One Missing Letter: The Amino Acid Triage Beneath the Protein Number

The protein recommendation counts grams. The body counts letters. When one letter is short, the word that specific tissue was building stays unbuilt, and the recommendation shows only the number that hid the shortage.

The letter that is not there

Consider a game of Scrabble. Your rack is full. Seven tiles, none missing. You reach for the letter you need to complete the word, and it is not there. You have all the others, in duplicate, in triplicate. It changes nothing. The word you were trying to spell stays unspelled. Every other letter you hold has become inert, because the word required a specific set, and one member of that set was not present.

This is the way protein synthesis works inside the body, and it is the way the recommended dietary allowance for protein does not describe protein synthesis. The RDA counts grams. The tissue counts letters. When one letter is short, the word that specific tissue was building stays unbuilt. The other letters, present in abundance, wait in the free pool until they are used for something else, or are catabolized. The gram total the label reports is unaffected. The synthesis the tissue needed is not.

The protein number is not the protein

The prevailing nutritional convention treats protein as a single entity indexed by mass. The RDA sits at 0.8 grams per kilogram of body weight per day, established in the 1940s by William Cumming Rose using the nitrogen balance method [1]. Recent scholarship converges on a higher figure. Indicator amino acid oxidation studies from the Elango, Pencharz, Ball, and Wolfe groups put the true adult minimum near 1.0 to 1.2 g/kg [2]. Athletic and clinical recommendations reach 1.6 to 2.2 g/kg [3][4]. The debate over the correct number, entirely legitimate at its own level, obscures a more fundamental problem. Whichever number one chooses, the number is silent about which letters it contains.

Every protein a tissue synthesizes is a specific sequence of amino acids joined in a specific order. Twenty amino acid types provide the alphabet. Roughly twenty thousand protein-coding genes provide the vocabulary. Nine of the twenty letters are indispensable, meaning the body cannot make them from carbon skeletons and nitrogen alone, and they must be supplied by the diet. Several others are conditionally indispensable, meaning the body can make them in principle but not always in the quantities its own metabolism demands, a distinction to which this piece will return. The stoichiometry that governs whether a specific tissue can build a specific protein is not a mass balance. It is a letter-by-letter accounting.

The mainstream nutritional science that produced the RDA framework knows this in theory. The concept of the limiting amino acid has been in the textbooks for decades. Protein quality scoring, in the form of the PDCAAS and DIAAS scales, was invented to weight dietary proteins by their capacity to supply the letters in the ratios human physiology requires. What has been slow to enter the frame is the observation that even a diet with a perfectly balanced amino acid profile can produce tissue-specific insufficiency, because certain tissues consume certain letters at rates that exceed what a “complete” protein at the recommended total will deliver. The letter is not missing from the diet. It is missing from what remains after the sinks have taken their share.

Eight sinks that do not appear on the label

The examples are not marginal. They are the arithmetic of ordinary metabolism. Each of the following is a case in which a single tissue or pathway consumes a substantial fraction of the daily supply of one amino acid, before the free pool becomes available for other syntheses. The list is conservative. Cases are included only where the flux has been measured or credibly estimated to consume at least ten percent of dietary intake, or at least thirty percent of endogenous synthesis capacity when the amino acid is conditionally essential.

Glycine, taken by collagen. The pioneering audit by Meléndez-Hevia and collaborators, published in 2009 and extended by de Paz-Lugo and collaborators in 2018 and 2023, established that endogenous glycine synthesis provides roughly three grams per day, while the metabolic demand, dominated by collagen synthesis at approximately ten grams per day and including glutathione (~1.5 g), creatine (~1.7 g), heme, bile acid conjugation, and purine synthesis, approaches fifteen grams per day when total demand is summed [5][6][7]. Collagen alone accounts for roughly sixty-five percent of this demand. The deficit persists even on a diet meeting the protein RDA. The paper’s authors state the conclusion in a form no textbook has yet absorbed: “glycine synthesis is indeed insufficient… it should be taken as a nutritional supplement to guarantee a healthy metabolism” [5].

Glutamine, taken by the gut. The intestinal mucosa oxidizes glutamine as its principal metabolic fuel. The work of Windmueller and Spaeth, initiated in the 1970s, established that enterocytes extract approximately twenty to thirty percent of splanchnic glutamine delivery, roughly ten to fourteen grams per day at rest [8][9]. Activated lymphocytes and macrophages consume glutamine at rates four to ten times higher than resting immune cells [10]. Under catabolic stress (burns, sepsis, major surgery), plasma glutamine falls thirty to fifty percent, and negative nitrogen balance follows the drop [11]. Glutamine is nominally non-essential, because the body synthesizes it from glutamate and ammonia. In the presence of these sinks, the nominally non-essential becomes conditionally indispensable, a term the intensive care literature has adopted without altering the underlying framework.

Threonine, taken by mucin. The neonatal gut extracts the majority of dietary threonine on first pass, using it to synthesize the MUC2 mucin that protects the intestinal epithelium. When threonine becomes limiting, extraintestinal protein synthesis is sacrificed to preserve mucin. The work of the Bertolo, Reeds, Ball, and Burrin group, using isotope tracer methods in piglets, quantified the phenomenon with rigor [12][13]. The estimated adult flux is 1.5 to 2 grams per day, roughly half of dietary threonine intake. The parallel to the glycine-collagen case is exact. A single high-turnover tissue-specific structural protein claims the bulk of the amino acid before any other synthesis can proceed.

Cysteine, taken by glutathione. Whole-body glutathione synthesis consumes on the order of 1.5 grams per day of cysteine, a mass comparable to typical dietary cysteine intake, before hair, nail, taurine synthesis, coenzyme A, and other cysteine-consuming pathways receive any share [14]. Under inflammation, transsulfuration flux from methionine dominates cysteine supply, and cysteine competes with keratin synthesis for methionine’s carbon skeleton [15]. See Right to Adequate Glutathione for the full physiology of the substrate whose demand this sink expresses.

Tryptophan, taken by the kynurenine pathway. Ninety to ninety-five percent of ingested tryptophan is catabolized through the kynurenine pathway toward niacin and quinolinate, leaving one to two percent for serotonin synthesis and small quantities for other neurotransmitters [16]. Immune activation, through IDO induction by IFN-γ, diverts even more tryptophan away from serotonin, producing the depressive symptomatology that chronic inflammation carries with it [17]. The RDA for tryptophan (~280 mg/day for a 70 kg adult) is calibrated against nitrogen balance in an uninflamed subject. The inflamed subject faces a different arithmetic entirely.

Lysine, taken by carnitine synthesis. Carnitine, indispensable for the transport of long-chain fatty acids into mitochondria, is synthesized from trimethyl-lysine released by protein turnover. The rate-limiting step is not lysine intake per se but the turnover-dependent release of TML from methylated proteins. Skeletal muscle, which stores the largest reservoir of TML precursor, is also the tissue with the highest carnitine demand, a self-limiting loop in which the tissue that needs the substrate is also its reservoir [18].

Methionine, taken by methylation and creatine. The SAM cycle is one of the most active biochemical loops in the body. Roughly forty to fifty percent of all methyl groups transferred via SAM are consumed by a single reaction: the methylation of guanidinoacetate to creatine, catalyzed by GAMT [19]. The daily SAM turnover approaches 1.5 grams, and the methionine required to sustain this cycle exceeds typical dietary intake, sustained only by the recycling of homocysteine through remethylation by folate and B12. Creatine supplementation demonstrably lowers the methylation load on the body, because the dominant methyl consumer is spared [20]. See Right to Adequate Folate for the co-substrate that sustains this cycle.

Arginine, taken by creatine, nitric oxide, and polyamines. The three concurrent sinks consume, together, approximately 2.5 to 3 grams per day out of a typical dietary intake of four grams, sixty to seventy-five percent of intake, before the free pool becomes available for protein synthesis, urea cycle regeneration (which is nominally arginine-neutral but not always in practice), or any other use. The largest single sink is creatine synthesis at approximately 1.7 grams per day of arginine, delivered as the amidino donor via AGAT [19]. Creatine supplementation spares arginine and simultaneously lightens the methylation load on the body, a triple economy that no protein RDA describes.

These eight cases are not the entire list. Contextual sinks, real but activated only under specific physiological conditions, include glutamine under catabolic stress, aspartate in rapidly proliferating tissues [21], histidine in muscle carnosine stores among vegetarians and older adults [22], and serine in the sphingolipid synthesis whose failure produces the peripheral neuropathy of HSAN1 and the deoxysphingolipid accumulation of diabetes 2 [23]. Each of these is real. Each becomes clinically dominant when its trigger appears. None of them appears in the protein number.

The catalog also excludes purported sinks that fail the quantitative test. The most seductive is the case of tyrosine, drawn into the synthesis of thyroid hormones on the small gland’s thyroglobulin scaffold. The image is intuitive. The thyroid does trap tyrosine and iodine to make T4 and T3. But the flux is trivial. Daily T4 production of approximately one hundred micrograms represents forty-seven micrograms of tyrosine incorporation, or 0.0016 percent of a typical three-gram daily tyrosine intake. Three orders of magnitude below any threshold that would qualify the thyroid as a sink. Similar reckoning applies to catecholamines (two to four milligrams per day) and to melanin incorporated into hair and skin (well under one hundred milligrams per day). The mechanism is real. The quantity is not. A discipline that names sinks correctly cannot afford to add cases that fail on arithmetic, however evocative they are on physiology.

What nitrogen balance did not see

The methodological reason the RDA does not register these sinks is not obscure. The nitrogen balance method estimates protein requirement by measuring nitrogen intake and nitrogen loss, and identifying the intake at which loss no longer exceeds it. It is agnostic about which amino acids the nitrogen came from and about which specific proteins were, or were not, synthesized. A subject in nitrogen balance may still be tissue-deficient in a specific amino acid, provided that some other syntheses have compensated in the aggregate mass reported. The method was designed to prevent frank protein-calorie malnutrition. It was not designed to detect the sinks this piece has named, and its floor value has been carried into modern recommendations without the correction that a more granular view would require.

The indicator amino acid oxidation method, developed more recently, does look at individual amino acid balance, and consistently reports requirements thirty to fifty percent higher than the nitrogen balance floor [2]. For lysine, IAAO estimates are roughly three times the WHO recommendation. That divergence is itself a signal. An independent method, applied to individual letters rather than to the total mass, finds systematically that the aggregate mass estimate was low. The revision has been slow to enter policy. In the interim, the recommendation whose calibration was performed by a method that could not see the sinks continues to be quoted as if it were adequate for a subject whose tissues are subject to them.

The problem is not that the RDA is wrong in some categorical sense. It is that the RDA answers a different question. It answers: at what intake will a healthy subject remain in aggregate nitrogen balance? It does not answer: at what intake will each of a specific subject’s tissues, with their specific composition and their specific turnover rates, receive the letters they need to build the specific proteins their function requires? A framework of physiological rights that takes tissue-level function as its criterion cannot rest on the first answer while the second remains unaddressed. This is the argument Two Doses, One Molecule makes for micronutrients. The same asymmetry applies to macronutrients when their internal composition is examined at the resolution the physiology actually uses.

Ames extended: an amino acid triage

Bruce Ames formulated the triage theory in 2006 to describe how the body allocates scarce micronutrients: when a vitamin or mineral falls into insufficient supply, evolution preserves the enzymes with lower Michaelis constants (those indispensable for short-term survival and reproduction) and starves the higher-Km enzymes whose functions serve long-term maintenance [24]. The insidious damage that follows accumulates over decades, unseen by the acute clinical eye. McCann and Ames applied the framework to vitamin K in 2009 [25], to selenium in 2011 [26], and Ames extended the analysis to a proposed class of “longevity vitamins and proteins” in 2018 [27].

The amino acid sinks catalogued here fit the same logic exactly. When an amino acid is limiting, the body preferentially maintains the syntheses whose failure would threaten immediate survival (contractile protein for movement, immune protein for defense, gluconeogenic substrate for glucose homeostasis) and defers the syntheses whose failure produces slow, cumulative damage. Collagen turnover slows before myosin turnover. Glutathione synthesis in the erythrocyte is preserved before glutathione replenishment in the aging skin. Carnitine biosynthesis, whose failure impairs β-oxidation but not immediate ATP production from glycolysis, is deprioritized before creatine synthesis, whose failure would embarrass the muscle in acute exertion.

The image of what an aging tissue looks like under sustained amino acid triage is not hypothetical. It is the phenotype of ordinary aging. Skin thins as dermal collagen turnover fails to keep pace with degradation. Vasculature stiffens as elastin cross-links are not replenished. Joint cartilage erodes as chondrocyte glycine supply cannot sustain the proteoglycan and collagen matrix. Sarcopenia advances as amino acid triage sacrifices maintenance for movement, a pattern this resource has already traced under a different angle in Sarcopenic Obesity: The Strength They Measure, the Mass They Miss. Each of these is a mode of maintenance failure whose kinetics match the low-Km/high-Km partition the Ames framework describes, applied not to vitamins but to the very substrates from which the tissues themselves are built.

The extension has not been drawn in the literature. Ames’s own writings do not make the amino acid application. Meléndez-Hevia describes the glycine-collagen deficit as a stoichiometric shortage, not as a triage. The framing this piece proposes, that the sink hypothesis and the triage hypothesis are the same hypothesis applied to a different class of substrates, has consequences. It suggests that supplementation of the sinks (glycine, collagen peptides, glutamine under stress, cysteine as N-acetylcysteine, creatine, arginine) may act as a triage relief, restoring maintenance syntheses that had been deferred, and producing effects across multiple systems that a single-nutrient RCT protocol would fail to see. It also suggests that the diseases of aging are not, or not only, diseases of vitamin triage. They are also diseases of amino acid triage, prosecuted by the same evolutionary logic on a different substrate.

The dipeptide exception: why collagen peptides work

Two features of the collagen peptide literature have long puzzled observers who read them from the standpoint of standard biochemistry. First, the two dipeptides prolyl-hydroxyproline (Pro-Hyp) and hydroxyprolyl-glycine (Hyp-Gly) persist in human plasma for hours after ingestion of collagen hydrolysate, at concentrations of micromoles per liter [28]. Second, they act as trophic signals on dermal fibroblasts, chondrocytes, and other cells, in vitro and apparently in vivo, at concentrations achievable by dietary intake [29][30]. Neither observation is explicable if collagen protein is simply digested to free amino acids and absorbed as such. Something else is happening.

The mechanism is now understood. The hydroxyproline residue confers unusual resistance to intestinal peptidases. The dipeptide is absorbed intact via PEPT1, the H+-coupled peptide transporter of the enterocyte brush border. Once in plasma, the dipeptide is degraded slowly enough to persist for hours and to reach target tissues in bioactive form. The dietary supply of collagen, in this reading, is not merely a source of glycine and proline for the free amino acid pool. It is a delivery vehicle for two specific signaling molecules that reach fibroblasts and chondrocytes intact, at concentrations sufficient to modulate their behavior. The signal is a food.

The apparent uniqueness of this mechanism has been a puzzle in the collagen literature itself. Why should collagen alone deliver bioactive dipeptides, when other tissue-specific structural proteins presumably contain their own signature sequences? Part of the answer is chemical. Hydroxyproline is a post-translationally modified residue that appears in few other proteins, and its resistance to peptidases is a special case. Part of the answer, however, is that the question has not been systematically asked. The bulk of the research has been financed by an industry with commercial interest in collagen supplementation. Other dipeptide candidates from other tissue-specific proteins have not received the same investment.

The one clean documented analog is carnosine (β-alanyl-L-histidine), and its methylated congener anserine (β-alanyl-1-methyl-L-histidine). Both are absorbed intact via PEPT1, both reach muscle and brain, and both have documented biological activity: metal chelation, antioxidant behavior, anti-glycation effects, pH buffering [31][32]. The reason carnosine has not generated a market comparable to collagen is not that the dipeptide fails to reach tissue. It is that human serum carnosinase, encoded by CNDP1, hydrolyzes plasma carnosine with a half-life of fifteen to thirty minutes, a species-specific asymmetry that rodents lack [33]. Pro-Hyp persists for hours in human plasma. Carnosine persists for minutes. The industry followed the exposure that would deliver measurable clinical effect, and the exposure was in collagen.

The framing that unites these dipeptides is worth naming: intact-absorbed food-derived dipeptides delivering amino acids to their end-use tissue by bypassing free-pool competition. Two members of the class are documented. Others almost certainly exist, and their absence from the literature reflects the funding structure of nutritional research more than the biochemistry of digestion. The claim is not that every tissue has its trophic dipeptide waiting to be discovered. It is that the mechanism by which collagen peptides act is not a curiosity. It is one implementation of a principle, dietary delivery of tissue-specific building blocks with signal-scale bioactivity, that the free-amino-acid model of digestion has systematically obscured.

The framework must include the letters

A right that recognizes protein only as an aggregate mass is a right calibrated for a subject whose sinks are quiet, whose losses are unelevated, and whose tissues, on average, will build enough. It is not, in this respect, different from the right calibrated for micronutrient sufficiency at the RDA: a right of prevention against frank deficiency in a subject who was never in deficit.

The subject who lives with an elevated sink does not receive the protein number as an answer. Whether that subject is the aging body whose collagen matrix degrades faster than glycine supply can rebuild it, the recovering athlete whose glutamine pool is drawn down by gut and immune demand, the vegetarian whose muscle carnosine and creatine stores are below optimum, the diabetic whose sphingolipid synthesis has failed for want of serine, the inflamed subject whose cystine is siphoned into glutathione at the expense of connective tissue and taurine, or the chronic user of proton pump inhibitors whose B12 recycling has been silently disabled and whose one-carbon metabolism no longer regenerates the methyl groups the SAM cycle requires, the answer they receive is a number that describes a state they are not in.

The physiological right this piece defends extends the argument Two Doses, One Molecule made for micronutrients into the substrate class where the mainstream nutritional framework has been slowest to see it. It is the right to have one’s protein needs indexed to the specific tissues one is trying to build and maintain, to the specific sinks that will draw on the supply before those tissues receive their share, and to the specific losses that continued sink activity imposes. It includes the right to formulations (glycine, collagen peptides, glutamine, N-acetylcysteine, creatine, arginine) that address the sinks directly, at doses the aggregate protein number does not describe. The right presupposes, first, that the sinks be named.

The Scrabble player who realizes their rack is missing the letter that would have made the word does not conclude that they need more tiles in general. They conclude that the specific tile is missing, and that no amount of the others will substitute for it. A recommendation that reports total grams and does not report the letters is a recommendation designed to be scored without ever revealing whether the word was built.

References

1. Rose WC. The amino acid requirements of adult man. Nutrition Abstracts and Reviews. 1957;27(3):631-647.
2. Elango R, Humayun MA, Ball RO, Pencharz PB. Evidence that protein requirements have been significantly underestimated. Curr Opin Clin Nutr Metab Care. 2010;13(1):52-57. PubMed 19841581
3. Phillips SM, Van Loon LJ. Dietary protein for athletes: from requirements to optimum adaptation. J Sports Sci. 2011;29(sup1):S29-S38.
4. Bauer J, Biolo G, Cederholm T, et al. Evidence-based recommendations for optimal dietary protein intake in older people: a position paper from the PROT-AGE Study Group. J Am Med Dir Assoc. 2013;14(8):542-559.
5. Meléndez-Hevia E, De Paz-Lugo P, Cornish-Bowden A, Cárdenas ML. A weak link in metabolism: the metabolic capacity for glycine biosynthesis does not satisfy the need for collagen synthesis. J Biosci. 2009;34(6):853-872. Link
6. de Paz-Lugo P, Lupiáñez JA, Meléndez-Hevia E. High glycine concentration increases collagen synthesis by articular chondrocytes in vitro. Amino Acids. 2018;50(10):1357-1365. PMC6153947
7. de Paz-Lugo P, Lupiáñez JA, Sicilia J, Meléndez-Hevia E. Control analysis of collagen synthesis by glycine, proline and lysine in bovine chondrocytes in vitro. J Theor Biol. 2023. ScienceDirect
8. Windmueller HG, Spaeth AE. Uptake and metabolism of plasma glutamine by the small intestine. J Biol Chem. 1974;249(16):5070-5079.
9. Reeds PJ, Burrin DG. Glutamine and the bowel. J Nutr. 2001;131(9):2505S-2508S.
10. Newsholme EA, Crabtree B, Ardawi MS. The role of high rates of glycolysis and glutamine utilization in rapidly dividing cells. Biochem J. 1985;239(1):121-125.
11. Rodas PC, Rooyackers O, Hebert C, Norberg Å, Wernerman J. Glutamine and glutathione at ICU admission in relation to outcome. Clin Sci (Lond). 2012;122(12):591-597.
12. Bertolo RFP, Chen CZ, Law G, Pencharz PB, Ball RO. Threonine requirement of neonatal piglets receiving total parenteral nutrition is considerably lower than that of piglets receiving an identical diet intragastrically. J Nutr. 1998;128(10):1752-1759.
13. Law GK, Bertolo RF, Adjiri-Awere A, Pencharz PB, Ball RO. Adequate oral threonine is critical for mucin production and gut function in neonatal piglets. Am J Physiol Gastrointest Liver Physiol. 2007;292(5):G1293-G1301.
14. Stipanuk MH, Ueki I. Dealing with methionine/homocysteine sulfur: cysteine metabolism to taurine and inorganic sulfur. J Inherit Metab Dis. 2011;34(1):17-32.
15. Yin J, Ren W, Yang G, et al. L-Cysteine metabolism and its nutritional implications. Mol Nutr Food Res. 2016;60(1):134-146.
16. Badawy AA. Kynurenine pathway of tryptophan metabolism: regulatory and functional aspects. Int J Tryptophan Res. 2017;10:1178646917691938.
17. Kim H, Chen L, Lim G, et al. Brain indoleamine 2,3-dioxygenase contributes to the comorbidity of pain and depression. J Clin Invest. 2012;122(8):2940-2954.
18. Vaz FM, Wanders RJA. Carnitine biosynthesis in mammals. Biochem J. 2002;361(3):417-429.
19. Brosnan JT, Brosnan ME. Creatine: endogenous metabolite, dietary, and therapeutic supplement. Annu Rev Nutr. 2007;27:241-261.
20. da Silva RP, Nissim I, Brosnan ME, Brosnan JT. Creatine synthesis: hepatic metabolism of guanidinoacetate and creatine in the rat in vitro and in vivo. Am J Physiol Endocrinol Metab. 2009;296(2):E256-E261.
21. Sullivan LB, Gui DY, Hosios AM, Bush LN, Freinkman E, Vander Heiden MG. Supporting aspartate biosynthesis is an essential function of respiration in proliferating cells. Cell. 2015;162(3):552-563.
22. Everaert I, Mooyaart A, Baguet A, et al. Vegetarianism, female gender and increasing age, but not CNDP1 genotype, are associated with reduced muscle carnosine levels in humans. Amino Acids. 2011;40(4):1221-1229.
23. Fridman V, Suriyanarayanan S, Novak P, et al. Randomized trial of l-serine in patients with hereditary sensory and autonomic neuropathy type 1. Neurology. 2019;92(4):e359-e370.
24. Ames BN. Low micronutrient intake may accelerate the degenerative diseases of aging through allocation of scarce micronutrients by triage. Proc Natl Acad Sci USA. 2006;103(47):17589-17594. PMC1693790
25. McCann JC, Ames BN. Vitamin K, an example of triage theory: is micronutrient inadequacy linked to diseases of aging? Am J Clin Nutr. 2009;90(4):889-907.
26. McCann JC, Ames BN. Adaptive dysfunction of selenoproteins from the perspective of the triage theory. FASEB J. 2011;25(6):1793-1814.
27. Ames BN. Prolonging healthy aging: Longevity vitamins and proteins. Proc Natl Acad Sci USA. 2018;115(43):10836-10844.
28. Iwai K, Hasegawa T, Taguchi Y, et al. Identification of food-derived collagen peptides in human blood after oral ingestion of gelatin hydrolysates. J Agric Food Chem. 2005;53(16):6531-6536.
29. Ohara H, Matsumoto H, Ito K, Iwai K, Sato K. Comparison of quantity and structures of hydroxyproline-containing peptides in human blood after oral ingestion of gelatin hydrolysates from different sources. J Agric Food Chem. 2007;55(4):1532-1535.
30. Shigemura Y, Iwai K, Morimatsu F, et al. Effect of prolyl-hydroxyproline (Pro-Hyp), a food-derived collagen peptide in human blood, on growth of fibroblasts from mouse skin. J Agric Food Chem. 2009;57(2):444-449.
31. Boldyrev AA, Aldini G, Derave W. Physiology and pathophysiology of carnosine. Physiol Rev. 2013;93(4):1803-1845.
32. Blancquaert L, Everaert I, Derave W. Beta-alanine supplementation, muscle carnosine and exercise performance. Curr Opin Clin Nutr Metab Care. 2015;18(1):63-70.
33. Peters V, Klessens CQ, Baelde HJ, et al. Intrinsic carnosine metabolism in the human kidney. Amino Acids. 2015;47(12):2541-2550.