
Glutathione’s first peptide bond is an amide between the γ-carboxyl group of glutamate and the α-amino group of cysteine, which makes the molecule γ-L-glutamyl-L-cysteinylglycine rather than the α-linked Glu–Cys–Gly a standard tripeptide would have[1][2]. The second bond, from cysteine to glycine, is an ordinary α-peptide bond[1]. That swapped carboxyl is the mechanistic answer: GCLC selects glutamate’s side-chain carboxyl, so the first linkage is a γ amide. This explains how the bond forms and what it does; it does not establish an evolutionary reason for the choice.
Two carboxyl groups, one non-standard choice
Glutamate is the only residue in glutathione that carries a second carboxyl group. Its α-carboxyl sits on the α-carbon next to the amino group; its γ-carboxyl sits at the end of the side chain. Standard peptide chemistry uses the α-carboxyl, and that is what happens between cysteine and glycine here. The glutamate–cysteine bond is different: it forms through the γ-carboxyl and leaves the glutamate α-carboxyl free[2].
The PubChem connectivity record for glutathione (CID 124886) gives the formula C10H17N3O6S and a molecular weight of 307.33, and its SMILES string places one amide on the glutamate side chain and a second, conventional amide between cysteine and glycine[3]. Read as structure rather than as biology, that string shows two free acid groups on the molecule — the glutamate α-carboxyl and the glycine C-terminus — plus a free α-amino group on glutamate. A conventional Glu–Cys–Gly tripeptide would have its glutamate α-carboxyl locked into the first amide instead.
How GCLC selects the γ-carboxyl

Glutathione is built in two ATP-dependent steps[1][2]. The first, rate-limiting step joins glutamate and cysteine and is catalyzed by glutamate–cysteine ligase (GCL, EC 6.3.2.2), a heterodimer of a catalytic subunit (GCLC, about 73 kDa) and a modifier subunit (GCLM, about 31 kDa) encoded by different genes[2]. GCLC carries the catalytic activity and the feedback inhibition by glutathione; GCLM has no catalytic activity of its own but lowers the enzyme’s Km for glutamate and raises its Ki for glutathione[2]. The 2009 review of the two subunits states the point directly: GCLC catalyzes a unique γ-carboxyl linkage from glutamate to cysteine and requires ATP and Mg2+ as cofactors, while GCLM raises Vmax and Kcat, lowers Km for glutamate and ATP, and increases Ki for glutathione-mediated feedback inhibition[4].
The kinetics collected in Lu’s review, attributed there to the primary studies it cites, put the Km for glutamate at 1.8 mM and for cysteine at 0.1–0.3 mM, with glutathione acting as a competitive inhibitor against glutamate at Ki = 2.3 mM[2]. Those values are enzyme-level kinetics; on their own they do not establish which substrate limits glutathione synthesis in a cell, because that depends on cellular substrate concentrations and flux measurements rather than on Km alone. The second step, addition of glycine to the cysteine C-terminus by glutathione synthetase (EC 6.3.2.3), is not feedback-inhibited by glutathione[2].
GCL conjugates the γ-carboxyl group of L-glutamate to the amino group of L-cysteine, and the proposed mechanism runs through phosphorylation of that γ-carboxylate by ATP: the α-amino group of L-cysteine attacks the resulting γ-glutamyl phosphate intermediate as a nucleophile, releasing γ-glutamylcysteine[5]. The binding side comes from crystals of the Saccharomyces cerevisiae enzyme — 2.1 Å with glutamate and MgCl2, 2.7 Å with glutamate, MgCl2 and ADP — in which glutamate sits in an extended conformation inside a funnel-shaped active site, Arg-313 and Tyr-362 position its α-carboxylate in its own binding pocket, and its γ-carboxylate occupies one coordination site of a bound Mg2+ close to Arg-472, with the ADP β-phosphate also sitting near that γ-carboxylate[5]. These substrate-binding structures support a model in which a separate pocket anchors the α-carboxylate while the γ-carboxylate is positioned for activation. The proposed mechanism uses ATP to phosphorylate the γ-carboxylate; proximity to ADP β-phosphate in the observed complex is not itself an observation of that phosphate-transfer step. The resulting peptide retains a free glutamate α-carboxylate. The same assignment fits the failure mode when cysteine runs short: γ-glutamyl phosphate is not consumed, autocyclizes to 5-oxoproline, and 5-oxoprolinase reopens the ring to glutamate at a cost of ATP[6]. Those structures are of the yeast enzyme plus a homology model of human GCLC, not a solved human GCLC–ATP–cysteine complex, so residue-level contacts in the human enzyme remain model-based.
The γ-glutamyl cycle and the one enzyme that opens the bond

Because the bond is γ rather than α, ordinary peptidases do not cleave it. Lu’s review states that γ-glutamyltranspeptidase (GGT) is the only enzyme that can hydrolyze this unusual bond, and that GGT is present only on the external surfaces of certain cell types[2]. That localization matters because glutathione is synthesized inside the cell, while GGT-dependent cleavage of the γ bond occurs at the external membrane of GGT-expressing cells[2].
In the γ-glutamyl cycle, glutathione leaves the cell and GGT transfers its γ-glutamyl moiety to an amino acid acceptor — cystine is the best acceptor — producing a γ-glutamyl amino acid and cysteinylglycine[2]. The γ-glutamyl amino acid is transported back in and converted to the free amino acid plus 5-oxoproline, which can be turned into glutamate for new glutathione; cysteinylglycine is split by a dipeptidase into cysteine and glycine[2]. The structural study of human GGT1 fills in the enzyme side of that reaction: Thr381 is the reactive nucleophile, the active site has an oxyanion hole and a loop forming the substrate entrance, and the glutamate analog 6-diazo-5-oxo-norleucine inactivates the enzyme irreversibly, with a second-order rate constant of 0.052 mM⁻¹ min⁻¹ and a Ki of 2.7 ± 0.7 mM[7]. That is active-site structure and inhibitor pharmacology, not a measurement of how fast GGT handles glutathione in a tissue.
A later review of the cycle argues that GGT’s main job is hydrolysis of glutathione and related γ-glutamyl peptides at apical membranes, with the products taken up and reused for glutathione and mercapturic acid synthesis, and that the cycle should not be read as the amino acid transporter Meister originally proposed but as part of redox and xenobiotic handling[8]. The distinction is between the described transfer reaction and the broader physiological role of extracellular hydrolysis and amino-acid reuse; the cycle is not simply a transporter for intact amino acids.
What the bond buys: stability, concentration, and a steep gradient
Glutathione reaches an unusually high concentration for a small peptide: 1–10 mM in mammalian tissues, highest in liver[2], or 0.5–10 mmol/L in animal cells[1]. More than 90% of the pool is normally in the reduced form, with the remainder as glutathione disulfide; the cytosol holds 80–85% of cellular glutathione against 10–15% in mitochondria[1]. Extracellular fluid sits at 2–20 µM, a gradient that can reach a thousand-fold[1].
The sources read here do not compare the intracellular half-lives of matched γ- and α-linked tripeptides, so the gradient cannot be attributed to the bond alone.
Conformation and redox chemistry
The γ linkage changes the positions of free groups and therefore the potential arrangement of hydrogen-bond donors and acceptors. This is a connectivity-based inference; the sources read here do not include NMR, crystallographic or simulation data that define a preferred glutathione conformation or a specific hydrogen-bond network in solution. Glutathione presents a glutamate α-carboxyl, a glycine carboxylate, a glutamate α-amino group, two amide groups and the cysteine thiol. The thiol is the redox-active part: two glutathione molecules oxidize to glutathione disulfide (GSSG), with a disulfide bond between their cysteines[1]. The γ-glutamyl bond itself is not oxidized in that reaction; the cysteine thiol is the redox-active group, while the γ linkage is part of the peptide framework that holds the three residues together.
Both linkages are amides, but the γ bond changes the path of the backbone rather than the amide chemistry itself: the glutamate side chain becomes part of the backbone, the distance between the glutamate α-carbon and the first amide nitrogen is longer than in an α linkage[2]. From that connectivity, the potential hydrogen-bonding groups occupy different positions along the peptide; the free α-carboxyl replaces the free side-chain carboxyl of the corresponding α-linked structure rather than adding an extra carboxyl group. The remaining difference is one of connectivity and local amide planarity[9].
Reading the evidence: what each kind of measurement shows
Readouts in this area measure different things, and mixing them up is the usual source of overclaiming. Glutathione concentration and the ratio of reduced to oxidized glutathione are pool measurements: they report how much thiol is present and how oxidized the pool is, not why it reached that state[1]. Pool size, oxidation state, enzyme activity and intermediate accumulation answer different questions: γ-glutamylcysteine is a synthesis intermediate and 5-oxoproline participates in the cycle, but neither concentration alone identifies a blocked reaction[2], and none of these static readouts is a flux measurement of how fast glutathione is synthesized and consumed.
The human clinical material available here is a case report and narrative review of pyroglutamic acidemia, in which a derangement in the γ-glutamyl cycle produces a high anion gap metabolic acidosis and a large elevation of pyroglutamate (5-oxoproline) in urine; most reported cases are pediatric and caused by hereditary enzyme deficiencies, and acquired cases occur in adults[10]. One case report shows that the pathway can fail in a way clinicians can observe. It does not establish incidence, diagnostic thresholds or any treatment effect, and it should not be read as evidence about glutathione status in general.
Method choice changes what a GSH/GSSG value means. The ratio is used as an oxidative-stress readout, and an enzymatic Tietze recycling assay adapted for blood microvolumes uses a standard spectrophotometer or microplate reader and volumes up to 10 µL[11]. The same assay is vulnerable to sample-handling artifact: GSSG can be raised after collection, which biases the ratio toward oxidation, so collection and processing conditions have to be reported before values are compared[11].
The evidence types answer different questions. Purified-enzyme and crystal-structure work explains molecular recognition and catalysis. Review-level cellular pool measurements describe concentrations but are not direct cell-culture experiments in this evidence set. Animal evidence is represented indirectly: Lu’s review reports that Gclm-knockout mice have tissue glutathione levels reduced by about 85–90%, supporting a regulatory role for GCLM without establishing a human effect[2]. The human case report concerns a particular pathway disturbance, and the blood-assay abstract concerns measurement performance. Neither demonstrates an intervention benefit. The concentration ranges and kinetic constants summarized above are reported values from reviews, not new measurements.
Related research: What Is Reduced Glutathione: GSH in Redox Research; Peptide Identity: Sequences, Fragments, Modifications, Salt Forms; Glutathione.
Research Use Only. This article is for laboratory research reference and analytical education, not a validated experimental protocol or medical or veterinary advice, and not for human or animal administration.
References