SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Objective

Oxidized ATP (oATP) is a selective inhibitor of the P2Z/P2X7 ATP receptor for extracellular ATP, which contributes to the antinociceptive effect. This study sought to determine the mechanism by which local administration of oATP is able to relieve inflammatory pain in arthritic rat paws.

Methods

Arthritis was induced in Wistar rats by injections of Freund's complete adjuvant into one hind paw. Nociceptive thresholds were measured before and after local injection of oATP into the inflamed paws. The influence on pain transmission due to the presence of recruited inflammatory cells at the site of inflammation was determined by inhibiting the initial phase of their migration (by intravenous treatment with fucoidin, which blocks the adhesion molecules of the selectin family). ATP intraplantar content was determined in the different experimental conditions. Histologic features of the hind paws were evaluated by using the anti–P2X7 receptor polyclonal antibody.

Results

Intraplantar administration of oATP into inflamed paws significantly relieved inflammatory pain. The antinociceptive effect of oATP was independent of the immune-cell recruitment. ATP levels in inflamed tissues were significantly reduced by oATP treatment. A variable presence of P2X7 receptors on cutaneous sensory nerves with respect to the different treatments was observed. Following oATP treatment, there was a reduction in P2X7 expression in the endings of peripheral nerves, as well as in endothelial cells.

Conclusion

Oxidized ATP inhibits inflammatory pain in arthritic rats by inhibition of the P2X7 receptor for ATP, which is localized on nerve terminals.

The detection of pain (nociception) following peripheral injury is regulated by central and peripheral nervous system pathways (1–4). In inflamed tissues, the enhancement of pain through the efferent activity of afferent neurons (the axon reflex of Lewis) may be partly responsible for the sensitization of peripheral nociceptors (5, 6). ATP may be released, together with substance P and calcitonin gene–related peptide, from some sensory nerves during “axon-reflex” activity, when antidromic impulses pass down collaterals supplying blood vessels (7); these neurotransmitters cause mast cells to degranulate, thereby producing further sensory and vasomotor effects (8). In addition, the site of injury is continuously invaded by immune cells that release cytokines, which further promote inflammation and pain transmission (9).

The possibility of pain relief induced by immune cells that have migrated to peripheral damaged tissues has been proposed (10). Thus, immunocytes, following selectin-induced migration to the inflamed hind paw of the rat, were shown to produce the opioid peptide β endorphin, which, by binding to opioid receptors on sensory nerves, was able to block the pain response (11). Such an effect was influenced by the presence of endogenous (stress) or exogenous (corticotropin-releasing factor) mechanisms. The existence of these local pain-regulatory factors prompted us to investigate the possibility of inducing local antihyperalgesic effects through other endogenous systems.

The release of ATP from sympathetic nerves innervating blood vessels acts to modulate pain at the level of the P2X subtype of P2 purinoceptors (12). Since P2X subunits are present in the cell body of nociceptive neurons, they may also occur in their peripheral terminals and thus be a target for extracellular ATP released from the cytoplasm of damaged cells (13). In this context, ATP has been proposed as a depolarizing agent of the sensory nerve terminals, with the ability to initiate a nociceptive signal. Experimental data have indicated that extracellular ATP possesses pronociceptive activity. In rats, acute nociception has been shown to be mediated by P2X receptor activation in the hind paw (14), and this is augmented by inflammation and inflammatory mediators (15). Distinct P2X receptors for ATP, which are present on pain-sensing and stretch-sensing neurons, have been identified (16).

Our studies have focused on the extracellular ATP receptor P2Z/P2X7, which mediates ATP cytolytic activity on macrophages (17). Activation of the P2Z/P2X7 receptor in microglial cells, by stimulation with bacterial endotoxin, has been shown to favor interleukin-1β release (18). This receptor is selectively blocked by periodate-oxidized ATP (oATP) (18, 19).

In our initial studies, we examined the effect of local administration of oATP into the rat hind paw on the peripheral nociceptive response following an inflammatory stimulus. We induced a unilateral inflammation in the rat hind paw by intraplantar injection of Freund's complete adjuvant (CFA), and showed that local treatment with oATP significantly reduced inflammatory pain (20). In the present study, we found that this antinociceptive effect is related to inhibition of the action of P2X7 receptors localized on nerve terminals and on endothelial cells, which become unable to release ATP. The effect was independent of the presence of immune cells. Such results demonstrate that ATP exerts a key role in the pathophysiology of peripheral pain related to inflammation, and that oATP may be effective in treating such inflammatory pain.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The experiments reported herein followed the guidelines of the International Association for the Study of Pain (21, 22).

Animals and induction of inflammation.

Male Wistar rats (Charles River Italia, Calco, Italy) weighing ∼250 gm were used. Standard rodent chow and water were available ad libitum. A 12/12-hour light/dark cycle was used, and testing was conducted in the light phase. The rats were prepared for injection by brief exposure to halothane anesthesia (Hoechst, Milan, Italy), and subsequently an intraplantar injection of 0.15 ml CFA (Sigma-Aldrich, Milan, Italy) was administered into the right hind paw.

Behavioral experiments.

Paw pressure thresholds (PPTs), or hyperalgesia, was assessed as the pressure required to elicit paw withdrawal, and a cut-off value of 250 gm was established. The rats were familiarized with the handling procedure 3 days before the antinociceptive test. Nociceptive thresholds were determined using a paw pressure analgesimeter (Ugo Basile, Comerio, Italy), as previously reported (10, 20). Before any injection, baseline PPTs were assessed in 3 consecutive trials, separated by 10-second intervals, and the average PPT was calculated. PPTs were then measured at various intervals (from 0.5 hours to 48 hours) after CFA injection.

Experimental protocol and drug administration.

Intraplantar and intravenous injections were given to the rats following brief exposure to halothane anesthesia. The oATP was a kind gift from Dr. S. Hanau (University of Ferrara, Italy). It was synthesized as previously reported (19). The oATP was dissolved in sterile saline in a volume of 0.15 ml. Previously, we studied the analgesic effect following intraplantar injection of oATP, at different concentrations, into inflamed hind paws (20) and the oATP was found to induce peripheral antinociception as evaluated using the paw pressure test (Italian patent application no. MI2000A001827 and PCT/EP01/08643). In the present study, the rats, 24 hours after CFA injection, received the same amount of oATP (168 μM) in both hind paws (inflamed and noninflamed). Another group of animals received the vehicle of the drug (0.15 ml of sterile saline) in both hind paws. The number of rats used per group was between 7 and 9.

Paw edema was measured with a plethysmometer (Ugo Basile), as previously reported (10), at 3, 12, and 24 hours following inoculation of saline or 168 μM oATP in the paws of rats in which inflammation was induced 24 hours earlier by CFA injection. We also tested the nociceptive function of ATP injection, both in the inflamed paws of rats 24 hours after CFA injection and in the noninflamed paws. ATP (0.01 mmoles; Sigma-Aldrich) was dissolved in 0.15 ml sterile saline and injected into rat paws.

In order to examine the possible role of inflammatory cells recruited to the site of injury (10), we inhibited immunocyte extravasation into inflamed tissues. Immunocyte recruitment is a multistep process that is based on sequential activation of adhesion molecules located on immune and endothelial cells. The adhesion molecules of the selectin family mediate the initial phase (“rolling”) of leukocyte diapedesis. In these experiments, PPT determinations were continued for ∼0.5 hours, and then rats were pretreated with the polymer of L-fucose, fucoidin (Sigma-Aldrich), which belongs to a group of sugar analogs that potently bind to selectins. Fucoidin (10 mg/kg body weight/0.2 ml sterile saline) was administered into the femoral vein of rats 0.5 hours before CFA injection (i.e., inflammation was induced at 1 hour from time 0) into the right hind paw. The PPT was measured in both paws, in which 168 μM oATP was injected 3 hours later (i.e., 4 hours from time 0). We measured PPTs for another 12 hours in inflamed and noninflamed paws.

Assay of ATP content in the rat paw.

We determined, in a separate group of rats, the modifications in ATP content induced in the plantar tissue by the inflammatory process and/or by oATP treatment. At established times, subcutaneous paw tissues were removed and rapidly frozen in liquid nitrogen, with the aim of blocking any metabolic activity. The frozen tissue was homogenized with a polytron (Kinematica, Lucerne, Switzerland) in ice-cold 6% (weight/volume) HClO4 to extract nucleotides. The homogenate was centrifuged and the supernatant was used for ATP determination, following the procedure previously described (23). ATP assay was performed using the luminescence method (24).

Immunohistochemistry.

Expression and localization of the ATP receptor P2X7 was tested on 16 formalin-fixed hind paws from 8 rats. Following administration of halothane anesthesia, 4 paws were treated with oATP, 3 with CFA and oATP, 3 with sterile saline, 3 with CFA, and 3 with ATP. Sections (4μ thick) were prepared on polylysined slides. After 30 minutes in xylol, sections were rehydrated using alcohol solutions (99%, 95%, 70%, and 50%) and incubated 30 minutes at room temperature in a 0.3% H2O2 solution to block endogenous peroxidases. After rinsing in phosphate buffered saline, sections were incubated in 1:20-diluted normal serum solution for 50 minutes at room temperature, and then for 2 hours with rabbit anti–P2X7 receptor–purified polyclonal antibody (Chemicon International, Temecula, CA) at room temperature. Binding was developed using peroxidase-conjugated streptavidin (Vectastain Elite ABC kit; Vector, Burlingame, CA), and peroxidase was revealed by 3′,3′-diaminobenzidine (Liquid Dab; BioGenex, San Ramon, CA).

A double-staining method was applied to examine the relationship between P2X7 and cutaneous terminal nerve fibers (25), using a monoclonal antibody to neurofilaments (Chemicon International). Sections were then treated as described for the single stain, from the beginning of the procedure until chromogen development. After staining with P2X7-specific polyclonal antibody, sections were sequentially incubated in 1:20-diluted normal serum, and then for 2 hours with 1:100-diluted monoclonal neurofilament antibody at room temperature, developed using an avidin–biotin–alkaline phosphatase complex (Vectastain Elite ABC kit) and revealed using 5-bromochloroindoxyl phosphate and nitroblue tetrazolium chloride (Dako, Copenhagen, Denmark). Experiments were performed by alternating incubation between the 2 antibodies. Sections were counterstained with Harry's hematoxylin. To obtain a negative control, the primary antibodies were routinely omitted.

Statistical analysis.

All data are expressed as the mean ± SEM. Elevations in the PPT are expressed as a percentage of the maximum possible effect, according to the following formula: (PPT postinjection − basal PPT) divided by (250 gm − basal PPT). For the evaluation of the parametric data (PPT and ATP measurements), we performed a one-way analysis of variance. The nonparametric data (PPT alterations [expressed as a percentage of the maximum possible effect]) were analyzed by Kruskal-Wallis test followed by Dunnett's test. Significance was assumed when P was less than 0.05.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Reduction of pain and edema in inflamed peripheral tissue by treatment with oATP.

Intraplantar injection of CFA into the right hind paw of each rat (n = 7) induced a unilateral inflammation, resulting in inflammatory signs such as increased paw volume (edema) and redness that were evident from 3 hours to 48 hours following inoculation. PPT values before inoculation with CFA were similar in the right and left hind paws (mean ± SEM 105 ± 8.2 gm and 101 ± 9.8 gm, respectively). After onset of inflammation, the basal PPT did not significantly change in noninflamed paws (102 ± 9.7 gm) but did decrease in inflamed paws, to 60 ± 4.1 gm.

We previously demonstrated that oATP treatment induced a significant increase in the PPT of inflamed paws, with an elevation in values up to >200 gm (20). In the present study, we measured the elevation in PPT as a percentage of the maximum possible effect at different times following intraplantar administration of the drug (time 0). Temporal evolution of the peripheral antinociception is shown in Figure 1. In inflamed paws, the first significant elevation in PPT upon oATP administration was observed at 0.5 hours. Thereafter, PPT values increased until 4 hours and remained significantly higher in oATP-treated inflamed paws until 48 hours. In noninflamed paws, oATP treatment did not significantly increase PPT values when compared with the values under saline treatment. Similar results were obtained when the inflammatory process was induced for 6 or 12 hours instead of 24 hours (data not shown). Paw volume (edema) significantly increased 24 hours after CFA injection, and was significantly reduced in oATP-treated paws (data not shown).

thumbnail image

Figure 1. Effect of oxidized ATP (oATP) intraplantar injection on peripheral antinociception. Alterations in the paw pressure threshold (PPT) are expressed as a percentage of the maximum possible effect (% MPE) at different times following intraplantar administration (time 0 hours) of oATP (168 μM) or saline in rat hind paws. Inflamed paws were injected with Freund's complete adjuvant 24 hours earlier and treated with oATP (open bars) or saline (solid bars), and values were compared with those of noninflamed paws that received oATP (diagonal hatched bars) or saline (horizontal hatched bars). Bars show the mean ± SEM of 7 rats. ∗ = P < 0.05 versus saline-treated paws and § = P < 0.05 versus noninflamed paws, by the Kruskal-Wallis test, followed by Dunnett's test.

Download figure to PowerPoint

Effect of local ATP treatment.

To evaluate the possible nociceptive effect due to local ATP administration, we studied PPT values following injection of ATP (0.01 mmoles at time 0) into inflamed or noninflamed rat hind paws. Before ATP administration, PPT values were significantly higher in noninflamed paws than in inflamed paws (Figure 2). A significant reduction of PPT values was progressively shown in noninflamed paws from 0.5 hours to 2 hours after ATP treatment, whereas no significant changes in PPT values were observed during the same time in inflamed paws. The results suggest that the levels of extracellular ATP are more elevated in inflamed tissues than in noninflamed tissues, and that further addition of ATP in inflamed paws had a minimal influence. PPT values similar to those measured at time 0 were observed at 4 hours after ATP injection.

thumbnail image

Figure 2. Effect of ATP intraplantar injection on peripheral nociception. Paw pressure threshold (PPT) values were determined at different times following intraplantar ATP (0.01 mmoles) administration (time 0 hours) in noninflamed hind paws (solid bars) or inflamed hind paws of rats that had received intraplantar inoculation of Freund's complete adjuvant 24 hours earlier (open bars). Bars show the mean and SEM of 8 rats. ∗ = P < 0.05 versus the respective inflamed paws and ° = P < 0.05 versus basal values at time 0, by one-way analysis of variance.

Download figure to PowerPoint

Independence of the analgesic effect of oATP from the recruitment of immunocytes in inflamed tissues.

Fucoidin injection did not significantly change the PPT values, and significant reductions in the PPT were observed following CFA injection (Figure 3). In inflamed paws, oATP treatment restored the PPT levels to those measured in untreated paws, and, with time, such thresholds further increased. In noninflamed paws, oATP injection did not significantly change the PPT values. This indicates that nociception due to CFA injection is not necessarily related to leukocyte recruitment; instead, the oATP analgesic effect may be related to the inhibition of P2X7 receptors bound on nociceptive nerve endings or on cells other than recruited immunocytes.

thumbnail image

Figure 3. Effect of fucoidin on peripheral antinociception. Paw pressure threshold (PPT) values in inflamed hind paws (solid circles) and noninflamed hind paws (open circles) before (time 0 hours) and after administration of fucoidin (10 mg/kg by intravenous injection at 0.5 hours), Freund's complete adjuvant (FCA; by intraplantar injection at 1 hour), and oxidized ATP (oATP; 168 μM by intraplantar injection at 4 hours). Bars show the mean ± SEM of 9 rats. ∗ = P < 0.05 versus noninflamed paws, by one-way analysis of variance.

Download figure to PowerPoint

Modification of ATP content in peripheral subcutaneous tissues following oATP treatment.

We measured ATP levels in inflamed paws (induced by 24-hour CFA treatment) and in noninflamed paws that were treated with 168 μM oATP for 6 hours and 12 hours (following oATP administration), and in the contralateral untreated paws. In noninflamed tissues, oATP treatment did not significantly change the ATP levels (Figure 4). In contrast, the levels of ATP in inflamed tissues were significantly higher than in noninflamed tissues, and these levels were significantly reduced by oATP treatment (Figure 4).

thumbnail image

Figure 4. Effect of intraplantar oxidized ATP (oATP) injection on ATP levels in inflamed or noninflamed paws. ATP content was measured at 6 hours and 12 hours following intraplantar oATP (168 μM) injection in A, noninflamed paws (horizontal stippled bars) (versus open bars [untreated]) and B, inflamed paws (24 hours after Freund's complete adjuvant administration) (diagonal stippled bars) (versus solid bars [untreated]). ∗ = P < 0.05 versus inflamed untreated paws, by one-way analysis of variance. Bars show the mean and SEM of 7 experiments.

Download figure to PowerPoint

Histologic expression of P2X7 receptors.

A strong expression of P2X7 receptors was observed in control rat hind paws, either in nerve-ending fibers or in the peripheral nerves. A similar, but less intense, expression of P2X7 receptors was observed in some capillaries and in the middle layer of the arterial wall (media) of small, peripheral arterial vessels (Figure 5). In paw tissues subjected to CFA injection, we observed a moderate P2X7 receptor expression in the nerve endings and peripheral nerves, which was similar in intensity and distribution to that seen in the controls; only a few of the arterial vessels had a weak P2X7 positivity (Figure 5).

thumbnail image

Figure 5. Immunohistologic localization of P2X7 receptors in rat hind paws. Sections were double-labeled with polyclonal antibody recognizing P2X7 and monoclonal antibody recognizing neurofilaments. A and A′, In rat hind paws treated with sterile saline (controls), immunoreactivity was present in nerve fibers (arrow; identified by the presence of neurofilaments) and in some vessels. B and B′, In inflamed hind paws treated with oxidized ATP (oATP), P2X7 expression was reduced in the nerve fibers (arrow). C, In paw tissues treated with Freund's complete adjuvant (CFA), a P2X7 immunoreactivity was present in nerve fibers (arrow) and vessels. D, In inflamed hind paws treated with CFA and oATP, P2X7 expression was reduced in the nerve fibers (arrow) and vessels. Asterisk in A and B denotes the area that is magnified in A′ and B′ (original magnification × 200; inset × 500).

Download figure to PowerPoint

An intense and homogeneous reduction in P2X7 expression was observed in the terminal-ending fibers close to the epidermis and in peripheral nerves both in the samples obtained from hind paws treated with oATP alone and in those treated with CFA and oATP (Figure 5). In the latter sections, vessels and some peripheral nerves more close to the site of inflammation revealed a reduction in P2X7 expression (Figure 5). In all specimens treated with CFA, an irregularly diffuse inflammatory infiltrate was present, with a prevalence of granulocytes and a variable amount of lympho-histiocytic cells. The cells bearing macrophage histologic features showed a weak, but constant, cytoplasmic positivity for P2X7, as previously reported (26).

The samples obtained from hind paws treated with only ATP had a P2X7 expression in both the nerves and the vessels that was similar to that in the controls; a focal and mild granulocyte inflammatory infiltrate was present (data not shown). In all specimens, sweat glands revealed a faint, but fairly diffuse P2X7 positivity (data not shown).

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Although many aspects of central pain modulation are understood, the control of peripheral nociceptive transmission by endogenous and exogenous mechanisms has still to be clarified (6). The control of peripheral inflammatory pain was previously related to the presence of immune cells that have migrated to injured sites (27). These immunocytes secrete β endorphins that activate peripheral opioid receptors, localized on sensory nerve terminals, to inhibit pain (28). Thus, pain is enhanced by measures that limit the migration of opioid-producing cells (10). Inflammatory pain is provoked by the release of many hyperalgesic mediators from migrant immune cells and from resident cells, such as prostaglandins, leukotrienes, and cytokines. In addition, neurotransmitters such as substance P and ATP are released by nerve endings of sensory fibers during “axon-reflex” activity (7), and these can further influence pain signaling.

We hypothesized that one approach to limit inflammatory pain could be to block ATP activity because of its cytolytic (17) and hyperalgesic (14, 15) effects. We used the same experimental model of inflammation in which the efficacy of peripheral opioid antinociception has been demonstrated, i.e., the intraplantar injection of CFA into the hind paw of the rat (10). We studied the antinociceptive function of oATP, a selective inhibitor of the P2X7 ATP receptor, and attempted to distinguish the role of P2X7 ATP receptors on migrating immunocytes compared with those on other sites (i.e., sensory nerve terminals).

We demonstrated, in the present study, that local treatment with oATP significantly increased PPT levels in inflamed paws, and that the antinociceptive effect (maximal at 4 hours following oATP injection) persisted for a long time (until 48 hours). Local treatment of noninflamed paws with oATP did not elicit significant changes in PPT values. P2X7 receptors are known to be present in nociceptive nerve endings (29). The presence of P2Z/P2X7 receptors has been demonstrated in lymphocytes (30) as well as in macrophages (31), microglial cells (18), and mast cells (32). Our histologic evaluations showed that nerve-ending fibers, peripheral nerves, and a few vessels expressed P2X7 receptors in samples obtained from inflamed tissues and from noninflamed controls. In samples obtained from hind paws treated with oATP, P2X7 receptor expression was reduced; it is possible that these receptors are not available to the antibody. However, ATP levels, which were significantly more elevated in inflamed tissues than in noninflamed tissues, were reduced by oATP treatment. This fact could indicate that oATP, by binding to P2X7 receptors, limits ATP production and release by cells bearing these receptors.

In our previous study (20), we have shown that a concentration of oATP higher than 168 μM induced a more elevated and more prolonged analgesic effect in inflamed rat paws. The results possibly depend on the level of saturation of P2X7 receptors; this saturation is progressively reached by increasing the oATP concentration. In addition, by increasing this concentration, the analgesic effect was also evident in noninflamed paws (Dell-Antonio G, et al: personal observations). It is known that inflamed tissues are more prone than noninflamed ones to the action of antinociceptive agents. So, treatment with antinociceptive molecules permits greater elevation of PPT values in inflamed rat paws compared with those observed in control, noninflamed paws.

We also determined whether the analgesic effect exerted by oATP was related to leukocyte migration in inflamed tissues and to their mediator release. We studied the influence exerted on peripheral antinociception by a block of selectin-dependent leukocyte extravasation into inflamed, peripheral subcutaneous tissues. Fucoidin, which binds to adhesion molecules of the selectin family and blocks the “rolling” of leukocytes, reduces leukocyte accumulation at inflammatory sites (10, 33–36). Our results showed that the presence or the absence of immune cells in inflamed tissues did not influence the antinociceptive activity of oATP. These findings are in accordance with the results of a previous study (10) which demonstrated that fucoidin did not affect hyperalgesia in inflamed tissues. Our results indicate that oATP does not act by favoring the release of mediators by recruited immune cells. Such observations suggest that oATP mainly inactivates nociceptive signals through binding of receptors, which are present on sensory nerves, as our histologic findings indicated.

Other data have suggested different ways to explain the analgesic effect due to oATP. Recently, it was shown that mice lacking P2X7 receptors were unable to release interleukin-1 in response to ATP from peritoneal macrophages (37). Impairment of the in vivo cytokine signaling cascade in P2X7-deficient mice suggests that the block of P2X7 receptors by oATP could reduce macrophage activation and the subsequent cytokine production, thereby limiting tissue damage. In addition, our histologic evaluations revealed the presence of P2X7 receptors at the level of capillaries. In endothelial cells, lipopolysaccharide-triggered ATP secretion, via P2X7 receptor activation, causing interleukin-1α release, has been demonstrated (38). We could hypothesize that the block of P2X7 receptors by oATP inhibits interleukin-1 as well as ATP release by endothelial cells. Recently, a role of E-selectin, but not of P-selectin, was demonstrated in the development of adjuvant-induced arthritis in the rat (39). E-selectin may contribute to the migration of antigen-reactive T cells to peripheral tissues.

Our data suggest that binding of oATP with receptors localized on many cells, and also on sensory nerve terminals, could competitively block the binding of extracellular ATP to the same structures, thus limiting ATP-related cytotoxicity and sensory-nerve activation and thereby inducing pain relief. Our results also indicate that oATP treatment of inflamed tissues limits further production of ATP by inflammatory and other cells, possibly through a block of their activation.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

We thank Dr. Mauro Bianchi for his helpful suggestions and B. Johnston for editing the manuscript.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES