Chaperone mRNA secondary structure does not strongly track thermal adaptation in bacteria, but groL coding sequence shortens in its disordered C-terminal tail

Ribosomal and transfer RNAs show clear compositional and structural signatures of adaptation to growth temperature, but messenger RNA does not, and the generality of this asymmetry at individual loci remains untested. We asked whether the two principal bacterial chaperone transcripts, dnaK and groL , encode systematically different mRNA secondary structures in thermophiles relative to mesophiles. Using a phylum-matched cohort of 50 bacterial strains (25 mesophiles, 25 thermophiles; balanced across Actinobacteria, Bacteroidetes, Firmicutes and Proteobacteria), we folded each coding sequence at 37°C and 55°C and compared fifteen thermodynamic and structural descriptors per gene. Raw folding metrics showed no significant group differences after multiple-testing correction. However, after controlling for dinucleotide composition with Altschul-Erickson shuffles, thermophile sequences, particularly groL , retained modestly more residual secondary structure than mesophile sequences. The strongest and cleanest signal was not structural: groL coding sequences are shorter in thermophiles (mean 1626.5 versus 1636.3 nt; Cliff's δ = 0.571, P = 5.1×10 -4 ), a relationship that scales continuously with optimal growth temperature and persists after phylogenetic generalised least squares correction (n = 50, Pagel's λ = 0.997, slope = -0.356 nt°C -1 , P = 6.0×10 -5 ). Protein-level alignment localised this difference: approximately three-quarters of the excess deletion signal falls within the terminal ~25 residues corresponding to the intrinsically disordered, glycine/methionine-rich C-terminal tail of GroEL. Thermal adaptation in these loci therefore appears to act primarily on a disordered peptide extension, with only a modest residual signature detectable in mRNA folding after composition is controlled.