Alternative splicing of ryanodine receptor subtype 3 (RYR3) may generate a short isoform (RYR3S) without channel function and a functional full-length isoform (RYR3L). The RYR3S isoform has been shown to negatively regulate the native RYR2 subtype in smooth muscle cells as well as the RYR3L isoform when both isoforms were coexpressed in HEK-293 cells. Mouse myometrium expresses only the RYR3 subtype, but the role of RYR3 isoforms obtained by alternative splicing and their activation by cADP-ribose during pregnancy have never been investigated. Here, we show that both RYR3S and RYR3L isoforms are differentially expressed in nonpregnant and pregnant mouse myometrium. The use of antisense oligonucleotides directed against each isoform indicated that only RYR3L was activated by caffeine and cADP-ribose in nonpregnant myometrium. These RYR3L-mediated Ca2+ releases were negatively regulated by RYR3S expression. At the end of pregnancy, the relative expression of RYR3L versus RYR3S and its ability to respond to cADP-ribose were increased. Therefore, our results suggest that physiological regulation of RYR3 alternative splicing may play an essential role at the end of pregnancy.
- ryanodine receptor
- smooth muscle
- alternative splicing
intracellular Ca2+ concentration ([Ca2+]i) controls the force of myometrial contractions. Two Ca2+ sources participate to increase [Ca2+]i: 1) Ca2+ entry through voltage-dependent Ca2+ channels responsible for spontaneous contractions and 2) Ca2+ release from intracellular stores activated by agonists such as oxytocin, endothelin, or acetylcholine (for reviews, see Refs. 41 and 26). The release of Ca2+ stored in the sarcoplasmic reticulum (SR) is assumed by two different families of channels: inositol 1,4,5-trisphosphate (InsP3) receptors (InsP3Rs) and ryanodine receptors (RYRs) (4, 5). Although the activation of InsP3Rs is well documented in the myometrium (22, 32, 35, 37), the activation and function of RYRs in the myometrium during pregnancy are still under debate.
Three subtypes of RYRs are encoded by three distinct genes [RYR1, RYR2, and RYR3 (for a review, see Ref. 13)], and the expression of RYRs in the myometrium at different stages of pregnancy has been explored (20). The expression of RYR3 has been reported in human, rat, and mouse nonpregnant myometrium and during pregnancy (23, 24, 28), whereas a very low level of RYR2 expression has been found in the pregnant myometrium, but only in the rat (24). RYRs were then suggested to be involved in a Ca2+-induced Ca2+ release mechanism in pregnant but not nonpregnant myometrium (39). However, in physiological solution, caffeine cannot activate RYRs in freshly isolated myocytes from the rat pregnant myometrium (2), although some responses were observed in cultured cells (24). In the mouse myometrium, the RYR3 subtype has been shown to be activated by caffeine only when the SR is Ca2+ overloaded (28). Several studies about RYR3 in other cell types have indicated that RYR3 is not able to induce an excitation-contraction mechanism (12) but can encode spontaneous Ca2+ release when overexpressed in HEK-293 cells (1, 34). However, several isoforms of RYR3 are expressed by alternative splicing (17, 25, 29). The short-length isoform (RYR3S) has been described to inhibit RYR2 (9, 17), and the full-length isoform (RYR3L) can form a functional channel in a heterologous system (17). In these studies, it was indicated that both isoforms of RYR3 are potentially expressed in the rabbit and mouse myometrium (9, 17). RYR3S is characterized by the alternative splicing of a 87-bp exon potentially encoding a putative transmembrane domain of the channel (17). Moreover, the RYR3 subtype has been proposed as the target for cADP-ribose (18), and both the expression of ADP-ribosyl cyclase and the production of cADP-ribose have been shown to be increased during pregnancy (3).
In the present study, we examined the expression and function of both isoforms of RYR3 in the nonpregnant and pregnant myometrium of the mouse. Using an antisense oligonucleotide strategy, we showed that RYR3L is activated by caffeine and is a target for cADP-ribose. For the first time, we showed that the expression of RYR3 is increased in the myometrium at the end of pregnancy and that RYR3L becomes the major isoform. These findings suggest an important role for RYR3L-mediated Ca2+ signaling at the end of pregnancy and the physiological relevance of the cADP-ribose pathway at this stage of gestation.
MATERIALS AND METHODS
This investigation conformed with European community and French guiding principles on the care and use of animals. Authorization to perform animal experiments (A-33-063-003) was obtained from the Préfecture de la Gironde. Nonpregnant and pregnant mice (10–16 wk old) were killed by cervical dislocation. The longitudinal layer of uterine smooth muscle was stripped and cut into several pieces, which were incubated twice for 15 min at 37°C in low-Ca2+ (40 μm) HBSS containing 0.8 mg/ml collagenase (EC 22.214.171.124), 0.2 mg/ml pronase E (EC 126.96.36.199), and 1 mg/ml BSA. Tissues were placed in an enzyme-free solution, and cells were mechanically released. Cells were seeded at a density of 103 cells/mm2 on glass slides and were maintained in short-term primary culture in medium 199 containing 5% FCS, 20 U/ml penicillin, and 20 μg/ml streptomycin. Cells were kept for 1–4 days in an incubator gassed with 95% air and 5% CO2 at 37°C.
Total RNA was extracted from uterine myocytes using the RNA preparation kit from Epicentre following the instructions of the supplier. The reverse transcription reaction was performed on 50 ng RNA using the Sensiscript-RT kit (Qiagen). The sense and antisense primer pairs specific for RYR1 and RYR2 were as previously described (8). The specific primer pair amplifying the dominant negative splicing site of RYR3 has been previously determinated in the mouse (9). PCRs were performed with a thermal cycler (Eppendorf). The annealing temperature was 60°C. Amplification products were separated by electrophoresis (2% agarose gel) and visualized by ethidium bromide staining. Gels were photographed with EDAS120 and analyzed with KDS1D 2.0 software (Kodak Digital Science). Relative amounts of amplicons were determined and normalized to that of the GAPDH fragment as previously described (31). After electrophoresis, the PCR products were cleaned and purified with the Qiaquick gel extraction kit (Qiagen). They were sequenced by the GenomExpress sequencing service to control RYR3 isoform sequences.
Western blot analysis.
The longitudinal layer of uterine smooth muscle from nonpregnant and pregnant mice was homogenized in an appropriate volume of 10% SDS. After centrifugation (300 g, 10 min), supernatants were collected, and the protein content was measured according to the method of Bradford (6). Equal amounts of proteins (12.5, 25, 50, and 100 μg) from nonpregnant and pregnant mice were heated at 95°C for 5 min in Laemmli buffer, separated by electrophoresis in a 4–12% gradient SDS-polyacrylamide gel (GeBaGel, Interchim), and electrically transferred to a polyvinylidene difluoride membrane (70 min, 100 V, 4°C). Nonspecific binding was blocked by incubating the membrane in phosphate buffer-Tween 20 (0.1%) containing 5% nonfat dry milk for 1 h, and blots were incubated (overnight, 4°C) with anti-RYR3 antibody (1:500). Primary antibody was detected with a horseradish peroxidase-coupled secondary antibody (1:3,000). RYR3 was detected using an enhanced chemoluminescence kit (Amersham Biosciences). After RYR3 revelation, the membrane was striped by an incubation in 0.7 mM β-mercaptoethanol and 0.4% SDS solution (30 min, 55°C) and revealed with anti-tubulin (1:8,000) or anti-CD38 (1:100) antibodies. All proteins were quantified using KDS1D 2.0 software.
Transfection of oligonucleotides.
The design of phosphorothioate antisense oligonucleotides used in the present study were based on RYR3 isoforms sequences, and they were chosen for their ability to totally abolish the PCR signal of one RYR3 isoform without modifying the expression of the other (9). The antisense oligonucleotide sequences targeting RYR3L and RYR3S were, respectively, 5′-GAACCTCAGGTTGTAGAA-3′ (asRYR3L) and 5′-CAGTGACCAATAAC-3′ (asRYR3S); the control scrambled antisense sequence was 5′-CAGCACTATCAGTACGAC-3′ (asSCR). Freshly isolated myocytes were electroporated with oligonucleotides in PBS by a T820 electroporator (BTX). The electroporation protocol was one 200-V pulse during 10 ms in a 4-mm gap cuvette at room temperature. Myocytes were then cultured for 4 days in culture medium, and glass slides were transferred into the perfusion chamber for physiological experiments. To localize transfected cells, antisense oligonucleotides were 5′-Cy5 indocarbocyanin labeled during synthesis, and fluorescence emitted at 680 ± 32 nm (excitation: 647 nm) was recorded before Ca2+ measurements.
Cytosolic Ca2+ measurements.
Cells were loaded by an incubation in physiological solution containing 2 μM fluo-4 AM for 20 min at 37°C. These cells were washed and allowed to cleave the dye to the active fluo-4 compound for 10 min. Images were acquired using the image series of a confocal Bio-Rad MRC1024ES setup as previously described (14). Briefly, fluo-4 was excited at 488 nm, and the emitted fluorescence was filtered and measured at 540 ± 30 nm. The image series was composed of images of the same confocal section of the cell taken at 1.2-s intervals. To analyze the variation of fluorescence, regions of interest (ROIs) were drawn around each myocyte and used for each frame of the series. The mean pixel value of fluorescence of each ROI in each frame (F) was divided by the mean fluorescence of the six first frames (baseline; F0) and reported as F/F0 in a time course graph. Analyses were performed with Laser Pix Software (Bio-Rad).
Caffeine was applied by pressure ejection from a glass pipette for the periods indicated in the figures. All experiments were carried out at 26 ± 1°C.
The immunostaining protocol has been described previously (9). Briefly, myocytes were washed with PBS, fixed with 4% (vol/vol) formaldehyde and 0.05% glutaraldehyde for 10 min at room temperature, and permeabilized in PBS containing 2% BSA and 1 mg/ml saponin for 20 min. Cells were then incubated with PBS, saponin (1 mg/ml), and anti-RYR3 antibody overnight at 4°C. Cells were washed and incubated with the appropriate secondary fluo probe 488 antibody during 45 min at room temperature. After being washed in PBS, slides were mounted in Vectashield (Valbiotech). Images of stained cells were obtained with a MRC 1024ES confocal microscope. Fluorescence was acquired on each cell from z-series analysis (15 ± 5 sections) using Lasersharp software (Bio-Rad). Fluorescence was estimated by gray-level analysis using IDL software (RSI) in 0.5-μm confocal sections and expressed as volume units. Cells were compared by keeping acquisition parameters constant.
To measure the Ca2+ signal evoked by cADP-ribose, myocytes were permeabilized as previously described (14). Briefly, the physiological solution [composed of (in mM) 130 NaCl, 5.6 KCl, 10 mM HEPES, 11 glucose, 2 CaCl2, and 1 MgCl2; pH 7.4 with NaOH] was replaced by a solution containing 140 mM KCl, 20 mM HEPES, 0.5 mM MgCl2, 0.1 mM ATP, and 10 μg/ml saponin (pH 7.4 with NaOH). cADP-ribose was applied by pressure ejection from a glass pipette for the periods indicated in the figures. All experiments were carried out at 26 ± 1°C.
Chemicals and drugs.
Collagenase was obtained from Worthington Biochemical. Fluo-4 AM and fluo probe 488 antibody were from Fluoprobes (Interchim). Caffeine was from Merck. BAY K 8644 was from Bayer. Ryanodine was from Calbiochem. Medium 199, streptomycin, and penicillin were from Invitrogen. Anti-RYR3 (SC-21330), anti-CD38 (SC-7325), and horseradish peroxidase-coupled secondary antibodies were from Santa Cruz Biotechnology. All primers and phosphorothioate antisense oligonucleotides were synthesized and purchased from Eurogentec. Anti-tubulin (T-5168) antibody and all other chemicals were from Sigma.
Data are expressed as means ± SE; n represents the number of tested cells. Significance was tested by means of Student's t-test to compare both myometrium physiological states and by one-way ANOVA to compared nonelectropored with electropored cells. P values of <0.05 were considered as significant.
Both RYR3 splice variants are expressed in mouse myometrial cells.
The investigation of RYR subtypes expressed in mouse myometrium myocytes was performed using RT-PCR. In nonpregnant and pregnant myometrium (18 days of pregnancy), RYR1 and RYR2 subtypes were not detected (Fig. 1A). Both RYR3S and RYR3L splice variants were detected in mouse pregnant and nonpregnant myometrium (Fig. 1B). To compare the expression levels of both isoforms in nonpregnant and pregnant myometrium cells, we also normalized the data using the GAPDH amplicon as an internal standard. Measurements of fluorescence emitted by each amplicon were performed, and the ratio of RYR3L to GAPDH and RYR3S to GAPDH were calculated for four different dissociations (Fig. 1C). The relative expression of RYR3L against RYR3S was evaluated (RYR3L-to-RYR3S ratio was 0.48 ± 0.07 and 0.74 ± 0.09 in nonpregnant and pregnant myometrium, respectively, n = 4, 35 cycles, P < 0.05). The RYR3L-to-RYR3S ratio was significantly increased in the pregnant myometrium, suggesting that RYR3L was more expressed at the end of pregnancy.
To validate this result, the expression of RYR3 protein was evaluated by Western blot analysis (Fig. 2). The expression of RYR3 protein was significantly increased in the pregnant myometrium (Fig. 2B, left). After the membrane had been stripped, the expression of tubulin and CD38 were revealed. As previously published (10, 19), the expression of tubulin appeared constant, whereas the expression of CD38 was increased at the end of pregnancy (compared with the nonpregnant myometrium; Fig. 2A). Tubulin could be used as an internal standard to evaluate the modification of RYR3 expression. The ratio of RYR3 to tubulin was calculated. This ratio was significantly higher in pregnant myometrium than in nonpregnant myometrium, confirming the increase of RYR3 expression at the end of pregnancy. The same results were observed on four different myometrium preparations with 25, 50, and 100 μg protein, thus excluding protein loading and revelation errors.
Effects of antisense oligonucleotides directed against RYR3 isoforms.
The ability of antisense oligonucleotides to specifically inhibit the targeted RYR3 isoform was measured by RT-PCR. The expression of both RYR3 isoforms was maximally inhibited 3 days after asRYR3 electroporation, whereas the electroporation of asSCR was not able to inhibit the expression of RYR3 splice variants. In a similar manner, expressions of RYR3L and RYR3S were undetectable in cells electroporated with asRYR3L and asRYR3S, respectively, under both pregnant (Fig. 3, A and B) and nonpregnant conditions (not shown). To detect the effect of antisense oligonucleotides at the protein expression level, immunostaining of RYR3 with a specific anti-RYR3 antibody that recognized both RYR3S and RYR3L isoforms was performed 3 days after electroporation of antisense oligonucleotides in myometrium cells. The result was confirmed at the protein level. The expression of RYR3 was revealed by Western blot analysis and relatively quantified using tubulin as the standard. The electroporation of asRYR3 drastically inhibited RYR3 expression, whereas asRYR3L and RYR3S partially but significantly inhibited RYR3 expression. It was noticeable that the electroporation of all antisense oligonucleotides did not affect the expression of other proteins, as showed with CD38 (Fig. 3, C and D). As shown in Fig. 4, RYR3 was largely expressed in the whole cell with the exception of the nucleus. Both asRYR3S and asRYR3L significantly decreased RYR3 immunostaining and asRYR3 totally inhibited it (data not shown). It was noticeable that the inhibitory effect of asRYR3S on RYR3 immunostaining was greater than the asRYR3L effect in nonpregnant myometrium cells (Fig. 4, A and C).
Increase of RYR3L expression increases the caffeine-induced Ca2+ response.
As the caffeine sensitivity of the RYR3 subtype has been reported to depend on the RYR3L isoform (9, 17), we tested the effects of caffeine (10 mM) on nonpregnant and pregnant myometrial cells. In nonelectroporated or asSCR-electroporated cells, caffeine-induced Ca2+ responses were small and detectable in only 10% of nonpregnant myometrium cells (Fig. 5A). In contrast, the caffeine-induced Ca2+ response was significantly increased and was detectable in 65% of pregnant myometrium cells (Fig. 5B). SR Ca2+ overloading (obtained after 15 min in 10 mM CaCl2 external solution) significantly increased the amplitude of caffeine-induced Ca2+ responses (1.50- and 1.21-fold in nonpregnant and pregnant myometrium cells, respectively) and the number of responding cells (17% and 73% in nonpregnant and pregnant myometrium cells, respectively) as previously described (data not shown and Ref. 28). Electroporation of asRYR3L totally suppressed the caffeine-induced Ca2+ release in nonpregnant and pregnant myometrial cells (Fig. 5, A and B), whereas the electroporation of asRYR3S increased the number of cells responding to caffeine in the nonpregnant myometrium (35% of the tested cells) but not in the pregnant myometrium (65% of the tested cells). In addition, the amplitude of caffeine-induced Ca2+ signals was significantly increased in asRYR3S-electroporated cells in both nonpregnant and pregnant myometrium (Fig. 5C). These results suggest that inhibition of RYR3S in the nonpregnant myometrium was sufficient to reveal a caffeine-induced Ca2+ response, whereas in the pregnant myometrium, asRYR3S increased the amplitude of the caffeine-induced Ca2+ response but not the ability of the cells to respond to caffeine. The viability of myometrium myocytes was tested by the application of acetylcholine, which is known to activate InsP3R activation. In all tested dissociations of pregnant or nonpregnant myometrium, the acetylcholine-induced Ca2+ response was measurable in 80–90% of cells (data not shown).
Antagonist roles of RYR3L and RYR3S in cADP-ribose-induced Ca2+ signaling.
RYR3 has been proposed as one of the targets of cADP-ribose (for a review, see Ref. 15), which is a second messenger produced by oxytocin during labor (3, 10). We tested the hypothesis that RYR3L, upregulated during pregnancy, could mediate the cADP-ribose-induced Ca2+ release. In permeabilized cells from the nonpregnant myometrium, the application of 1 μM cADP-ribose induced similar Ca2+ responses in nonelectroporated and asSCR-electroporated cells (12% of tested cells). Larger Ca2+ responses were measured in asRYR3S-electroporated myocytes (65% of tested cells), whereas cADP-ribose failed to activate Ca2+ release in asRYR3L-electroporated myocytes (Fig. 6). In pregnant myometrial cells, cADP-ribose evoked a Ca2+ signal in 75% of nonelectroporated and asSCR- and asRYR3S-electroporated cells. Moreover, in cells electroporated with asRYR3S, amplitudes of cADP-ribose-induced Ca2+ responses were increased by 2- and 2.4-fold in nonpregnant and pregnant myometrium, respectively (Fig. 6). Nevertheless, these highly responsive cells from the pregnant myometrium were unable to produce significant cADP-ribose-induced Ca2+ signals when electropored with asRYR3L (Fig. 6C). These results suggested that cADP-ribose was able to activate the SR Ca2+ release via activation of RYR3L.
In the present study, we report an increase of expression of RYR3L in the myometrium during pregnancy. The functional consequence of this regulation is an increase of the SR Ca2+ release activated by caffeine and cADP-ribose, an intracellular messenger potentially involved in labor induction.
Our results showed that, in the mouse myometrium at the end of pregnancy, the expression of RYR3 was increased at the protein level and the ratio of RYR3L to RYR3S was increased at the mRNA expression level. Myometrium expression of several plasma membrane channels such as Cav3.2 and Slo are also regulated during pregnancy (33, 42). A change in the RYR3L-to-RYR3S ratio suggests that alternative splicing might be under control of hormones and that this mechanism could play a significant role at the end of pregnancy by modulating RYR3-dependent Ca2+ signaling.
It has been reported that the RYR3S isoform did not display Ca2+ channel activity, whereas the RYR3L isoform was able to release Ca2+ from the endoplasmic reticulum in HEK-293 cells (17) as well as in native smooth muscle cells (9). In nonpregnant and pregnant myometrial cells expressing only RYR3S (electroporated with asRYR3L), Ca2+ responses to caffeine could not be recorded, further supporting the nonfunctional status of RYR3S as a Ca2+ release channel. In contrast, electroporation of asRYR3S increased the amplitude of caffeine-induced Ca2+ responses, suggesting that, in the myometrium, a functional interaction between RYR3L and RYR3S can exist in native cells and that RYR3S functions as a dominant negative toward RYR3L. These results differ from those obtained in native duodenal myocytes, where no functional interaction between RYR3S and RYR3L was observed (9). In these duodenum myocytes, different subcellular localization of RYR3L and RYR3S has been reported. In contrast, in myometrium cells, RYR3L and RYR3S were expressed all over the cytoplasm, allowing an interaction between both isoforms. Localization of RYR3 isoforms thus appears to be an important parameter that determines the Ca2+ signals encoded by RYR3L. We show here that the level of RYR3L expression controls the amplitude of caffeine-induced Ca2+ responses. A previous study (1) also reported that the ability of the cell to present spontaneous Ca2+ signals is a function of RYR3L expression. These results support the idea that the increased ability to produce Ca2+ signals at the end of pregnancy might be due to the increased level of RYR3L expression. The comparison of caffeine-induced Ca2+ responses between nonpregnant and pregnant myometrium myocytes indicates that the increase of RYR3L expression can be sufficient to increase the release of stored Ca2+ and the percentage of cells responding to caffeine in pregnant myocytes. The expression pattern of RYR3 splice variants appears important to modulate Ca2+ signaling in isolated myocytes, as reported in cardiac Purkinje cells for RYR subtypes (38) or in portal vein myocytes for RYR and InsP3R subtypes (8, 30).
The RYR3 subtype has been involved in cADP-ribose-induced Ca2+ release (15). In both pregnant and nonpregnant myometrium, cADP-ribose-induced Ca2+ responses were abolished in asRYR3L-electroporated cells and increased in asRYR3S-electroporated cells, indicating that RYR3L is activated by cADP-ribose and that RYR3S again negatively controls RYR3L. In the myometrium, the significance of the cADP-ribose pathway for labor is demonstrated by the ability of oxytocin to induce cADP-ribose production and the increase of CD38 expression during pregnancy (3, 10). Here, we suggest another argument: RYR3L, the channel target of cADP-ribose, is also more expressed during pregnancy.
Our results were obtained in cultured cells from the mouse myometrium. Several human and rat myometrium studies (2, 7, 16) concluded on the absence of RYR function in Ca2+ signals. In smooth muscles, the functions of RYRs are principally to encode Ca2+ sparks and Ca2+ waves to regulate contraction (21). In this study as in others, Ca2+ sparks were never observed (7) and caffeine activated Ca2+ waves in a weak proportion of cells in the nonpregnant myometrium under control conditions. The low level of RYR expression in the myometrium was proposed to explain why the RYR function was poorly or not involved in Ca2+ signaling (23). Here, we showed that the investigation of RYR-dependent Ca2+ signal in the myometrium requires measurement of tens to hundreds of cells to compensate for this weak proportion of answering cells. RYR function is also regulated by the level of SR Ca2+ loading (27, 28). Finally, the presence and amplitude of RYR-dependent Ca2+ response are dependent on the relative expression of dominant negative RYR3S compared with RYR3L (this study and Ref. 1) or the RYR2 subtype (9). At the end of pregnancy, the increase of RYR3L expression is responsible for the increases of activated Ca2+ signal amplitude and number of responding cells.
The functional significance of RYRs in the myometrium was recently reinforced by studies (3, 10, 40) demonstrating the activation and regulation of the cADP-ribose pathway during pregnancy in the mouse, rat, and human. Here, we indicate that the expression of the functional target of this transduction pathway is also increased.
The physiological function of RYR3 splice variants should be now investigated in myometrium tissues because some differences could be observed in Ca2+ signaling in isolated cells versus in muscle bundles (7).
In conclusion, we demonstrated that the RYR3S isoform is able to negatively regulate the RYR3L isoform in native cells expressing only the two RYR3 isoforms, and we suggest that splice variants of RYR3 may have important roles in the myometrium during labor. Indeed, we demonstrated that the increased expression of RYR3L (the functional RYR3 isoform) in myometrium myocytes at the end of pregnancy may support the increased ability of cADP-ribose to release stored Ca2+. The physiological regulation of RYR3 alternative splicing can be associated with labor preparation. In contrast, in nonpregnant myocytes, the RYR3L-to-RYR3S ratio allows the control of RYR3L activation by its dominant negative partner, RYR3S, to inhibit the release of stored Ca2+.
This work was supported by grants from the Centre National de la Recherche Scientifique and Centre National des Etudes Spatiales and by Agence Nationale de la Recherche Grant ANR 05-PCOD-029.
The authors thank N. Biendon and J. L. Lavie for technical assistance.
Present address of F. Dabertrand and N. Macrez: Centre de Neurosciences Intégratives et Cognitives, UMR5228 CNRS, Université Bordeaux 1 and Université Bordeaux 2, Avenue des Facultés, Talence 33405, France.
Present address of N. Fritz: Medical Biochemistry and Biophysics, Karolinska Institut, MBB, Scheeles väg 2, Stockholm 171 77, Sweden.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society