SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Objective

Obesity commonly causes hepatic lipid accumulation that may favor oxidative stress and inflammation with negative clinical impact. Acylated ghrelin (A-Ghr) modulates body lipid distribution and metabolism, and it may exert antioxidant effects in vitro as well as systemic anti-inflammatory effects in vivo. The impact of A-Ghr on liver triglyceride content, redox state and inflammation markers in diet-induced obesity was investigated.

Design and Methods

A-Ghr (200-μg/injection: HFG) or saline (HF) were administered subcutaneously twice-daily for 4 days to 12-week-old male rats fed a high-fat diet for 1 month (n = 8–10/group).

Results

Compared to lean animals, liver triglyceride accumulation occurred in HF despite enhanced phosphorylation of the lipid oxidation regulator AMPK and preserved mitochondrial enzyme activities. High triglycerides were accompanied by pro-oxidant changes in redox state and proinflammatory changes in NF-kB and TNF-alpha. A-Ghr limited liver triglyceride excess (P < 0.05 HF > HFG > Control) with concomitant activation of glutathione peroxidase and normalized redox state and cytokines. A-Ghr-induced liver changes were associated with higher plasma adiponectin and lower circulating fatty acids (P < 0.05 HFG vs. HF).

Conclusions

A-Ghr limits liver triglyceride accumulation and normalizes tissue redox state and inflammation markers in diet-induced obese rats. These results suggest a favorable impact of A-Ghr on hepatic complications of diet-induced obesity.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Obesity is often characterized by hepatic triglyceride accumulation that may progress to nonalcoholic steato-hepatitis and liver cirrhosis thereby resulting in enhanced metabolic and cardiovascular risk [1-4]. Mechanisms underlying these alterations remain only partly understood, and effective therapeutic strategies are limited. In human and experimental models, high fatty acid availability and impaired liver lipid oxidation might directly contribute to liver triglyceride deposition. Altered liver oxidative capacity has however not been confirmed in all available studies, and preserved or increased markers of hepatic oxidative metabolism have been reported in obesity [5, 6]. Liver oxidative stress and inflammation are also associated with obesity, could enhance tissue lipid accumulation and have been proposed as important contributors to the progression of obesity-induced hepatic metabolic alterations in nonalcoholic steato-hepatitis (NASH) [2, 3].

Acylated ghrelin (A-Ghr) is a gastric hormone with orexigenic and adipogenic effects [7-11] that has been also reported to modulate in vivo tissue lipid distribution. We and others demonstrated that A-Ghr administration leads to liver fat accumulation in lean rodents [11-13], and these effects were associated with impaired activation of the master regulator of lipid oxidation and mitochondrial function AMP-activated protein kinase (AMPK) [11, 12]. Antioxidant and anti-inflammatory effects of A-Ghr have however been also described [14-17], and A-Ghr was accordingly demonstrated to reduce liver pro-oxidant and proinflammatory changes in experimental models of oxidative damage, inflammation and fibrosis [14, 18]. Also importantly, chronic activation of liver insulin signaling at AKT level has been reported to directly contribute to liver fat accumulation, oxidative damage, and inflammation in rodent high-fat feeding [19], and A-Ghr was shown to lower hepatic insulin signaling under different experimental conditions [20, 21]. Total circulating ghrelin is reported to be reduced in patients with nonalcoholic fatty liver disease [22]. The impact of A-Ghr on liver lipid content, redox state, and inflammation in obesity remains however unknown.

In the current study we therefore investigated the potential interactions of high-fat feeding and A-Ghr in the regulation of liver triglyceride content and their impact on redox state and inflammation markers in vivo. We hypothesized that A-Ghr administration to diet-induced obese rats does not enhance liver triglyceride content and may improve tissue redox state and inflammation. Changes in AKT-mediated insulin signaling were also determined, and AMPK phosphorylation and marker mitochondrial enzyme activities were measured to determine their potential involvement in the observed effects. Circulating leptin and adiponectin concentrations were finally measured since high leptin-to-adiponectin ratio is associated with both liver fat accumulation and steato-hepatitis [23, 24]; circulating ghrelin is in turn associated with metabolically favorable adipokine profiles in obesity, but potential direct modulation of adipose tissue hormone secretion by A-Ghr remains largely undetermined [25, 26].

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Animals and experimental protocol

The experimental protocol was approved by the Committee for Animal Studies at Trieste University. Thirty 12-week-old male Wistar rats were purchased from Harlan-Italy (San Pietro al Natisone, Udine, Italy) and kept for 2 weeks in the Animal Facility of the University of Trieste in individual cages on a 12-h light/12-h dark cycle (0600 h/1800 h). Animals were then randomly assigned to undergo a high-fat feeding program with a diet containing 60% energy from fat (HF, n = 20) or a control diet containing 10% calories from fat (C, n = 10; both diets from Mucedola, Settimo Milanese, MI, Italy) for 1 month. All animals were weighed and food intake was monitored two times per week. After 30 days, animals in the HF group were randomized to receive eight subcutaneous injections of either A-Ghr (AnaSpec, San Jose, CA) (HFG, 200 μg ghrelin/injection) or saline, that was also administered to control rats. The ghrelin dose was chosen to be identical to the nonorexigenic dose that we already used in previous studies [11], thereby allowing us to compare the resulting effects in different experimental settings. The dose is also in the same order of magnitude of those used in longer infusion studies in chronic disease models [17], aimed at assessing A-Ghrelin's therapeutic potential in human disease. Injections were administered two times per day at 8 pm and 8 am, beginning in the evening of day 1 with the last injection on the morning of day 4. After the last injection, food was withdrawn for 3 h before the intraperitoneal injection of an overdose of sodium pentobarbital (80 mg kg−1). After achievement of adequate anesthesia, liver was dissected and quickly frozen in liquid nitrogen to be stored at −80°C before analyses. The epidydimal and retroperitoneal fat pads were then dissected and weighed, and blood was finally collected through cardiac puncture for plasma separation and storage at −80°C. The current study is part of a protocol aimed at investigating A-Ghr effects on tissue lipid metabolism in diet-induced obesity, and ghrelin-induced changes in skeletal muscle parameters has been reported elsewhere [27].

To directly assess the potential differential effects of A-Ghr on liver triglyceride storage and their impact on redox state and inflammation in nonobese animals, an additional group of lean 12-week-old male Wistar rats was treated with identical A-Ghrelin administration and compared to a saline-treated control group (n = 8 each).

Liver total and oxidized glutathione, glutathione peroxidase activity

Liver total and oxidized glutathione were measured using the HT Glutathione Assay Kit (Trevigen, Gaithersburg, MD) according to manufacturer's instructions (27; Figure 4). Briefly, ∼50 mg of liver were homogenized in ice cold 5% (w/v) metaphosphoric acid (20 ml g−1 tissue). Homogenates were then centrifuged (12,000g for 15′) and the supernatant was used for total glutathione measurement after appropriate dilution. For oxidized glutathione, samples were pretreated with 2M 4-vinylpyridine. Reduced glutathione was calculated by subtracting the oxidized fraction from the total. To further determine potential diet- and ghrelin-induced liver changes in enzyme activity of the antioxidant scavenging enzyme glutathione peroxidase, the latter was also determined by spectrophotometry and a commercially available kit following the manufacturer's instructions (HT Glutathione Peroxidase Assay Kit, Trevigen, Gaithersburg, MD).

Protein analysis

Total tissue protein was extracted from homogenized tissue as described [20, 27]. Tissue phosphorylated and total levels of AKT and AMPK were determined as referenced [27, 28]. Nuclear p65 NFkB subunit (TransAM, Active Motif North America, Carlsbad, CA) and tissue TNF-alpha (Pierce Biotechnology, Rockford, IL) protein levels were measured using commercially available kits [27]. Briefly, nuclear protein extracts were obtained as follows: 100 mg tissue were homogenized in lysis buffer (10 mM HEPES pH 7.9, 10 mM KCl, 1.5 mM MgCl2, 1 mM DTT, 0.5 mM PMSF) containing protease inhibitors, followed by addition of 1% Nonidet P-40. The homogenate was incubated on ice for 20 min and then centrifuged at 12,500 rpm 4°C for 30 s. Pellets were resuspended in 100-μl extraction buffer (20 mM HEPES pH 7.9, 420 mM NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 5% glycerol, 1 mM DTT, 0.5 mM PMSF) added with protease inhibitors and the tube was gently rocked on ice for 30 min. The mixture was then spun at 12,500 rpm at 4°C for 15 min and the resulting supernatant containing nuclear proteins was stored at −80°C until analyses. Protein concentration in all samples was measured by spectrophotometer (BCA Protein Assay Reagent, Pierce, Rockford, IL).

Cytochrome c oxidase (COX) and citrate synthase (CS) activity

COX and CS enzyme activities were measured spectrophotometrically from tissue homogenates as referenced [11].

Liver triglycerides and plasma metabolic profile

Liver triglycerides were measured as previously referenced [11]. Plasma insulin concentration was measured by ELISA using a commercially available kit (Insulin (Rat), DRG Instruments GmbH Frauenbergstr) [27]. Plasma leptin and adiponectin concentrations were also measured using ELISA kits (rat Leptin, Assay Designs; Mouse/Rat Adiponectin, B-Bridge International). Plasma FFA was determined spectrophotometrically using an acyl-CoA oxidase-based colorimetric kit (NEFA-C, WAKO Pure Chemical Industries).

Statistical analysis

Results in the three groups were compared using One-Way ANOVA. Student's t test for unpaired data was then used to compare results between two groups. P values <0.05 were considered statistically significant.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Body weight, plasma metabolic profile

Initial body weight was comparable in the three experimental groups (not shown), while final body weight was higher in the two high-fat fed groups, resulting from higher weight gain during the 1-month dietary treatment (final BW control: 378 ± 7 g, HF: 412 ± 12, HFG: 408 ± 8; P < 0.05 HF-HFG vs. control: +9% and +8%, respectively). The weight of epidydimal and retroperitoneal fat were also similar in HF and HFG and higher than in control animals (retroperitoneal: control 5.7 ± 0.4 g, HF 9.4 ± 0.7, HFG 9.4 ± 0.5; epidydimal: control 5.3 ± 0.4 g, HF 9.2 ± 0.5, HFG 8.4 ± 0.5, both P < 0.05 control vs. HF-HFG: +65–74 and +65%, respectively). As expected, caloric intake during the last 4 days of the study was elevated in both high fat-fed groups (control: 505 ± 13 calories, HF: 631 ± 11, HFG: 639 ± 12; P < 0.05, control vs. HF-HFG: +25% and +27%, respectively). Blood glucose was moderately higher with similar plasma insulin in HF compared to control group. In HFG, both blood glucose and plasma insulin were similar to control group and not different from HF (blood glucose: control 107 ± 3 mg dl−1, HF 118 ± 4, HFG 114 ± 3; P < 0.05 HF vs. control and +10%; insulin: control 6.8 ± 1.4 ng ml−1, HF 5.9 ± 0.8, HFG 7.2 ± 1, P = NS among groups). Plasma free fatty acids were markedly lower in HFG than in the HF group although they remained moderately higher than in control animals (control 0.21 ± 0.026 mmol l−1, HF 0.59 ± 0.062, HFG 0.39 ± 0.047, P < 0.05 HF > HFG > control, HFG = −34% HFG). Plasma growth hormone (GH) was not statistically different among groups (control 12.6 ± 2.1 ng ml−1, HF 11.1 ± 2.8, HFG 12.8 ± 3.4, P = NS among groups).

Liver triglyceride content, redox state, and inflammation markers

HF expectedly lead to higher liver triglyceride content compared to control group (+135%, Figure 1). This alteration was associated with higher liver GSSG/GSH ratio (+60%), resulting mainly from low reduced glutathione levels, while there were no statistically significant changes in antioxidant activity of glutathione peroxidase (Figure 2). Moderate but statistically significant elevations were also observed in HF for both nuclear p65 NFkB subunit and tissue TNF-alpha levels compared to Control animals (+18–20%, Figure 3).

image

Figure 1. Effects of 1-month high-fat feeding alone (HF) or in combination with 4 days of ghrelin treatment (HFG) on liver triglycerides. Different letters denote statistically significant differences (P < 0.05 ANOVA and post-hoc tests).

Download figure to PowerPoint

image

Figure 2. Effects of 1-month high-fat feeding alone (HF) or in combination with 4 days of ghrelin treatment (HFG) on liver reduced glutathione (A), oxidized glutathione (B) and their ratio (C) and on glutathione peroxidase activity (d). Different letters denote statistically significant differences (P < 0.05 by ANOVA and post-hoc tests).

Download figure to PowerPoint

image

Figure 3. Effects of 1-month high-fat feeding alone (HF) or in combination with 4 days of ghrelin treatment (HFG) on liver nuclear p65 NFkB subunit (A) and TNF-alpha protein (B). Different letters denote statistically significant differences (P < 0.05 ANOVA and post-hoc tests).

Download figure to PowerPoint

In contrast to saline-treated HF, A-Ghr administration in HFG limited diet-induced liver triglyceride accumulation (−34% vs. HF, Figure 1), and resulted in normalization of tissue redox state and inflammation markers (Figures 2 and 3) with a statistically significant increase in glutathione peroxidase activity (+25% vs. HF, Figure 2).

Liver AKT phosphorylation and signaling

AKT phosphorylation and signalling were measured since AKT activation under basal, noninsulin-stimulated conditions has been reported to cause liver lipid accumulation and may contribute to oxidative stress and inflammation during high-fat feeding (19; Figure 4). A-Ghr may in turn reduce AKT activation in vitro and in nonobese rodents [20, 21]. HF and Control animals had comparable levels of AKT and GSK phosphorylation. HFG was however associated with a decline in P/T AKT ratio as well as downstream P/T GSK ratio compared to both HF (−24–50%) and Control groups (−40–41%).

image

Figure 4. Effects of 1-month high-fat feeding alone (HF) or in combination with 4 days of ghrelin treatment (HFG) on liver AKT and GSK phosphorylation. Different letters denote statistically significant differences (P < 0.05 ANOVA and post-hoc tests). For each protein, representative bands for phosphorylated (top) and total (bottom) forms in the three different experimental groups are presented (control: left; HF: middle; HFG: right band).

Download figure to PowerPoint

Liver AMPK phosphorylation, transcript levels of modulators of fatty acid oxidation and synthesis and mitochondrial enzyme activities

HF was not associated with altered activities of rate-limiting enzymes of tricarboxylic acid cycle (citrate synthase) and respiratory chain (cytochrome c oxidase) (Table 1). Preserved mitochondrial enzyme activities were associated with enhanced activating phosphorylation of the master lipid oxidation regulator AMP-activated protein kinase (AMPK) (+340%), high transcriptional expression of fatty acid oxidation rate-limiting enzyme carnitine-palmitoyl transferase-I (+325%) and no change in transcript levels of lipogenic enzymes acetyl-CoA carboxylase and fatty acid synthase (Figure 5). Under the current experimental conditions, HFG did not change HF-associated alterations.

image

Figure 5. Effects of 1-month high-fat feeding alone (HF) or in combination with 4 days of ghrelin treatment (HFG) on liver AMPK phosphorylation (A), carnitine palmitoyl transferase-I (CPT-I) (B), acetyl-CoA carboxylase (ACC) (C), and fatty acid synthase (FAS) (D) mRNA levels. Different letters denote statistically significant differences (P < 0.05 ANOVA and post-hoc tests). For AMPK, representative bands for phosphorylated (top) and total (bottom) forms in the three different experimental groups are presented (Control: left; HF: middle; HFG: right band).

Download figure to PowerPoint

Table 1. Liver enzymes activities of mitochondrial cytochrome c oxidase (COX) and citrate synthase (CS)
 ControlHFHFG
  1. Values are expressed in μmol/min mg protein in the three experimental groups.

  2. Data are mean ± SE. No statistically significant differences were observed by ANOVA and post hoc tests.

COX280±22315±19296±29
CS190±11177±13224±36

Plasma adipokines

HF expectedly leads to an increase in plasma leptin concentration, with no modifications in plasma adiponectin (Figure 6). HFG did not modify leptin but it lead to higher plasma adiponectin (+26%) and lower leptin/adiponectin ratio (−25%) compared to HF.

image

Figure 6. Effects of 1-month high-fat feeding alone (HF) or in combination with 4 days of ghrelin treatment (HFG) on plasma leptin (A), adiponectin (B) and leptin/adiponectin ratio (C). Different letters denote statistically significant differences (P < 0.05 ANOVA and post-hoc tests).

Download figure to PowerPoint

Effects of A-Ghrelin on liver triglycerides, redox state and inflammation markers in nonobese rats

We [11] and others [12, 13] previously showed that A-Ghr leads to liver triglyceride accumulation with low AMPK activation [11] in nonobese animals (Figure 7). In additional experiments we therefore aimed at confirming this finding and at assessing whether A-Ghr-induced liver triglyceride accumulation is accompanied by altered redox state and inflammation markers. Compared to a saline-treated control group, A-Ghr lead to higher liver triglycerides (saline 0.084 ± 0.005 mg g−1 dry weight, ghrelin 0.109 ± 0.006, P = 0.048) in the presence of lower AMPK phosphorylation (−26%, P < 0.05) and preserved mitochondrial enzyme activities (not shown). No changes were observed in plasma glucose (saline 109 ± 11, ghrelin 115 ± 8 mg dl−1), insulin (saline 5.9 ± 1.8, ghrelin 6.9 ± 1 ng ml−1) and FFA (saline 0.26 ± 0.019, ghrelin 0.24 ± 0.012). Under the above conditions triglyceride accumulation was however not associated with pro-oxidative changes in redox state and inflammation markers. On the other hand and similar to obese animals, A-Ghr lead to the activation of liver glutathione peroxidase (+36%, P < 0.05) and resulted in lower oxidized-to-reduced glutathione (−16%, P < 0.05) with no changes in NFkB and TNF-alpha.

image

Figure 7. Effects of 4 days of saline (S) or ghrelin treatment (G) on liver reduced glutathione (a), oxidized glutathione (b) and their ratio (c) and on glutathione peroxidase activity (d) as well as liver nuclear p65 NFkB subunit (e) and TNF-alpha protein (f). Different letters denote statistically significant differences (P < 0.05 by ANOVA and post-hoc tests).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

In the current experimental setting, 1-month high fat feeding lead to triglyceride accumulation with pro-oxidant and proinflammatory changes in the liver. Peripheral A-Ghr administration to diet-induced obese animals was able to limit liver triglyceride accumulation and to normalize tissue redox state and inflammation markers. The current results therefore identify novel beneficial metabolic effects of A-Ghr administration in diet-induced obesity, and they suggest that A-Ghr administration may represent a novel potential intervention for high-fat diet-induced hepatic complications.

We and others previously demonstrated that A-Ghr modulates tissue lipid distribution and may lead to liver fat accumulation in non-obese rodents [11-13]. Hepatic effects of A-Ghr were intriguingly reported to be dependent on receptor GHS-R1a in one paper [13], although its expression in the liver has not been demonstrated in other studies [28]. The current results however further demonstrate that the effects of A-Ghr on liver triglyceride content are profoundly affected by nutritional state. Differential effects in obese animals could involve the diet-induced activation of AMPK that could directly favor fatty acid oxidation over lipogenesis [11]. Interestingly, this finding is consistent with preserved or elevated markers of liver oxidative capacity and AMPK activation in human obesity and type 2 diabetes [4-6, 28-31], likely representing an adaptive mechanism aimed at limiting excess tissue lipid deposition through yet unidentified mechanisms. Higher circulating adiponectin in A-Ghr-treated obese rats could have specifically contributed to prevent ghrelin-induced AMPK inactivation, since adiponectin is reported to directly activate AMPK in liver and muscle [31, 32]. A synergistic antilipogenic effect could be hypothesized for suppression of insulin signaling at AKT level that was also reported in A-Ghr-treated hepatocytes [21] and therefore likely represents a direct hormone effect that was also reported in lean animals in vivo [20]. Finally and importantly, downregulation of circulating fatty acids in ghrelin-treated obese animals also likely contributed to limit hepatic triglyceride storage through reduced substrate availability. Adipogenic, lipogenic, and antilipolytic A-Ghr effects in adipose tissue [9, 10] could have lead to enhanced adipose storage of excess fatty acids; enhanced mitochondrial oxidative capacity in skeletal muscle previously reported in the same experimental model [27] could have also contributed to enhanced fatty acid disposal.

A relevant study aim was to assess the interaction between A-Ghr, liver redox state, and inflammation markers, since the early onset of oxidative stress and inflammation is commonly proposed as a major mediator of progression from fatty liver to steatohepatitis in humans [1, 2]. Antioxidant and anti-inflammatory effects of A-Ghr including stimulation of glutathione peroxidase activity have been recently reported in vitro and in models of liver damage [14, 15]. The current results indicate that A-Ghr treatment of diet-induced obese animals indeed results in a model of liver triglyceride accumulation with normalized markers of redox state and inflammation. Oxidative stress and inflammation have been proposed to substantially contribute to the onset of nonalcoholic steatohepatitis in obese patients with fatty liver disease, and A-Ghr treatment could therefore potentially limit or prevent the onset of fatty liver metabolic complications. Stimulation of antioxidant glutathione peroxidase likely contributed at least in part to improved redox state and indirectly to lower inflammation markers, in agreement with previous in vitro reports [14, 15]. Importantly, lack of changes in inflammation markers and paradoxically improved redox state with activated glutathione peroxidase were also observed in A-Ghr-treated lean animals despite liver triglyceride accumulation, thereby further supporting the above hypothesis.

In previous studies, effects of acute ghrelin administration were investigated in lean humans, and A-Ghr was reported to enhance lipolysis and circulating fatty acids [33]. These effects may have been at least in part mediated by GH elevation that did not occur in the current experiments of chronic administration, and they were accordingly attenuated in patients with GH deficiency [33]. The same authors interestingly demonstrated the lipolytic A-Ghr effect to occur specifically in skeletal muscle [34]. Because ghrelin had been previously reported to inhibit lipolysis in adipose tissue, skeletal muscle could therefore be responsible for acute A-Ghr-induced elevation of circulating fatty acids [34]. On the other hand, limited muscle triglyceride stores may not be available to sustain circulating fatty acid availability during sustained ghrelin administration as observed in the current study.

It should finally be pointed out that A-Ghrelin infusion was reported to also acutely increase D-Ghrelin in humans through yet undefined mechanisms [35]. The potential independent metabolic effects of D-Ghr remain largely to be characterized, although differential effects of A- and D-Ghr have been described on hepatic glucose production [35, 36]. Previous studies however indicated that ghrelin effects on hepatic lipid metabolism require receptor GHS-R1a and therefore appear to be primarily mediated by A-Ghr [13]. In addition, changes in glutathione peroxidase activity, glutathione balance, inflammation and hepatic insulin signalling in the current study are consistent with previous reports specifically involving A-Ghr effects in vitro [14-16, 21]. Finally, adipogenic changes in adipose tissue [9, 10] as well as stimulation of muscle insulin signalling [37, 38] that could contribute to lower circulating fatty acids have been attributed to both acylated and desacylated ghrelin forms, and it is therefore unlikely that d-ghrelin elevation would substantially alter acylated hormone effects. While the potential independent metabolic role of d-ghrelin should be further investigated, the above observations suggest that the current results are likely largely independent of potential changes in desacylated hormone.

In conclusion, this study demonstrated that A-Ghr administration in vivo to diet-induced obese rats limits obesity-associated liver triglyceride accumulation and normalizes altered redox state and inflammation markers. The data support a potential novel role of A-Ghr in prevention and treatment of diet-induced hepatic metabolic complications. The impact of A-Ghr on liver lipid storage appears to be profoundly influenced by nutritional state, but potential antioxidant effects with preserved redox state and inflammation are also observed in lean animals despite A-Ghr-induced liver fat accumulation.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

The authors gratefully acknowledge the skilful assistance of Ms Anna De Santis in performing experiments.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References