Inhibitors targeting bacterial enzymes should not interfere with enzymes of the host, and knowledge about structural determinants for selectivity is important for designing inhibitors with a therapeutic potential. We have determined the binding strengths of two hydroxamate compounds, galardin and compound 1b for the bacterial zinc metalloproteases, thermolysin, pseudolysin and auerolysin, known to be bacterial virulence factors, and the two human zinc metalloproteases MMP-9 and MMP-14. The active sites of the bacterial and human enzymes have huge similarities. In addition, we also studied the enzyme-inhibitor interactions by molecular modelling. The obtained Ki values of galardin for MMP-9 and MMP-14 and compound 1b for MMP-9 are approximately ten times lower than previously reported. Compound 1b binds stronger than galardin to both MMP-9 and MMP-14, and docking studies indicated that the diphenyl ether moiety of compound 1b obtains more favourable interactions within the S´1-subpocket than the 4-methylpentanoyl moiety of galardin. Both compounds bind stronger to MMP-9 than to MMP-14, which appears to be due to a larger S´1-subpocket in the former enzyme. Galardin, but not 1b, inhibits the bacterial enzymes, but the galardin Ki values were much larger than for the MMPs. The docking indicates that the S´1-subpockets of the bacterial proteases are too small to accommodate the diphenyl ether moiety of 1b, while the 4-methylpentanoyl moiety of galardin enters the pocket. The present study indicates that the size and shape of the ligand structural moiety entering the S´1-subpocket is an important determinant for selectivity between the studied MMPs and bacterial MPs.

Compounds containg catechol or bisphosphonate were tested as inhibitors of the zinc metalloproteases, thermolysin (TLN), pseudolysin (PLN) and aureolysin (ALN) which are bacterial virulence factors, and the human matrix metalloproteases MMP-9 and À14. Inhibition of virulence is a putative strategy in the development of antibacterial drugs, but the inhibitors should not interfere with human enzymes. Docking indicated that the inhibitors bound MMP-9 and MMP-14 with the phenyl, biphenyl, chlorophenyl, nitrophenyl or methoxyphenyl ringsystem in the S 1 0-subpocket, while these ringsystems entered the S 2 0-or S 1-subpockets or a region involving amino acids in the S 1 0-and S 2 0-subpockets of the bacterial enzymes. An arginine conserved among the bacterial enzymes seemed to hinder entrance deeply into the S 1 0-subpocket. Only the bisphosphonate containing compound RC2 bound stronger to PLN and TLN than to MMP-9 and MMP-14. Docking indicated that the reason was that the conserved arginine (R203 in TLN and R198 in PLN) interacts with phosphate groups of RC2.