The effects of endogenous and exogenous nitric oxide on gut motility in zebrafish Danio rerio embryos and larvae

Information about the development of gut motility in vertebrates is relatively sparse, comprising a few studies on mammals(Morriss et al., 1986; Bisset et al., 1988; Sase et al., 2000; Acosta et al., 2002; Oyachi et al., 2003; Anderson et al., 2004) and teleosts (Rønnestad et al.,2000; Holmberg et al.,2003). Zebrafish Danio rerio is becoming a valuable model animal for developmental studies, not the least when studying development of gut motility in vivo. The external fertilization and development of zebrafish eggs and embryos, and the transparency of the gut wall of the developing zebrafish, allow us to use a non-invasive video technique to study the gut motility from the earliest stages(Holmberg et al., 2003; Holmberg et al., 2004). Thus we have obtained new physiological data that together with existing molecular and morphological information can contribute to a better understanding of gut development.

In a previous study on zebrafish, we observed sporadic non-propagating contractions from the first day investigated (3 days post fertilization,d.p.f.), and at 4 d.p.f. spontaneous, regular contraction waves were seen(Holmberg et al., 2003). These propagating contraction waves coincide with the first occurrence of a continuous sheet of circular smooth muscle around and along the gut(Wallace et al., 2005). This is before the zebrafish start exogenous feeding, but gut motility might be of importance in young non-feeding zebrafish to transport slugged off intestinal material or to prevent bacterial overgrowth. A similar pattern of sweeping contractions occurs in-between meals in adult vertebrates, the so-called migrating motor complexes (MMCs)(Szurszewski, 1969).

Gut motility is primarily controlled by the enteric nervous system. A variety of neurotransmitters have been identified in the developing gut in mammals (Gintzler et al.,1980; Rothman and Gershon,1982; Larsson et al.,1987; Timmermans et al.,1994; Brandt et al.,1996; van Ginneken et al.,1998), birds (Epstein et al.,1985; Rothman et al.,1986; Balaskas et al.,1995), amphibians (Holmberg et al., 2001; Maake et al.,2001; Badawy and Reinecke,2003) and teleosts (Reinecke et al., 1997; Villani,1999; Poon et al.,2003; Holmberg et al.,2004; Holmqvist et al.,2004; Pederzoli et al.,2004). In the zebrafish, enteric neurons are present in the gut before the onset of exogenous feeding(Raible et al., 1992; Bisgrove et al., 1997; Holmberg et al., 2003). Only a few of the above studies have correlated neurotransmitter findings with their effect on gut motility at an early stage. However, in zebrafish, neurons expressing NKA (neurokinin A) and PACAP (pituitary adenylate cyclase-activating polypeptide) have been detected in the developing gut from 2 d.p.f., and from 5 d.p.f. the frequency of the anterograde contraction waves were increased and decreased by the application of NKA and PACAP,respectively. In addition, a cholinergic tonus was observed from 4 d.p.f.(Holmberg et al., 2004).

Another important transmitter is nitric oxide (NO), which generally has an inhibitory effect on smooth muscle cells, including those of the teleost gut(Karila and Holmgren, 1995; Olsson and Holmgren, 2000)(reviewed in Olsson and Holmgren,2001). Further, NOS (nitric oxide synthase), the enzyme responsible for NO synthesis, has been identified in gut neurons of several adult vertebrates including teleosts (Li and Furness, 1993; Olsson and Karila, 1995) (reviewed in Olsson and Holmgren, 2001). NOS is also present in interstitial cells of Cajal (ICCs) in mammals and are thought to be the pacemaker cells of the gut, but also act as relay cells for both excitatory and inhibitory neurotransmission to gut smooth muscle. In developing zebrafish larvae, neuronal NOS is expressed in nerves innervating peripheral organs, including enteric ganglia(Poon et al., 2003; Holmqvist et al., 2004). However, the function of NO in the developing gut is not known.

Our aim was to study the development of the nitrergic (i.e. nitric oxide releasing) control system in zebrafish gut, in vivo, by using micro-applications and a video technique.

It was hypothesized that endogenous/exogenous NO is inhibitory on gut motility in zebrafish, as in most other vertebrates. If so, and if the appropriate receptors are present, treatment with the NO donor SNP (sodium nitroprusside) will reduce the frequency of gut contraction waves. Furthermore, if the application of the NOS blocker l-NAME(N^G-nitro-l-arginine methyl ester), stimulates the gut motility, this would indicate an endogenous nitrergic tonus in the animal.

The study protocols were approved by the Animal Ethics Committee of Göteborg.

Adult zebrafish Danio rerio Hamilton were purchased from a local supplier and were kept in a controlled environment (25°C, 12 h:12 h light:dark).

Embryos (3 and 4 d.p.f.) and larvae (5 and 6 d.p.f.) from several batches were used in the study (N=6–12 for each d.p.f. stage). Mature zebrafish were allowed to breed spontaneously in an aquarium with a breeding box. On the day of fertilization (0 d.p.f.) eggs were collected and transferred to small beakers kept at a water temperature of 28°C. During the experimental period, the embryos and larvae were not fed. Active feeding at 28°C usually starts around 5–6 d.p.f., and at 6 d.p.f. most of the yolk is depleted.

Embryos and larvae were anaesthetized in phosphate-buffered MS222(3-aminobenzoic acid ethyl ester; Sigma, St Louis, MO, USA, 75–100 mg l^–1, pH 7) and embedded lying on one side in an agarose solution (type VII, Sigma, 1%, gelling point 26–30°C, dissolved in phosphate-buffered MS222). The agarose was allowed to set at room temperature and was covered with the MS222 solution in order to keep the fish anaesthetized. The gut was observed in vivo, using an inverted microscope (Nikon, 10× magnification).

The Optimas imaging system has been described in detail earlier(Schwerte and Pelster, 2000). Live video recordings of the gut were made and a digital film sequence was created by extracting still images (1 s^–1). For the making and analysis of the film sequences, the Optimas program package (Media Cybernetics, Gleichen, Germany) was used. From the film, the number of propagating anterograde and retrograde contractions waves along the gut were counted (Fig. 1) and the average frequency was calculated as cycles min^–1. The data were exported to Excel for further analysis.

The l-arginine analogue l-NAME(N^G-nitro-l-arginine methyl ester; Sigma,10^–3 mol l^–1) that blocks the enzyme NOS was used to reduce the production of endogenous NO, while SNP (sodium nitroprusside; Sigma, 10^–4 mol l^–1) was used as a NO donor.

To avoid possible activation of sensory systems by stretching the gut, the drugs were applied immediately outside the body wall and allowed to diffuse into the experimental animal. Boluses between 150 and 300 nl of drug solution were applied next to the abdomen at three different locations along the gut. Effects of the application per se and normally occurring variations in gut frequency over time were studied by adding the same volume of saline(NaCl, 0.9%).

After a 5-min control period, saline was added and the activity recorded for 5 min. Subsequently, the NOS inhibitor l-NAME(10^–3 mol l^–1) was applied and the effects on contraction frequency were recorded for 20 min. Thereafter, the NO donor SNP (10^–4 mol l^–1) was applied and the effects were recorded for 9 min. For control purposes, the effects of repeated applications and time were studied in a second set of experiments, where saline was applied instead of l-NAME.

The basal contraction frequency during the control period as well as after saline application was calculated over 5 min. No significant change in contraction frequency was observed before or after saline application. Therefore the contraction frequency during the first saline application was subtracted from all further effects obtained, in order to standardise the response to drug or second saline application. 9 min after l-NAME or the second saline application, the maximal contraction wave frequency over a 6-min period was calculated and was statistically analysed in comparison to the previous saline application.

The inhibitory effect of SNP was compared to the increased frequency obtained after exposure to l-NAME. Hence, animals that showed no activity before the SNP applications were discarded.

To see the effects of the application per se, the frequency during a 3-min period directly after each saline application was compared with the first 3 min of the control period. For normal variation over time, the first 6 min after the second saline application were compared with the maximal contraction frequency 9 min after saline application (during 6 min).

Mean values ± s.e.m. were calculated for each d.p.f. group. Values were analysed statistically by using Wilcoxon Signed Ranks, matched-pairs test(SPSS 12.0 for Windows). Repetitive uses of experimental groups were taken into account. Differences in mean values were regarded as significant at P<0.05.

The middle intestine (MI), posterior to the intestinal bulb, was dissected out (N=7) and placed in cold zebrafish Ringer's solution (composition in mmol l^–1: NaCl 116, KCl 2.9, CaCl₂ 1.8, Hepes 5, glucose 11, pH 7.2). Ring preparations (3–4 mm wide) were mounted in organ baths containing zebrafish Ringer's solution (22°C, bubbled with 0.3% CO₂ in air). The force developed by the smooth muscle was recorded using a force displacement transducer (model FT03, Grass Instruments,West Warwick, RI, USA) connected to a polygraph (model 7, Grass). An initial force of 0.5 mN was applied and the preparations were left for 1–2 h to develop a steady baseline (resting tone). A single dose of l-NAME(3×10^–4 mol l^–1) was applied to the organ bath. When maximal response to l-NAME was obtained, after approximately 20–25 min, SNP was applied at increasing concentrations in a cumulative fashion, allowing maximal response to each concentration to be obtained before addition of a higher concentration(10^–7–10^–5 mol l^–1).

Changes in force developed by the strip preparations were sampled on a computer (Labview, acquisition software). A control period of 1 min was recorded before the addition of drug. The basal frequency was subtracted from all further calculations. The response to each drug added was calculated as the mean force during 1 min of peak response to the drug. The effects of l-NAME were calculated in relation to the control (spontaneous)activity, which was set to 100%. The inhibitory effect of SNP was calculated in relation to the increased tonus obtained after exposure to l-NAME. Differences in mean values were regarded as significant at P<0.05.

Embryos and larvae from 2, 3, 5 and 7 d.p.f. (N=4 from each stage)were anaesthetized in 0.01% MS222 and fixed for 24 h in Zamboni's fixative(1.5% picric acid, 2% formaldehyde in 0.1 mol l^–1 phosphate buffer, pH 7.2). Adult zebrafish were anaesthetized in 0.1% MS222, decapitated and the proximal (PI), middle (MI) and distal intestine (DI) were dissected out and fixed as above. The fixative was removed by repeated washing in 80%ethanol, followed by dehydration in 95 and 99.5% ethanol, xylene treatment,and rehydration in an ethanol series (99.5%, 95%, 80%, 50%) to phosphate-buffered saline (0.1 mol l^–1 PBS, 0.9% NaCl) for 30 min for each step. The fixed tissues were stored in a PBS with 30% sucrose solution at least overnight before preparation for sectioning.

Whole embryos and larvae were positioned in 1.5% agarose–5% sucrose solution, which was left to set in room temperature. The agarose blocks were placed in PBS–sucrose solution at 4°C until they had sunk to the bottom and were then quick-frozen in isopentane chilled by liquid nitrogen. Tissue pieces from adult intestines were embedded in embedding medium (OCT;Sakura, Zoeterwoude, The Netherlands) and frozen as above. Preparations were sectioned on a cryostat (Zeiss Micron International GmbH, Walldorf, Germany),at 16 μm (embryos and larvae) or 4 μm (adults) and picked up on gelatine-coated slides, left to dry in darkness overnight and then stored at–20°C.

The sections were pre-incubated with normal donkey serum (10%, Jackson ImmunoResearch Laboratories, West Grove, PA, USA) for 30 min in a moist chamber to reduce non-specific staining. They were then incubated with an antibody raised against rat neuronal NOS (1:100, 31030, Transduction Laboratories, Pharmingen, Germany) for 2 days in a moist chamber at room temperature. The specificity of the antisera has been reported in earlier studies on fish (e.g. Karila et al.,1997). Excess antibody was washed away with phosphate buffer containing 2.0% NaCl three times before incubating with a secondary antibody conjugated to either indocarbocyanine (Cy3; 1:800, 711-165-157, Jackson ImmunoResearch Laboratories) or fluorescein isothiocyanate (FITC; 1:100,711-095-152, Jackson ImmunoResearch Laboratories) for 1 h. The sections were washed as described above, mounted in Vector H-1000 medium and examined with a Nikon Eclipse E1000 digital fluorescence microscope equipped with a Nikon digital camera DXM1200 and Nikon's software, ACT1. Contrast and brightness were adjusted and mounts were made using Adobe PhotoShop 6.0.

General observations showed a regular spontaneous motility pattern of anterograde and retrograde anterior contraction waves from 4 d.p.f. However,as noted before, in all age groups investigated there were specimens with no activity during the control period. This is probably due to a normal individual difference in the development(Holmberg et al., 2003).

In agreement with earlier studies(Holmberg et al., 2004),single or repeated applications of saline per se did not affect the frequency of contraction waves compared to the control period. Furthermore, no changes in gut motility over time were seen after the second saline application in comparison with first saline application (anterograde, 4 d.p.f., 0.15±0.13, N=8, P=0.40; 5–6 d.p.f.,0.42±0.41, N=8, P=0.64: retrograde 4 d.p.f.,–0.35±0.33, N=8, P=0.54; 5–6 d.p.f.,0.76±0.33, N=8, P=0.1).

Anterograde contraction waves originate immediately behind the intestinal bulb and spread in an anal direction along the middle intestine(Fig. 1). The basal contraction frequency was 1.05±0.15 (N=18) and 0.95±0.13(N=20) contractions min^–1 at 4 and 5–6 d.p.f.,respectively. No propagating contractions were seen at 3 d.p.f.

Prevention of endogenous NO formation by l-NAME (three boluses of 50–100 nl, 10^–3 mol l^–1) increased the anterograde contraction wave frequency (compared to saline application)from 4 d.p.f. and onward, indicating an endogenous nitrergic tonus before the zebrafish commence exogenous feeding (Fig. 2). The increase at 4 d.p.f. was 0.50±0.10 cycles min^–1 (N=10)(Fig. 2A), and at 5–6 d.p.f. the increase was 0.31±0.08 cycles min^–1(N=12) (Fig. 2B). No effect of l-NAME was seen at 3 d.p.f. (0.15±0.12 cycles min^–1; N=6, P=0.283).

The application of the NO donor SNP (three boluses of 50–100 nl,10^–4 mol l^–1, after a previous application of l-NAME) reduced the contraction frequency from 4 d.p.f.onward,showing that functional reaction pathways for NO are present before onset of exogenous feeding (Fig. 2). At 4 d.p.f. the reduction was –0.71±0.20 cycles min^–1 (N=7) (Fig. 2A) compared to 3 min prior to application (from an increase of 0.60±0.10 cycles min^–1 after l-NAME to–0.12± 0.21 cycles min^–1 in comparison to saline). At 5–6 d.p.f., SNP decreased contraction frequency with–0.64±0.12 cycles min^–1 compared to 3 min prior to SNP application (from an increase of 0.42±0.14 cycles min^–1 to –0.21±0.12 cycles min^–1 in comparison to saline, N=11)(Fig. 2B). At 3 d.p.f., no gut motility was observed; therefore no conclusion of an inhibitory effect of SNP at this stage can be drawn from the experiments.

The anterior retrograde contraction waves originate from the same general area as the anterior anterograde contraction waves, i.e. immediately behind the intestinal bulb (Fig. 1). The retrograde contraction waves travel in an oral direction along the proximal intestine (intestinal bulb). The basal contraction frequency was 2.06±0.39 cycles min^–1 (N=18) and 1.62±0.28 cycles min^–1 (N=20) at 4 and 5–6 d.p.f., respectively. No propagating contractions were seen at 3 d.p.f.

At 5–6 d.p.f., application of l-NAME (three boluses of 50–100 nl, 10^–3 mol l^–1) increased the frequency of retrograde waves with 0.39±0.15 cycles min^–1 compared to saline (N=12)(Fig. 2D), indicating the tonic release of (inhibitory) endogenous nitric oxide in the larvae. In the younger stages l-NAME did not induce an increase in contraction frequency in comparison to saline (3 d.p.f., 0.39±0.57 cycles min^–1, N=6, P=0.18; 4 d.p.f.,0.33±0.26 cycles min^–1, N=9, P=26)(Fig. 2C). At 5–6 d.p.f.,4 out of 12 individuals were unaffected by l-NAME.

The application of the NO donor SNP (three boluses of 50–100 nl,10^–4 mol l^–1, after a previous application of l-NAME) reduced the contraction frequency to 1.61±0.36 cycles min^–1 compared to the frequency 3 min prior to the SNP application at 5–6 d.p.f. (from an increase of 0.75±0.20 to a decrease of 0.86±0.44 in comparison to saline)(Fig. 2D). Before the stage of onset of feeding, SNP did not affect the frequency of retrograde contraction waves (3 d.p.f., from 0.33±1.26 to 0.00±0.00, N=6: 4 d.p.f., from 0.96±0.42 to –0.54±0.34, N=4, P=0.07, all values are in comparison with saline)(Fig. 2C).

Strip preparations adopted a tonus of 12.8±1.7 mN (N=7). However, spontaneous rhythmic activity was generally not observed in the preparations. Prevention of NO formation by l-NAME(3×10^–4 mol l^–1, N=7)increased the mean force exerted by the preparations to 130.8±6.7% of control, mainly by increasing the basal tonus. In the majority of strip preparations, the rhythmic contraction pattern that was induced by l-NAME persisted after washout of l-NAME. Subsequent addition of SNP, 10^–5 mol l^–1 (N=5)decreased the l-NAME induced tension to 71.2±27.0% of control period (Fig. 3) while lower concentrations had no effects.

At 2 d.p.f., no NOS-like immunoreactive material could be detected along the gut while at 3 d.p.f., neurons showing NOS-like immunoreactivity were distinguished in the middle and distal parts of the intestine(Fig. 4A). From 4 d.p.f.,NOS-positive neurons were found throughout the gut, however, there was a higher number of neurons in the more distal parts of the gut compared to the proximal part (Fig. 4B–D). This difference persisted until 7 d.p.f. when no major difference could be observed between the different parts of the gut. In the adult animal, NOS-like immunoreactivity was detected in nerve fibres and myenteric nerve cell bodies throughout the gut. Nerve fibres were seen in all layers of the gut, with the highest density in the myenteric plexus and the circular muscle layer (Fig. 4E).

To our knowledge, this is the first report of a nitrergic tonus in the gastrointestinal tract of any vertebrate around the onset of exogenous feeding. Both endogenously released and exogenously applied nitric oxide reduces gut motility in zebrafish larvae. We also report an inhibitory effect of NO on adult zebrafish intestine, which is in agreement with earlier studies on adult vertebrates, including teleosts(Karila and Holmgren, 1995; Olsson and Holmgren, 2000)(reviewed in Olsson and Holmgren,2001).

In the present study, NOS was detected in neurons throughout the gut before onset of exogenous feeding. Similar results have been reported from studies in prenatal mammals (Timmermans et al.,1994; Brandt et al.,1996; van Ginneken et al.,1998) and birds (Balaskas et al., 1995) as well as in amphibian(Holmberg et al., 2001) and teleost embryos and larvae (Villani,1999; Poon et al.,2003; Holmqvist et al.,2004). The latter studies include zebrafish; however, the difference in time of appearance of NOS positive neurons between the middle-distal and proximal intestine observed in this study has not been reported previously. This variation may be reflected in the physiological development of motility (see below).

The early appearance of neurotransmitters in the gut seems to be a common feature to several neurotransmitters, within and between species. The first observation of detectable levels of NOS in the present study correlates well with the occurrence of other putative neurotransmitters, both excitatory (NKA,acetylcholine) and inhibitory (PACAP), in zebrafish larvae(Holmberg et al., 2004). In addition, a variety of inhibitory and excitatory neurotransmitters have been observed in other vertebrate species including mammals, amphibians and teleosts before or around the onset of exogenous feeding(Gintzler et al., 1980; Rothman and Gershon, 1982; Saffrey et al., 1982; Epstein et al., 1985; Huang et al., 1986; Saffrey and Burnstock, 1988; Reinecke et al., 1997; Salvi et al., 1999; Holmberg et al., 2001; Maake et al., 2001; Badawy and Reinecke, 2003; Holmberg et al., 2004; Pederzoli et al., 2004).

One limitation of most of the above studies is that they are strictly morphological, and functional data are lacking. Even if the neurotransmitters are present at an early stage, they might not be released, or the appropriate receptors on smooth muscle cells, ICCs or neurons might not be expressed or fully functional. In contrast, in our previous study we could correlate the effects of two putative neurotransmitters, with their appearance in neurons. The results showed that in vivo application of NKA and PACAP did not affect anterograde contractions until 4 d.p.f., i.e. at least 2 days later than they were immunohistochemically detected in neurons(Holmberg et al., 2004). Similarly, in the present study, NOS is expressed in neurons at least 1 day before NO has an effect on propagating gut motility. This time lapse between occurrence and effect of the neurotransmitters might be explained by the fact that the smooth muscle cells are not aligned to form a continuous circular smooth muscle sheet until 4 d.p.f.(Wallace et al., 2005). Consequently, contraction waves cannot propagate for longer distances at earlier stages. This is in agreement with our previous work showing that at 3 d.p.f., non-propagating irregular gut activity is seen in the zebrafish gut(Holmberg et al., 2003).

For a fully functional gut motility, a fine-tuned cooperation between excitatory and inhibitory mechanisms is needed. Earlier studies have shown the presence of an excitatory cholinergic tonus that modulates contraction frequency from 4 d.p.f. in developing zebrafish larvae(Holmberg et al., 2004). Likewise, a functioning cholinergic regulation of gut motility was observed in foetal guinea pig and rabbit (Gintzler et al., 1980; Acosta et al.,2002; Oyachi et al.,2003). In the zebrafish, our present results suggest that an inhibitory nitrergic tonus starts to modulate anterior anterograde contraction waves at the same time as the stimulatory cholinergic tonus develops. These two mechanisms probably cooperate to keep the gut activity at a desired contraction frequency.

It is notable that the onset of the effects of NO on retrograde contraction waves (spreading orally over the intestinal bulb) is lagging 1 day behind the effects on anterior contraction waves. A similar lag is seen for the appearance of NOS positive neurons. Such cells were first observed in the middle and distal part at 3 d.p.f. while occurrence in the proximal intestine was delayed 1 day. Further studies are needed to elucidate the significance of the difference in reactivity and expression.

The function of spontaneous contractions in non-feeding fish larvae is so far unknown. They can be considered as training for future mixing and transport of foodstuff. However, there is probably also a need for transportation of slugged-off intestinal material and prevention of bacterial overgrowth, even before feeding starts. The anus and mouth open up at 3 d.p.f., and bacteria, etc. can thus enter the gut before the animal has started to actively take in food. Similar housekeeping activity in the gut is known to occur in e.g. non-feeding adult mammals(Szurszewski, 1969). Migrating, rhythmic contractions are part of migrating motor complexes or MMCs. Anterogradely propagating contraction waves have also been identified in isolated intestine of fasting adult cod(Karila and Holmgren, 1995). The migration velocity was similar to mammalian MMCs and the frequency was approximately 0.5 cycles min^–1. In the present study, the frequency in zebrafish embryos and larva in vivo was somewhat higher,with approximately 1 cycle min^–1 in the anterograde direction and 2 cycles min^–1 for the retrograde contractions. Whether this just reflects the different experimental set-up (in vitro vs in vivo) or depends on species differences is so far difficult to tell. However, it is likely that the gut control system is immature in the embryos and larvae and hence the differences in contraction frequency are due to age and developmental stage. In mammals, ICCs, nerves and smooth muscle have not reached adult maturity at birth, and although slow waves are present at that stage gut motility will continue to develop during the period immediately after birth (Daniel and Wang,1999; Torihashi et al.,1997).

ICCs are believed to be involved in the control of gut motility in both the fed and fasted state in mammals. In mouse, ICCs in the muscular layer (ICC-IM)express NOS and enhance nitrergic neurotransmission by releasing NO(Ward et al., 2000; Burns el al., 1996). In the present study, NOS appears to be present in neurons only but further studies are needed to elucidate all sources of the endogenous nitric oxide in zebrafish.

In conclusion, the present study shows that besides an excitatory cholinergic tonus (Holmberg et al.,2004) there is a nitrergic inhibiting tonus present in zebrafish from just before or at the onset of exogenous feeding. There is probably a co-functionality between these two pathways to balance sweeping gut contractions to a desired frequency.

 
• DI

  distal intestine

 
• d.p.f.

  days post fertilisation

 
• ICCs

  interstitial cells of Cajal

 
• l-NAME

  N^G-nitro-l-arginine methyl ester

 
• MI

  middle intestine

 
• MMCs

  migrating motor complexes

 
• MS222

  3-aminobenzoic acid ethyl ester

 
• NKA

  neurokinin A

 
• NO

  nitric oxide

 
• NOS

  nitric oxide synthase

 
• PACAP

  pituitary adenylate cyclase-activating polypeptide

 
• PI

  proximal intestine

 
• SNP

  sodium nitroprusside

The authors are grateful to Prof Margareta Wallin and Ms Elisabeth Norström for the use of experimental set-up for the in vivostudies and for technical advice. This study was supported by grants from The Swedish Research Council to S. Holmgren and from the Helge Ax:son Johnsons and the Längmanska foundations to A. Holmberg.