A series of novel xanthine derivatives 2a-l incorporating benzoic acid moieties were rapidly generated by using strategy of scaffold-hopping from our previously reported scaffold uracil to xanthine, a scaffold of approved drug linagliptin. After systematic structure-activity relationship (SAR) study around benzoic acid moieties, 5 novel DPP-4 inhibitors with low picomolar potency range (IC<sub>50</sub> < 1 nM) and excellent selectivity against various DPP-4 homologues were identified, in which the best one, compound 2f, with the IC<sub>50</sub> value of 0.1 nM for DPP-4, showed 22-fold improvement in inhibitory activity compared to lead compound uracil 1, its activity was 45-fold more potent than alogliptin. 2e, 2f, 2i and 2k were selected for pharmacokinetic evaluation, and 2f and 2i showed the better pharmacokinetic profiles after iv administration, but poor oral bioavailability. To improve the oral pharmacokinetic profile, prodrug design approach was performed around 2f and 2i. Esters of 2f and 2i were synthesized and evaluated for stability, toxicity and pharmacokinetics. Compound 3e, the methyl ester of compound 2f, was identified to demonstrate good stability, low toxicity and improved oral bioavailability, with 3-fold higher blood concentration compared to 2f in rats. The following in vivo evaluations revealed 3e provided a sustained pharmacodynamics effect for 48h, and robustly improved glucose tolerance in normal ICR and db/db mice in dose-dependent manner. Chronic treatments investigations demonstrated that 3e achieved more beneficial effects on fasting blood glucose levels and glucose tolerance than alogliptin in type 2 diabetic db/db mice. The overall results have shown that compound 3e has the potential to efficacious, safety and long-acting treatment for T2DM.