The application of stable‐isotope tracers to study human musculoskeletal protein turnover: a tale of bag filling and bag enlargement

Abstract

The nutritional regulation of protein and amino acid balance in human skeletal muscle carried out by the authors with Mike Rennie is reviewed in the context of a simple physiological model for the regulation of the maintenance and growth of skeletal muscle, the “Bag Theory”. Beginning in London in the late 1970s the work has involved the use of stable isotopes to probe muscle protein synthesis and breakdown with two basic experimental models, primed‐dose continuous tracer infusions combined with muscle biopsies and arterio‐venous (A‐V) studies across a limb, most often the leg, allowing both protein synthesis and breakdown as well as net balance to be measured. In this way, over a 30 year period, the way in which amino acids and insulin mediate the anabolic effect of a meal has been elaborated in great detail confirming the original concepts of bag filling within the muscle endomysial “bag”, which is limited by the “bag” size unless bag enlargement occurs requiring new collagen synthesis. Finally we briefly review some new developments involving ²H₂O labelling of muscle proteins.

Introduction

Our purpose here is to review the application of stable‐isotope tracers to the study of human musculoskeletal protein turnover from the specific focus of a simple model for the growth and maintenance of the muscle mass identified in the 1980s as a “bag” theory (see Millward, 1995). Muscle differs from many other tissues in that its cells, the myofibres, are encased in an inelastic connective tissue framework, through which the transmission of the force of muscle contraction occurs and which defines myofibre volume. During growth, remodelling of muscle connective tissue occurs in response to the passive stretch of bone lengthening allowing myofibre mass to increase as a fixed function (4th power) of the increases in length of the long bones so that a proportionality between strength and body weight is maintained. In the adult, after epiphyseal fusion of the long bones, height growth stops with muscle mass fixed at its phenotypical size, which is a function of height. No further growth occurs unless muscle is subject to passive stretch through resistance exercise. The extracellular matrix connective tissue sheath can be conceived as a series of “bags” which surround the individual myofibres (endomysium), bundles of myofibres (perimysium) and the entire muscle (epimysium), so that muscle growth and maintenance is conceived in simple terms as bag enlargement and bag filling with intracellular proteins. Because humans exhibit a diurnal cycle of fasting and feeding in which body protein is lost and replaced, muscle protein homeostasis in the adult involves postprandial protein deposition, bag refilling, after bag emptying in the postabsorptive state.

When the work began in the labs of Dave Halliday, Mike Rennie and Millward in London in the late 1970s we had an evidence base of the nutritional regulation of muscle growth in the rat (Millward & Waterlow, 1978), and muscle hypertrophy in an avian model (Laurent & Millward, 1980), and some information on whole‐body protein turnover in humans, but no information on skeletal muscle. The re‐emergence of the use of stable isotopes allowed that work to start. In fact there is a large literature on the use of stable isotopes in human metabolic research, of which a 1999 review by Mike Rennie was written at a level suitable for students and new workers to the field (Rennie, 1999), while a more recent review by Daniel Wilkinson reviews both historical and contemporary stable‐isotope tracer approaches to studying protein metabolism in great detail (Wilkinson, 2018). Waterlow's update of his original protein turnover book remains an important reference source in terms of protein turnover (Waterlow, 2006) whilst Bob Wolfe's book (1992) deals with stable isotope terminology and kinetic analysis.

All of the work to be described has involved two basic approaches. Once muscle biopsies became an accepted standard clinical tool, tracer incorporation studies of muscle protein synthesis (MPS) during primed‐dose continuous tracer infusions became widely used allowing not only accurate precursor–product calculations of MPS, but also measurement of the molecular signalling systems which mediate responses. Measurement of net balance and tracer dilution with arterio‐venous (A‐V) studies across a limb, most often the leg, allows both protein synthesis and breakdown to be assessed in the limb (LPS and LPB). This can avoid the need for muscle biopsies, although when combined with muscle biopsies, the A‐V approach provides a more complete description of the turnover process. However, this is a complex model requiring measurement of blood flow, amino acid balance, tracer dilution, and when [1‐¹³C]leucine is used, α‐ketoisocaproate (KIC) exchange and [¹³C]bicarbonate enrichment. Tracer dilution is used to calculate amino acid release, MPB, with MPS the sum of net balance and breakdown. A‐V studies are difficult to mount and interpretation of results must take into account potential influences of non‐skeletal muscle tissues, skin, bone and adipose tissue. Nevertheless important advances have been made in our understanding of protein turnover in muscle with the A‐V system since it was clear from the start that both protein synthesis and proteolysis are highly regulated processes. Furthermore, given that the ultimate goal of most studies of muscle protein metabolism is to achieve net gain where possible, this is only measurable with the A‐V model.

Initial studies of human muscle protein synthesis

Response to feeding and in muscle wasting disease

This phase of the work ran for a 20‐year period from the early 1980s to 2000. It was enabled by the introduction of gas chromatography–mass spectrometry with selective ion monitoring (SIM‐GCMS) (Holmes et al. 1973; Bier & Christopherson, 1979; Matthews et al. 1979), which allowed rapid automated analysis of ¹³C‐ and ¹⁵N‐labelled amino acids and metabolites, although the low level enrichment of protein‐bound amino acids isolated from muscle biopsies required the greater sensitivity and precision of gas‐phase isotope ratio mass spectrometry (IRMS) with laborious wet chemistry techniques needed to separate amino acids and generate the ¹³CO₂ or N₂ gas from the biological samples (Halliday & Read, 1981).

After an initial pioneering study of muscle protein synthesis by Halliday & McKeran (1975) involving muscle biopsies during a 20–30 h constant i.v. infusion of l‐[¹⁵N]lysine, a series of 7–8 h studies involving primed‐dose continuous infusions of [1‐¹³C]leucine were mounted in London by Rennie, Halliday and Millward. With no SIM‐GCMS facilities in Europe at the time, collaboration with Matthews in St Louis was required. The studies involved healthy fasted and fed adults (Rennie et al. 1982a) and adolescent boys with Duchenne muscular dystrophy (DMD; Rennie et al. 1982b). [1‐¹³C]Leucine allowed both whole‐body protein turnover, as synthesis, breakdown and leucine oxidation, to be determined from the leucine flux, indicated by plateau enrichment of α‐ketoisocaproate, [1‐¹³C]KIC, a surrogate measure of leucyl‐tRNA, the precursor for protein synthesis, and leucine oxidation calculated from the rate of expired ¹³CO₂ in breath. MPS was indicated by tracer uptake into protein‐bound leucine. [1‐¹³C]KIC was measured by chemical ionization‐SIM‐GCMS (Schwarz et al. 1980) with the protein‐bound [¹³C]leucine from the muscle biopsy analysed by IRMS of ¹³CO₂ in Halliday's lab after laborious preparative ion exchange chromatography and reaction with ninhydrin.

These studies showed that, in the adult men, MPS doubled with feeding indicating a sensitivity of MPS to nutrition, comparable to that observed in animal studies (Millward & Waterlow, 1978), but which also indicated, somewhat surprisingly, that muscle contributed more than half of whole‐body protein synthesis as indicated by the plasma leucine flux.

In the boys with DMD, MPS was dramatically reduced (by two‐thirds), implying that muscle protein breakdown must also have been reduced since the rate of muscle wasting was less than what would have been the case if MPB had not been reduced. At the time the common view was that in wasting diseases, like muscular dystrophy, wasting reflected an elevated rate of muscle proteolysis, with many working on proteolytic inhibitors as therapeutic agents. Furthermore, in the boys with DMD, their urine contained a threefold increase in the ratio of N ^τ (or 3)‐methyl‐histidine (3‐MH) to creatinine. This ratio was assumed to indicate the rate of muscle protein breakdown because 3‐MH, as a post‐translationally modified amino acid mainly in actin, was excreted unmodified after release by proteolysis. Thus its urinary excretion rate in relation to creatinine (an index of muscle mass), had become widely used as an indicator of the rate of muscle proteolysis. In fact, rat kinetic studies with ¹⁴C tracers conducted in Millward's and Rennie's laboratories clearly showed that small, rapidly turning‐over actin pools in the gastrointestinal tract, skin and other non‐skeletal muscle tissues accounted for 75% of urinary 3‐MH excretion (Millward et al. 1980; Rennie & Millward, 1983). In the boys with DMD with their reduced skeletal muscle mass, both creatinine and muscle‐derived 3‐MH would have been reduced so that the normal non‐muscle sources of urinary 3‐MH meant that an elevated 3‐MH/creatine ratio was to be expected. Subsequently Rennie reported direct measurements of a reduced efflux of 3‐MH from the legs of postoperative patients in whom urinary 3‐MH and nitrogen excretion was increased, confirming that changes in urinary 3‐MH excretion did not reflect changes in skeletal muscle protein breakdown (Rennie et al. 1984).

The rates of MPS identified in these studies were higher than anticipated from the 1975 [¹⁵N]lysine study. Subsequently Halliday's lab acquired both GCMS and new derivatisation approaches (Ford et al. 1985), and several improved IRMS instruments enabling twofold improvement in the sensitivity, accuracy and precision of the protein‐bound [¹³C]leucine enrichments. This meant that by 1988 Halliday and Rennie could publish more studies of MPS in the fed and fasted state and in patients with a variety of myopathies (Halliday et al. 1988). They reported a pattern of responses of MPS rates for quadriceps muscle protein to feeding and muscle wasting disease which was the same as in the 1982 studies (an increase with feeding and much lower in disease), but the absolute rates of MPS were lower and considered to be more accurate.

Mechanisms of the feeding response

The role of amino acids and insulin

The relative role of amino acids and insulin in these feeding responses of human MPS became an important issue in the early 1980s. Rat studies with radioactive tracers had clearly shown that the feeding response of MPS was mediated by a direct effect of insulin on the translational phase of protein synthesis with little evidence for an important stimulatory role for amino acids (Millward et al. 1976, 1983; Jefferson et al. 1977; Millward & Waterlow, 1978). Although feeding induced increases in the essential amino acids, they increased to a much greater extent when food was withdrawn (Millward et al. 1974) and MPS was reduced (Garlick et al. 1973). This was in contrast to in vitro studies in muscle which had suggested a role for leucine in the stimulation of MPS (e.g. Buse & Reid, 1975; Fulks et al. 1975).

In fact the work to identify the relative importance of insulin and amino acids in the feeding response of MPS in human adults was to prove extraordinarily difficult, requiring work over several decades. William Bennet in Rennie's group in Dundee carried out the first series of studies deploying both biopsy and A‐V studies across the leg. While a 4 h mixed amino acid infusion during a two‐phase (4 h + 4 h) [1‐¹³C]leucine constant infusion, muscle biopsy study (Bennet et al. 1989) should have been straightforward, interpretation was complicated since neither the [¹³C]leucine enrichment in the muscle free or protein‐bound pools changed significantly. However, a modest 35% increase in MPS was observed when either plasma [¹³C]KIC or leucine was used as surrogate precursor, since their enrichment was lower, indicating a stimulation of MPS by amino acids. This was confirmed with a two‐phase (3 h + 3 h) A‐V leg study with l‐[¹⁵N]phenylalanine and l‐[1‐¹³C]leucine in healthy men and women (Bennet et al. 1990b): a mixed amino acid infusion improved amino acid balance compared with the postabsorptive state through 24–31% increases in MPS with no change in MPB indicated by both tracers. Further support for the stimulation of MPS by amino acids came from studies in young pigs in which a mixed amino acid infusion, with insulin clamped, increased MPS as indicated by [1‐¹³C]leucine incorporation into muscle protein compared with muscle [¹³C]leucyl‐tRNA, the true precursor for protein synthesis (Watt et al. 1992).

As for the role of insulin, having to infuse amino acids with the insulin to avoid a reduction in amino acid concentrations was a complicating factor in study design. During a biopsy study in insulin‐withdrawn Type 1 diabetic men (Bennet et al. 1990c), amino acids were infused throughout and insulin was infused during the last 4 h when the amino acid infusion rate was increased. MPS did not change in response to insulin regardless of whether the precursor surrogate was plasma leucine, KIC or intramuscular leucine. However, the increased amino acid supply during the insulin infusion was not enough to prevent reductions in some intramuscular amino acids (e.g. a 75% fall for tyrosine). This most likely reflected an inhibition of MPB by insulin, as indicated by the fall in whole‐body protein breakdown calculated from the changes in the leucine flux. Thus while an inhibition of MPB seemed likely, lack of stimulation of MPS could have reflected a substrate limitation.

With A‐V leg studies during continuous amino acid supply, insulin's potential influence was studied comparing physiologically low insulin with a euglycaemic hyperinsulinaemia during an increased rate of amino acid infusion, to enable an unambiguous influence of insulin on MPS to be identified (Bennet et al. 1990a). In fact these results were not entirely satisfactory for two reasons. Firstly the increased rate of amino acid infusion during the hyperinsulinaemia prevented separation of their influences on any change in MPS. Secondly the two tracers indicated different mechanisms for the net positive balance in response to the insulin. [¹⁵N]Phe indicated a marked increase in uptake (protein synthesis) with minimal effect on release (protein breakdown), while [1‐¹³C]Leu indicated a marked reduction in release (protein breakdown) and a slight increase in MPS, although this was not entirely clear because of concern for the accuracy of increased leucine oxidation across the leg, much greater (×4) than observed in the whole body (×2). Given the better confidence of the phenylalanine results due to the simplicity of the model, i.e. Phe is not metabolised in muscle making consequent analyses easier, it was concluded that insulin, given with sufficient amino acids, may stimulate MPS and inhibit MPB, but delineating its role required further investigation.

Further explorations of the amino acid influence on muscle protein synthesis

There are a number of important issues which emerged related to the stimulation of MPS by amino acids within the anabolic effect of feeding: (a) which amino acids were involved, (b) which free amino pool (extracellular (EC) or intracellular (IC)) was sensed for its anabolic influence, and (c) what was the time course and dose response.

Evidence on the first issue had accumulated from studies of the effects of large doses of single amino acids on MPS. A flooding dose of 3.5 g of 20 atom% [1‐¹³C]leucine, aimed at equalising enrichment of all free amino acid pools, had been introduced as a method of measuring MPS over a short 90 min period (Garlick et al. 1989). This indicated much higher basal rates of MPS than observed with trace level infusions, with no effect of feeding (McNurlan et al. 1993), suggesting that the large dose of leucine stimulated MPS. In a prolonged series of studies, which included a collaborative study with Garlick's group (Smith et al. 1992b), Smith and Rennie investigated the influence of large doses of individual amino acids on MPS. The large dose was given during the last 90 min of 7.5 h primed‐dose constant infusions of tracer amounts of ¹³C‐labelled amino acids and MPS was measured in muscle biopsies before (30 min to 6 h) and after (6–7.5 h) the flooding dose. Flooding doses of both leucine (Smith et al. 1992a) and valine, phenylalanine and threonine stimulated MPS, but not serine, glycine, alanine or proline (Smith et al. 1998). Thus leucine and other essential amino acids were initiating an anabolic response supported by amino acids from the free pool.

As to the the issue of where amino acid sensing occurred, this was investigated in dose–response studies of MPS to amino acids by Rennie in Bob Wolfe's laboratory in Galveston, Texas, during a 2‐year sabbatical (1998–2000). With primed constant infusion–muscle biopsy studies of young men and women with D₅‐Phe (or in one case D₃‐KIC), infusions of mixed amino acids were given at increasing rates so that blood amino acid concentrations were raised by up to 2‐ to 3‐fold basal (Bohe et al. 2003). With muscle free D₅‐Phe or D₃‐Leu as precursor surrogates, MPS increased by 30–90%. However, these increased rates of MPS were accompanied by a decreasing concentration of intramuscular essential amino acids (EAAs) at low amino acid infusion rates, but with an obvious hyperbolic relationship with extracellular (plasma) EAA concentrations. This indicated that MPS stimulation depended on the sensing of the concentration of EC rather than IC EAAs. It was proposed that the fall in intramuscular EAAs observed when blood amino acid concentrations rise modestly reflected the stimulation of MPS by the EC sensor with increased deposition of amino acids into muscle protein at a faster rate than the inward membrane transport of amino acids. As to the nutritional significance of the observed dose response of MPS to infused EAAs, it suggested that near‐maximal stimulation was achieved by an infusion rate of 162 mg kg⁻¹ h⁻¹ which for adults of average weight 55–75 kg would be equivalent to about 9–12 g of protein in a meal. However, in a study in young men of the dose response to a single whey protein (WP) dose of 0, 10, 20 or 40 g myofibrillar MPS increased maximally at 20 g even though both insulin and plasma leucine were higher at 40 compared with 20 g WP (Witard et al. 2014). Finally, and somewhat surprisingly, studies of MPS in elderly women comparing 20 or 40 g of whey protein with low doses of EAAs (1.5, 3 g and 6 g), all containing 40% leucine, showed that maximal stimulation of MPS was achieved by the lowest dose of 1.5 g EAAs containing only 600 mg of leucine (Bukhari et al. 2015; Wilkinson et al. 2017b). Although the nutritional significance of this is not easy to explain, especially given the anabolic resistance in the elderly (see below) it clearly shows the high sensitivity of MPS in muscle to leucine supply. The potential mechanisms of the amino acid‐mediated response are discussed later.

Further explorations of insulin's role on muscle protein turnover

While an inhibition of muscle proteolysis was clear from Bennet's studies described above any permissive or enhancement of the EAA stimulation of MPS was not clearly identified. In the amino acid dose–response studies described above (Bohe et al. 2003) insulin was not controlled but its availability was not different during infusion at the two lowest rates, when MPS increased by ∼30 and ∼60%, and its increase at the highest amino acid infusion rate was not associated with any further increase in MPS. This suggested that insulin had little or no part in the increase of MPS with the increased amino acid levels (although this did not exclude a permissive effect at low concentrations).

The role of insulin was further investigated in more detail in two further important studies. In the first, a cross‐over study in eight healthy young men (Greenhaff et al. 2008), in which both A‐V and biopsy studies with [²H₅]Phe and [1‐¹³C]Leu were deployed, the response of LPS, LPB and MPS to insulin clamped at four different concentrations from basal to supraphysiological was assessed during a high rate of amino acid infusion, i.e. a total of 32 infusions. The results were quite clear showing that compared with the postabsorptive state the increased amino acid levels resulted in a doubling of both LPS and MPS, with no influence of insulin, while LPB fell only when insulin was raised above basal (from 5 to 30 μIU l⁻¹), with no further dose response (at 70 or 170 μIU l⁻¹).

Subsequently in studies examining the effect of a very modest increase in insulin on MPB in healthy young and elderly men and women by means of A‐V studies with D₅‐Phe and [1,2‐¹³C₂]Leu during euglycaemic, isoaminoacidaemic clamp procedures (Wilkes et al. 2009), increasing plasma insulin concentrations very modestly, from 5 to 15 μIU ml⁻¹, lowered LPB maximally, by 47%, in the young subjects with no change in either leg LPS or muscle MPS.

Taken together the above studies made it quite clear that the anabolic feeding response to a protein‐containing meal involved both increases in MPS in response to the increased availability of the essential amino acids in the extracellular compartment which signalled the response with some individual essential amino acids, especially leucine, playing an important role, with insulin's role limited to no more than permissive, together with an inhibition of MPB mediated by very modest increases in insulin. Since these studies a systematic review and meta‐analysis of all studies of insulin's role in the regulation of human skeletal muscle protein metabolism indicated its permissive role in MPS in the presence of elevated amino acids, and a direct role in reducing MPB independent of amino acid availability (Abdulla et al. 2016).

Protein turnover in muscle in the context of models of muscle homeostasis and growth

Primed dose constant infusion [1‐¹³C]leucine studies in the fed and fasted state at various levels of habitual protein intake had allowed whole‐body postprandial protein utilisation to be assessed in terms of [1‐¹³C]leucine balance (intake–oxidation) as a surrogate for nitrogen balance (see Millward & Pacy, 1995). This had revealed that adaptive changes in amino acid oxidation with increasing protein intakes results in an increasing amplitude of the diurnal losses and gains (see Fig. 1) in which muscle mass is likely to participate disproportionately (e.g. Biollo et al. 1995). Thus homeostasis is maintained at the phenotypic muscle mass at any level of protein intake over a wide range as long as food intake stimulates sufficient protein deposition to replace postabsorptive losses. As indicated above this can be conceptualised as “bag filling” until it is limited by a “bag full, or muscle full” signal preventing further deposition by either inhibiting MPS or stimulating MPB. Any growth subsequent to resistance exercise is assumed to occur after bag enlargement.

Figure 1. The regulation of the lean body protein mass throughout the diurnal feed–fast cycle in subjects adapted to increasing habitual protein intakes.

Body protein is regulated at a phenotypically fixed muscle mass although there may be an increasing splanchnic protein mass associated with digestion absorption and metabolism as protein intakes increase. The increasing postabsorptive losses result from the adaptive increases in the capacity of the amino acid oxidation pathways to avoid potentially toxic expansions of the relatively small cellular pool sizes of the branched chain, aromatic and sulphur amino acids. The increased oxidative capacity is maintained throughout the diurnal feed–fast cycle with only a very slow adaptation (weeks) with a change to a lower intake. Thus in the postabsorptive state more muscle protein is lost on a higher than lower protein intake. The “bag” or “muscle full” phenomenon should be observed in the postprandial state at protein intakes consistent with the habitual protein intakes of the subjects. (See Millward, 1995; Millward & Pacy, 1995.)

Whilst we recognise this to be a very simplified description of a more complex process, nevertheless, to the extent that it does have some merit, several predictions flow from these concepts.

• The postprandial anabolic response should be of limited extent, i.e. a “muscle full” state in terms of an anabolic response to feeding should be observed.

• Muscle collagen turnover should normally be very limited apart from during any bag enlargement during muscle hypertrophy (e.g. with resistance exercise).

• An adequate anabolic response is required for maintenance of muscle mass and might not occur when muscle loss occurs as in the sarcopenia observed with ageing or in chronic disease where muscle wasting is a feature.

Bag filling: demonstration of the muscle full state

This was first demonstrated using an infusion of D₃‐KIC to measure MPS during Rennie's sabbatical in Bob Wolfe's laboratory (Bohe et al. 2001). With this tracer they were able to measure both precursor and product, D₃‐leucine, in muscle protein by GCMS alone. In response to a high‐dose mixed amino acid infusion over 6 h (∼70 g protein), MPS was stimulated by ∼2.8‐fold after a ∼30 min latent period, but this only lasted for 90 min before returning to baseline despite continuing elevated concentrations of amino acids for a further 4 h.

Ten years later the stable‐isotope technology was much improved with GC‐combustion‐IRMS allowing incorporation of [1,2‐¹³C₂]leucine into muscle biopsy protein with sufficient precision to analyse uptake over 45 min intervals with confidence. A bolus dose of 48 g whey protein tripled MPS after a 45 min latency before rapidly returning to postabsorptive rates after 90 min, despite the continued availability of amino acids (Atherton et al. 2010). At the same time as these studies Drummond et al. (2010) showed that after stimulating MPS in young adults with 10 g EAAs by 2.4‐fold from baseline, the rate had returned to baseline by 2 h even though muscle free leucine remained elevated.

More recently Mitchell et al. (2015) showed the muscle full response in healthy young men with 15 g of a mixture of just the nine EAAs, either given as a bolus or as 4 × 3.75 g portions given every 45 min over 2 h during 7 h infusions of l‐[ring‐¹³C₆]phenylalanine. Biopsies taken at −2, 0, 1.5, 3 and 4 h showed identical responses of MPS: i.e. a ∼90 min latency, a 60–70% increase after 90 min, and a return to baseline after 180 min, although, as would be expected, there were marked differences in the post‐meal amino acid and insulin profiles with the spread pattern maintaining increased leucine concentrations for at least 4 h. According to the adaptive increases in postabsorptive losses with increasing intakes (see Fig. 1), the maximum postprandial deposition in muscle should increase with habitual protein intake, but this has yet to be investigated.

Finally, most recently, Mitchell et al. (2017) used 7 h infusions of [¹³C₆]phenylalanine in active older males to demonstrate that 15 g of EAA doubled MPS between 90 and 180 min, after which it returned to baseline and was refractory to a topping‐up dose of 3 g of leucine at 90 min, which maintained elevated leucine until the end of the study. Insulin peaked at 30 min and returned to baseline by 90 min in both cases.

Although measurement of MPS is not the same as net protein accretion in muscle, with the possibility of early decreases in MPB occurring in all of these studies (except with the spread protocol of Mitchell et al. (2015) when insulin responses were minimal), it is clear that the muscle full phenomenon of an inability to sustain increases in MPS by amino acids after about 3 h is reproducible.

Bag enlargement: collagen turnover in resting and exercised muscle

Increased muscle collagen synthesis had been observed in a weighted‐wing fowl model of muscle hypertrophy (Laurent et al. 1978) with a laborious [¹⁴C]proline infusion method. Laurent subsequently developed a flooding dose method using [³H]proline in a rabbit model (Laurent, 1982). Because Smith and Rennie (Smith et al. 1998) had shown that a flooding dose of proline did not stimulate protein synthesis, they were able to use flooding doses of either ¹⁵N‐ or ¹³C‐labelled proline to investigate collagen synthesis over 2 h with incorporation into hydroxyproline isolated from collagen compared with the area under the blood proline decay curve following the flooding dose. This was in addition to primed dose trace infusions of [1‐¹³C]leucine.

Studies performed by Rennie's group in collaboration with Michael Kjaer in Copenhagen showed that muscle collagen synthesis rates were very low as would be expected, and were unresponsive to feeding (Babraj et al. 2005). Subsequently, collagen synthesis in both muscle and tendon was shown to be stimulated by strenuous exercise (one‐legged kicking) with the same time course, and this was interpreted as a coordinated musculotendinous adaptation (Miller et al. 2005).

Work involving a different exercise model known to induce muscle hypertrophy (maximal shortening and lengthening contractions) in Stewart Phillips’ lab in Hamilton, Ontario, showed a similar 5‐fold increase in muscle collagen synthesis (Moore et al. 2005).

The inadequate anabolic response to feeding with ageing: “anabolic resistance”

The sarcopenia of ageing is one of the most important practical problems taxing muscle physiologists in terms of what causes it and what can be done about it. Studies of whole‐body protein turnover or postprandial protein utilisation had not revealed obvious age‐related changes (Millward et al. 1997), but others and animal studies had suggested that there was a reduced postprandial stimulation of MPS through reduced sensitivity toward amino acids in the ageing muscle (see Dorrens & Rennie, 2003). This was indeed shown to be the case in an extraordinary (and highly cited) clinical study in 44 healthy young and old men exposed to various increasing doses of essential amino acids during an insulin and glucose clamp with MPS, amino acid sensing/signalling proteins and inflammatory markers measured over 3 h (Cuthbertson et al. 2005). While basal rates of MPS were indistinguishable in the elderly from the young, the elderly showed a lower increase in MPS, decreased anabolic signalling and increased inflammation‐associated pathways. Subsequently anabolic resistance with ageing of the inhibitory influence of insulin on MPB was also identified (Wilkes et al. 2009). With A‐V leg studies of MPS and MPB during a euglycaemic, iso‐aminoacidaemic clamp, a small increase in insulin levels (from 5 to 15 μIU ml⁻¹) improved amino acid balance across the leg with a 50% reduction in LPB in the young, but with no significant change in the elderly. Changes in insulin‐related signalling mirrored the decreases in LPB.

Mechanisms of bag filling

The above findings confirm the muscle full phenomena at least in terms of changes in MPS and MPB in response to amino acids and insulin but do not explain it mechanistically.

With respect to human MPB this is very poorly understood and, although some progress has been made in terms of understanding the relative importance of the different proteolytic systems in muscle (see Pasiakos & Carbone, 2014), attempts to link insulin's inhibition of MPB with its post‐receptor signalling via protein kinase B (PKB)/AKT and expression of elements of proteolytic systems have not always identified concordance (e.g. Greenhaff et al. 2008; Wilkes et al. 2009).

As for MPS, the identification of the extracellular amino acid pool as the source of the amino acid‐sensing mechanism (Bohe et al. 2003) is a very important observation implicating amino acid transporters at the start of the signalling cascade, i.e. the sensed amino acid pool reflecting both food supply and exchange with various tissues, before it is subject to intracellular amino acid catabolism. This is important in the context of the work of Haussinger (see Schliess et al. 2006) which raises the possibility that an important component of the anabolic response and its limitation associated with the muscle full signal could be related to cellular hydration and swelling. They showed that cell swelling acts like a pleiotypic anabolic signal with cell shrinkage being catabolic, an effect which Rennie had also observed with muscle (Low et al. 1997a,b). What this means is that at the end of the postabsorptive state, loss of muscle protein would allow some myofibre swelling in response to anabolic signalling by amino acids and insulin initiated by feeding, with the muscle full signal being a loss of responsiveness to anabolic signals once total myofibre volume fills its available space within its “bag”.

The work of Hundal and Taylor in Dundee is very relevant to this. Following on from studies on glutamine transport initiated by Rennie in the 1980s (Rennie et al. 1986), they have reviewed the evidence for the concept of dual‐function amino acid transporter/receptor (‘‘transceptor’’) proteins initiating anabolic signalling linked to maintaining high intracellular glutamine concentrations and promoting cell swelling (Hundal & Taylor, 2009; Nicklin et al. 2009; Taylor, 2014). The transporter system L operates as an obligatory 1:1 heteroexchanger facilitating uptake of leucine in exchange for glutamine, which is accumulated via secondary (Na⁺ linked) active transporters (e.g. SNAT2). The uptake of leucine by such a system requires maintenance of the high glutamine gradient and glutamine reuptake will establish an osmotic gradient for the passive influx of water that promotes cell swelling and activates anabolism (Schlies et al. 2006). In addition to this anabolic influence, SNAT2 appears to be capable of sensing amino acid availability in the extracellular pool and directly signalling to intracellular signalling pathways, including mammalian target of rapamycin (mTOR), possibly through phosphoinositide 3‐kinase‐dependent mechanisms and/or by means of a specific signalling protein (Pinilla et al. 2011). Thus, Hundal & Taylor (2009) argue that leucine's influence on cell signalling critically depends on maintaining both cellular hydration and the glutamine transmembrane concentration gradient (and other SNAT2 amino acid substrates), identifying glutamine as physiologically indispensable from the perspective of preserving the primacy of leucine in terms of mTORC1 signalling (Nicklin et al. 2009). Inhibition of SNAT2 in L6 skeletal muscle cells with methyl aminoisobutyric acid, acidosis or silencing SNAT2 expression depleted intracellular l‐glutamine and l‐leucine, and inhibited mTOR signalling and protein synthesis (Evans et al. 2007), and stimulated proteolysis (Evans et al. 2008). This is consistent with the direct relationship between muscle glutamine concentration and ribosomal capacity and activity and MPS in rat muscle we reported some years ago (Jepson et al. 1988).

As for the muscle full anabolic switch‐off when cell swelling can no longer occur, one possibility involves integrin signalling, which is known to play an important regulatory role in insulin action on muscle (Huang et al. 2006), possibly involving mTORC2 signalling to the actin cytoskeleton (Sarbassov et al. 2004). If increases in degradation were part of the muscle full response, then this could account for the apparent return to baseline of MPS while leucine concentrations and mTORC1 signalling remained elevated, i.e. newly synthesised polypeptides could be degraded, possibly through endoplasmic reticulum‐associated proteolysis (Hampton, 2002) preventing their incorporation into muscle protein.

These ideas are shown in Fig. 2. A more detailed account of the intracellular signalling associated with responses to both amino acids and insulin by the MTOR, GCN2C and AKT/PKB pathways are beyond the scope of this paper but are discussed in Wilkinson et al. (2018) and the accompanying review article by Crossland et al. (2019).

Figure 2. Anabolic influences of amino acids and insulin on human skeletal muscle: “bag filling”.

Muscle responds to food with transient increases in protein synthesis and decreases in proteolysis mediated by amino acids in the extracellular fluid (ECF) and insulin after which it is unresponsive. The concept discussed in this review is that because muscle is encased in a series of inelastic connective tissue sheaths like bags, muscle size is limited by the volume of these bags with one of them, the endomysium, shown. For the healthy adult, bag emptying and refilling occurs during the diurnal cycle of fasting and feeding with muscle mass regulated at a phenotypically fixed “muscle full” size (see Fig. 1), identified by its unresponsiveness to anabolic influences of amino acids and insulin. Potential mechanisms involve an anabolic volume sensor signalling both increases in MPS and decreases in MPB in response to cell swelling associated with an inward flow of water and net amino acid influx through the Na⁺‐coupled system A/SNAT2 (SLC38A2) transporter (II in the figure), with glutamine uptake particularly important. The large glutamine gradient allows inward transport of other neutral amino acids like leucine by the system L (SLC7A5) transporter (III in the figure), via exchange with glutamine. SNAT2 has been identified as a “transceptor,”sensing amino acid supply in the ECF and signalling both increases in MPS and decreases of MPB possibly through an attached peptide (which may generate or transmit the transceptor signal) (SP in the figure). The mTORC1 system activates MPS in response to signals from SP, the volume sensor and upstream signals from amino acids in the intracellular fluid such as leucine. Insulin's role appears to mainly involve an inhibition of MPB via the PKB/AKT system. It is suggested that a loss of responsiveness to anabolic signalling, “muscle full”, is mediated by the volume sensor once total myofibre volume fills its available space within its “bag”. See text for further details and references.

Mechanisms of bag enlargement

Miller et al. (2005) cited a cellular “tensegrity” model to explain how cells can respond coordinately to mechanical stress through integrins embedded in the plasma membrane and proteins connecting the extracellular matrix to the cytoskeleton (Ingber et al. 1994; Maniotis et al. 1997; Mayer, 2003).

Another possibility is that the growth hormone/insulin‐like growth factor‐1 (GH/IGF‐1) axis, which is activated during exercise‐induced muscle hypertrophy, mediates the coordinated responses. However, work in Phillips’ laboratory showed that the increase in MPS following elbow flexor exercise was not enhanced when accompanied by increases in testosterone, growth hormone, IGF‐1 and cortisol induced by high volume intense heavy leg exercise following the arm exercise (West et al. 2009). This was consistent with work showing that GH supplementation did not further enhance the improved anabolism and function of muscle with resistance training (Yarasheski et al. 1992) apart from potentially increasing collagen synthesis in connective tissue (Rennie, 2003). The issue was clarified to some extent in studies in Denmark involving a 14 day trial of the effect of recombinant human GH on collagen synthesis in quadriceps muscle and patella tendon in healthy young men (Doessing et al. 2010). Collagen content and synthesis, measured 2 h after combined flooding doses of [¹⁵N]proline and [1–¹³C]proline, was increased in tendon and muscle but muscle myofibrillar MPS was unchanged. This reinforced the view that GH/IGF‐I may be more important for strengthening the supportive extracellular matrix in tissues than for muscle cell hypertrophy per se, i.e. potentially important in the treatment of traumatic musculoskeletal injuries but not as a means of increasing muscle mass.

Future prospects: heavy water, an old tracer with a new application

While deuterium as a stable‐isotope tracer was used in the very first tracer studies (Schoenheimer & Rittenberg, 1938) it has only recently re‐emerged as a suitable tracer to study muscle protein metabolism (Previs et al. 2004; Busch et al. 2006; Brook et al. 2017a; Wilkinson et al. 2017a). Indeed, Mike Rennie was involved in the initial work to develop pyrolysis methods (currently used by the group) for measuring deuterium by continuous flow IRMS, as part of studies of energy expenditure using the doubly labelled water approach to energy expenditure measurement (Begley & Scrimgeour, 1996).

Following a bolus of ²H₂O, ²H exchange occurs with H atoms in multiple biological substrates including amino acids. Thus with alanine for example, of its four carbon–hydrogen bonds, an average of 3.7 are exchanged with ²H, enabling precursor product approaches to MPS with ²H₂O and protein‐bound alanine. The assumption that the ²H₂O and free ²H alanine maintain their relationship with ²H‐alanyl‐tRNA seems to be the case (Belloto et al. 2007).

Orally administered ²H₂O, which is relatively inexpensive, has obvious advantages enabling volunteers to be studied in free‐living conditions, over extended time periods, providing a cumulative measure of MPS, e.g. myofibrillar protein synthesis measured over 12 days (MacDonald et al. 2013). At Nottingham, ²H₂O has been deployed to study day‐to‐day changes in muscle protein synthesis in humans and the response to resistance exercise training (RET) in a unilateral exercise model which increased both myofibrillar and collagen synthesis (Wilkinson et al. 2014). With higher doses of ²H₂O, MPS can be measured over short, 3 h periods. However, because of a potential problem with nausea and dizziness with doses in excess of 150 ml, higher doses need to be given by multiple small doses over short periods. Thus 400 ml, given over 6 h, 24 h prior to the study, allowed direct comparison between MPS measured by ²H₂O labelling, and a primed infusion of l‐[ring ¹³C₆]‐phenylalanine, over 3 h in the postabsorptive and postprandial state after 20 g EAAs (Wilkinson et al. 2015). This showed good agreement in terms of mean values. However, there was some variation on an individual basis to the MPS rates and responses, possibly due to the better precision of the [¹³C]Phe analysis by GC‐combustion‐IRMS than that of [²H]alanine by GC‐pyrolysis‐IRMS.

The ²H₂O approach, involving an initial dose with subsequent smaller weekly top‐up doses, has been employed to investigate muscle hypertrophy over 3‐week intervals during 6 weeks of unilateral RET in young men. The hypertrophy at 3 weeks was associated with increased MPS but this had returned to control leg rates at 6 weeks (Brook et al. 2015). In elderly men this exercise programme had minimal effects on muscle strength or size with no changes in MPS at 6 weeks (Brook et al. 2016).

The potential for measurement of turnover with ²H₂O in a wide range of macromolecules including ribosomal (r)RNA is an exciting feature of ²H₂O studies. At a very early stage of our animal studies of MPS, changes in rRNA were shown to be a major determinant of changes in MPS (Millward et al. 1973). rRNA turnover was measured, on the basis of ²H incorporation into the ribose moiety of muscle rRNA, following 6‐week RET in young healthy males, revealing a clear relationship between rRNA synthesis and MPS (Brook et al. 2017b).

Thus ²H₂O tracer studies probably represent the most important development in stable‐isotope studies in recent decades, although its potential has only just begun to be explored. For example proteome dynamics is an exciting new approach in which the recent advances in LC‐MS‐MS instrumentation (Lehmann, 2017) are coupled with ²H₂O studies of fractional synthesis rates (FSRs). For example the FSRs of a wide range of muscle proteins in sedentary and exercise‐trained subjects has been monitored showing variable increased FSRs with training (Shankaran et al. 2016). Furthermore by assessing the turnover of muscle‐derived enzymes in blood such as creatine kinase M‐type and carbonic anhydrase 3, which correlate closely with FSRs of muscle proteins of various ontologies, the sampling of muscle proteins by biopsy can be avoided. Clearly further validation of this approach is required, since the expected relationships are not always observed (Murphy et al. 2018) and there are limitations. One is that it would not allow the study of interventions in an individual limb or muscle. Nevertheless it remains an innovative and potentially very powerful approach with widespread applications to measure muscle protein turnover in a minimally invasive manner.

Conclusions

Although Mike Rennie's work was wide ranging, protein metabolism in skeletal muscle in health and disease was at its core. Of this, we have focused here on its nutritional regulation, which he investigated using stable isotopes, demonstrating that the concepts we discussed at the very beginning of the work, based on what we knew about muscle structure and the phenomenology of the stability of muscle mass, can indeed be demonstrated to be true. Clearly the bag hypothesis is a very crude analogy but as we have tried to show, in the context of what is now known about cellular hydration and swelling as an important pleiotropic signal, much of the regulation of muscle homeostasis and hypertrophy can be explained. It is also gratifying to see that much of Mike's efforts in trying to characterise amino acid transporter mechanisms in muscle in the 1980s with Hundal and Taylor has come to fruition with the novel transceptor concepts we highlighted herein.

Although great strides have been made, with the research teams Mike established at Dundee and Nottingham at the international forefront of the work of understanding the regulation of muscle amino acid and protein metabolism, many challenges still lie ahead. One is that of relating studies of muscle MPS responses to feeding to actual maintenance of muscle throughout the diurnal cycle of fasting and feeding. The A‐V approach, which indicates at least acute net balance, MPS and MPB responses, has an advantage in this context and there are many examples of biopsy studies of MPS where A‐V studies would have provided a much more complete account.

The physiological and nutritional significances of anabolic resistance are not easy to reconcile. As recently pointed out by Mitchell et al. (2016), there is a marked mismatch between the magnitude of the blunted anabolic responses and the very slow rates of muscle loss with sarcopenia in healthy subjects, imperceptibly slow on a daily basis. This suggests that some other compensating mechanisms are involved. We showed some years ago that postabsorptive losses were lower in the elderly compared with younger adults at similar protein intake levels (Millward et al. 1997). While it is clear that RET is a beneficial influence on the course of sarcopenia, whether nutritional influences, especially dietary protein supply, will unequivocally be shown to have benefit remains to be seen. Given that sarcopenia occurs in healthy, well‐nourished, aged master athletes (see Millward, 2012), it seems unlikely to have a nutritional aetiology and therefore equally unlikely to have a simple nutritional remedy. Although claims are made that sarcopenia is less with higher protein intakes, usually on the basis of a single highly cited study (Houston et al. 2008), it has been pointed out that the authors’ conclusions of dietary protein being a modifiable risk factor for sarcopenia in older adults was not robustly supported by the data they presented. The greater postabsorptive losses with increased habitual protein intakes (see Fig. 1) negate any benefit of any increased postprandial gain, and raise the question of whether investigation of the extent and time course of the development of “bag emptying” in the postabsorptive state would be fruitful, given its equal importance to the overall balance mechanism.

Additional information

Competing interests

None declared.

Author contributions

D.J.M. and K.S. have contributed equally to writing this paper with no others qualifying for authorship. Both authors have approved the final version of the manuscript and agree to be accountable for all aspects of the work.

Funding

No funding was received by D.J.M. or K.S. for writing this paper.

Biographies

D. Joe Millward is Emeritus Professor of Human Nutrition at the University of Surrey where he held the chair of Human Nutrition from 1992 to 2005 after working in the nutrition department at the London School of Hygiene and Tropical Medicine from 1970 to 1992. His research has focused on the regulation of muscle and bone growth, on macronutrient requirements, especially for protein and amino acids and protein quality, and for energy, and on the aetiology and management of obesity and coronary heart disease risk.

Ken Smith is Professor of Metabolic Mass Spectrometry at the University of Nottingham, and manages the Stable Isotope Mass Spectrometry Facility at the Derby Campus. His research interests focus on the application of novel stable isotope approaches to investigate metabolism in human skeletal muscle in health, ageing and disease: investigating the impact of nutrition (protein/amino acids), exercise and hormones on the regulation of muscle mass.