LINK TO ORIGINAL ARTICLE Payne et al. recently reported an excellent overview of a target-based approach to new antibacterial development and the lack of new antibacterial drugs in late-stage develop- ment several years ago 1 . This observation has also been made by the participants in the recent forum 2 of anti-infective research and development. Also, the Infectious Diseases Society of America recently identi- fied six top-priority dangerous pathogens — extended-spectrum β-lactamase (ESBL)- producing Enterobacteriaceae, Acinetobacter baumannii, Pseudomonas aeruginosa, vanco- mycin-resistant Enterococcus faecium, methicillin-resistant Staphylococcus aureus and Aspergillus species — for which there are few or no drugs in late-stage development. This could further limit future safe and effective choices for treating these infections 3 . Three of these six pathogens are anti- biotic-resistant Gram-negative bacteria. Recently, antibacterial drugs against ESBL- producing Gram-negative bacteria accounted for ~15% (2 out of 13) of all antibacterial drugs undergoing development in Phase II trials or later clinical studies 3 . However, there are no drugs being developed against class C ESBL-producing Gram-negative bacteria. Here, we draw attention to important aspects of urgently needed antibacterial drugs against class C ESBL-producing Gram-negative bac- teria, which have been overlooked by these reports. We also suggest that the category of ESBLs should be expanded. ESBLs are a group of enzymes for which the substrate spectrum has extended to third-generation oxyimino-cephalosporins (for example, cefotaxime and ceftazidime) 4 . Most of the known ESBLs are class A and D β-lactamases, but recently, several class C ESBLs were reported in Gram-negative bac- teria: KL 5 , HD 6 , CMY-10 (REF. 7) and CMY-19 (REF. 8). The hydrolytic efficiency (k cat /K m ) of class C ESBLs for ceftazidime was higher than or similar to that (0.029 µM –1 s –1 ) of SHV-38 (SHV stands for sulphydryl variable) 9 , a typical class A ESBL. Most of the class C β-lactamases have hydrolysing activity against cephamycins (that is, second-generation cephalosporins: cefoxitin and cefotetan), which are not hydrolysed by class A or D ESBLs 4–8 . Cefepime (a fourth-generation oxyimino- cephalosporin) was also inactivated by KL, HD and CMY-19 ESBLs 5,6,8 . Rubinstein and Zhanel have noted that physicians are increasingly being forced to use the carbap- enems (for example, imipenem or mero- penem) and fluoroquinolones (for example, ciprofloxacin or levofloxacin) as first-line therapy for ESBL-producing Gram-negative bacteria; indeed, the situation will become even more severe as ESBL-producing organ- isms increasingly become concomitantly resistant to the fluoroquinolones 2 . However, we recently found that the CMY-10 ESBL had higher imipenem- hydrolysing activity than OXA-23, a class D carbapenemase 7,10 . Gram-negative bacteria producing such class C ESBLs could present a major therapeutic challenge, and so new antibacterial drugs against class C ESBL- producing Gram-negative bacteria are urgently needed. To develop these antibacterial drugs, it is necessary to understand the operative mechanism of class C ESBLs to extend their substrate spectrum. Our kinetic data and crystal structure 7 of a plasmid-encoded class C ESBL (that is, CMY-10) clarify this mechanism. The region responsible for the extended substrate spectrum is the R2-loop (amino-acid residues 289–307) 7 . Our sequence alignment of four class C ESBLs shows that the R2-loop includes all regions responsible for the extended substrate spectrum in all class C ESBLs, com- pared with P99 (a class C non-ESBL) (FIG. 1): • three amino-acid deletion (residues 303–305) of CMY-10 (REF. 7); • four amino-acid deletion (residues 293–296) of HD 6 ; • the single amino-acid substitution (L296H) of KL 5 ; • the single amino-acid substitution (A292S) of CMY-19 (REF. 8). These natural mutations in the R2-loop can change the architecture of the active site in class C ESBLs, thereby affecting their hydrolysing activity. Owing to the deletion in CMY-10, for example, the R2-loop in the R2 active site (that is, the region that accommodates the R2 side-chain at C3 of the β-lactam nucleus in oxyimino-cephalo- sporins) displays noticeable structural alterations. The shortened path of the connection R2-loop between α10 and β11 induces the ~2.5 Å shift of α9 and α10 rela- tive to the adjacent helix α11 in CMY-10 compared with both P99 (REF. 11) and GC1 (REF. 12) β-lactamases, thereby opening the gap between α9–α10 and α11 (REF. 7). Therefore, the bulky R2 side-chain of oxy- imino-cephalosporins could fit snugly into the significant widening of the R2 active site in this way. Clinically available β-lactamase inhibi- tors (for example, clavulanic acid, sulbactam or tazobactam) co-administered with less A lack of drugs for antibiotic- resistant Gram-negative bacteria Jung Hun Lee, Seok Hoon Jeong, Sun-Shin Cha and Sang Hee Lee Figure 1 | A sequence alignment of amino-acid residues near the H-9 (α9) and H-10 (α10) helix of class C β-lactamases with extended substrate spectrum. Alignment among CMY-10 and P99 β-lactamases for which structures are available is performed based on their superimposed structures. The image above the sequence alignment indicates secondary structure annotation of CMY-10. A partial amino-acid sequence alignment of CMY-10 (Enterobacter aerogenes K9911729; GenBank accession no. AF357598; PDB code, 1ZKJ); CMY-19 (Klebsiella pneumoniae HKY466; GenBank accession no. AB194410); HD (Serratia marcescens HD; GenBank accession no. AY336102), KL (Escherichia coli KL; GenBank accession no. AY533244); and P99 (Enterobacter cloacae P99; GenBank accession no. X07274; PDB code, 2BLT) is shown. The R2-loop of residues 289–307 is shaded. CMY-10, CMY-19, HD and KL are class C extended-spectrum β-lactamases (ESBLs), whereas P99 is a class C non-ESBL. NATURE REVIEWS | DRUG DISCOVERY www.nature.com/reviews/drugdisc CORRESPONDENCE © 2007 Nature Publishing Group