New insights into creatine function and synthesis

The role of creatine and creatine phosphate in energy metabolism has long been recognized. The creatine kinase (EC 2.7.3.2) reaction, (ATP+creatine↔creatine phosphate+ADP) provides the basis for this role. However, in recent years, our view of the function served by this reaction has changed from one in which the primary role is that of buffering ATP levels or the ATP/ADP ratio to one of rapid, intracellular translocation of high-energy phosphate compounds from their mitochondrial sites of production to their cytosolic sites of utilization. This latter view is supported by the subcellular location of different creatine kinase isoenzymes, in the mitochondria as well as in myofibrils, sarcoplasmic reticulum and plasma membranes (Schlattner et al., 2006).

This view of the creatine/creatine phosphate system is illustrated in Fig. 1. It is often suggested that the major advantage provided by this energy translocation model over the more conventional system, where ATP and ADP are thought to diffuse between sites of production and utilization, is the higher diffusivity of creatine phosphate compared to ATP, as the molecular weight of creatine phosphate is less than half that of ATP. In fact, the diffusion coefficient for creatine phosphate is only about 30% higher than that for ATP (Ellington, 2001). It is much more likely that the key difference is in the diffusivity of creatine compared to ADP and is provided less by the difference in their molecular weights (ADP=427; creatine=131) than by very great differences in their cytosolic concentrations (Wallimann et al., 1992). Free cytosolic concentrations of ATP, creatine phosphate and creatine in skeletal muscle are, respectively, about 8, 27 and 13 mM (Veech et al., 1979). The free cytosolic ADP concentration is about 20 μM (Nioka et al., 1992). It should, of course, be emphasized that the two postulated roles of the creatine/creatine phosphate system are not mutually exclusive and it is likely that both temporal energy buffering and spatial energy translocation occur, to different degrees, in different cell types. It appears that temporal buffering is more important in fast-twitch muscles whereas the energy translocation function predominates in slow-twitch muscles (Wallimann et al., 1992).

The creatine/creatine phosphate system is not equally important in every cell. Indeed, some cells with high rates of oxidative metabolism (e.g. hepatocytes) do not employ this system. It is apparent that it is primarily present in cells with high and intermittent energy demands, such as skeletal muscle (where energy output can vary by a factor of 100), heart (where there are substantial differences in energy demand from systole to diastole) and the brain (where certain tasks require some neurons to fire very quickly) (Wallimann et al., 1992).

In addition to these physiological roles of creatine, this substance is also consumed in quite substantial amounts both by athletes, as an ergogenic supplement (Kraemer and Volek, 1999) and by patients suffering from a variety of neurological disorders (Baker and Tarnopolsky, 2003). Annual consumption of creatine is estimated at about four million kilograms per annum with a value, in the United States alone, of more than $200 million.

The whole body pool of creatine plus creatine phosphate in 70 kg young adults is of the order of 120 g. Both creatine and phosphocreatine are non-enzymatically and irreversibly degraded to creatinine at a rate of about 1.7% of the total body pool per day (Wyss and Kaddurah-Daouk, 2000). This creatinine is lost in the urine. The amount of this loss is proportional to muscle mass and is, therefore, greater in males than in females and in young adults than in the elderly. Daily creatinine excretion (and creatine loss) varies from about 1.7 g to about 0.9 g in 70 kg males of ages 18–29 and 70–79, respectively (Stead et al., 2006). In vegetarians, all of this creatine needs to be replaced via synthesis. However, for individuals on a typical North American omnivorous diet, we have calculated that approximately half of the replacement can be provided via food creatine and the other half requires endogenous synthesis. Thus, we have calculated that men, aged 20–39, need to synthesize about 1.0 g creatine/day; women of the same age group synthesize about 0.75 g/day (Stead et al., 2006).

Creatine synthesis occurs via an exceedingly simple metabolic pathway, consisting of only two enzymes (Fig. 2). The first enzyme, l-arginine:glycine amidinotransferase (AGAT) (EC 2.1.4.1) catalyzes the formation of guanidinoacetate and ornithine from arginine and glycine. The second enzyme, glycine N-methyltransferase (EC 2.1.1.2), employs S-adenosylmethionine to methylate guanidinoacetate, producing creatine and S-adenosylhomocysteine. Although most of the body's creatine is found in skeletal muscle, there is very little evidence for appreciable creatine synthesis in this tissue. Rather, creatine, whether of dietary origin or synthesized endogenously, is taken up from the plasma via a specific transporter (CreaT). CreaT is found in a variety of tissues, including skeletal muscle, kidney, heart, forebrain, cerebellum and liver (Snow and Murphy, 2001). It transports creatine, together with sodium and chloride ions, against a considerable concentration gradient. Additional complexity is provided by the occurrence of at least two different isoforms of CreaT, most likely as a result of alternative mRNA splicing (Guerrero-Ontiveras and Wallimann, 1998).

Creatine synthesis represents a considerable metabolic load, in particular with regard to amino acid utilization. A comparison of rates of creatine synthesis with the dietary intake of glycine, arginine and methionine is revealing. The synthesis of 1.0 g creatine/day amounts to 7.6 mmol. This compares to a dietary intake of about 78 mmol glycine, 37 mmol of arginine and 19 mmol of methionine (these figures are based on an intake of 80 g protein/day with the amino acid composition of beef protein). Therefore, creatine synthesis consumes about 10% of dietary glycine as the entire glycine molecule is incorporated into creatine. It is not certain whether similar arguments can be made for methionine and arginine since only the methyl group of methionine and the amidino group of arginine are incorporated into creatine. In addition to methionine, the diet provides labile methyl groups in the form of choline and betaine and new methyl groups (methylneogenesis) can be produced via the remethylation of homocysteine (Stead et al., 2006). Comparison of rates of creatine synthesis with isotopically measured rates of transmethylation reveal that the synthesis of this molecule makes very great demands on S-adenosylmethionine. Approximately 40% of S-adenosylmethionine's methyl groups are consumed by creatine synthesis (Stead et al., 2006).

The ornithine produced by AGAT could, in theory, be reconverted to arginine by the reactions of the urea cycle (Brosnan and Brosnan, 2004). However, there is, at present, no compelling evidence that this actually occurs. Therefore, it is possible that creatine synthesis could consume as much as 20% of dietary arginine. Regardless of the uncertainties inherent in these calculations, it is apparent that creatine synthesis places a considerable demand on amino acid metabolism. This demand is even greater in vegetarians where creatine synthesis is double that in omnivores and where protein intake is, typically, less.

Inborn errors involving mutations in each of the three proteins required for creatine synthesis and transport are now known. Perhaps the most remarkable feature of these disorders is that they present, in children, with a very similar constellation of neurological symptoms, i.e. mental retardation, epilepsy and difficulties in language development. The symptoms are invariably more severe in the GAMT-deficient children and this has been attributed to the accumulation of guandinoacetate, which is thought to exert specific neurotoxic actions. Because of its methyl group, creatine is readily detected non-invasively by ¹H-MRS. Children who present with these inborn errors display a massive depletion of brain creatine. In addition, those with AGAT deficiency have decreased plasma creatine and guanidinoacetate while those with GAMT deficiency have markedly increased plasma guanidinoacetate and lowered creatine. In the case of patients with creatine transporter deficiency, plasma guanidinoacetate is in the normal range while creatine is elevated (Stromberger et al., 2003).

Treatment of children with these enzyme defects with creatine largely or completely restores brain creatine over time and results in a marked clinical improvement. The general experience with creatine supplementation, however, has been that the children never fully recover; learning and language functions remain impaired. However, until recently, it has never been clear whether the neurological deficits could have been avoided if creatine supplementation had begun at birth. One recent case-study gives encouraging information on this point. The affected child was homozygous for the W149X mutation in AGAT that also affected two older siblings. A very early diagnosis could, therefore, be made. Treatment with creatine monohydrate was begun at 4 months, upon weaning. This child's growth and development was entirely normal at the age of 18 months (Battini et al., 2006). In contrast to the improvement found upon creatine supplementation of children with the enzyme deficiencies, the transporter-deficient patients are refractory to treatment with creatine. There is no accumulation of creatine in the brain and no clinical improvement (Stromberger et al., 2003).

The creatine-deficiency syndromes clearly emphasize the importance of creatine synthesis. We have, therefore, conducted a series of studies on creatine synthesis, employing the rat as a model. Previous work has indicated that the two enzymes of creatine synthesis occur in a variety of tissues. In particular, AGAT is most highly active in the rat kidney but this enzyme is also found in pancreas, spleen, brain, testis and thymus (Wyss and Kaddurah-Daouk, 2000). A particularly vexing question concerns the possible occurrence of AGAT in the liver. McGuire et al. (1986) using immunofluorescence microscopy reported AGAT protein in hepatocytes. However, we have been unable to detect enzyme activity in the liver. The highest activity of GAMT is found in the rat liver but activity is also reported in pancreas, spleen and testis (Wyss and Kaddurah-Daouk, 2000). Based on the fact that the highest abundance of AGAT is in the kidney whereas that of GAMT is in the liver, we have postulated the occurrence of an inter-organ pathway for creatine synthesis (Fig. 3) by which guanidinoacetate is released by the kidney, taken up by the liver and methylated there to produce creatine.

We have tested this inter-organ hypothesis in a variety of ways. Direct measurements of arterio-venous (A-V) differences across rat kidneys reveal guanidinoacetate production (arterial concentration 5 μM, renal venous concentration 9 μM, p<0.01). When these studies are carried out on rats fed purified diets (i.e. no dietary creatine), we found that guanidinoacetate production by the kidney was not significantly different from creatinine loss across the same kidney. This implies that the kidney is the organ responsible for the great bulk of guanidinoacetate production in the rat. We have also examined the production of creatine, from guanidinoacetate in isolated hepatocytes. These cells produced creatine from guanidinoacetate (half maximal rates at [guanidinoacetate]=14 μM). The rate of creatine production was doubled by the provision of a source of methyl groups (e.g. methionine or betaine in the presence of homocysteine). Maximal rates of creatine synthesis by rat hepatocytes are about 25 nmol/h/mg dry weight.

We have also examined the regulation of creatine synthesis. It is known that renal AGAT activity is down-regulated by dietary creatine (McGuire et al., 1984). We have confirmed this in rats that were fed a diet containing 0.4% creatine for 2 weeks—renal AGAT activity was decreased by 85%; renal AGAT mRNA abundance also fell, though not as much. In turn, this decreased renal AGAT activity translated into a reduced A-V difference for guanidinoacetate across the kidney (1 vs. 4 μM) and a markedly reduced circulating guanidinoacetate concentration (1.5 vs. 5 μM). We suggest that the primary regulation of creatine synthesis occurs at the level of renal AGAT expression and its modulation by dietary creatine. We suggest that little active regulation occurs in the liver. Half-maximal creatine production by hepatocytes is found at about 14 μM guanidinoacetate, which is appreciably higher than circulating levels (about 5 μM). Therefore, we suggest that hepatic creatine synthesis can respond to increased or decreased renal guanidinoacetate production in a simple concentration-dependent manner. This is not to suggest, however, that our knowledge of the hepatic aspect of creatine synthesis is complete. In particular, we need to understand the transport mechanisms whereby guanidinoacetate enters hepatocytes and by which creatine exits. There is the well-described creatine transporter in tissues such as muscle, which takes up creatine against a considerable gradient. However, creatine transport in hepatocytes occurs in the opposite direction and may require a unique transporter.

Creatine synthesis accounts for approximately 40% of all S-adenosylmethionine-derived methyl groups (Stead et al., 2006). This, in turn, implies that it is responsible for the production of about 40% of the body's homocysteine. Elevated plasma homocysteine has been implicated as a risk-factor for a number of chronic illnesses, including cardiovascular disease (Stampfer et al., 1992), Alzheimer's disease (Seshadri et al., 2002) and fractures (van Meurs et al., 2004). There is, therefore, considerable interest in ways of reducing plasma homocysteine levels. Given that we have shown that dietary creatine down-regulates endogenous creatine synthesis, it was important to determine whether dietary creatine could decrease plasma homocysteine levels. We, therefore, measured plasma homocysteine in rats that had been fed a diet containing 0.4% creatine for 2 weeks (Stead et al., 2001). Fig. 4 shows that plasma homocysteine decreased by approximately 25%. There was no effect on methionine levels.

Does dietary creatine decrease plasma homocysteine in humans? Certainly, there is evidence that it down-regulates guanidinoacetate production as the plasma concentration of this compound decreases in subjects who ingested creatine (Derave et al., 2004). Two groups have examined plasma homocysteine with discordant results. Steenge et al. (2001) found no decrease in plasma homocysteine in subjects who ingested 3 g creatine monohydrate/day for 61 days. On the other hand, Korzun (2004) found a 10% decrease in subjects who, each day, ingested an amount of creatine equal to twice their daily creatinine excretion for 28 weeks. We consider it likely that the decrease in homocysteine that we observed in rats also occurs in humans although its magnitude may be somewhat less.

As the great bulk of the body's total creatine is found in muscle, attention has primarily centered on its function in this tissue. However, the fact that children with inborn errors in creatine synthesis or transport present with profound neurological impairments obliges us to focus both on how the brain acquires its creatine as well as its function in this organ. The occurrence of the neurological symptoms implies that dietary creatine is insufficient in these children. However, the almost complete absence of brain creatine is somewhat puzzling since infant formulas that are based on cow's milk contain substantial quantities of creatine. However, our own analyses show that soy-based formulas are devoid of creatine. It is possible, therefore, that these children's diet may play some role in the severity of their symptoms. The fact that creatine supplements are effective in restoring brain creatine in these children indicates that the brain can take up blood-borne creatine. Nevertheless, both AGAT and GAMT have been reported to be ubiquitously located, in neuronal and glial cells, throughout the rat brain (Braissant et al., 2001). The possible role of brain creatine synthesis, vis-à-vis creatine uptake from the circulation, needs to be clarified.

What role(s) might creatine play in the brain? At the moment, we have no reason not to suppose that the traditional roles of creatine phosphate and creatine kinase are paramount. The brain has a very high rate of oxygen consumption; the adult human brain weighs less than 1% of body weight but accounts for about 10% of the total oxygen consumption (Burton, 1966). In addition, individual rapidly firing neurons have very high energy requirements.

An intriguing study from Australia examined the effect of creatine supplementation on performance in a number of cognitive tests (Rae et al., 2003). The subjects, students from the University of Sydney, were either vegetarians or vegans because it was reasoned that their creatine status might be impaired due to little or no dietary creatine ingestion. A double-blind, placebo-controlled, cross-over trial was carried out in which subjects ingested a daily dose of either 5 g of creatine monohydrate or placebo for 6 weeks, Then, after a 6-week wash-out period, the doses were reversed. A number of cognitive tests that require speed of processing were administered at the beginning and after each 6-week period. Creatine ingestion resulted in remarkably improved performance in these tests. This remarkable result emphasizes the importance of creatine in non-muscle tissues and suggests that brain creatine acquisition and function is an important field for future research.