Nitrogen-containing heterocycles
(N-heterocycles) are essential
in drug discovery, with a majority of best-selling small-molecule
drugs containing at least one N-heterocycle. Various enzymes have
been characterized or engineered for N-heterocycle biosynthesis, but
these attempts are frequently restricted to specific N-heterocycle
skeletons and substitutes. Multidomain carboxylic acid reductase (CAR)
is gaining attention for its ability to reduce amino acids to amino
aldehydes, which can be converted into a variety of N-heterocycle
skeletons through nonenzymatic dimerization and rearrangement. This
unique cyclization process could expand the biocatalytic potential
of CARs beyond those of specific N-heterocycles. Here, we applied
domain-based genome mining to identify 32,735 CAR enzymes, providing
the most comprehensive landscape of CARs to date. Through in vivo and in vitro assays, we characterized
an untapped CAR family that exhibited substrate tolerance on amino
acids and efficiently catalyzed the formation of various N-heterocycles
from amino acids. By harnessing the potential of promiscuous CARs
for one-step biosynthesis and diversification of N-heterocycles, we
created a library of N-heterocycles for bioassays, identifying multiple
quorum sensing (QS) inhibitors and protease inhibitors. Our discovery
of untapped promiscuous CARs offers a promising alternative for the
biosynthesis and diversification of N-heterocycles, making them a
valuable resource for chemical and synthetic biologists.