The application of rational design in reallocating metabolic flux to overproduce desired chemicals is always restricted by the native regulatory network. Here, we demonstrated that in vitro modular pathway optimization combined with in vivo multiplexed combinatorial engineering enables effective characterization of the bottleneck of a complex biosynthetic cascade and improves the output of the engineered pathway. As a proof of concept, we systematically identified the rate-limiting step of a five-gene malate biosynthetic pathway by combinatorially tuning the enzyme loads of a reconstituted biocatalytic reaction in a cell-free system. Using multiplexed CRISPR interference, we subsequently eliminated the metabolic constraints by rationally assigning an optimal gene expression pattern for each pathway module. The present engineered strain Escherichia coli B0013-47 exhibited a 2.3-fold increase in malate titer compared with that of the parental strain, with a yield of 0.85 mol/mol glucose in shake-flask culture and titer of 269 mM (36 g/L) in fed-batch cultivation. The strategy reported herein represents a powerful method for improving the efficiency of multi-gene pathways and advancing the success of metabolic engineering.