Volume 62, Issue 5 p. 357-362
Critical Review
Free Access

Enzymology of the carnitine biosynthesis pathway

Karin Strijbis

Karin Strijbis

Department of Medical Biochemistry, Academic Medical Center, Amsterdam, The Netherlands

Search for more papers by this author
Frédéric M. Vaz

Frédéric M. Vaz

Laboratory Genetic Metabolic Diseases, Academic Medical Center, Amsterdam, The Netherlands

Search for more papers by this author
Ben Distel

Corresponding Author

Ben Distel

Department of Medical Biochemistry, Academic Medical Center, Amsterdam, The Netherlands

Tel: + 31205665127. Fax: +331206915519.

Department of Medical Biochemistry, Academic Medical Center, Meibergdreef 15, Amsterdam 1105AZ, The NetherlandsSearch for more papers by this author
First published: 19 March 2010
Citations: 62


The water-soluble zwitterion carnitine is an essential metabolite in eukaryotes required for fatty acid oxidation as it functions as a carrier during transfer of activated acyl and acetyl groups across intracellular membranes. Most eukaryotes are able to synthesize carnitine endogenously, besides their capacity to take up carnitine from the diet or extracellular medium through plasma membrane transporters. This review discusses the current knowledge on carnitine homeostasis with special emphasis on the enzymology of the four steps of the carnitine biosynthesis pathway. © 2010 IUBMB IUBMB Life, 62(5): 357–362, 2010


L-carnitine is an important metabolite which function is indispensable for intermediary metabolism in eukaryotic cells. Its prime function is to act as a carrier for the transport of activated long-chain fatty acids from the cytosol into the mitochondrial matrix where β-oxidation takes place. After activation of fatty acids with a CoA group, the formed acyl-CoA units are converted into carnitine esters, a reaction catalyzed by carnitine palmitoyl-transferase (CPT1). Carnitine esters can subsequently cross the membrane by the additional aid of a transporter. In mammals, very long chain fatty acids undergo shortening by the peroxisomal β-oxidation, while shorter fatty acids are directly imported into mitochondria (1). Carnitine is therefore involved in both the transport of activated long-chain fatty acids from the cytosol to mitochondria as well as the transfer of the products of peroxisomal β-oxidation, that is shortened fatty acids and acetyl units, to the mitochondria (Fig. 1). Carnitine-dependent transport routes between intracellular compartments are slightly different in yeasts like Saccharomyces cerevisiae and Candida albicans. Fatty acid β-oxidation is completely peroxisomal in these unicellular eukaryotes (1) and therefore acetyl units are the only products that need to be transported from peroxisomes to mitochondria. The enzymes involved in the reversible coupling of acetyl units to carnitine, that is carnitine acetyl-transferases, have been well characterized in yeasts and were shown to localize to peroxisomes and mitochondria (2, 3). In addition to the oxidation of fatty acids, yeasts also employ carnitine-dependent transport during growth on C2 carbon sources like ethanol and acetate (3, 4) (Fig. 1). During growth on these carbon sources, acetyl units are produced in the cytosol and need to be transported to the mitochondrial TCA cycle and into the peroxisomal/cytosolic glyoxylate cycle. While the mammalian and yeast mitochondrial carnitine/acyl-carnitine translocases have been identified (CACT and CRC1, respectively) (57), the identity of the (putative) peroxisomal transporter is not known. Additional roles of carnitine in mammalian energy metabolism besides transport of metabolites between intracellular compartments include modulation of the acyl-CoA/CoA ratio, storage of energy in the form of acetyl-carnitine and excretion of poorly metabolizable, toxic, acyl groups (8).

Details are in the caption following the image

Function of carnitine in intracellular transport of acyl- and acetyl-CoA Yeast- and mammalian specific pathways are depicted in dashed boxes. Abbreviations: acetyl-CoA: acetyl coenzyme A, acyl-CoA: activated fatty acid, ALD: peroxisomal transporter ABCD1, Cat: carnitine acetyl-transferase, Cpt: carnitine palmitoyltransferase, TCA cycle: tricarboxylic acid cycle, Yat: yeast carnitine acetyl-transferase.

Carnitine was discovered as a nutrient essential for optimal growth of larvae of the insect Tenebrio molitor and tentavily named vitamin BT (9). In mammals, carnitine is considered a “conditionally-essential” nutrient, that is it can be synthesized by the organism but most is taken up from the diet (Fig. 2B). It has been estimated that 75% of total body carnitine levels comes from the diet and only 25% from endogenous synthesis (10). Meat and dairy products contain high levels of carnitine, while food sources derived from plants contribute very little and therefore total carnitine intake is highly dependent on diet (11). In humans the major metabolites of orally administered carnitine are trimethylamine (excreted primarily in urine) and γ-butyrobetaine (excreted primarily in feces) (12). These metabolites are most likely produced by indigenous bacteria in thegut as carnitine is not degraded by enzymes of eukaryotic origin (13).

Details are in the caption following the image

The enzymes of the carnitine biosynthesis pathway and their contribution to cellular carnitine homeostasis. (A) Enzymatic conversions of the carnitine biosynthesis pathway. TML: N6-trimethyllysine, TMLD: TML dioxygenase, HTML: 3-hydroxy TML, HTMLA: HTML aldolase, TMABA: 4-N-trimethylaminobutyraldehyde, TMABADH: TMABA dehydrogenase, γ-BB: 4-N-trimethylaminobutyrate, BBD: γ-BB dioxygenase, PLP: pyridoxal 5′-phosphate. (B) schematic representation of carnitine uptake and biosynthesis. Carnitine uptake is mediated by the OCTN2 transporter in humans or the Agp2 transporter in S. cerevisiae. Degradation of proteins containing trimethylated lysine residues results in availability of TML as a substrate for intracellular carnitine biosynthesis. (C) Comparison of the enzymatic reaction catalyzed by HTMLA and threonine aldolase (TA). Both enzymes perform a similar aldolytic cleavage of their substrate into glycine and an aldehyde (TMABA and acetaldehyde for HTMLA and TA, respectively).


Primary carnitine deficiency is caused by mutations in the sodium-dependent plasma membrane carnitine transporter OCTN2 (encoded by the SLC22A5 gene), which is essential in controlling the body pool of carnitine (1416). Over 99% of body carnitine is intracellular (17) and OCTN2 is responsible for maintaining the large tissue gradient for carnitine in muscle and for setting the renal threshold for carnitine excretion. OCTN2 deficiency, an autosomal recessive disorder, presents with progressive cardiomyopathy and skeletal muscle weakness at age 2–4 (18). Carnitine treatment in these patients raises muscle carnitine concentrations to only 5–10% of normal, but this seems sufficient for normal function (19). Acquired deficiency of carnitine can be caused by chronic administration of certain drugs such as pivalate, a nonmetabolizable branched-chain fatty acid (17, 20), and valproic acid [reviewed by (21)]. Defects in fatty acid oxidation also result in acyl-carnitine abnormalities in combination with a deficiency in free carnitine. However, these defects are considered “secondary” carnitine deficiencies, as carnitine abnormalities occurring in these cases are the consequence, rather than the cause, of the impairment in fatty acid oxidation.


Besides the ability to take up carnitine from diet, many eukaryotes possess a functional carnitine biosynthesis pathway. Humans have been convincingly shown to synthesize carnitine from the labeled substrate 6-N-trimethyllysine (TML) at a very early age (22). A functional carnitine biosynthesis pathway has been identified in mammals like mice and rats (23) and blast analysis showed that also the fruitfly Drosophila melanogaster and round worm Caenorhabditis elegans have orthologs of the carnitine biosynthesis genes (24). Yeasts like Neurospora crassa and Candida albicans do express a functional carnitine biosynthesis pathway (25, 26), while Saccharomyces cerevisiae, on the other hand, is unable to synthesize the molecule endogenously but takes carnitine up from the medium and uses it metabolically (7, 27). The chemical characterization of the carnitine biosynthesis pathway was performed in the 1970s in experiments with labeled substrates and subsequent detection of carnitine, carnitine biosynthesis intermediates and byproducts in the filamentous fungus Neurospora crassa and rat (25, 2831). TML, the substrate for carnitine biosynthesis, is the product of lysosomal or proteasomal degradation of proteins containing N-methylated lysines. N-methylation is a post-translational modification carried out by methyl-transferases that use S-adenosylmethionine as a methyl-donor (32). In mammals certain proteins such as calmodulin, myosin, actin, cytochrome c and histones contain such N-methylated lysine residues and TML is generated by degradation of these proteins by lysosomes or the proteasome. The scheme of carnitine biosynthesis from its precursor TML is depicted in Fig. 2A. It is essentially the same as the scheme published in 1977 by Kaufman and Broquist (25), underscoring the fact that the identity of the intermediates was firmly established more than 30 years ago. The more recent contributions to the pathway are the genes encoding the enzymes that catalyze the four steps of carnitine biosynthesis. Three of the four enzymes have been characterized at the molecular level in mammals (3335), but only recently the identity of the last remaining enzyme was revealed. Genetic studies in the yeast Candida albicans led to the identification of the aldolase that catalyzes the second step of the carnitine biosynthesis pathway (26). The four enzymatic steps of the carnitine pathway, TMLD, HTMLA, TMABADH and BBD are separately described below.


The first enzyme of the carnitine biosynthesis pathway is TML dioxygenase (TMLD), which hydroxylates TML to yield 3-hydroxy-TML (HTML). The human, rat, mouse and C. albicans genes encoding TMLD were identified (26, 34). TMLD is a nonhaem ferrous-iron dioxygenase, which requires 2-oxoglutarate, Fe2+ and molecular oxygen as cofactors. Dioxygenases react with molecular oxygen through their associated iron and use this iron-bound oxygen to hydroxylate the substrate. The remaining oxygen atom is incorporated into 2-oxoglutarate, thereby forming succinate and simultaneously releasing CO2 (36). In addition to these cofactors, TMLD also requires the presence of ascorbate (vitamin C) for enzymatic activity, presumably to maintain the iron in the ferrous state. While TML is produced by protein degradation in the cytosol, TMLD was shown to localize to the mitochondrial matrix in rat kidney (37). The other carnitine biosynthesis enzymes are thought to reside in the cytosol and therefore this compartmentalization would suggest transport of metabolites over the mitochondrial membrane.


The second enzyme of carnitine biosynthesis performs a pyridoxal 5′-phosphate (PLP)-dependent aldolytic cleavage of HTML, resulting in 4-trimethylaminobutyraldehyde (TMABA) and glycine and was dubbed HTML aldolase (HTMLA). The identity of the gene encoding HTMLA has long remained a mystery. It was speculated that HTMLA might be identical to serine hydroxymethyltransferase (SHMT), as purified SHMT was able to convert HTML to TMABA and glycine (38, 39). SHMTs are a class of PLP-dependent enzymes that perform the interconversion between serine and glycine. Another class of enzymes that exhibit a reaction mechanism that is very similar to the aldolytic cleavage performed by the putative HTMLA is the group of threonine aldolases (Fig. 2C). Threonine aldolases are PLP-dependent enzymes that convert threonine to acetaldehyde and glycine. Investigations into the carnitine biosynthesis pathway of C. albicans led to the eventual identification of the HTMLA-encoding gene. The C. albicans genome encodes two threonine aldolases orthologs and detailed biochemical and genetic experiments showed that the previously uncharacterized orf19.6306 is actually the HTMLA of the carnitine biosynthesis pathway (26). However, the enzymatic conversion of HTML to TMABA seems more promiscuous than the other steps in thecarnitine biosynthesis pathway. C. albicans strains lacking either the first, third or fourth enzyme of the pathway are unable to grow on nonfermentable carbon sources such as fatty acids and acetate due to their inability to transport acetyl units. However, the C. albicans htmla null strain shows residual growth under these conditions. Subsequent disruption of the true threonine aldolase gene GLY1 (26, 40) in the htmla null mutant background further reduced but still not completely blocked growth on fatty acids while the gly1 single deletion strain showed a wild growth rate on this carbon source (Table 1). These results suggest that the conversion of HTML to TMABA was still taking place in the htmla/gly1 double null strain, albeit with strongly reduced efficiency compared to the HTMLA (or Gly1) catalyzed reaction. We can only speculate on the identity of the third protein involved, but the cytosolic SHMT of C. albicans seems a likely candidate.

Table 1. Specific growth rates (h−1) of C. albicans wild type and mutant strains on minimal oleate or minimal acetate media. Specific growth rates are calculated as ln OD600 T2-ln OD600 T1/T2-T1 using the values from the experiment displayed in Fig. 6C and 6D of reference26
C. albicans strain Min. oleate Min. acetate
Wild type 0.283 0.100
gly1Δ/Δ 0.278 0.091
htmlaΔ/Δ 0.138 0.066
htmla/gly1Δ/Δ 0.080 0.067

The human gene encoding HTMLA has thus far not been identified; neither have other mammalian HTMLA genes. Threonine aldolases now seem the likely candidates for the mammalian HTMLA, based on the recent studies in C. albicans. This hypothesis is supported by observations that mammals might lack a true threonine aldolase enzyme [reviewed by (41)] and that some mammalian SHMTs also possess low threonine aldolase activity (42). Single threonine aldolase orthologs can be found for both human and mouse and the genes are called THA1P and GLY1, respectively. However, identification of the human HTMLA is not as straightforward as it appears, as human THA1P has been reported to be a pseudogene. Two single nucleotide deletions cause a predicted premature in-frame stop and the THA1P mRNA was not detectable by 5′ and 3′ RACE methods on human liver RNA, whereas the mouse GLY1 mRNA was present in several adult tissues, including prostate, heart and liver (43). Nevertheless, Rebouche and Engel showed already in 1980 that HTMLA enzymatic activity can be detected in several human tissues, including liver, brain, kidney, heart and skeletal muscle homogenates (44). The pseudogene status of the THA1P gene now leads to the hypothesis that in humans an enzyme like SHMT with low HTMLA activity could catalyze the conversion of HTML to TMABA and glycine. In rat and mice on the other hand, the closest threonine aldolase ortholog GLY1 is a functional gene (43), and therefore total flux through the carnitine biosynthesis pathway might be higher in rodents compared to humans. This hypothesis is supported by the observation that administration of TML to humans mainly leads to increased excretion of TML with only marginal synthesis of carnitine from the exogenous TML (22), while TML administered to rats is almost completely converted to carnitine (45).


The third reaction of the carnitine biosynthesis pathway is the dehydrogenation of TMABA to 4-N-trimethylaminobutyrate or γ-butyrobetaine (γ-BB), a reaction catalyzed by the NAD+-dependent TMABA dehydrogenase (TMABADH). The high number of aldehyde dehydrogenases encoded in mammalian and yeast genomes and their considerable sequence similarity has complicated the identification of the TMABADH gene simply by its sequence. Therefore, the TMABADH enzyme was purified from rat liver and based on the amino acid sequence the encoding cDNA was identified (33). In the same article the human aldehyde dehydrogenase 9 (ALDH9) was shown to be the likely human ortholog of rat TMABADH, as substrate specificities of the two enzymes are very similar. In C. albicans the TMABADH gene was identified by its proximity to the HTMLA gene: the genes are chromosomal neighbors that share a 1,000 base pair upstream region (26). It remains to be investigated whether this region contains common promoter elements required for coordinated expression of TMABADH and HTMLA.


In the fourth enzymatic step of the carnitine biosynthesis pathway, γ-BB is hydroxylated by γ-butyrobetaine dioxygenase (BBD) to yield L-carnitine. BBD, like TMLD, is a dioxygenase that incorporates an oxygen molecule in its substrate and succinate. Both enzymes require the same cofactors and have considerable overall sequence homology (46). Interestingly, BBD was first purified from the bacterium Pseudomonas sp. AK 1, that can grow on butyrobetaine as the sole source of carbon and nitrogen (47) but presumably lacks a functional carnitine biosynthesis pathway. BBD was purified from various species, including human kidney (48) and the BBD cDNA was found to be expressed in human kidney, liver and brain (35).


Experiments in rat and C. albicans have convincingly shown that the flux through the carnitine biosynthesis pathway is limited by the availability of the substrate, TML (26, 45). To solely rely on the degradation of certain trimethylated proteins for the utilization of an important energy source seems risky, but on the other hand very low levels of carnitine seem to be sufficient for normal function (see OCTN2 deficiency, described above). Nevertheless, carnitine is widely used in vitamin preparations and as food supplement. Although direct evidence for the effect of carnitine supplementation is lacking, some beneficial effects on physical fatigue and exercise recovery have been reported (24, 49, 50). Fatty acid oxidation, and therefore also the availability of carnitine, has been shown to play an important role during fetal development [discussed by (51)]. OCTN2 is present in the human placenta and mediates fetal uptake of carnitine, but both the human fetal liver, kidney, spinal cord and placenta where also shown to synthesize carnitine (51). However, the availability of TML might be limiting as carnitine supplements have been shown to increase fetal growth in pigs (52, 53) and human maternal plasma carnitine levels decrease during early pregnancy (54). Both of these findings indicate a high demand for additional carnitine during fetal development.

Intriguingly, so far no patients have been identified with a defect in carnitine biosynthesis. On the one hand, requirement for carnitine during fetal development suggests that a carnitine biosynthesis defect might be lethal to the embryo. On the other hand, the absence of patients with a defect in the carnitine biosynthetic pathway indicates that uptake of carnitine from the diet or placenta in combination with an increase in renal reabsorption is sufficient to maintain carnitine homeostasis both during fetal development and throughout life. Patients with a deficiency in the carnitine transporter OCTN2 exemplify the importance of carnitine uptake in carnitine homeostasis (1416). Even in these patients, carnitine deficiency presents itself at a relatively late age (2–4 years), supporting the notion that carnitine levels probably need to drop below 10% of normal before they become a limiting factor in carnitine-dependent processes [reviewed by (55)]. Together these observations suggest that dietary carnitine is probably sufficient for normal function and possibly only a strict vegetarian diet in combination with a defect in carnitine biosynthesis would manifest in symptoms of genuine carnitine deficiency.