Su_2026_Front.Syst.Biol_6_1914536

INTRODUCTION: Recycling polyethylene terephthalate (PET) from cotton-blended fabrics is constrained by substrate access. Textile PET is more crystalline and occluded than the amorphous film on which most PET hydrolases are optimized, and whether engineered enzymes retain their advantage as substrate complexity rises remains untested. METHODS: Six Thermobifida fusca cutinase 2 (TfCut2) variants were designed by structure-guided modelling and machine-learning prediction. Activity was screened with p-nitrophenyl butyrate (pNPB) and on PET film across 35-70 degreesC and pH 7.0-8.5, then assayed on pure PET fiber, knit and mesh textiles, and cotton-PET blends, with and without pretreatment in 15% NaOH at 80 degreesC for 4 h. Degradation was quantified as absorbance at 260 nm (A260), a relative measure of soluble product release. RESULTS: Under a unified condition (51.8 degreesC, pH 7.5), five variants were active on PET film, and MC-nc27 was inactive throughout. Advantage over wild-type widened with substrate complexity, reaching 14.3-fold for A65H/L90A/I213S on pretreated mesh textile and an A260 of 1.043 +/- 0.186 on pretreated CP55/45 against -0.007 +/- 0.033 for wild-type. Esterase ranking did not predict textile performance. DISCUSSION: Fold improvements over wild-type were preserved whether or not textiles were pretreated, with no hierarchy change detected across substrate formats, indicating that crystallinity is not the dominant variable governing relative variant performance. Physical features that persist regardless of crystallinity state are the more probable constraints on enzyme-substrate engagement. Wild-type gained little from crystallinity reduction alone. These results suggest that the capacity to engage a physically constrained polymer surface may be a productive target for engineering textile-active PET hydrolases.