanical allodynia, was induced by intrathecal (i.t) administration of PGF, as well as PGE2, to the subarachnoid space of conscious mice (Minami et al. 1992, 1994b). PGF induces long-lasting mechanical allodynia, but not hyperalgesia, while PGE2 induces both mechanical allodynia and hyperalgesia (Uda et al. 1990; Minami et al. 1994b). Capsaicin treatment of neonatal mice suggests that capsaicin-insensitive, primary afferent Aβ-fibres are involved in PGF-induced mechanical allodynia, whereas capsaicin-sensitive C-fibres are involved in PGE2-induced allodynia (Minami et al. 1999). Further studies using mice with disrupted N-methyl-d-aspartate (NMDA) receptor genes confirm that there are two pathways and mechanisms for the mechanical allodynia (Minami et al. 1994a, 1999, 2001; Ito et al. 2001). NMDA receptor ɛ4 subunit is involved in PGF-induced allodynia, which is insensitive to morphine; whereas NMDA receptor ɛ1 subunit is involved in PGE2-induced allodynia, which is morphine-sensitive. As PG-induced mechanical allodynia was blocked by DL-threo-β-benzyloxyaspartate (TBOA), a non-transportable blocker of the glutamate transporter (Shimamoto et al. 1998), the glutamate transporter is also essential for the induction and maintenance of mechanical allodynia (Minami et al. 2001).

Mechanical allodynia is pain generated by innocuous tactile stimuli via primary afferent sensory neurons that terminate in the dorsal horn of the spinal cord. Thus, it is highly expected that certain cells in the spinal cord express FP. However, there are difficulties in detecting FP-expressing cells by in situ hybridization due to the very low abundance of FP mRNA, and by immunohistochemistry due to the lack of a suitable anti-FP antibody. Here, we showed that FP mRNA was detected in the spinal cord by reverse transcription-polymerase chain reaction (RT-PCR) and, then, functionally characterized FP-expressing cells that were involved in PGF-induced mechanical allodynia in the mouse spinal cord using antisense oligodeoxyribonucleotides (hereafter, oligonucleotides). Finally, we detected PGF-responsive cells in spinal cord slices by measuring [Ca2+]i and demonstrated that the FP antisense oligonucleotide affected [Ca2+]i increase.

Materials and methods

Animals

Male ddY mice (4 weeks old, weighing 22 ± 2 g; Japan SLC Ltd, Shizuoka, Japan) were used for all the experiments except [Ca2+]i measurement. The animals were maintained in climate- and light-controlled rooms, and each animal was used on one occasion only. All the experiments were conducted with the approval of the animal experimentation committees of Kansai Medical University and Osaka Medical College and in accordance with the guidelines of the Ethics Committee of the International Association for the Study of Pain (Zimmermann 1983).

Chemicals and drugs

Oligofectamine reagent and Fura 2-AM were purchased from Invitrogen (Carlsbad, CA, USA) and Wako Pure Chemicals (Osaka, Japan), respectively. λ-Carrageenan, NMDA and strychnine hemisulfate were obtained from Sigma (St Louis, MO, USA). PGE2 and PGF were generous gifts of Ono Central Research Institute (Osaka, Japan).

RT-PCR analysis of mRNA

Mice were decapitated before and at 1, 3, 6, and 12 h after intraplantar injection of λ-carrageenan, and the lumbar spinal cord without roots and meninges was subjected to RNA isolation and cDNA synthesis, as described previously (Doi et al. 2002). The lumber spinal cords were excised from the mice after the repetitive oligonucleotide administration and the allodynia assessment, and total RNA and cDNA were prepared in the same way. The high-stringency step-down RT-PCR (Nishizawa et al. 2000) was performed with the FP primers, AF and AR (Table 1), and the primers for glyceraldehyde-3-phosphate dehydrogenase (GAPDH; Nishizawa et al. 2000) as a control. The PCR protocol applied was: 10 cycles of (94°C, 1 min; 72°C, 2 min); 15 cycles of (94°C, 1 min; 65°C, 2 min; 72°C, 30 s); and 20 cycles of (94°C, 1 min; 60°C, 2 min; 72°C, 30 s) for FP, or 5 cycles for GAPDH. The PCR products (318 bp for FP, and 461 bp for GAPDH) were electrophoresed on 2.5% agarose gels. Linearity of the cDNA amplification was confirmed, and the DNA sequences were verified by sequencing, as described previously (Nishizawa et al. 2000). The nucleotide sequence data will appear in the DDBJ/EMBL/GenBank databases with the accession numbers AB073965, AB073966 and AB094411.

Table 1.  Oligonucleotides used in this study
OligonucleotideStrandPositionSequence (5′→ 3′)
  1. Oligonucleotides for antisense knockdown of FP and the RT-PCR analysis are listed. AS1 to AS5 are the antisense oligonucleotides targeting FP mRNA. Se5, Ma5, Scr5 and Random oligonucleotides were negative controls to AS5. AF and AR were a pair of primers for the RT-PCR analysis of FP mRNA. The position of nucleotides is numbered from A of the translation initiation codon of mouse FP cDNA. Sequences are depicted in the 5′-to-3′direction. Letters in lower case represent phosphorothioate-blocked nucleotides. Mutated bases of Ma5 are underlined. N of Random oligonucleotides denotes the mixture of bases A, C, G and T.

AS1antisense30–11cacTGGCTGCTTGGAActgT
AS2antisense329–310ataCTGCACAGGATGTttgA
AS3antisense418–399gatTGGTGACTCCTATacaC
AS4antisense645–626aacACCAAGAGCTAAGagtC
AS5antisense997–978agaTATGCAAGCTGATgatG
Se5sense978–997catCATCAGCTTGCATatcT
Ma5mutant997–978agaTATCCCAGCTAATgatG
Scr5scramble gtaGTACTGCAGATGAtagA
Randomrandom nnnNNNNNNNNNNNNNnnnN
AFsense791–812CCTTTCTGGTAACAATGGCCAA
ARantisense1108–1087CATTGATTTACAGTCCAGCTTC

Design of FP antisense oligonucleotides

The 20-mer antisense oligonucleotides AS1-AS5 that targeted the entire coding sequence of mouse FP mRNA, various negative controls (Table 1) were designed according to the method described by Brysch (1999). Briefly, many candidate sequences without problematic motifs, such as ‘CG’ and ‘GGGG’, were designed from the mouse FP cDNA sequence (accession No. D17433 and AB094411) and screened by the blast program to eliminate candidates having unexpected homology to other mRNAs in the DDBJ/EMBL/GenBank databases. Finally, selected antisense oligonucleotides AS1-AS5 were administered to mice. Similarly, mutant oligonucleotides were designed by introducing three base mismatches to the antisense AS5 sequence and screened by the homology search to select the Ma5 oligonucleotide. The scramble control Scr5 was selected by base shuffling without changing the base composition of AS5 and by the following homology search. Random oligonucleotides were a mixture of random 20-mer oligonucleotides. All the oligonucleotides blocked by three phosphorothioate bonds at each terminus were purchased from Sigma Genosys Japan (Ishikari, Japan).

Cannulation of tubing to mouse spinal subarachnoid space

For repetitive administration of oligonucleotides and drugs to mice, we modified the previous method (Yaksh and Rudy 1976), which is optimized for rats and rabbits. Under deep anaesthesia by sodium pentobarbital (50 mg/kg), intervertebral ligaments between L5 and L6 were exposed after incision of skin and muscles and pierced by a 27-gauge needle. An 85-mm Teflon tube (inner diameter, 0.1 mm; outer diameter, 0.4 mm; Eicom, Kyoto, Japan) was curved by heat at 5 mm from an end and filled with sterile saline. The shorter part of the tube was inserted rostrally to the subarachnoid space through the intervertebral opening and fixed to muscles using 3–0 silk threads. Then, skin between the ears was longitudinally incised, and the longer part of the tube was subcutaneously pull out using a 16-gauge venous cannula (Terumo, Tokyo, Japan), and the tube was attached to the skin with silk threads. The dead volume of this tubing was about 2 μL. For administration to mice, 7 μL (per mouse) of each oligonucleotide-Oligofectamine mixture or drug was injected to the opening of the tubing between the ears using a 27-gauge needle (Terumo) and a sterile glass syringe. Mice showing signs of motor disturbance or infection (less than 3% of mice operated) were omitted.

Administration of FP antisense oligonucleotides through tubing

Saline (80 μL), containing 80 pmol of an oligonucleotide and 3.2 μL of Oligofectamine reagent, was incubated at room temperature (22–25°C) for 20 min. After addition of saline to a final volume of 200 μL, this mixture (7 μL per mouse) was injected through tubing. On day 2 and 3, the same amount of the freshly prepared oligonucleotide-Oligofectamine mixture was administered once per day. Mechanical allodynia was evaluated on day 4. As controls, saline alone or Oligofectamine reagent in saline without oligonucleotides (7 μL per mouse per day) was administered to mice for 3 days. Prior to [Ca2+]i measurement on day 4, saline alone or the AS5-Oligofectamine mixture (7 μL per mouse per day) was administered for 3 days in the same fashion.

Evaluation of mechanical allodynia

On day 4, PGF in saline (1 μg per mouse) was administered to mice through tubing. Five minutes after PGF injection, mechanical allodynia was assessed 10 times every 5 min, as described previously (Minami et al. 1992, 1994a,b). The mouse flank was lightly stroked with a paintbrush, and the allodynic response was ranked as follows: 0, no response; 1, mild squeaking with attempts to move away from the stroking probe; and 2, vigorous squeaking evoked by the stroking probe, biting at the probe, or strong efforts to escape. The scores were cumulated over a 50-min period, and thus the possible maximum cumulative score is 20 per mouse. Each cumulative score was expressed as a per cent of possible maximum cumulative score. If necessary, saline or PGE2 (10 ng per mouse) was injected as negative and positive controls, respectively. After mechanical allodynia assessment, 7 μL of 33 mm strychnine hemisulphate, a glycine receptor antagonist, was injected through tubing to confirm that the cannulated tubing was open to the spinal subarachnoid space. Unless spontaneous agitation was observed within 1 min after strychnine injection, the allodynia data of the mouse were discarded. More than 75% of mice operated responded to strychnine. The spinal cord of mice responded to strychnine was subjected further to the RT-PCR analysis to estimate the FP mRNA level.

[Ca2+]i measurement of mouse spinal cord slices

The AS5 antisense oligonucleotide or saline (control) was administered to 3-week-old mice (16–19 g) through tubing once per day for 3 days, and PGF-induced allodynia was evaluated on day 4, as described above. The spinal cord of mice showing expected allodynia responses was quickly excised under anaesthesia by pentobarbital and immersed in ice-cold artificial cerebrospinal fluid (ACSF; 120 mm NaCl, 3.1 mm KCl, 2.0 mm CaCl2, 1.0 mm MgCl2, 1.25 mm NaH2PO4, 25 mm NaHCO3, 5.0 mm glucose, 2.0 mm sodium pyruvate, 0.5 mmmyo-inositol, and 0.02 mm ascorbic acid). ACSF used in all the procedures of this experiment was equilibrated with 95% O2 and 5% CO2. Roots and meninges were removed from the spinal cord. The lumbar cord was embedded in 3% low-melting temperature agarose in ACSF and sliced transversely with a vibrating blade microtome (Leica VT-1000S, Nussloch, Germany). The slices (350 μm in thickness) at L4–L6 levels were placed in the recording chamber, mechanically fixed using an overlaying grid of nylon threads attached to a platinum ring, and superfused by ACSF at a flow rate of 3 mL/min. The slices were loaded by 10 μm Fura 2-AM in ACSF at room temperature for 2 h. The chamber was mounted on an inverted fluorescence microscope (Olympus IX-70, Tokyo, Japan) equipped with a dichroic mirror (505 nm) and an emission filter (515–550 nm). Fura 2 was excited at 340 and 380 nm using a high-speed excitation wavelength switcher (Hamamatsu Photonics, Hamamatsu, Japan), and the fluorescence emission signal from cells was monitored by the Aquacosmos-Ratio imaging system (Hamamatsu Photonics) with a cooled charge-coupled device camera. [Ca2+]i was expressed as a ratio of fluorescence emission intensity at 340–380 nm. To the superfusing ACSF buffer in the chamber, TBOA was added to a final concentration of 10 μm, and then PGF was added to a final concentration of 20 μm. After washing the chamber, NMDA was added to a final concentration of 100 μm to stimulate [Ca2+]i increase as a positive control. Cells which did not respond to NMDA were omitted.

Data analyses

The statistical analyses were carried out by Mann–Whitney U-test for multiple comparisons in Figs 2, 3, and 4, and Student's t-test in Fig. 6.

Figure 2.

Effects of various antisense oligonucleotides targeting FP mRNA on PGF-induced mechanical allodynia. (a) Schematical representation of FP mRNA and the FP antisense oligonucleotides. The coding sequence of mouse FP mRNA is shown as an open square and seven regions encoding transmembrane domains (TM1 to TM7) are indicated by closed boxes. The FP antisense oligonucleotides (AS1 to AS5) are depicted by short lines beneath the coding sequence. (b) Effects of the FP antisense oligonucleotides on PGF-induced mechanical allodynia. The antisense oligonucleotides AS1 to AS5, Random oligonucleotides (Table 1), and saline were administered to mouse spinal subarachnoid space through tubing for 3 days, and then PGF-induced allodynia was evaluated, as described in Materials and methods. Each column represents the mean ± SEM (n = 3–6) of the percentage of the possible maximum cumulative score for 50 min. Statistically significant mean difference versus saline-administered mice with **p < 0.01.

Figure 3.

Specific knockdown of FP by the antisense AS5 oligonucleotide. (a) Effects of saline, PGF and PGE2 on mechanical allodynia of AS5-administered mice. AS5 and Oligofectamine reagent (AS5 + Oligofectamine) were administered through tubing, and mechanical allodynia was assessed after sequential injection of saline, PGF and PGE2 to the same mice, as described in Materials and methods. Each column represents the mean ± SEM (n = 5) of the percentage of the possible maximum cumulative score for 50 min. Statistically significant mean difference versus saline-administered mice with *p < 0.05. (b) Effect of Oligofectamine reagent on PGF-induced mechanical allodynia. Oligofectamine reagent without oligonucleotides (Oligofectamine alone) was administered through tubing as a control, and mechanical allodynia was assessed after sequential injection of saline and PGF to the same mice. Each column represents the mean ± SEM (n = 5) of the percentage of the possible maximum cumulative score evaluated for 50 min. Statistically significant mean difference versus saline-administered mice with *p < 0.05.

Figure 4.

Disappearance of PGF-induced mechanical allodynia and reduction of FP mRNA by administration of the antisense AS5 oligonucleotide through tubing. (a) Effects of AS5 and negative controls on mechanical allodynia. AS5, various negative controls to AS5 (Se5, Ma5, Scr5 and Random), and saline were administered through tubing, and then PGF-induced allodynia was evaluated. Each column represents the mean ± SEM (n = 6) of the percentage of the possible maximum cumulative score for 50 min. Statistically significant mean difference versus saline-administered mice with **p < 0.01 and *p < 0.05. (b) The FP mRNA content after administration of AS5 and negative controls. We evaluated PGF-induced allodynia and confirmed that tubing was open to the spinal subarachnoid space by strychnine, and the spinal cord from mice responded to strychnine was subjected to the RT-PCR analysis. Upper and lower panels show the bands for FP and GAPDH mRNAs, respectively.

Figure 6.

Summary of antisense effects on [Ca2+]i increase of spinal PGF-responsive cells. (a) Population of PGF-responsive cells in the spinal cord slices. All the data of [Ca2+]i measurement are summarized. Spinal cord slices were prepared from mice to which saline or the antisense AS5 oligonucleotide was administered through tubing, and fluorescence image of the dorsal horn was monitored, when TBOA and PGF and then NMDA were added. Cells showing [Ca2+]i increase by both PGF and NMDA were counted as a PGF-responsive cell. The number of PGF-responsive cells was divided by the number of NMDA-responsive cells, and was expressed as the percentage. (b) The peak ratio of PGF-stimulated [Ca2+]i increase to NMDA-stimulated increase. Spinal cord slices were prepared from mice to which saline or AS5 was administered. The peak height of PGF-stimulated [Ca2+]i increase was divided by that of NMDA-stimulated [Ca2+]i increase (peak ratio of PGF/NMDA) in each PGF-responsive cell. Each column represents the mean ± SD (n = 93 for saline, and n = 37 for AS5) of the percentage of peak ratio of PGF/NMDA. Statistically significant mean difference versus saline with #p < 0.001.

Results

Expression of FP mRNA in mouse spinal cord

To examine the presence of FP mRNA in mouse spinal cord, we analyzed the FP mRNA expression by the RT-PCR technique. As shown in Fig. 1, FP mRNA was detected in the spinal cord of untreated mice (0 h, lane 1). The amplification of FP cDNA in the spinal cord required a total of 45 cycles in PCR, while that of GAPDH cDNA required 30 cycles. Thus, the expression level of FP mRNA seemed to be much lower than that of GAPDH mRNA. Intraplantar injection of carrageenan induces inflammation of paws and marked hyperalgesia, as well as the increase of prostacyclin receptor and cyclo-oxygenase 2 mRNAs (Seibert et al. 1994; Hay and de Belleroche 1997; Doi et al. 2002). When carrageenan was injected in mice, there were not apparent changes in the expression level of FP mRNA for up to 6 h following the carrageenan injection, and the level of FP mRNA declined at 12 h (lanes 2–5). These results demonstrated that FP mRNA was expressed at a low level in the mouse spinal cord and the expression was not induced by carrageenan, suggesting that FP mRNA is constitutively expressed in the spinal cord.

Figure 1.

Expression of FP mRNA in mouse spinal cord. Total RNA was extracted from the spinal cord before (lane 1) and 1, 3, 6 and 12 h (lanes 2–5) after intraplantar carrageenan injection, as described in Materials and methods. The RT-PCR analysis was performed with total RNA, and the PCR products were resolved by agarose gel electrophoresis. Upper panel shows the 318-bp band for the FP mRNA, and lower panel shows the 461-bp band for the GAPDH mRNA as a control.

Antisense knockdown of FP in vivo

Next, we tried to knockdown FP mRNA in the spinal cord of living conscious mice and to examine whether PGF-induced mechanical allodynia was suppressed. The antisense oligonucleotides targeting mouse FP mRNA, AS1 to AS5, were designed to cover the entire coding region of FP mRNA (Table 1 and Fig. 2a). The ‘CG’ and ‘GGGG’ motifs causing non-specific biological effects (Brysch 1999), which are often seen in nucleotide sequences encoding transmembrane domains, were avoided. Single-shot i.t. administration of the FP antisense oligonucleotides with Oligofectamine reagent did not give reproducible results in PGF-induced allodynia (data not shown). Therefore, we tried to administer the oligonucleotides to mouse spinal subarachnoid space repetitively. As the previous method (Yaksh and Rudy 1976) is optimized for drug delivery to rats and rabbits, we modified and adapted it to mice. Through the tubing cannulated to the spinal subarachnoid space, the FP antisense oligonucleotide-Oligofectamine mixture was injected once per day for 3 days, as described in Materials and methods. PGF for mechanical allodynia assessment was also injected through the same tubing. After the evaluation of PGF-induced allodynia, strychnine, an antagonist at glycine receptors, was injected through the same tubing. When spontaneous agitation was observed within 1 min after strychnine injection, it was confirmed that the cannulated tubing was open to the subarachnoid space.

By our newly established method, the mice administered by the antisense AS5 oligonucleotide gave marked decrease of allodynia scores, compared with the saline-administered mice (Fig. 2b). In contrast, the administration of AS1, AS2, or Random oligonucleotides, which were a mixture of random 20-mer oligonucleotides as a negative control, did not diminish allodynia scores significantly. As the antisense oligonucleotide inhibits translation by sequence-specific binding to mRNA and subsequent digestion of mRNA with endogenous RNase H (Brysch 1999), AS1 near the translation initiation codon is theoretically the most effective. However, AS5 close to the termination codon was experimentally the most effective on PGF-induced mechanical allodynia. For further analyses, AS5 was selected.

We examined whether the antisense AS5 oligonucleotide functioned specifically to PGF-induced mechanical allodynia. The AS5-administered mice were assessed by sequential injection of saline, PGF and PGE2 through tubing. As expected, mechanical allodynia did not appear for 50 min after saline injection (Fig. 3a). After PGF injection to the same mice, mechanical allodynia disappeared over the 50-min period, indicating that PGF-induced allodynia was suppressed. When PGE2 was then injected, allodynia scores increased significantly. These results clearly demonstrated that the antisense AS5 oligonucleotide suppressed specifically PGF-induced allodynia. In our experiments, Oligofectamine reagent, a transfection reagent for oligonucleotides, was used to improve the incorporation of oligonucleotides into spinal cord cells. To exclude the possibility that this reagent non-specifically induced mechanical allodynia or blocked PGF-induced allodynia, we performed the next experiment (Fig. 3b). Oligofectamine reagent in saline without oligonucleotides was prepared and administered to mice for 3 days. On day 4, saline and PGF were sequentially administered through tubing for mechanical allodynia assessment. When saline was injected, mechanical allodynia was not observed. After PGF injection to the same mice, allodynia scores increased significantly. These data indicated that Oligofectamine reagent itself did not have any non-specific effects on mechanical allodynia. Together with these data, the Oligofectamine-mediated administration of antisense AS5 oligonucleotide to mice specifically suppressed PGF-induced mechanical allodynia.

To ensure the specificity in the AS5 function, we administered negative controls to mice, evaluated PGF-induced mechanical allodynia, and estimated the FP mRNA level in the spinal cord (Fig. 4). All the negative controls used for this study had low cross-homology to mouse mRNAs except FP mRNA by searching mouse nucleotide databases of the DDBJ/EMBL/GenBank. Se5, a sense oligonucleotide to AS5, had the same sequence of FP mRNA. Unexpectedly, it suppressed PGF-induced allodynia, and the level of FP mRNA was significantly lower than that of the saline-administered mice. Ma5, which was one of mutant oligonucleotides having three mismatches in the AS5 sequence, was administered to mice. Ma5 also reduced the allodynia score and the FP mRNA content. Finally, the scramble Scr5 oligonucleotide was designed by base shuffling without changing the base composition of AS5, as the scramble (or nonsense) oligonucleotide is thought to be an ideal negative control (Brysch 1999). As shown in Fig. 4, Scr5 slightly reduced the allodynia score and the FP mRNA content. Random oligonucleotides, a negative control for any mRNAs, did not suppress mechanical allodynia, or lower the level of FP mRNA. Thus, Random oligonucleotides functioned as the best negative control to AS5. Together with these results, administration of the antisense AS5 oligonucleotide targeting FP mRNA through tubing suppressed specifically PGF-induced mechanical allodynia and decreased FP mRNA in mouse spinal cord.

Antisense effects on [Ca2+]i in mouse spinal cord cells

To investigate the distribution of FP-expressing cells in the spinal cord, we measured [Ca2+]i in situ using spinal cord slices. For control mice, saline was administered through tubing, and mechanical allodynia was evaluated. When PGF-induced allodynia was observed, slices at the lumbar spinal cord were prepared from the mice. When PGF alone was added to the superfusion buffer, only one cell showed [Ca2+]i increase among 105 cells examined. The fluorescence signal for PGF-stimulated [Ca2+]i increase might be weak and transient, and it was possibly masked by rapid uptake of glutamate released from synaptic terminals. To prolong the activation of the glutamate NMDA receptor by PGF, we added TBOA, a blocker of the glutamate transporter (Shimamoto et al. 1998), prior to PGF addition. In the presence of TBOA, the [Ca2+]i fluorescence of many cells in the deeper layer (lamina III and deeper) of the dorsal horn increased immediately after PGF addition (Figs 5 and 6a). There were very few cells showing [Ca2+]i increase in the superficial layer, i.e. laminae I and II, of the dorsal horn (data not shown). As most subunits of the Ca2+-permeable NMDA receptor are widely expressed in the spinal cord (Watanabe et al. 1992, 1994; Mori and Mishina 1995), marked [Ca2+]i increase was observed when NMDA was added as a positive control (Fig. 5b). In summary, 93 PGF-responsive cells were observed among 170 NMDA-responsive cells using seven spinal slices (54.7%, Fig. 6a).

Figure 5.

[Ca2+]i increase in cells of mouse spinal cord slices. (a) Schematic representation of transverse section of mouse spinal cord at L5 level. Top, dorsal; bottom, ventral. Laminae in the grey matter are separated by broken lines. The box depicts the region to monitor the fluorescence image of cells in the dorsal horn, which corresponds to laminae III to V. (b) [Ca2+]i increase by PGF and NMDA in a spinal cord cell from a saline-administered mouse. As a control, saline was administered through tubing, as described in Materials and methods. From mice showing PGF-induced allodynia, spinal cord slices were prepared for [Ca2+]i measurement. Fluorescence image of each cell was monitored in the region (box in Fig. 5a). [Ca2+]i is expressed as a ratio of fluorescence emission intensity at 340–380 nm. TBOA, PGF and then NMDA (as a positive control) were added to the superfusion buffer, as indicated by bars on the bottom. This panel shows a typical example of a cell responsive to both PGF and NMDA, i.e. PGF-responsive cell. (c) [Ca2+]i increase by PGF and NMDA in a spinal cord cell from an AS5-administered mouse. The antisense AS5 oligonucleotide was administered through tubing. From mice that did not show PGF-induced allodynia, spinal cord slices were prepared for [Ca2+]i measurement. This panel shows a typical example of a PGF-responsive cell.

Then, we administered the AS5 antisense oligonucleotide through tubing. From the mice showing disappearance of PGF-induced allodynia, spinal cord slices were prepared and subjected to [Ca2+]i measurement by sequential addition of TBOA, PGF, and then NMDA (Fig. 5c). PGF-stimulated [Ca2+]i increase was observed in only 37 cells among 342 NMDA-responsive cells in the deeper layer using 11 spinal slices (10.8%, Fig. 6a). Furthermore, the peak height of PGF-stimulated [Ca2+]i increase reduced in PGF-responsive cells (Fig. 5c). When the peak ratio of PGF-stimulated [Ca2+]i increase to NMDA-stimulated increase was calculated, the ratio using spinal cord slices from AS5-administered mice was significantly lower than that from saline-administered mice (Fig. 6b). Therefore, the antisense AS5 oligonucleotide suppressed the FP expression in the spinal cord cells both quantitatively and qualitatively. Together with these data, it is strongly suggested that the FP-expressing cells, which are involved in PGF-induced mechanical allodynia, are present in the deeper layer of the dorsal horn in the spinal cord.

Discussion

The central role of PGF and FP in pain transmission has been behaviourally studied since i.t. injection of PGF was found to induce mechanical allodynia in conscious mice (Minami et al. 1992, 1994a,b, 1999). It is hardly possible to detect FP-expressing cells in the spinal cord, even by very sensitive in situ hybridization, due to the low level expression of FP. Here, we clearly demonstrated that FP-expressing cells were present in mouse spinal cord by RT-PCR analysis of FP mRNA (Fig. 1). Furthermore, we newly established the antisense knockdown technique for conscious mice, validity of which was evaluated by allodynia responses, mRNA levels and [Ca2+]i increase. To repeatedly administer antisense oligonucleotides with Oligofectamine reagent, we used the tubing cannulated to the mouse spinal subarachnoid space. The reproducible results were obtained from the mice operated without any detectable motor disturbances and behavioural changes, except pain responses, for several days. Using this technique, we identified the most effective AS5 among five FP antisense oligonucleotides which covered the entire coding sequence of FP mRNA and did not include the problematic ‘CG’ and ‘GGGG’ motifs. On the basis of the action that the antisense oligonucleotide inhibits mRNA translation, AS1 near the initiation codon was theoretically the most effective. The results demonstrated that AS5 close to the termination codon had marked knockdown effects on both PGF-induced allodynia and the FP mRNA level in the spinal FP-expressing cells (Figs 2 and 4). Effective antisense oligonucleotides (AS3, AS4 and AS5) theoretically hybridize to single-stranded regions of FP mRNA, while an ineffective AS2 hybridizes to the hairpin structure. It suggests the correlation between the antisense effect and the hybridization potentials to the mRNA single-stranded regions.

The [Ca2+]i measurement using Fura 2 with mouse spinal cord slices provided the interesting results that most PGF-responsive cells located in the deeper layer in the dorsal horn (Fig. 5). Moreover, the combination with our antisense knockdown of FP in mice indicated that there were PGF-responsive cells in the dorsal horn and that population and [Ca2+]i increase of these cells reduced by the antisense AS5 oligonucleotide. These results coincide with the reports that signals from cutaneous low-threshold mechanoreceptors, which confer touch and innocuous sensation, are transmitted by primary afferent Aβ-fibres and that Aβ-fibres terminate at laminae III to V of the grey matter (Dray et al. 1994; Woolf 1994). Our results strongly suggest that PGF-responsive cells are involved in PGF-induced mechanical allodynia.

Are all the ‘PGF-responsive cells’ FP-expressing cells in the spinal cord? The present study demonstrated that FP-expressing cells existed in the spinal cord (Fig. 1). PGF-responsive cells in the dorsal horn were functionally characterized as the cells responsive to both PGF and NMDA by [Ca2+]i measurement (Figs 5 and 6). PGF potentiates the activation of the glutamate receptors by innocuous stimuli. It is most likely that PGF activates an FP-expressing cell (neuron or interneuron), and glutamate released from this cell further activates a post-synaptic, NMDA receptor-expressing neuron to elevate [Ca2+]i. This agrees with the results that the fluorescence signal of [Ca2+]i increase stimulated by PGF alone was weak and transient and that TBOA increased PGF-responsive cells. It is possible that the PGF-responsive cells express both FP and the NMDA receptor. These hypotheses well explain PGF-induced allodynia involving the glutamatergic system (Minami et al. 2001). The third possibility remains that the PGF-responsive cell is a glial cell, as some type of astrocytes have FP (Ito et al. 1992). The NMDA receptor, however, is expressed mainly in neurons (Mori and Mishina 1995), and PGF-induced allodynia cannot be well explained by this hypothesis. It is crucial to clarify in which cell FP and the glutamate receptors are expressed. To answer the question, detailed studies using electrophysiological techniques and the single-cell PCR method are required.

The features of PGF-induced mechanical allodynia are quite similar to those of causalgia and reflex sympathetic dystrophy after peripheral nerve injury in humans (Fields 1994) and those of allodynia induced by L5–L6 nerve ligation in rats (Plummer et al. 1992). PGF-induced allodynia has a distinct pathway and mechanism from PGE2-induced allodynia, although the glutamatergic system and glutamate transporters mediate induction and maintenance of both mechanical allodynia (Minami et al. 1999, 2001; Ito et al. 2001). Our newly established technique for mice, namely antisense knockdown through tubing combined with allodynia evaluation and [Ca2+]i measurement, can be applied to analyze genes encoding molecules that are involved in pain transmission. The advantage of this method is that the whole neural structure and network in the spinal cord are restored. Furthermore, many knockout mice showing various phenotypes can also be analyzed. In our laboratory, genes involved in PG-induced mechanical allodynia are now under investigation using this method to distinguish PGF-induced allodynia from PGE2-induced allodynia. Recently, multiple different mechanisms including pre-synaptic interactions between Aβ- and C-fibres are proposed for induction of mechanical allodynia (Bennett 1994; Cervero and Laird 1996). Further analyses of spinal cord cells expressing FP and the glutamate receptor, as well as allodynia-involving molecules, will provide an insight to elucidate the mechanism of PGF-induced allodynia.

Acknowledgements

This work was supported in part by Grants-in-Aid for Scientific Research on Priority Areas and Scientific Research (B)(2) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan; and by a grant from the Science Research Promotion Fund of the Japan Private School Promotion Foundation.

Ancillary