How Seven Tiny Mutations Break a Metabolic Enzyme and Cause a Rare Brain and Heart Syndrome

Deep inside every cell’s mitochondria, an enzyme called serine hydroxymethyltransferase 2, or SHMT2, quietly performs one of life’s most fundamental chemical chores: it hands single carbon atoms over to the machinery that builds DNA letters and keeps cellular metabolism running. When this enzyme falters, the consequences can be devastating. In 2020, researchers linked seven different genetic variants of SHMT2 to a previously unknown developmental syndrome affecting the brain and heart, seen in pediatric patients presenting with intellectual disability, microcephaly, spastic paraparesis, ataxia, and hypertrophic cardiomyopathy. Now, a team of structural biologists and biochemists at Sapienza University of Rome and Italy’s National Research Council has dissected, mutation by mutation, exactly how each of these tiny genetic changes sabotages the same protein in strikingly different ways.

The new study, published in Frontiers in Chemical Biology, combines enzyme kinetics, thermal stability measurements, analytical ultracentrifugation, and X-ray crystallography into a systematic portrait of the seven disease-associated variants: P157S, T186R, G375del, N379D, G423S, Q435P, and P499A. The findings reveal that there is no single way to break SHMT2. Instead, the mutations attack different layers of the enzyme’s architecture, from its four-subunit assembly to the precise positioning of its vitamin B6-derived cofactor, and even the delicate balance between structural rigidity and the conformational flexibility that catalysis demands.

SHMT2 belongs to a large family of pyridoxal 5′-phosphate, or PLP, dependent enzymes. Its canonical job is to catalyze the reversible transfer of a one-carbon unit from serine to tetrahydrofolate, producing glycine and methylenetetrahydrofolate, a versatile carrier that feeds the synthesis of purines and thymidine, the metabolism of amino acids and neurotransmitters, and the maintenance of redox balance in mitochondria. But SHMT2 is no one-trick enzyme. Eukaryotic SHMTs are tetrameric proteins built as dimers of dimers, and this higher-order assembly underpins a suite of moonlighting functions: the enzyme can bind RNA and be regulated by it, participate in inflammatory cytokine signaling through its interaction with the BRISC complex, and even engage with KIRREL3, a synaptic cell-adhesion protein whose own defects are tied to neurological and cognitive disorders.

The enzyme’s conformational landscape is central to all of this. Without PLP bound, the active-site region is largely unstructured, and the dimeric building blocks can adopt either an open, catalytically inert conformation or a closed one. Only closed dimers can assemble into the active tetramer. PLP binding drives a disorder-to-order transition that stabilizes the closed state and, with it, the tetrameric assembly, which is cemented by an extensive network of intersubunit contacts centered on a histidine residue, His158, at the dimer-dimer interface. Mutations that destabilize either interface push the equilibrium back toward the open dimer, which in turn lowers the enzyme’s grip on its cofactor. It is precisely this equilibrium that the disease variants perturb.

When the Roman team expressed and purified the mutant proteins, one variant immediately stood out. Q435P, which replaces a glutamine with a rigid proline in a loop connecting two alpha helices, proved poorly soluble and prone to aggregation, and had to be excluded from further biochemical analysis. The researchers suspect that the proline insertion disrupts a crucial hydrogen bond between Q435 and the backbone of W43 in the partner subunit, destabilizing the dimeric interface and promoting misfolding. The remaining six variants all folded and could be studied, but they fell into three functional classes: P499A remained essentially fully active at about 95 percent of wild-type activity; G375del, N379D, and P157S were mildly impaired, retaining roughly 45 to 65 percent activity; and G423S and T186R were severely compromised, with G423S keeping only about 20 percent activity and T186R showing no measurable catalytic function at all, even at tenfold higher enzyme concentrations.

The structural basis of the two most damaged variants emerged from crystallography. T186R, which swaps a threonine for a bulky, positively charged arginine at the dimeric interface, elutes mainly as a dimer in solution and retains only about a third of the PLP-associated signal seen in the wild-type enzyme. Its cofactor affinity is roughly 42-fold worse than wild-type, with a dissociation constant of 5.50 micromolar versus 0.13 micromolar. In the crystal, the oversized arginine side chain creates steric hindrance that unfolds the entire PLP-interacting loop, residues 165 through 186, leaving the cofactor invisible in the electron density and the active-site surroundings in disarray. The tetramer that does form is a distorted assembly of two semi-open dimers, held together by only half of the normal interface. P157S, located at the tetrameric interface, tells a subtler story. Its mutation flips the carboxyl group of the neighboring His158, dismantling the aromatic stacking and hydrogen-bond cluster that anchors the tetramer, and shifts an adjacent alpha helix by about 1.2 angstroms, enough to force the PLP-binding loop to unfold. Remarkably, the cofactor itself was caught in a never-before-seen relaxed conformation, its pyridine ring tilted by 90 degrees and its phosphate group flipped to the opposite side, where it occupies a position normally reserved for stabilizing the substrate-bound intermediate. The team interprets this as an auto-inhibited state, and notes that such a cofactor tilt had only recently been glimpsed in a cryo-EM structure of a soybean SHMT isoform.

Perhaps the most paradoxical finding comes from G423S, a mutation lying some 20 angstroms away from the PLP-binding region. Despite losing 80 percent of its catalytic activity, this variant is more thermally stable than the wild-type enzyme, with an apparent melting temperature 4 degrees Celsius higher and a folding free energy more favorable by 1.71 kilocalories per mole. The crystal structure, solved at 1.8 angstroms, explains the contradiction: the serine side chain clashes with cysteine 412, nudging its backbone and triggering a domino effect that disorders the folate-binding loop, residues 413 through 420, at the mouth of the active site. In the native enzyme, this loop appears to sit in a frustrated, energetically suboptimal conformation, a common feature of protein regions dedicated to binding and catalysis. By relaxing that frustration, the mutation buys stability at the cost of function, a vivid illustration of the evolutionary trade-off between a protein’s fold and its job. With the loop disordered, the active site becomes more solvent-exposed, PLP affinity drops elevenfold, and tetrahydrofolate can no longer be positioned correctly, crippling the reaction in both directions.

The remaining variants cluster in the enzyme’s C-terminal domain, a region whose dynamics govern allosteric regulation and RNA binding in SHMT enzymes. N379D buries a negative charge in a packed hydrophobic pocket, perturbing a hydrogen-bonding network built around two histidines near the active site, yet the overall damage is mild. G375del shortens a surface loop and breaks the interaction between the carbonyl oxygen of S374 and R230, distorting the geometry of the active-site cavity in much the same way. P499A is the strangest of all: fully active and stable, yet with an eightfold reduction in PLP affinity. Because the mutant does not shed its cofactor during chromatography the way P157S does, the authors propose that its weakened binding reflects a slower association rate, likely caused by restricted access to the active site, rather than faster dissociation, a hypothesis that will require further experiments to confirm.

Beyond cataloguing molecular defects, the study delivers a broader message about how disease emerges from protein architecture. Rather than converging on a single failure mode, the seven SHMT2 variants impair the enzyme through distinct mechanisms: broken oligomerization, destabilized active-site loops, weakened cofactor binding, disrupted folate positioning, and outright misfolding. The work also provides the first direct crystallographic visualization of a frustrated cofactor conformation in SHMT, reinforcing the idea that the enzyme’s catalytic power depends on holding PLP in a precisely tuned, energetically strained position. As clinical sequencing continues to surface new SHMT2 variants in patients, this structural framework offers a rational starting point for interpreting each mutation’s pathogenic potential, and underscores how subtle perturbations of a protein’s conformational landscape can cascade into profound developmental disease.

Subject of Research: Structural and biochemical characterization of SHMT2 genetic variants linked to a neurodevelopmental syndrome

Article Title: Structural and functional defects of mitochondrial serine hydroxymethyltransferase genetic variants responsible for a novel neurodevelopmental syndrome

Article References: Boumis, G., Breccia, S., Pistoia, G., Spizzichino, S., Zamparelli, C., Troilo, F., Di Matteo, A., Travaglini Allocatelli, C., Contestabile, R., Tramonti, A., Cutruzzolà, F., & Giardina, G. (2026). Structural and functional defects of mitochondrial serine hydroxymethyltransferase genetic variants responsible for a novel neurodevelopmental syndrome. Frontiers in Chemical Biology, 5, Article 1941005. https://doi.org/10.3389/fchbi.2026.1941005

Image Credits: AI Generated

DOI: 10.3389/fchbi.2026.1941005

Keywords: SHMT2, one-carbon metabolism, pyridoxal phosphate, mitochondria, neurodevelopmental syndrome, X-ray crystallography, protein oligomerization, enzyme kinetics, conformational frustration, genetic variants, tetrahydrofolate, structural biology