Enzymes capable of processing a variety of compounds enable plants to adapt to diverse environmental conditions. PRISEs (progesterone-5β-reductase/iridoid synthase-like enzymes), examples of such substrate-promiscuous enzymes, are involved in iridoid and cardenolide pathways and demonstrate notable substrate promiscuity by reducing the activated C=C double bonds of plant-borne and exogenous 1,4-enones. In this study, we identified PRISE genes in Plantago media (PmdP5βR1) and Plantago lanceolata (PlP5βR1), and the corresponding enzymes were determined to share a sequence identity of 95%. Despite the high sequence identity, recombinant expressed PmdP5βR1 was 70 times more efficient than PlP5βR1 for converting progesterone. In order to investigate the underlying reasons for this significant discrepancy, we focused on specific residues located near the substrate-binding pocket and adjacent to the conserved phenylalanine “clamp”. This clamp describes two phenylalanines influencing substrate preferences by facilitating the binding of smaller substrates, such as 2-cyclohexen-1-one, while hindering larger ones, such as progesterone. Using structural analysis based on templates PDB ID: 5MLH and 6GSD from PRISE of Plantago major, along with in silico docking, we identified positions 156 and 346 as hot spots. In PlP5βR1 amino acid residues, A156 and F346 seem to be responsible for the diminished ability to reduce progesterone. Moreover, the double mutant PlP5βR_F156L_A346L, which contains the corresponding amino acids from PmdP5βR1, showed a 15-fold increase in progesterone 5β-reduction. Notably, this modification did not significantly alter the enzyme’s ability to convert other substrates, such as 8-oxogeranial, 2-cyclohexen-1-one, and methyl vinyl ketone. Hence, a rational enzyme design by reducing the number of hotspots selectively, specifically improved the substrate preference of PlP5βR1 for progesterone.