Rational Design of Oncocin Derivatives with Superior Protease Stabilities and Antibacterial Activities Based on the High‐Resolution Structure of the Oncocin‐DnaK Complex

Despite the success story of antibiotics, which began nearly a hundred years ago, bacterial infections are still a major cause of death worldwide. The emergence of multiple-drug-resistant (MDR) bacterial pathogens in hospitals (nosocomial infections) presents a global problem of growing importance, with an estimated annual death toll of 50 000 in the EU and 60 000 in the USA. More recently, MDR bacteria have also caused severe community-acquired infections, indicating that we will soon face bacterial strains with the ability to overcome existing antibiotic treatments, and which will therefore represent a severe global threat. Agents responsible for important resistance mechanisms in Gram-negative bacteria include extended spectrum b-lactamases (ESBLs) in Enterobacteriaceae (e.g. , E. coli, K. pneumoniae, and Enterobacter cloacae) or broad-range b-lactamases (e.g. , KPC in Klebsiella pneumoniae or metallo-b-lactamases in Pseudomonas aeruginosa). MDR Acinetobacter baumannii, associated with invasive infections such as pneumonia, meningitis, and bacteremia, has been found to be responsible for outbreaks in intensive care units, including “panresistant” A. baumannii clones susceptible only to polymyxin. To provide effective future treatment options, novel antimicrobial drug classes with novel modes of action are urgently needed. Inducible, gene-encoded antimicrobial peptides (AMPs) represent such a promising alternative, having been selected and optimized by evolution over millions of years. Although AMPs that kill bacteria by lytic effects on the membrane are often toxic to human cells at higher doses, the class of small prolinerich AMPs (PR-AMPs) expressed in mammals and insects has attracted considerable interest. These peptides specifically target intracellular components in Gram-negative bacteria with no indication of any resulting toxic side effects. Despite their favorable antibacterial spectrum against Enterobacteriaceae and nonfermenting species (e.g. , A. baumannii and P. aeruginosa), there are multiple obstacles to be overcome in their further development for therapeutic consideration. We have recently used rational design to optimize the 19-residue-long PR-AMP oncocin (VDKPPYLPRPRPPRRIYNR-NH2) as a potential means of countering the five human pathogens discussed. Substitution of Arg15 and Arg19 by ornithine drastically improved the half-life in mouse serum. Mechanistically, oncocin freely penetrates the bacterial membrane and distributes homogenously within E. coli cells. Here we report its further optimization, based on a positional Ala-scan to deduce residues critical to its antibacterial activity and the crystal structure of an oncocin-DnaK (ligand–target) complex. The new lead compounds were highly resistant against serum (t1=2 8 h) and E. coli proteases (t1=2 >10 h). The mode of action assumed for oncocin and all other PRAMPs has not been worked out in detail, but most likely consists of at least three steps: passive penetration of the bacterial outer membrane, active transport from the periplasmic space into the cytoplasm, and inhibition of DnaK and maybe other targets. 12] A lead optimization strategy therefore has to consider all aspects and cannot focus only on the target binding. We first identified residues crucial for the antibacterial activity of oncocin by determining the minimal inhibitory concentrations (MICs) and inhibition zones for all 19 peptides resulting from a positional Ala-scan (see Figure S1 in the Supporting Information). Substitution of residues 1, 2, 4, 5, 10, and 12–19 reduced the antibacterial activity slightly, whereas substitutions at positions 3, 6–9, and 11 abolished it almost completely. Although the target of oncocin has not yet been identified, the high sequence homology to other insect-derived PR-AMPs, especially pyrrhocoricin, suggests that it might be the bacterial chaperone DnaK. Full-length DnaK was therefore expressed in E. coli and purified in order to study oncocin binding by fluorescence polarization. The binding constants for oncocin and its analogue O2, with a 5(6)-carboxyfluorescein unit at the N terminus, were 1.0 0.2 mm and 4.0 1.0 mm, respectively, whereas all-d oncocin did not bind (Table 1, Figures S2 and S3). These values correspond to binding constants reported for other DnaK-binding sequences, ranging from 0.1 to 10 mm. Cocrystallization of oncocin O2 with the substrate binding domain of DnaK (residues 389 to 607), demonstrated that oncocin residues 4 to 10 (PPYLPR) bound to the peptide binding site of DnaK, whereas the remaining terminal residues of the peptide were flexible (Figures 1 and S4–S8). Interestingly, this sequence stretch matched the residues identified by the Ala-scan as crucial for antibacterial activity. The antibacterial activity of oncocin thus mostly depends on its DnaK binding site, although the activity is also abolished by shortening the sequence, either from the N or the C terminus. Such inactivation, which can occur through the action of proteases either in the bacteria or in blood, should be minimized for systemic applications. Because the peptide O2 is rel[a] D. Knappe, M. Zahn, Prof. Dr. N. Str ter, Prof. Dr. R. Hoffmann Institut f r Bioanalytische Chemie Biotechnologisch-Biomedizinisches Zentrum Fakult t f r Chemie und Mineralogie, Universit t Leipzig Deutscher Platz 5, 04103 Leipzig (Germany) Fax: (+ 49) 341-97-31339 E-mail : hoffmann@chemie.uni-leipzig.de [b] U. Sauer, Dr. G. Schiffer AiCuris GmbH & Co KG Friedrich-Ebert-Strasse 475, Building 302, 42117 Wuppertal (Germany) Supporting information for this article is available on the WWW under http ://dx.doi.org/10.1002/cbic.201000792.