Increased expression of GDF-15 may mediate ICU-acquired weakness by down-regulating muscle microRNAs

Rationale The molecular mechanisms underlying the muscle atrophy of intensive care unit-acquired weakness (ICUAW) are poorly understood. We hypothesised that increased circulating and muscle growth and differentiation factor-15 (GDF-15) causes atrophy in ICUAW by changing expression of key microRNAs. Objectives To investigate GDF-15 and microRNA expression in patients with ICUAW and to elucidate possible mechanisms by which they cause muscle atrophy in vivo and in vitro. Methods In an observational study, 20 patients with ICUAW and seven elective surgical patients (controls) underwent rectus femoris muscle biopsy and blood sampling. mRNA and microRNA expression of target genes were examined in muscle specimens and GDF-15 protein concentration quantified in plasma. The effects of GDF-15 on C2C12 myotubes in vitro were examined. Measurements and main results Compared with controls, GDF-15 protein was elevated in plasma (median 7239 vs 2454 pg/mL, p=0.001) and GDF-15 mRNA in the muscle (median twofold increase p=0.006) of patients with ICUAW. The expression of microRNAs involved in muscle homeostasis was significantly lower in the muscle of patients with ICUAW. GDF-15 treatment of C2C12 myotubes significantly elevated expression of muscle atrophy-related genes and down-regulated the expression of muscle microRNAs. miR-181a suppressed transforming growth factor-β (TGF-β) responses in C2C12 cells, suggesting increased sensitivity to TGF-β in ICUAW muscle. Consistent with this suggestion, nuclear phospho-small mothers against decapentaplegic (SMAD) 2/3 was increased in ICUAW muscle. Conclusions GDF-15 may increase sensitivity to TGF-β signalling by suppressing the expression of muscle microRNAs, thereby promoting muscle atrophy in ICUAW. This study identifies both GDF-15 and associated microRNA as potential therapeutic targets.


Online Repository
Increased expression of GDF-15 may mediate ICU acquired weakness by downregulating muscle microRNAs.

Measurement of mid thigh muscle layer thickness
Muscle layer thickness of the mid thigh (MLT) was measured using a previously described technique [1]. Despite being validated in ICU patients [1], the MLT has potential flaws, including the impact of variable extracellular water/oedema on the measurement. However, the alternative rectus femoris cross sectional area (RF csa ) was not measurable in all the participants, because the whole muscle was often not visualised in a single US image. B-mode US imaging with a 10MHz 12L-RS probe was used (Logiq E, GE Healthcare, UK). The patient was positioned supine on the bed with their legs flat and leg muscles relaxed. The anterior superior iliac spine (ASIS) and the point 60% of the distance from the ASIS to the superior border of the patella were identified. The US probe was positioned perpendicular to the axis of the leg. Three separate consecutive images were taken and the MLT estimated using Image-J software (National Institutes of Health, USA). The average of these three measurements was used. Inadequate ultrasound images were defined as those where the edges of the RF muscle could not be defined well enough to calculate the MLT.

Muscle biopsy and blood sampling
Rectus femoris biopsy was taken by two methods. Open biopsy was carried out by the surgical team for the control group at the time of surgery and in those patients who were anti-coagulated or undergoing general anaesthesia for another surgical intervention, for example tracheotostomy. In the rest of the patients a closed biopsy was carried out using a Bergstrom needle as previously described [2].
Muscle samples were flash frozen in liquid nitrogen or mounted on cork in O.C.T. (VWR, UK) and frozen. Plasma was separated from blood collected into EDTA sample tubes and centrifuged at 1,500g (3500 rpm) for 10 minutes. Plasma and muscle samples were stored at -80 o C until they were processed. Where the biopsy was small, samples were prioritised for RNA and microRNA quantification, and histology was only carried out in samples that were macroscopically of good quality. Adequate histology specimens were available in 4 out of 7 controls and 7 out of 20 patients of whom 5 were male; 4 were respiratory patients, 3 were recovering from cardiac surgery, and therefore can be considered representative of the whole patient group.

Measurement of muscle fibre diameter
Mounted samples were stained with haematoxylin and eosin. Images were photographed at 10x magnification and fibre diameter measured using Image-J software. The minimal Feret's diameter, defined as the smallest distance between parallel edges of each fibre (http://www.curecmd.org/wp-content/uploads/scientists/sop/MDC1A_M.1.2.002.pdf), was measured. This distance should therefore reflect the true diameter if the fibre has been cut cross-sectionally or longitudinally. In a separate analysis sections were stained for nuclear localisation of p-SMAD 2/3 (primary antibody, Santa Cruz, UK) and imaged at 20x magnification. The protocol was modified, sections were fixed with 4% paraformaldehyde and then permeabilised with 0.3% Triton X100 (Sigma, UK) in PBS-Tween, to ensure nuclear permeability to the antibody, prior to blocking with 5% bovine specific albumin (BSA). Antibodies were also prepared in BSA.

Blood Sample analysis
Plasma level of GDF-15 was quantified using a commercially available ELISA kit (R&D systems, Abingdon, UK).

Transfection of luciferases and over expression of mir-181a
C2C12 myoblasts were seeded as above. For over-expression of miR-181a, on day 1 post seeding (in a 96 well plate at 6250 cells per well) cells were transfected with miR-181a mirVANA mimic (0.5µl of a 20µM stock) or negative control (Life Technologies) using lipofectamine 2000 (Life Technologies). The following day p(CAGA) 12 luciferase and pRL-TK (Renilla luciferase) plasmids (2 µg DNA per well) were transfected with lipofectamine (Life Technologies) as previously described [4]. After 24h the cells were treated with TGF-β 2.5ng/ml (R&D Systems) or control for 6 hours. Cells were lysed in passive lysis buffer (Life Technologies, UK) and the luciferase activity measured using the dual luciferase reporting system (Promega, UK).   Figure E2: Rectus femoris muscle mRNA expression of different mRNA in patients with ICUAW (n=20) and controls (n=7) for Myostatin, MuRF-1 (muscle ring finger protein-1) and IGF-1 (insulin like growth factor -1). No significant differences were seen (Mann-Whitney). Data presented as median and error bars represent inter-quartile range.