Impaired Ca2+ release contributes to muscle weakness in a rat model of critical illness myopathy
- 869 Downloads
Critical illness myopathy is an acquired skeletal muscle disorder with severe myosin loss and muscle weakness frequently seen in intensive care unit (ICU) patients. It is unknown if impaired excitation-contraction coupling contributes to the muscle weakness.
We used a unique ICU model where rats were deeply sedated, post-synaptically pharmacologically paralyzed, mechanically ventilated and closely monitored for up to ten days. Single intact fibers from the flexor digitorum brevis muscle were isolated and used to measure force and free myoplasmic [Ca2+] ([Ca2+]i) during tetanic contractions.
Fibers from ICU rats had 80 % lower tetanic [Ca2+]i and produced only 15 % of the force seen in fibers from sham-operated (SHAM) rats. In the presence of 5 mM caffeine, tetanic [Ca2+]i was similar in fibers from ICU and SHAM rats but force was 50 % lower in fibers from ICU rats than SHAM rats. Confocal imaging showed disrupted tetanic [Ca2+]i transients in fibers from ICU rats compared to SHAM rats. Western blots showed similar levels of Na+ channel and dihydropyridine receptor (DHPR) protein expression, whereas ryanodine receptor (RyR) and sarco-endoplasmic reticulum Ca2+ ATPase 1 (SERCA1) expression was markedly lower in muscle of ICU rats than in SHAM rats. Immunohistochemical analysis showed that distribution of Na+ channel and DHPR protein on the sarcolemma was disrupted in fibers from ICU rats compared with SHAM rats.
These results suggest that impaired SR Ca2+ release contributes to the muscle weakness seen in patients in ICU.
KeywordsCritical illness myopathy Intensive care unit Skeletal muscle weakness Ca2+ handling Excitation-contraction coupling
Critical illness myopathy (CIM) is a neuromuscular disorder frequently observed in intensive care unit (ICU) patients exposed to long-term mechanical ventilation and immobilization. It is characterized by severe muscle weakness in limb and respiratory muscles, muscle wasting and a preferential loss of myosin and myosin-associated proteins [1, 2, 3, 4, 5]. Treatment of patients who develop CIM is expensive and CIM is associated with higher mortality rates .
The mechanisms that underlie the altered muscle function in ICU patients are incompletely understood [5, 7]. Recently, a rat model was introduced that mimics the ICU conditions of long-term sedation, mechanical ventilation and muscle unloading but avoids confounding factors related to age, gender, pre-existing disease or drug treatment. This ICU model develops a phenotype that closely resembles the key features of CIM in humans with loss of myosin, dysregulation of myofibrillar protein turnover, muscle atrophy and loss of specific force [8, 9].
The myosin loss and muscle atrophy in CIM will lead to muscle weakness. However, it is currently unknown whether intrinsic defects in excitation-contraction coupling (ECC) such as changes in action potentials or Ca2+ release and re-uptake by the sarcoplasmic reticulum (SR) also contribute to the muscle weakness [5, 10]. It is reported that in critically ill patients and in animal models the skeletal muscles have reduced muscle membrane excitability due to either depolarization and/or inactivated Na+ channels [11, 12, 13, 14]. The dihydropyridine receptor (DHPR) communicates depolarization of the sarcolemma to the ryanodine receptors (RyR) in the sarcoplasmic reticulum (SR) membrane causing them to open and release Ca2+ into the myoplasm. However, there are few data on changes in DHPR or RyR protein or function aside from one study, which showed increased DHPR and RyR1 protein expression in skeletal muscle in an ICU rat model . To date, there are no data on intracellular Ca2+ transients evoked by action potentials in the muscles of patients on ICU. In the present study, we hypothesized that in skeletal muscle from the ICU model rats, ECC is impaired and this contributes to the muscle weakness. Our results show clearly that in muscle fibers from rats subjected to 8–10 days of ICU simulation, action potential-induced SR Ca2+ release is significantly reduced.
The ethical committee at Uppsala University approved all aspects of this study and it complied with the recommendations in the Guide for the Care and Use of Laboratory Animals .
Animals and tissue collection
Eight sham-operated controls (hereafter called SHAM) and seven experimental female Sprague–Dawley rats (weight: 235–299 g, Taconic, Denmark) were included in this study. The experimental (hereafter called ICU) rats were anesthetized with isoflurane, treated with the neuromuscular blocker α-cobratoxin and mechanically ventilated for up to 10 days as described in detail previously . Briefly, the following preparatory surgery was carried out under sterile conditions. Precordial silver wire electrocardiogram (ECG) electrodes were implanted subcutaneously. Arterial blood pressure was monitored with an aortic catheter (28-gauge Teflon) inserted into the left carotid artery. A 0.9-mm Renathane catheter was threaded into the left jugular vein to administer parental solutions. Three subcutaneous electroencephalogram (EEG) needle electrodes were placed into the skull above the right and left temporal lobes, and a third reference electrode was placed in the neck region to record brain activity. Temperature was measured by a vaginal thermistor and servo-regulated at 37 °C. A silicone cannula was inserted in the urethra to continuously record urine output. SHAM rats underwent the same interventions as the ICU rats, except that they were not pharmacologically paralyzed with α-cobratoxin and breathed spontaneously. SHAM rats were killed within 2 hours after the anesthesia had started.
During surgery or other potentially irritating interventions, the isoflurane level was kept at >1.5 %. After the initial surgery, the isoflurane concentration was gradually reduced over the next 1–2 days and kept at <0.5 % for the rest of the experiment. Rats were ventilated through a coaxial tracheal cannula at 72 breaths/minute with an inspiratory-expiratory ratio of 1:2 and a minute volume of 180–200 ml and gas concentrations of 50 % O2, 47 % N2, and 3 % CO2. Expiratory CO2 was monitored continuously. Intermittent respiratory hyperinflations (six per hour at 15 cm H2O) were delivered throughout the experiment. Neuromuscular blockade was induced (100 μg α-cobratoxin injected intravenously) and maintained by continuous infusion (250 μg/day intravenously). Mechanical ventilation was initiated immediately after the neuromuscular blocker induction. Experiments were terminated after eight or ten days. In no case did the anesthetized rats show any sign of infection, septicemia or systemic inflammation.
Immediately after euthanasia by removal of the heart, the flexor digitorum brevis (FDB) and tibialis anterior (TA) muscles were dissected from both legs. The TA muscles were quickly frozen in liquid propane pre-cooled in liquid nitrogen and stored at −160 °C for further analyses. The FDB muscles were placed in Tyrode solution supplemented with 0.2 % fetal bovine serum and transported to the laboratory for single fiber isolation.
Specific force and [Ca2+]i measurements
The FDB muscles were split into individual digits and the central three digits were used in these experiments. Fibers were stimulated with a handheld homemade point stimulation electrode that delivered 1 ms, 6–10 V pulses. The electrode was used to stimulate fibers at various points along their length. The overwhelming majority of fibers from SHAM rats contracted when stimulated at any point along their length. In contrast, less than half of the fibers from ICU rats contracted. Using iris scissors and jeweler’s forceps we dissected only those FDB fibers that contracted when electrically stimulated. Aluminum or platinum T-clips were attached to the tendons of the isolated fiber. The fiber was then placed horizontally in a superfusion chamber between a force transducer (AE801, Kronex Technologies, CA, USA) and an adjustable holder. The fiber was stimulated with supramaximal electrical pulses (0.5–1 ms in duration) delivered via platinum electrodes placed parallel to the long axis of the fiber. The length of the fiber was adjusted until tetanic force was maximal, and the fiber diameter at this length was measured to calculate the cross-sectional area.
The mean diameter of dissected fibers was 35 ± 1 and 30 ± 1 for fibers from SHAM and ICU rats, respectively. Tetanic force was normalized to cross-sectional area and expressed as specific force (kN/m2). Fibers were superfused with Tyrode solution containing (in mM) 121 NaCl, 5.0 KCl, 1.8 CaCl2, 0.5 MgCl2, 0.4 NaH2PO4, 24.0 NaHCO3, 0.1 EDTA, and 5.5 glucose at room temperature (24–26 °C). Fetal bovine serum (0.2 %) was added to the solution. The solution was bubbled with 95 % O2 - 5 % CO2, giving a pH of 7.4. When used, 5 mM caffeine was dissolved in Tyrode and super-fused over the fiber for one minute before electrical stimulation.
To measure resting and tetanic [Ca2+]i, fibers were pressure injected with the fluorescent indicator indo-1 (Molecular Probes/Invitrogen, Carlsbad, CA, USA). Indo-1 was excited at 360 ± 5 nm and the light emitted at 405 ± 5 and 495 ± 5 nm was measured with a pair of photomultiplier tubes (Photon Technology International, Wedel, Germany). Indo-1 fluorescence signals were converted into [Ca2+]i as described previously . Mean resting [Ca2+]i was measuring during the 200 ms preceding a tetanus. Tetanic [Ca2+]i was measured as the mean of the final 250 ms of the 350 ms tetanus, and tetanic force was measured as the peak force during the tetanus. Fluorescence and force signals were sampled online and stored on a computer for subsequent data analyses.
In some experiments, enzymatically isolated FDB fibers were loaded with the fluorescent [Ca2+]i indicator fluo-3 (4 μM, 20 minutes at room temperature) and spatial and temporal aspects of [Ca2+]i transients were examined during 70 Hz electrical stimulation. A BioRad MRC 1024 confocal unit equipped with a Calypso laser (Cobolt, Solna, Sweden) attached to a Nikon Diaphot 200 inverted microscope with a Nikon Plan Apo × 20 objective lens was used to obtain line scan images (6 ms per line) of changes in the fluo-3 signal. The fiber was placed to ensure that the line scan was performed across the diameter of the fiber. Fluo-3 was excited at 491 nm, and the emitted light was collected through a 515 long-pass filter. Images were analyzed using ImageJ software (National Institutes of Health, Bethesda, MD, USA; https://imagej.nih.gov/ij/). Background fluorescence was subtracted and the fluorescence during electrically induced contractions is expressed relative to that measured before electrical stimulation (F/F0).
Repeated stimulation protocol
Fatigue was induced by stimulating fibers with 150 Hz, 500 ms tetani repeated at 1 s intervals until force decreased to 40 % of the initial force. Fibers from ICU rats contracted very weakly and to get a sizeable force at the start of fatigue runs, these fibers were exposed to 5 mM caffeine for one minute and then fatigued in the continued presence of caffeine .
Isolated muscles were homogenized with a ground glass homogenizer in ice-cold homogenisation buffer at pH 7.4 (20 μl per mg wet weight) consisting of (mM): Hepes, 20; NaCl, 150; EDTA, 5; KF, 25; Na3VO4, 1; and 20 % glycerol, 0.5 % Triton X-100, and protease inhibitor cocktail (Roche, Basel, Switzerland), one tablet/50 ml. The homogenate was centrifuged at 700 g for 10 minutes at 4 °C. Protein content of the supernatant was determined using the Bradford assay (#500-0006, Bio-Rad, Hercules, CA, USA). Samples were diluted 1:1 in Laemmli buffer (Bio-Rad) with 5 % 2-mercaptoethanol and heated to 95 °C for 5 minutes.
Ten μg protein was run on a 4–12 % precast Bis–Tris gel (NP0336PK2, NuPAGE, Invitrogen Carlsbad, CA, USA) and transferred onto polyvinylidine fluoride membranes (Immobilon FL, Millipore, Billerica, MA, USA). Membranes were then blocked with Li-Cor Blocking buffer (LI-COR Biosciences, Lincoln, NE, USA) followed by incubation overnight at 4 °C with the following antibodies diluted in blocking buffer; mouse anti-dihydropyridine receptor (DHPR; Abcam, Cambridge, UK, ab2864, 1:1000), mouse anti-ryanodine receptor 1(RyR1; Abcam, ab2868, 1:1000), anti-Na+ channel (Millipore, Catalog number 06-811, 1:1000), anti-sarco-endoplasmic reticulum Ca2+ ATPase 1 (SERCA1; Abcam, ab109899, 1:1000). Membranes were then washed and incubated with secondary antibody IRDye 680-conjugated donkey anti-mouse IgG and IRDye 800-conjugated donkey anti-rabbit IgG (926–68072, 926–32213, LI-COR). Immunoreactive bands were visualized using infrared fluorescence (IR-Odyssey scanner, LI-COR Biosciences). Band density was measured using Image Studio v 2.0.38 (LI-COR Biosciences) and normalized to subsequent Coomassie protein staining (Additional file 1: Figure S1) of the same membranes (#161-0436, Bio-Rad).
Cross-sections or longitudinal sections, 10 μm thick, were cut from the tibialis anterior (TA) muscle in a cryostat at −23 °C and placed on glass slides. Longitudinal sections were used to visualize the distribution of Na+ channels and DHPR in the sarcolemma while transverse sections were preferred to identify the distribution of the RyR and SERCA1 proteins embedded in the sarcoplasmic reticulum membrane. The tissue sections were air dried at room temperature and stored at −80 °C. The sections were fixed for 20 minutes in cold 2 % formaldehyde diluted in phosphate-buffered saline system (PBS) (Sigma-Aldrich), and then washed with 0.1 % saponin (Sigma) at pH 7.4 (referred to as PBS-Sap) for 10 minutes. The blocking step was performed with 5 % normal goat serum (Sigma-Aldrich) for 1 hour at room temperature. After washing with PBS-Sap, sections were stained with anti-ryanodine receptor 1 (RyR1; Abcam, ab2868, 1:600), anti-sarco-endoplasmic reticulum Ca2+ ATPase 1 (SERCA1; Abcam, ab109899, 1:200), anti-Na+ channel (Millipore, Catalog number 06-811, 10 μg/ml), and anti-DHPR (Abcam, ab2864, 1:600) diluted in 1 % goat serum + PBS-Sap and incubated overnight at 4 °C in a humid chamber. After washing with PBS-Sap, secondary antibodies (Alexa 488 goat anti-rabbit IgG, or Alexa 488 goat anti-mouse IgG, 1:1000, Invitrogen, Oregon, USA) diluted in 1 % goat serum + PBS-Sap were applied to the sections for 30 minutes in the dark. A negative control for each run was used (Additional file 2: Figure S2). Slides were mounted using the anti-fade mounting medium for fluorescence Vectashield (Vector, Burlingame, CA). Slides were examined using a Bio-Rad MRC 1024 confocal unit with a Calypso dual laser (Cobolt, Solna, Sweden) attached to a Nikon Diaphot 200 inverted microscope and a Nikon Plan Apo × 40 oil-immersion objective, N.A. 1.3 (Nikon, Tokyo, Japan). The stained sections were coded and evaluated in a descriptive manner by independent observers (MLD, JDB, VS and AJC). Images were prepared using ImageJ software (National Institutes of Health, Bethesda, MD; https://imagej.nih.gov/ij/).
Data are presented as mean ± SEM. Statistical analyses were performed using the SigmaPlot software (Systat Software, Inc., CA, USA). Unpaired t tests were performed to establish significant differences between groups and a p value <0.05 was considered to be statistically significant.
Tetanic [Ca2+]i and specific force in FDB fibers are reduced after 8–10 days of the ICU intervention
To determine whether the lower tetanic [Ca2+]i and specific force recorded fibers from ICU rats were due to less Ca2+ being stored in the SR, the fibers were electrically stimulated in the presence of 5 mM caffeine, which potentiates SR Ca2+ release and releases all stored free Ca2+ . The presence of caffeine significantly increased tetanic [Ca2+]i (p < 0.05) in fibers from both ICU and SHAM rats and there was no longer any statistically significant difference in tetanic [Ca2+]i between fibers from ICU and SHAM rats (Fig. 1b), indicating that SR Ca2+ stores were not markedly reduced in fibers from ICU compared to SHAM rats. However, despite the marked increased in tetanic [Ca2+]i induced by the presence of caffeine, mean specific tetanic force was still approximately 50 % lower ((Fig. 1d, p < 0.05) in fibers from ICU than in SHAM rats, reflecting the fact that maximum force-generating capacity was lower in the fibers from ICU rats than in the SHAM rats.
Some fibers were fatigued by repeated tetanic stimulation and fibers from ICU rats underwent this stimulation in the presence of caffeine in order to have a sizeable starting force. Figure 1e-h shows representative records of [Ca2+]i (Fig. 1e, f) and specific force (Fig. 1g, h) during repetitive stimulation of fibers from SHAM and ICU rats. The fiber from ICU rats clearly responded to repeated tetanic stimulation, but the amplitude of [Ca2+]i and force transients decreased more rapidly than those in the fiber from SHAM rats. The changes in resting [Ca2+]i during the period of repetitive stimulation was similar in fibers from SHAM and ICU rats indicating that SR Ca2+ pumping increased in both the ICU and the SHAM rats. Overall, these results show that fibers from ICU rats increase ATP production to cope with the increased energy demands of repeated tetanic [Ca2+]i and force transients.
Examination of ECC proteins in SHAM and ICU rat muscle
The western blot data in Fig. 3 show that protein expression level of the SR Ca2+-release channel, RyR, and the SR Ca2+ pump, SERCA1 were markedly reduced in ICU rat compared to SHAM rat muscle. This finding was confirmed by the immunofluorescence staining, which shows that both the extent and the number of fibers with clear positive staining (marked with white asterisks) for RyR1 or SERCA1 was markedly lower in ICU than in SHAM muscle fibers (Fig. 4 e-h).
Previous studies have clearly demonstrated that in critically ill patients and in ICU rat models there is a massive preferential loss of myosin from the muscles [5, 20]. In addition to this myosin loss, the present study shows clearly that in an ICU rat model the muscle fibers respond to electrical stimulation with small and non-uniform tetanic [Ca2+]i transients that underlie the impaired force production.
The decreased tetanic [Ca2+]i transients in ICU rat muscle fibers led us to hypothesize that proteins involved in one or more steps in ECC were affected. The decreased protein expression of the RyR receptor shown in our western blots (Fig. 3) is likely to contribute to the smaller tetanic [Ca2+]i transients in ICU rat fibers. Stores of SR Ca2+, which provide the driving force for Ca2+ to leave the SR did not seem to be significantly different between ICU and SHAM rat muscle as caffeine, which augments SR Ca2+ release, increased tetanic [Ca2+]i to similar levels in ICU and SHAM rat fibers. There are few existing data on changes in RyR in CIM or ICU models. Increased RyR immunostaining has been reported in a combined surgical denervation-steroid treatment rat model . It should be noted that this model involves both partial denervation and sustained motor activity and differs markedly from both patients in ICU who develop CIM and the normally innervated, unloaded rat ICU model used in the present study.
The decreased tetanic [Ca2+]i in ICU fibers measured during electrical stimulation might also be related to impaired activation of RyR. Earlier electrophysiological data indicate that there is decreased membrane excitability, reflecting reduced expression/activation of Na+ channels in the sarcolemma in patients, [11, 12]. Our western blot data show that protein expression of the Na+ channel is similar in ICU and SHAM rat muscles. On first observation, this may appear puzzling. However, the western blot data indicate only that the denatured Na+ channel is present in the muscle, and tell us nothing about the state of these channels or how they are distributed along the sarcolemmal or transverse tubule membrane in the intact muscle fibers.
The immunofluorescent images show clearly that the distribution of the Na+ channel staining is not normal in fibers from ICU rats, with only approximately 40 % of the muscle fibers having a pattern resembling that seen in SHAM rat fibers. It can be speculated that a variable and abnormal distribution of Na+ channels in the sarcolemma of ICU rat fibers leads to a variety of action potential transmission problems and ultimately impaired Ca2+ release from the SR as reflected by the reduced tetanic [Ca2+]i transients recorded with indo-1. In addition, the confocal images showed that in a majority of ICU rat muscle fibers, there was an abnormal uneven spatial release of Ca2+ from the SR. In about 40 % of ICU fibers, ECC operated normally close to the sarcolemma but deeper in the fiber; there was clearly no SR Ca2+ release in ICU fibers. This is suggestive of subtle derangement of the DHPR or Na+ channel function in the transverse tubules.
Western blot data showed that ICU rat fibers had markedly decreased expression of SERCA1 (Fig. 3). However, the resulting reduction in total SR Ca2+ pumping capacity in ICU rat fibers did not have any major functional impact, as first, the resting [Ca2+]i and Ca2+ stored in the SR were similar in ICU and SHAM rat fibers, and second, [Ca2+]i rapidly returned to the baseline level when ICU rat fibers were subjected to repeated tetanic contractions (see Fig. 1).
A further contributor to the decreased tetanic [Ca2+]i in ICU rat fibers that must be considered is the DHPR protein, either in terms of quantity or its ability to respond to depolarization by the conformational change in its protein structure that is required to open the RyR channel. It is well-known that the DHPR can exist in either an active state where it responds to depolarization, or in an inactivated/blocked state, which does not respond to depolarization, e.g., as previously described . As pointed out by Friedrich and colleagues , aside from a study using a denervation-steroid treatment rat model , there have been no previous investigations of DHPR function in an ICU model. Our western blot data suggest that DHPR protein expression is not greatly different in ICU and SHAM rat muscle, while immunofluorescent images indicate that the pattern of DHPR staining differs in fibers from ICU rats compared to those from SHAM rats.
The present study focused only on those fibers that contracted upon electrical stimulation and did not examine those fibers that were electrically unexcitable. Under carefully controlled experimental conditions, the majority of ICU rat muscle fibers developed marked disturbances in excitation-contraction coupling that results in reduced amplitude and duration of tetanic [Ca2+]i transients and lower force generation. The confocal images showed that there were marked differences in tetanic [Ca2+]i transients in different fibers taken from the same rat. The precise reasons for this variability in response remain unclear to us. Differences in the degree of expression or localization of the Na+ channel or DHPR in individual fibers and/or their functional response to changes in the membrane potential are likely to underpin this variability in tetanic [Ca2+]i transient responses. It is unlikely that a difference in fiber types between SHAM rat and ICU rat muscle plays any role. Rat FDB muscles are composed predominantly of type II fibers (almost 90 %) and while they develop significant atrophy after 2 weeks of hind limb unloading, they show no evidence of fiber transformation .
In conclusion, our data demonstrate that in addition to the myosin loss, defective ECC due to impaired activation of RyR, and less action-potential-induced SR Ca2+ release contribute to the muscle weakness observed in ICU rats.
[Ca2+]i, free myoplasmic [Ca2+]; CIM, critical illness myopathy; DHPR, dihydropyridine receptor; ECC, excitation-contraction coupling; ECG, electrocardiogram; FDB, flexor digitorum brevis; ICU, intensive care unit; PBS, phosphate-buffered saline; RyR1, ryanodine receptor; SERCA1, sarco-endoplasmic reticulum Ca2+ ATPase 1; SHAM, sham-operated; SR, sarcoplasmic reticulum; TA, tibialis anterior
The authors thank Ann-Marie Gustafson and Yvette Hedström, Department of Neuroscience, Clinical Neurophysiology, Uppsala University Hospital for excellent technical assistance. The study was supported by grants from the Swedish Research Council to HW and LL and the Swedish Foundation for International Cooperation in Research and Higher Education (STINT) and the Stockholm County Council (ALF project) to LL.
MLD, AJC, NI, HW, NC, LL and JDB were involved in the study design and conception of the manuscript. NC and LL conducted the animal experiments and tissue harvesting. MLD, AJC, VS, NI, WJ and JDB were responsible for data collection and analysis. MLD, AJC, WJ, NI, HW, VS, NC, LL and JDB contributed to the writing of the manuscript and preparation of the figures. All authors read and approved the final manuscript.
The authors declare that they have no competing interests.
- 1.Larsson L, Li X, Edström L, Eriksson LI, Zackrisson H, Argentini C, Schiaffino S. Acute quadriplegia and loss of muscle myosin in patients treated with nondepolarizing neuromuscular blocking agents and corticosteroids: mechanisms at the cellular and molecular levels. Crit Care Med. 2000;28:34–45.CrossRefPubMedGoogle Scholar
- 6.Hermans G, Van Mechelen H, Clerckx B, Vanhullebusch T, Mesotten D, Wilmer A, Casaer MP, Meersseman P, Debaveye Y, Van Cromphaut S, et al. Acute outcomes and 1-year mortality of intensive care unit-acquired weakness. A cohort study and propensity-matched analysis. Am J Respir Crit Care Med. 2014;190:410–20.CrossRefPubMedGoogle Scholar
- 8.Ochala J, Gustafson AM, Diez ML, Renaud G, Li M, Aare S, Qaisar R, Banduseela VC, Hedstrom Y, Tang X, et al. Preferential skeletal muscle myosin loss in response to mechanical silencing in a novel rat intensive care unit model: underlying mechanisms. J Physiol. 2011;589:2007–26.CrossRefPubMedPubMedCentralGoogle Scholar
- 16.National Research Council (US) Committee for the Update of the Guide for the Care and Use of Laboratory Animals. Guide for the Care and Use of Laboratory Animals. 8th edition. Washington (DC): NationalAcademies Press (US); 2011. Available from: http://www.ncbi.nlm.nih.gov/books/NBK54050/. doi: 10.17226/12910.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.