Bacterial resistance to beta-lactam antibiotics, a clinically worrying and recurrent problem, is often due to the production of beta-lactamases, enzymes that efficiently hydrolyze the amide bond of the beta-lactam nucleus. Imipenem and other carbapenems escape the activity of most active site serine beta-lactamases and have therefore become very popular drugs for antibacterial chemotherapy in the hospital environment. Their usefulness is, however, threatened by the appearance of new beta-lactamases that efficiently hydrolyze them. This study is focused on the structure and properties of two recently described class A carbapenemases, produced by Serratia marcescens and Enterobacter cloacae strains and leads to a better understanding of the specificity of beta-lactamases. In turn, this will contribute to the design of better antibacterial drugs. Three-dimensional models of the two class A carbapenemases were constructed by homology modeling. They suggested the presence, near the active site of the enzymes, of a disulfide bridge (C69-C238) whose existence was experimentally confirmed. Kinetic parameters were measured with the purified Sme-1 carbapenemase, and an attempt was made to explain its specific substrate profile by analyzing the structures of minimized Henri-Michaelis complexes and comparing them to those obtained for the "classical" TEM-1 beta-lactamase. The peculiar substrate profile of the carbapenemases appears to be strongly correlated with the presence of the disulfide bridge between C69 and C238.