SEARCH

SEARCH BY CITATION

Keywords:

  • afferent;
  • cannabinoids;
  • endocannabinoid;
  • lipopolysaccharide;
  • neuron;
  • sensory

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Funding
  9. Disclosures
  10. Author contribution
  11. References

Background  Endocannabinoids influence immune function and nociceptive signaling. This study examines cannabinoid modulation of sensory signaling from the GI tract following an acute inflammatory response triggered by systemic administration of bacterial lipopolysaccharide (LPS).

Methods  A segment of proximal jejunum was intubated, to measure intraluminal pressure, in anesthetized rats. Afferent impulse traffic was recorded from a single isolated paravascular nerve bundle supplying the jejunal loop. Drugs and LPS were administered intravenously and changes in afferent firing were determined.

Key Results  The non-selective cannabinoid agonist, WIN55,212-2 (1 mg kg−1 i.v.) and the anandamide transport inhibitor, VDM11 (1 mg kg−1 i.v.) but not the fatty acid amide hydrolase (FAAH) inhibitor, URB597 (0.3 mg kg−1) caused a significant increase in afferent activity. The WIN55,212-2-induced afferent response was mediated by activation of CB1 receptors whereas the VDM11 response was mediated by both CB1 and CB2 receptor mechanisms. LPS (10 mg kg−1) evoked an increase in afferent activity which was significantly reduced in the presence of WIN55,212-2 and VDM11 but not URB597. The inhibitory effect of WIN55,212-2 was prevented by CB1 but not CB2 receptor antagonism. In contrast, the inhibitory effect of VDM11 remained unaltered after CB1 or CB2 receptor blockade.

Conclusions & Inferences  Endocannabinoids play a role in modulating afferent signaling and may represent a target for the treatment of visceral hypersensitivity. In contrast to the effects of blocking endocannabinoid uptake (VDM11), inhibiting breakdown of endocannabinoids (URB597) had no effect on baseline or LPS induced afferent firing. Therefore, uptake of cannabinoids rather than breakdown via FAAH terminates their action in the GI tract.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Funding
  9. Disclosures
  10. Author contribution
  11. References

The gastrointestinal (GI) tract has an elaborate immune system.1 Macrophages and mast cells express receptors on their cell membrane, such as toll-like receptors (TLRs), which recognize conserved pathogen-associated molecular patterns (PAMPs) including the gram negative cell wall glycoprotein, lipopolysaccharide (LPS).2–5 Ligand binding to the TLR4 complex activates a cascade of events which ultimately leads to the production and release of a variety of proinflammatory cytokines, chemokines, prostanoids, leukotrienes, and nitric oxide.6 These mediators in turn trigger an acute inflammatory response with leukocyte infiltration, extravasation, and neuronal activation leading to sickness behavior including fever, anorexia, and hyperalgesia5,7–9 which may ultimately lead to multiple organ failure, shock, and death.10 These CNS-induced behavioral responses are mediated by direct effects of circulating cytokines on neuronal and glial cell function and indirect effects via activation of peripheral afferent inputs to homeostatic brain circuits. Recent studies have demonstrated a marked activation of GI afferents following systemic LPS leading to mechano- and chemo-hypersensitivity and c-fos expression in the brainstem.11,12

Cannabinoids have been shown to play a protective function during inflammation and nociception via an action mediated by two distinct receptors, CB1 and CB2.13 There is extensive evidence that cannabinoids suppress immune cell functions such as recruitment, adhesion, migration, and proliferation.14,15 Cannabinoids have also been shown to suppress production and release of cytokines and chemokines.16–18 Cannabinoid (CB1) receptors are expressed on GI sensory neurons 19 expression of which has been shown to increase during GI inflammatory states.20,21 Moreover, the levels of endogenous cannabinoids are also increased during inflammation 22 suggesting a functional adaptation of the endocannabinoid system to restore homeostasis. In this respect experimental colitis is exacerbated in the CB1 knockout mouse while in wild-type animals colitis is similarly enhanced by treatment with CB1 receptor blockade.21 In addition, treatment with CB2 receptor agonists, JWH133 or AM1241 significantly reduced experimental colitis via a CB2 receptor mediated mechanism while the CB2 receptor antagonist, AM630, exacerbated colitis.23 The action of endogenous cannabinoids is terminated by an intracellular enzyme fatty acid amide hydrolase (FAAH).24 FAAH knockout mice have elevated anandamide (AEA) levels and reduced levels of histological and myeloperoxidase (MPO) markers of inflammation.21 The non-selective cannabinoid agonist, WIN55,212-2 has been shown to reduce LPS induced sickness behavior via a CB1 receptor mechanism an effect which coincided with a WIN55,212-2-induced reduction in the pyogenic factor IL-6.25 LPS-induced inflammation has been shown to cause an increase in GI transit which is dose dependently inhibited by the CB2 receptor agonist JWH133.26

Understanding the mechanisms involved during an inflammatory event and the hypersensitivity and therefore pain this causes could potentially help in the discovery of new therapies for GI disorders such as irritable bowel syndrome (IBS) and inflammatory bowel disease (IBD). Our aim therefore was to examine the potential for cannabinoids to modulate sensory signaling from the GI tract under both basal conditions and following an acute inflammatory response triggered by the systemic administration of bacterial LPS.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Funding
  9. Disclosures
  10. Author contribution
  11. References

Anesthesia

Experiments were performed using male Wistar rats (250–500 g) purchased from Charles Rivers. Rats were housed for at least 1 week, with free access to food, before terminal experiments were performed at The University of Sheffield in accordance with the UK animals (scientific procedures) Act 1986. Rats were anesthetized with an intraperitoneal (i.p.) injection of pentobarbitone sodium (60 mg kg−1) and anesthesia was maintained throughout the experimental procedure by infusion of pentobarbitone as required (0.5–1 mg kg−1 min−1 i.v.). Depth of anesthesia was assessed using hindlimb withdrawal and corneal blink reflexes. Experiments were terminated with an i.v. overdose of anesthetic.

Surgery

Once anesthetized, rats were placed on an electric blanket to maintain body temperature at 37 °C, controlled by an anal probe (Harvard Apparatus, Homeothermic Blanket Control Unit, Kent, UK). The jugular vein and carotid artery were canulated for maintenance of anesthetic/drug administration and monitoring arterial blood pressure, respectively. The trachea was canulated to enable spontaneous respiration. A midline laparotomy was performed and the cecum excised to provide more space within the abdominal cavity. A segment of jejunum extending 2–3 mesenteric arcades was isolated, cut on the anterior mesenteric border and flushed with isotonic saline. The isolated jejunum was then canulated at both ends, the oral catheter was attached to an infusion pump to deliver saline at the rate of absorption (∼0.8 mL h−1) and the aboral catheter contained a smaller polyethylene tube which was attached to a pressure transducer (DTXTM plus DT-XX, Becton Dickson, Singapore) to monitor intraluminal pressure (maintained at ∼3.5 mmHg). The jejunal catheters were securely sutured in the abdominal wall using a purse stitch and the original abdominal incision (muscle and skin) was sutured to a steel ring producing a well which was filled with prewarmed liquid paraffin oil.

Nerve preparation and recordings

A black Perspex platform was placed under one of the mesenteric bundles. A mesenteric nerve was teased out of the surrounding fat and connective tissue. Once a nerve had been isolated it was cut distal to the jejunum to eliminate efferent nerve traffic and the peripheral end attached to one arm of a bipolar wire recording electrode for afferent recordings. A piece of connective tissue was attached to the other arm of the electrode to act as a differential. The electrodes were connected to a headstage (NL100; Digitimer Ltd, Welwyn Garden City, Herefordshire, UK). Signals then passed through a differential amplifier (NL104, Neurolog system, Digitimer Ltd, Welwyn Garden City, Herefordshire, UK) and filters (NL125, Neurolog system and Hum Bug, Quest Scientific, Digitimer Ltd, Welwyn Garden City, Herefordshire, UK). The nerve, blood pressure, and jejunal pressure signals were then captured on a computer via a 1401 analogue to digital converter and Spike 2 software [Version 5.16; Cambridge Electronic Design (CED), Cambridge, UK]. Raw multiunit nerve activity was quantified using a spike processor which counted the number of action potentials crossing a preset threshold. A threshold level for spike counting was set at the peak of the smallest identifiable spike (at roughly twice the baseline noise level). Recordings were closely monitored for changes in noise levels or electrode drift throughout the experiments.

Protocols

Once the blood pressure, luminal pressure and afferent firing had stabilized the following protocols were applied:

Protocol 1  A bolus dose of the non-selective cannabinoid agonist, WIN55,212-2 (0.01, 0.03, 0.1, 0.3, and 1 mg kg−1), the fatty acid amide hydrolase inhibitor, URB597 (0.3 mg kg−1) or the AEA transport inhibitor, VDM11 (1 mg kg−1) was administered intravenously. The receptor mechanisms were investigated by applying either the CB1 selective antagonist (SR141716A, 1 mg kg−1 i.v.) or the CB2 selective antagonist (SR144528, 1 mg kg−1 i.v.) 30 min prior to drug treatments. These doses were chosen on the basis of preliminary findings and previous literature.25,27–29

Protocol 2  Lipopolysaccharide (10 mg kg−1) was infused intravenously with the aid of a perfusion pump to deliver the dose over a 3-min period to avoid profound cardiovascular shock ultimately leading to death.11 The effects were then observed for 2 hours. To investigate the effect of cannabinoids on this LPS-induced response either WIN55,212-2 (1 mg kg−1 i.v.), URB597 (0.3 mg kg−1 i.v.), or VDM11 (1 mg kg−1 i.v.) was applied 30 min before LPS administration. The receptor mechanisms were investigated by applying either the CB1 selective antagonist (SR141716A, 1 mg kg−1 i.v.) or the CB2 selective antagonist (SR144528, 1 mg kg−1 i.v.) 30 min prior to drug treatments followed 30 min later by LPS.

Analysis of data

The afferent nerve responses were measured as a mean change from baseline measured 10 min before and after drug administration. The change from baseline over time was measured where the mean afferent firing was taken over consecutive 1-min periods for 10 min. The LPS experiments were analyzed for 2 h after LPS administration, in these experiments the mean change or mean percentage change in afferent firing was taken every 10 min and represented as either a change in afferent firing over time or as an area under the curve histogram. Analysis included one-way anova with Tukey’s post-test and two-way anova with Bonferroni post-test. Data are presented as Mean ± SEM with n = 6.

Drugs

WIN55,212-2 and VDM11 were purchased from Tocris Bioscience (Bristol, UK). SR141716A and SR144528 were obtained from GSK, Harlow. URB597 and LPS (Escherichia coli 026:B6) were purchased from Sigma (Poole, UK). WIN55,212-2, SR141716A, and SR144528 were dissolved as follows 1:1:1:7 DMSO:Ethanol:Tween80:Saline. URB597 was dissolved as follows 1:1:8 Tween80:1methyl-2-pyrrolidinone:Saline. VDM11 was dissolved in 20% ethanol and LPS was dissolved in saline.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Funding
  9. Disclosures
  10. Author contribution
  11. References

The synthetic cannabinoid receptor agonist WIN55,212-2 caused a dose dependant increase in jejunal afferent nerve activity (Fig. 1B). 1 mg kg−1 WIN55,212-2 evoked a marked and prolonged increase in jejunal afferent firing (30.5 ± 5.2Spikes s−1) which returned to baseline after 22.5 ± 3.2 min (Fig. 1A,B). The action of endogenous cannabinoids is terminated by two classes of compounds that interfere either with their uptake or intracellular breakdown. Drugs targeting these processes provide a mechanism of raising endogenous levels to indirectly activate cannabinoid receptors. Studies have shown that using inhibitors of endocannabinoid uptake and degradation such as VDM11 and URB597, respectively, cause an increase in endocannabinoid levels in the GI tract.30,31 These drugs have recently been shown to reduce MPO levels and macroscopic damage scores in TNBS induced colitis.32 These data suggest that inhibiting endocannabinoid uptake and degradation may provide a possible therapeutic target for various inflammatory conditions such as IBD. The aim of this study therefore was to investigate the effect of URB597 and VDM11 on the hypersensitivity driven by an acute inflammatory LPS insult. In this respect, the AEA transport inhibitor VDM11 (1 mg kg−1) caused a significant increase in afferent nerve activity (Fig. 1C) of a magnitude (17.5 ± 4.4 Spikes s−1) similar to that of WIN55,212-2 (P = 0.09) which returned to baseline after 15.1 ± 0.7 s. In contrast, the fatty acid amide hydrolase inhibitor URB597 (0.3 mg kg−1) had no significant effect on afferent nerve activity (Fig. 1D).

image

Figure 1.  The effect of various cannabinoid drugs on jejunal afferent nerve activity. (A) Representative trace to illustrate the response to 1 mg kg−1 WIN55,212-2. Raw multiunit afferent nerve activity and rate histogram of number of spikes per second which pass a predetermined threshold showing baseline activity and the response to 1 mg kg−1 WIN55,212-2 above which are examples of the actual nerve recordings before (left) and after (right) application of WIN55,212-2. On application of WIN55,212-2 there is a significant increase in afferent firing. Mean change in nerve activity in response to (B) 0.01, 0.03, 0.1, 0.3, and 1 mg kg−1 WIN55,212-2, (C) 1 mg kg−1 VDM11 and (D) 0.3 mg kg−1 URB597.

Download figure to PowerPoint

The WIN55,212-2-induced increase in afferent nerve activity was significantly attenuated following pretreatment with the CB1 receptor antagonist, SR141716A (Fig. 2A), but not with the CB2 receptor antagonist, SR144528 (Fig. 2B). In contrast, the VDM11 induced increase in afferent nerve activity was significantly reduced by both, SR141716A (Fig. 2C) and SR144528, (Fig. 2D) receptor antagonists. Neither antagonist alone had any significant effect on baseline afferent nerve activity (data not shown).

image

Figure 2.  Receptor mechanisms involved in the afferent response to WIN55,212-2 and VDM11. The afferent response to WIN55,212-2 was significantly reduced in the presence of the CB1 receptor antagonist, SR141716A (A), but not the CB2 receptor antagonist, SR144528 (B). The afferent response to VDM11 was significantly reduced in the presence of the CB1 receptor antagonist, SR141716A (C), and the CB2 receptor antagonist, SR144528 (D). Neither antagonist alone had any significant effect on afferent nerve activity.

Download figure to PowerPoint

Lipopolysaccharide (10 mg kg−1 i.v.) evoked a profound increase in afferent activity which reached a maximum at 60 min after LPS administration (Fig. 3).

image

Figure 3.  The effect of LPS on jejunal afferent nerve activity. (A) Representative trace to illustrate the response to 10 mg kg−1 LPS. Raw nerve activity above which are traces of the actual nerve recording at the times indicated (A–C) and a histogram of firing frequency, showing the number of action potentials that pass a predetermined threshold level. LPS (10 mg kg−1 i.v.) caused a biphasic increase in mesenteric afferent firing, a small transient increase in afferent firing followed by a prolonged increase in mesenteric afferent nerve firing.

Download figure to PowerPoint

Lipopolysaccharide administrated 30 min after WIN55,212-2, at a time when afferent discharge had returned to baseline, evoked an afferent response that was significantly reduced compared with control (Fig. 4A). This effect of WIN55,212-2 was reversed by pretreatment with SR141716A but not SR144528 (Fig. 4B). Neither antagonist alone had any significant effect on the afferent response to LPS (data not shown). The inhibitory effect of WIN55,212-2 on LPS-evoked afferent firing was mimicked by VDM11 although this effect was not reversed by either the CB1 or CB2 receptor antagonist (Fig. 5). In contrast URB597 had no significant effect on the afferent response to LPS.

image

Figure 4.  Cannabinoids and their effect on the jejunal afferent response to LPS. Afferent nerve responses to LPS (10 mg kg−1 i.v.) represented as a change from baseline over time (2 h). (A) Pretreatment with WIN55,212-2 (1 mg kg−1) significantly reduced the afferent response to LPS. (B) The WIN55,212-2 induced attenuation of the afferent response to LPS was reversed in the presence of the selective CB1 receptor antagonist, SR141716A (1 mg kg−1), but not the selective CB2 receptor antagonist, SR144528 (1 mg kg−1).

Download figure to PowerPoint

image

Figure 5.  Endocannabinoids and their effect on jejunal afferent responses to LPS. (A) Histogram showing area under the curve (AUC) for the LPS afferent response in control animals and following treatment with the FAAH inhibitor, URB597 (0.3 mg kg−1 i.v.). The afferent response to LPS was not significantly altered by URB597. (B) Histogram showing the effect of the anandamide transport inhibitor, VDM11 (1 mg kg−1), on the afferent nerve response to LPS. VDM11 causes a significant reduction in the afferent response to LPS, a response that was independent of both CB1 and CB2 receptor mechanisms.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Funding
  9. Disclosures
  10. Author contribution
  11. References

These data demonstrate a profound effect of CB1 receptor activation on jejunal afferent firing under baseline conditions and following an acute inflammatory insult with LPS. Surprisingly these effects are different with baseline excitation vs inhibition of the LPS mediated afferent activation. Another striking observation is that the effect of the synthetic agonist WIN55,212-2 is mimicked by the AEA transport inhibitor, VDM11, while blocking the enzyme responsible for intracellular breakdown of endocannabinoids, using the FAAH inhibitor, URB597, was without effect. These data suggest that endocannabinoids play a role in modulating afferent signaling and may therefore represent a target for the treatment of visceral hypersensitivity such as occurs in IBS.

WIN55,212-2 (1 mg kg−1) caused a significant and prolonged increase in mesenteric afferent nerve activity which was attenuated in the presence of the CB1, but not the CB2, receptor antagonist. This is consistent with previous studies demonstrating CB1 receptor selectivity of WIN55,212-2 on cardiovascular and central neuronal function.33–40 However, in all of these previous studies CB1 receptor activation caused inhibition rather than excitation, consistent with the link between CB1 and Gi/o.41 Neuronal excitation was also observed using the selective CB2 receptor agonist, AM1241, when recording from mouse mesenteric afferents in vivo.42 However, since the response persisted in the CB2 receptor knockout it is likely that it was mediated by a CB1 receptor mechanism similar to that described here. Alternative coupling has been described including G-protein-coupled receptors coupled specifically to the Gs subtype.43–45 A recent study by Yüce et al.46 failed to find any significant effect of WIN55,212-2 on mesenteric afferent nerve activity recorded in vitro. It is therefore possible that the excitation observed in the current in vivo study arose indirectly triggered by a rise in systemic mediators delivered to the gut via the vascular system. However, this would seem unlikely since we have been able to reproduce our in vivo findings using an in vitro mouse model in which both WIN55,212-2 and LPS induce an increase in afferent nerve activity suggesting the afferent response may be independent of blood supply.47,48

Cannabinoids have been shown to have anti-inflammatory properties.20–22 We therefore sought to determine the effects of cannabinoids on the afferent response to LPS, a component of gram negative bacteria, that has been used extensively in studies of acute inflammation.11,12,25,26,49,50 The response to LPS was attenuated by WIN55,212-2 an effect reversed by the selective CB1 receptor antagonist but not the selective CB2 receptor antagonist. A contribution from CB2 might have been anticipated given their expression on various immune cells51 at levels 10–100 fold greater than the CB1 receptor.52 Although many studies have shown that CB2 receptor expression is increased during inflammation it has been shown that in response to LPS CB2 mRNA levels are downregulated.53–55 Recently, it has been shown that CB2 receptors are also localized on neurons of the ENS and that the receptor expression levels are unaltered in the presence of LPS.56

Drugs that prevent the breakdown of endocannabinoids have been the focus of much interest over the last few years and offer several advantages over treatment with receptor agonists. Firstly, they modulate existing cannabinoid signaling at specific sites of action and secondly, they might avoid the characteristic psychotropic effects seen with cannabis use and therapeutic application of cannabinoid agonists. The FAAH inhibitor, URB597, prevents the intracellular enzymatic breakdown of endocannabinoids. Although the mechanism of uptake is still controversial, the most widely accepted mechanism is via a carrier mediated transporter, however, this transporter is yet to be cloned. Nevertheless, a number of transport inhibitors have been developed, including VDM11, which by blocking the re-uptake of endocannabinoids prolongs their action in the synapse.57 Studies have shown that inhibitors of endocannabinoid uptake and degradation cause an increase in endocannabinoid levels in the GI tract30,31 and reduce markers of inflammation and hypersensitivity.32,58 These data suggest that inhibiting endocannabinoid uptake and degradation may provide a possible therapeutic target for various inflammatory conditions such as IBD.

In the current study URB597 (0.3 mg kg−1) had no significant effect on either baseline afferent nerve activity or the afferent response to LPS. Some studies have suggested that FAAH is the driving force needed for endocannabinoid uptake.59 However, cells can still accumulate AEA even in the absence of FAAH 60,61 with accumulation being aided by intracellular sequestration in lipid compartments or by binding to proteins.62 Moreover, AEA levels in the duodenum of FAAH knockout are comparable to levels in wild-type animals.27 Other catalytic enzymes have been identified63 which might compensate for the loss of FAAH. N-acylethanolamine-hydrolyzing acid amidase (NAAA) is sparse in the brain but present at high levels in the intestine suggesting that it may have an important role in cannabinoid metabolism in the gut.63,64 In the GI tract the levels of 2-AG are substantially higher than those of AEA.65 The 2-AG is preferentially broken down by the enzyme monoacylglycerol lipase (MAGL) which has been shown to be present throughout the small intestine.66 A recent article has shown that URB602 significantly inhibits whole gut transit via a CB1 receptor mediated mechanism.66 Future studies need to focus on the role of these other enzymes on modulating sensory function.

VDM11 (1 mg kg−1) caused a significant increase in mesenteric afferent nerve activity which was reversed by both CB1 and CB2 receptor antagonists. This suggests that blocking re-uptake leads to elevated levels of endocannabinoids active at both CB1 and CB2. Storr et al. 2008 found that the VDM11-induced decrease in MPO activity and macroscopic damage score in response to DNBS colitis was abolished in CB1 and CB2 receptor knockout mice suggesting that both receptors are required to confer the protective effects.32 Although CB2 receptors have not been found on GI extrinsic afferents the once widely accepted hypothesis that CB2 receptors were only present on immune cells has now been disputed with recent studies showing that they are present in the brain 29 and on nerves of the ENS.66 VDM11 also caused a significant reduction in the afferent response to LPS. However, this response could not be attenuated using the CB1 or CB2 receptor antagonists that reversed the effect on baseline firing. Thus, while CB1 and CB2 receptors contribute to the activation of baseline firing neither receptor is involved in the inflammation mediated increase in afferent firing.

In conclusion, these data provide evidence that the cannabinoid system is involved in afferent signaling of the GI tract. Cannabinoids can cause extrinsic afferent nerve excitation which was differentially mediated by both CB1 and CB2 receptor mechanisms depending on whether the response was mediated by exogenous ligands or elevated levels of endocannabinoids following blockade of re-uptake using VDM11. These data also suggest that cannabinoids may have a role in modulating the sensitivity of intestinal afferents during inflammation triggered by LPS. WIN55,212-2 caused a CB1 receptor-dependent reduction in the afferent response to LPS whereas the VDM11-induced effect was not mediated by CB1 or CB2 receptor mechanisms indicating the involvement of other receptors. The mechanism involved in endocannabinoid re-uptake is controversial, however, our data show that transport of endocannabinoids into the cell, rather than breakdown by the FAAH inhibitor, is important in rendering endocannabinoids inactive at the cannabinoid receptors.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Funding
  9. Disclosures
  10. Author contribution
  11. References

This work was supported by a special BBSRC Capacity Building Award in integrative mammalian biology.

Author contribution

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Funding
  9. Disclosures
  10. Author contribution
  11. References

JD: designed and performed the research, analyzed the data, and wrote the article; DG: designed the research study and wrote the article.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Funding
  9. Disclosures
  10. Author contribution
  11. References