Enzyme engineering toward catalytic-tetrad residues usually results in activity loss. Unexpectedly, we found that a directed evolution campaign yielded a beneficial residue A100 in KmCR (a carbonyl reductase from Kluyveromyces marxianus ZJB14056), which is a residue of catalytic tetrad and conserved according to multiple sequence alignment. Inspired by this finding, we performed saturation mutagenesis on all the four residues of catalytic tetrad of KmCR. A number of variants with improved enzymatic activities were obtained. Among them, the variant KmCR_A100S exhibited increased catalytic efficiency (k cat /K M = 47.3 s −1 •mM −1 ), improved stereoselectivity (from moderate selectivity (de P = 66.7%) to strict (S)-selectivity (de P > 99.5%)), and extended substrate scope, compared to those of KmCR_WT. In silico analysis showed that a relay system was rebuilt in KmCR via the beneficial residue S100. Furthermore, comparison of 11 protein engineering campaigns indicated that the beneficial position is easily overlooked due to the long distance (>10 Å) from ketone substrates. Since CRs share similar catalytic mechanism, the knowledge gained from this study has universal significance to CR engineering.