Development of improved antimicrobial agents will be necessary to combat the prevalence of multi-drug resistant bacteria. A number of biochemical approaches have been taken to identify functionally important hotspots on the ribosome that do not overlap with previously known antibiotic binding sites (Laios et al., 2004
; Llano-Sotelo et al., 2009b
; Yassin et al., 2005
; Yassin and Mankin, 2007
). In addition, a recent study has developed an assay using fluorescently labeled ribosomal proteins to monitor binding of small molecules, such as antibiotics, to the ribosome, which is amenable to high-throughput screening (Llano-Sotelo et al., 2009a
). Here we present several translation-related assays utilizing high-throughput 96- or 384-microtiter plate formats that have been used to screen a library of thiopeptide precursor compounds for their abilities to inhibit one or more aspects of translation and/or reverse the inhibition of known thiopeptide antibiotics. These screens identified four distinct families of precursor compounds, termed PA-PD, which could act as potential lead compounds for development of novel antimicrobials.
Two of the families identified, PA and PD, contain a six-membered nitrogen heterocycle core (PA, dehydropiperidine; PD, pyridine) analogous to the thiopeptide antibiotics thiostrepton and GE2270A (). The crystal structures of thiopeptides bound to the ribosome (Harms et al., 2008
) and of GE2270A bound to EF-Tu (Parmeggiani et al., 2006
) reveal the importance of the heterocycle core of these compounds for interaction with their respective targets, and allows modeling of how PA and PD members are likely to interact with the ribosome and/or EF-Tu ( and ). The resulting models are consistent with the rescue of translation in the presence of ThS and GE2270T by family members, such as PA2 and PD2, probably by direct competition for binding between the precursor compound and the thiopeptide antibiotic. In addition, PA1 and PA2 displayed inhibitory activity against translational GTPases IF2 (), and EF-G and Tet(M) (), respectively. However, compared to the parent thiopeptide compounds, much higher concentrations of the precursor compounds were necessary to exhibit similar effects, most likely indicating the much lower binding affinity of the precursors. The ineffectiveness of precursor compound PA2 as a direct inhibitor was surprising, since this compound has been previously reported to exhibit antimicrobial activity against methicillin-resistance Staphlococcus aureus
and vancomycin-resistant Enterococcus faecalis
with a minimal inhibitory concentration (MIC) of 5 µM (Nicolaou et al., 2005b
). Our results suggest therefore that the inhibitory effect of PA2 in vivo
may in fact not be related to translation, but verification of this point will require further investigation.
The other two families identified in our screen, PB and PC, have not, to our knowledge, been previously reported to target the translational machinery. PB1 is chemically similar to the thiazolidine precursor compound used to generate the pyridine core of amythiamicins (Nicolaou et al., 2008a
), which target EF-Tu analogously to GE2270A (Parmeggiani et al., 2006
; Parmeggiani and Nissen, 2006
). The PC series of compounds contain a protected β-hydroxy-α-aminoacid, which is a precursor in the synthesis of GE2270A/T/C1. Curiously, the PB and PC families display much higher specificity for the ribosome than for EF-Tu, as evidenced by the ability of PB1 and PC1 to restore translation more efficiently in the presence of ThS, as compared with GE2270A (). Although PB1 and PC1 are structurally distinct (), we believe the common aromatic/cyclic nature of both these compounds is important for ribosome binding. Accommodation of EF-G on the ribosome involves the insertion of domain V of EF-G into the crevice between H43/44 and L11-NTD. Inhibition by Class I thiopeptides has been proposed to stem in part from their physically linking L11-NTD to H43/44, thereby locking the cleft shut (Harms et al., 2008
). We suggest that PB1 and PC1 can also span the L11-rRNA crevice () and perform this locking function, analogous to ThS/PA2 () and NoS/PD1 (). Similarly to PA/D, the high concentrations of PB/C required to inhibit the ribosome-dependent GTPase activity of EF-G are indicative of their low binding affinities for the ribosome. Such low affinity may allow facile displacement of precursors from the ribosome, as a result of translation factors (IF2 or EF-G) binding, or from EF-Tu, during ternary complex formation, thus explaining the absence of any direct inhibitory effect of any of the precursors on GFP synthesis. The differential effects of the precursors on the GTPase assays compared to the TT assay is probably related to the ribosome concentrations in the GTPase assays being ~10× – 100× less (30 × 300 nM) compared to the TT assay (~2 µM) and to the putative higher affinity of EF-G for translating rather than empty ribosomes (Sergiev et al., 2005
The majority of clinically used antibiotics targeting the ribosome bind either to the decoding region on the small subunit or within either the peptidyltransferase center or the adjacent peptide exit tunnel of the large subunit, where they interact almost exclusively with ribosomal RNA (Spahn and Prescott, 1996
; Wilson, 2004
). The Class I thiopeptide compounds, however, are distinct in that they target a different region of the ribosome, namely the GTPase-associated region or translation-factor binding site, where they interact with both rRNA and ribosomal protein L11. As a consequence, no cross-resistance has been found between thiopeptide antibiotics and other clinically important drugs. The compounds such as PA-PD identified in our study provide lead structures for the development of novel antimicrobial agents that target this region of the ribosome. Furthermore, the ability of some precursor compounds, such as PA1 and PD1, to bind both EF-Tu and the ribosome suggests the feasibility of developing antimicrobials that are dual inhibitors of ribosomes and ternary complex formation.