RNase P as a Drug Target
The indispensability of RNase P for cell survival and its distinct architecture in Bacteria and Eukarya qualify this ribonucleoprotein enzyme as a potential drug target, although natural inhibitors of bacterial RNase P have not yet been identified. We report on the various attempts pursued so far to explore RNase P as a drug target. After an introduction into the topic and a brief historic synopsis, we will discuss antisense-based strategies, will detail recent advancements with respect to aminoglycoside-arginine conjugates, and will describe in silico-based high-throughput screening procedures that target the bacterial RNase P protein. The reader will be further updated on low molecular weight compounds that inhibit the activity of RNase P from the slime mold Dictyostelium, an amoebic eukaryote that might serve as a model system for some human pathogens. The chapter will finally be closed by mentioning ligands that bind to tRNA substrates as well as the macrolides which were reported to activate bacterial RNase P.
KeywordsAntisense Inhibitor Peptidyl Transferase Central Cleft Bacterial RNase Eukaryotic Pathogen
Historically, the first inhibitor of bacterial RNase P described in the literature was puromycin (Vioque 1989; see Fig. 13.5a). Puromycin inhibited precursor tRNA (ptRNA) processing by Escherichia coli RNase P RNA (P RNA), binding to P RNA with an affinity (K d ∼ 1 mM) only one order of magnitude lower than the drug’s interaction with the ribosome. Next, Kirsebom and coworkers analyzed several aminoglycosides in the ptRNA processing reaction catalyzed by E. coli P RNA (Mikkelsen et al. 1999). Among those tested, neomycin B inhibited the reaction most efficiently (50% inhibition of cleavage activity at 35 μM in the RNA-alone and 60 μM in the holoenzyme reaction), and suppression of Pb2+-induced cleavage of P RNA by neomycin B indicated that the aminoglycoside interferes with the binding of divalent metal ions to the RNA (Mikkelsen et al. 1999). Later, some or all of the six amino groups of neomycin B were converted to guanidinium groups or decorated with lysyl or arginyl groups to improve the inhibitory potential (Eubank et al. 2002; Kawamoto et al. 2008; Sect. 13.3). Apart from aminoglycosides such as neomycin B, paramomycin and kanamycin B, sisomycin and particularly 5-epi-sisomycin were observed to inhibit some RNAs at low micromolar or even submicromolar concentrations (Walter et al. 1999; Mikkelsen et al. 2001).
After the publication of the X-ray structure (Stams et al. 1998) of the Bacillus subtilis RNase P protein (P protein), two pharmaceutical companies became involved in exploring bacterial RNase P. Researchers at Smith Kline Beecham (now GlaxoSmithKline) reported an NMR structural analysis of the P protein from Staphylococcus aureus (Spitzfaden et al. 2000). This study substantiated conclusions drawn from the X-ray structure of the B. subtilis P protein (Stams et al. 1998) in conjunction with crosslinking results (Niranjanakumari et al. 1998), namely that the protein’s central groove, the bottom of which is formed by its β-sheet, functions as a binding interface for single-stranded RNA as present in 5′-leader sequences of ptRNA substrates of RNase P (Spitzfaden et al. 2000). A second putative RNA binding epitope has been localized to helix α2, involving the most conserved region of bacterial P proteins and harbouring the so-called “RNR” motif; this region is thought to interact with P RNA (Stams et al. 1998; Niranjanakumari et al. 2007). Although these investigations laid the structural basis for drug discovery, no further respective efforts by GlaxoSmithKline have become public since then.
The second company that had bacterial RNase P on their agenda of drug targets was Message Pharmaceuticals. Screening compound libraries for inhibition of the in vitro assembled RNase P holoenzyme of Neisseria gonorrhoeae (US patent, Giordano et al. 2006; see also http://statusreports.atp.nist.gov/reports/95-01-0098.htm), this company introduced high-throughput screening into the field of RNase P: processing reactions, performed in microtiter plate format, were terminated by addition of stop buffer containing EDTA for Mg2+ chelation, as well as a DNA oligonucleotide with a 5′-terminal TAMRA fluorescent probe for hybridization to the complementary 10-nt long 5′-leader of the substrate, a precursor tRNAGln from Synechocystis (Pascual and Vioque 1999). The extent of processing was then quantified by fluorescence polarization, exploiting the differences in molecular mobility of tRNA-linked and released 5′-leader. These enzyme activity assays identified the guanylhydrazones as potential inhibitors of bacterial RNase P (see Sect. 13.3). Some of these compounds selectively inhibited growth of certain bacteria, such as S. aureus or Streptococcus pyogenes. However, Message Pharmaceuticals suspended operations in early 2004.
Since these initial attempts, RNase P has been of continuous interest as a drug target, giving rise to several studies that follow up on the aminoglycosides, and explore other compounds as putative drugs. These studies are summarized in the following sections.
13.2 Antisense Inhibitors
The catalytic moiety of bacterial RNase P being a nucleic acid, inhibition of the enzyme by antisense oligonucleotide inhibitors provides a straightforward approach, once suitable target regions within the RNA have been identified. Four approaches have so far been pursued towards antisense inhibition of RNase P: (1) oligonucleotide-directed misfolding (Childs et al. 2003), (2) a rational approach addressing the L15/16 loop of type A P RNAs (Willkomm et al. 2003), (3) computer-based design according to rules established for targeting mRNAs (Willkomm et al. 2003), and (4) a SELEX strategy based on oligonucleotide-induced shifts in gel mobility of P RNA (own unpublished results).
In an independent study, a rational approach (Willkomm et al. 2003) pinpointed the L15/16 loop of E. coli P RNA as a target region particularly amenable to antisense inhibitors. This region was initially chosen because it is exposed within the tertiary structure, forms Watson–Crick base pairs with the substrate, requiring it to be accessible for base pairing interactions, and because it is part of the catalytic core. Starting from an RNA hairpin construct that would allow an initial loop–loop interaction between oligonucleotide and P RNA, our successive optimization of the inhibitor (Willkomm et al. 2003; Gruegelsiepe et al. 2003) led to a single-stranded RNA 14-mer, fully complementary to nucleotides 291–304 of E. coli P RNA (Fig. 13.2a). This RNA oligonucleotide inhibited the RNA-alone reaction with an IC50 value of 2.2 nM, and showed a K d of 0.7 nM for its target site, bound to P RNA over its entire length at the complementary site as predicted, and was sensitive to minor changes within the target sequence, thus qualifying for species-specific applications. Ribozyme inhibition was inferred to occur at four mechanistic levels: (1) direct blockage of base pairing of ptRNA 3′ NCCA ends to the P15 loop, (2) perturbation of the coordination of catalytically relevant Mg2+ [as a consequence of (1)] (Brännvall et al. 2003), (3) disruption of the P15 helix as part of the catalytic core, and (4) arresting the P RNA in an inactive conformation, in line with the study of Childs et al. (2003). Antisense inhibition via the nt 291–304 region of E. coli P RNA was also seen in reactions catalyzed by the holoenzyme reconstituted with the E. coli P protein (Willkomm et al. 2003).
The approach of computer-aided design of antisense inhibitors (Willkomm et al. 2003) employed rules deduced from designing antisense oligonucleotides that target mRNAs (Patzel et al. 1999; Kretschmer-Kazemi Far et al. 2001). After choosing potentially favourable target sites according to available secondary and tertiary structure models of E. coli P RNA, DNA antisense oligonucleotides of 18–20 nucleotides were designed to have their 3′ or 5′ end in the center of the putatively accessible target region. From in vitro RNA-alone reactions in the presence of these oligonucleotides, oligonucleotides targeting the P10-L11/12 region emerged as the most efficient inhibitors (Fig. 13.2a), with inhibition efficiency further increased when RNA variants of these DNA oligonucleotides were used.
Whereas all previous approaches had been restricted to or were successful only for P RNAs of the structural type A, we recently collaborated with J. Kjems (University of Aarhus) to apply a SELEX strategy to both types of P RNA, A and B (unpublished results). For this purpose, a plasmid encoding the P RNA gene was fragmented to generate a T7 expression library in a multistep procedure involving DNA linker ligation, Mme I restriction endonuclease digestion for fragment size standardization and PCR (Jakobsen et al. 2004). T7 transcripts from this library included a central 20-nt stretch of varying sequence derived from the plasmid, flanked with 18 and 28 nt of unvaried linker sequence on the 5′- and 3′-side, respectively. Our selection was then based on excising bands of P RNA that were shifted in the presence of the T7 transcript library on native polyacrylamide gels at Mg2+ concentrations of 2.5 or 5 mM. Excised library transcripts bound to P RNA were amplified by RT-PCR via their 5′- and 3′-terminal linker sequences, followed by the next round of in vitro transcription and gel mobility shift assay. After 3–4 rounds, the enriched pool was sequenced. We thus identified an inhibitory 19-mer sequence complementary to nucleotides 115–133 of E. coli P RNA (Fig. 13.2a) – interestingly the same target region was identified in our previous computer-based screening. Applied to B. subtilis P RNA as a representative of the structural type B, most prominently two overlapping oligonucleotide binders emerged, complementary to the P5.1 stem-loop (nucleotides 52–72 and 58–77 of B. subtilis P RNA; Fig. 13.2b). Unique to type B RNAs, this structural element is known to form a long-range interaction with L15.1, according to the crystal structure of P RNA from Bacillus stearothermophilus (Kazantsev et al. 2005). Indeed, a 20-mer RNA complementary to nucleotides 58–77 (oligo B2.5K5, Fig. 13.2b), displaying an IC50 of ∼40 nM in the RNA-alone reaction and of ∼300 nM in the holoenzyme reaction, disrupted this interaction, and in addition, perturbed P RNA structure upstream of its binding site. Further, this had a long-range effect on L15, reducing the accessibility of G258/G259 involved in base pairing with ptRNA 3′-CCA ends (unpublished results).
An important issue when applying oligonucleotides as therapeutic agents in vivo is rapid degradation of DNA and RNA oligonucleotides. To circumvent this problem, “third generation” antisense agents have been devised that are highly resistant to nucleases (Wahlestedt et al. 2000; Good et al. 2001; Kurreck 2003): LNA (locked nucleic acids, with a methylene bridge between the 2′-oxygen and 4′-carbon atom that preorganizes and fixes the sugar pucker in A-type helical conformation) and PNA (peptide nucleic acid, with the nucleobases attached via methylene carbonyl linkages to an uncharged peptide backbone). PNA has been reported to be able to invade stable stem-loop structures not accessible to natural oligonucleotides (Egholm et al 1993). The thermal stability of LNA-RNA duplexes exceeds that of RNA–RNA helices (Singh and Wengel 1998), which entails the capacity of LNA oligonucleotides to invade stable RNA secondary structures (Nulf and Corey 2004).
To assess the effects of the chemical nature of the antisense agents on inhibition, we compared all-LNA (all residues LNA), all-DNA, all-PNA and all-RNA variants of the 14-mer inhibitor complementary to nucleotides 291–304 of E. coli P RNA, devised in our earlier study (Gruegelsiepe et al. 2006). IC50 values were similar for RNA and LNA (around 3 nM), followed by PNA (13 nM) and DNA (25 nM), the fairly high IC50 for PNA possibly due to some aggregation of PNA oligomers. Helix stability was highest for the all-LNA 14-mer, at the cost of specificity, which was substantially lower for LNA than for PNA or RNA. The association rate k on was highest for PNA (22 × 104 M−1 s−1) vs. 2 × 104 M−1 s−1 for DNA, 7 × 104 M−1 s−1 for RNA, and 9 × 104 M−1 s−1 for LNA, at negligible dissociation rates for all except DNA. The PNA 14-mer also effectively inhibited the E. coli RNase P holoenzyme (Gruegelsiepe et al. 2006; Fig. S4 therein).
Finally, the major hurdle when applying antisense inhibitors to live bacteria is their uptake into the bacterial cell. Here, conjugates of PNA oligonucleotides with invasive peptides, derived from antimicrobial peptides of eukaryal innate immune systems, have been shown to enter Gram-negative and Gram-positive bacteria, and to specifically interact with the target RNA in the cell (Good et al. 2001; Nekhotiaeva et al. 2004). Following this concept, we appended a decamer peptide (KFF)3K to the 14-mer PNA targeting E. coli P RNA, using either 2-amino-ethoxy-2-ethoxy acetic acid (AEEA) or monoglycine (G) for linkage (Gruegelsiepe et al. 2006). At 10 µM concentration of PNA-peptide conjugate in the medium and 180 min incubation time, all cells of the two E. coli test strains (K12 and AS19) were killed when the conjugate was built with the G-linker (PNA-G-peptide); inhibition effects were weaker with the AEEA conjugates. Sequence specificity of the approach was demonstrated in two ways: (1) by using a scrambled version of the PNA-G-peptide, and (2) via growth rescue of PNA-G-peptide-treated E. coli cells by simultaneous expression of a plasmid-encoded B. subtilis P RNA, which was not targeted by the antisense inhibitor owing to sequence and structure variation in the L15 target region. Also, as deduced from RT-PCR experiments, the cellular levels of E. coli P RNA were reduced in E. coli AS19 cells expressing B. subtilis P RNA and treated with the PNA-G-peptide. Since such a reduction was not observed when the same bacteria were treated with the scrambled PNA-G-peptide, it can be concluded that binding of the antisense inhibitor to E. coli P RNA induced the RNA’s degradation.
In conclusion, antisense inhibition appears to be a feasible approach for targeting P RNA. Accessible target regions have by now been identified in P RNAs of structural type A as well as B, and sequence specificity of inhibition has been demonstrated. Studies so far have shown that optimization of antisense inhibitors in terms of length and precise positioning is an important issue. While all types of antisense agents, DNA, RNA, LNA and PNA act as inhibitors, DNA is the least effective of all in vitro. For in vivo applications, to date, PNAs might be considered as the agent of choice due to their nuclease resistance, improved bacterial uptake and increased rates of association with target nucleic acids, the latter two features deriving from the neutral backbone of PNA. Yet LNA, owing to its enhancement of duplex stability, may open perspectives to reduce the length of antisense oligonucleotides in order to improve entry into bacteria. There is encouraging evidence that short LNA-oligomers (8- to 10-mers) maintain sufficient target selectivity without toxic effects (discussed in Gruegelsiepe et al. 2006). Lastly, linkage of the antisense agent to invasive peptides helps to overcome the uptake barrier of the bacterial cell envelope, where not only the type of invasive peptide, but also the nature of the linker appears to play an important role. Despite this progress brought about by the exploration of PNA, bacterial uptake of antisense inhibitors is still inefficient, a large hurdle to be taken before considering this approach for therapeutic purposes.
13.3 Aminoglycosides and Arginine Derivatives
Aminoglycosides bind to RNA molecules via (1) electrostatic interactions between their protonated amino groups and the RNA phosphate backbone, and (2) through shape complementarity combined with the capacity to form specific hydrogen bonds and van der Waals contacts with their RNA targets (Walter et al. 1999). The conformational flexibility of aminoglycosides further permits adaptation to different RNA target geometries. For example, neomycin-type aminoglycosides bind to the bacterial ribosomal A-site in a compact conformation, whereas a more extended conformer is present in the complex with TAR (Tor 2006). In case of the hammerhead ribozyme, a striking overlap in space between the charged amino groups of the aminoglycosides and the metal binding sites in the hammerhead was uncovered, suggesting that these compounds complement the negative electrostatic potential created by certain three-dimensional RNA folds (Hermann and Westhof 1998). Despite the considerable promiscuity of aminoglycosides, attributable to their highly charged nature and their conformational flexibility, they primarily target the ribosomal A-site. This has been explained by the A-site’s encapsulating architecture and its function as a riboswitch of fundamental importance for the cell. Notably, the devastating effect of aminoglycosides on bacterial protein synthesis is achieved despite moderate target affinity, which nonetheless suffices because the aminoglycosides do not have to compete with endogenous high affinity ligands (Tor 2006). With their conformational adaptability and a core decorated with positve charges, the aminoglycosides are a natural paradigm of successful drug design against RNA. This has stimulated numerous attempts to develop aminoglycoside derivatives with diversified target specificity.
In their follow-up investigation, Gopalan and coworkers compared RNase P inhibition by hexa-guanidinium (NeoG6) and -lysyl (NeoK6) conjugates of NeoB next to the penta-arginine derivative NeoR5 (Fig. 13.3a). IC50 values for inhibition of the E. coli RNase P holoenzyme were determined as ca 400, 6, 3 and 0.5 μM for NeoB, NeoG6, NeoK6 and NeoR5, respectively (Kawamoto et al. 2008). For further comparison, NeoR1 showed an IC50 of 4 μM in the same reaction (Berchanski and Lapidot 2008). Neither the presence of 1 mM arginine, lysine or guanidine, nor addition of poly(A) or elevated ptRNA concentrations affected the inhibitory efficacy of the tested AACs in the E. coli RNase P holoenzyme reaction. Inhibition by the most potent of the tested inhibitors, NeoR5, was largely attenuated at elevated Mg2+ concentrations: 1 μM NeoR5 reduced RNase P activity to ca 25% of the activity in its absence under conditions of 10 mM Mg2+, but inhibition was essentially abrogated at 30 mM Mg2+. Similar trends were seen with NeoG6 and NeoK6, substantiating the earlier findings that AACs compete with Mg2+ ions for binding to P RNA (Kawamoto et al. 2008). The 6 to 12 fold lower inhibitory potency of NeoK6 and NeoG6 relative to NeoR5 permits to extract some important inferences regarding the nature of AAC-target RNA interactions: (1) the adverse effect caused by shortening of side chains to only the guanidinium functional groups in NeoG6 vs. NeoR5 illustrates the importance of side-chain flexibility, length and three-dimensional spacing of guanidinium groups; (2) NeoK6 is very similar to NeoR5 in terms of side-chain length and flexibility, but exposes terminal protonated amines instead of the planar, resonance-stabilized guanidinium moieties, suggesting the Arg side chains to be involved in H bonding and possibly π–π interactions in addition to electrostatic interactions (Kawamoto et al. 2008).
The selectivity of AACs was investigated by comparing their effect on reconstituted archaeal RNase P holoenzymes from Methanothermobacter thermautotrophicus (Mth), Pyrococcus furiosus (Pfu) and Methanocaldococcus jannaschii (Mja) RNase P; Mth and Pfu represent the P RNA type A architecture, similar to that of E. coli, whereas Mja is a prototype for the archaeal type M, whose more simplified RNA structure resembles that of eukaryotic P RNAs. NeoR5 inhibited the Mth and Pfu enzymes (reduction of Mth activity to ca 40% at 10 μM NeoR5), but even slightly stimulated the Mja enzyme (to 120%) at the same concentration. It was previously shown that mutations in the tRNA 3′-CCA binding loop L15 of E. coli P RNA, which weaken Mg2+ binding to L15, cause a threefold increase in the IC50 for NeoB (Mikkelsen et al. 1999). This finding pinpointed the L15 loop as a candidate target site for AACs. The absence of a bacterial-like L15 loop in the archaeal Mja RNA would explain the failure to inhibit this enzyme by AACs. Based on these considerations, a deletion variant of E. coli P RNA (ΔL15/P16/P17), lacking the L15/P16/P17 module, was constructed. However, the mutant holoenzyme was inhibited with the same efficiency by NeoR as the wild-type enzyme (Kawamoto et al. 2008). This finding suggested that either the L15 loop is not a major AAC target site or multiple AAC binding sites exist on bacterial P RNA, making inhibition insensitive to the loss of a single target site.
The Lapidot group (Berchanski and Lapidot 2008) further explored the structure-function relationships of AACs in comparison to a new set of aminoglycoside-polyarginine conjugates (APACs). Those included Neam-r9 and Neo-r9, neamine and NeoB derivatives with a D-arginine nonapeptide conjugated to ring I (Fig. 13.3a). The authors pursued a bioinformatic multistep docking approach to predict binding modes of AACs and APACs to B. subtilis P RNA, using the 3D model developed by the Westhof and Gopalan groups (Tsai et al. 2003) because of its good aggreement with the crystallographic data (Kazantsev et al. 2005; Torres-Larios et al. 2005) and since it avoids complications of the crystal structures owing to unresolved regions and crystal packing artifacts. The docking approach involved (1) energy minimization of AAC and APAC 3D structures, (2) geometric, geometric-electrostatic and geometric-hydrophobic Molfit scans, (3) defining the intersection from the three scans to predict putative binding sites, and (4) final refinement of predicted complexes by Discover3 (Berchanski and Lapidot 2008). Based on the docking results, three possible mechanisms of RNase P inhibition by AACs and APACs were suggested: competition with (1) the P protein and (2) ptRNA for binding to the P2/3/4-J19/4 region, which may entail displacement of Mg2+ ions from the P4 helix, and (3) interaction with the P15 region to displace catalytically important Mg2+ ions and to interfere with P protein and ptRNA binding to P RNA. The intermolecular energy was about tenfold lower for the analyzed APACs relative to the AACs, indicating stronger interactions of APACs than AACs with P RNA. Also, the preference for the aforementioned binding sites was less pronounced in the case of the AACs, and even more disperse binding was predicted for NeoB.
In this context, it is interesting to note that the apical guanylhydrazone moieties of the compounds investigated by Message Pharmaceuticals (see also Sect. 13.1) are chemically closely related to the arginine moieties of the AACs and APACs (see Fig. 13.3a and c). Thus, the mode of RNase P inhibition may be similar for the guanylhydrazones and the AACs/APACs.
In summary, the AACs and APACs (and possibly compounds with multiple guanylhydrazone moieties) appear to be promising inhibitors of bacterial RNase P, with up to several orders of magnitude higher affinities compared to their parental aminoglycoside core structures. Even selectivities have emerged: eukaryotic-like RNase P enzymes are much less sensitive to AACs than their bacterial counterparts. It will be intriguing to see how this class of compounds affects the second architectural type of bacterial RNase P, type B, as present in the Firmicutes such as S. aureus.
One potential strategy to increase the selectivity of AACs and APACs may be to constrain their conformation by covalently linking individual rings of the aminoglycoside core, as has been done for neomycin-type antibiotics (Tor 2006). Assuming that the affinity and selectivity of AACs and APACs can be further improved, one eagerly awaits studies on their uptake into bacterial and eukaryotic cells and the phenotypes caused, using assays such as the curing of HeLa cell cultures from E. coli infections (Good et al. 2001). This will help to solve critical questions, such as AAC and APAC selectivity at physiological Mg2+ concentrations (ca 1 mM free Mg2+), and in a cellular context with many potential RNA targets being present simultaneously. Also, it will be interesting to see if toxic effects, such as the oto- and nephrotoxicity of neomycin B, are mitigated or exacerbated for these compounds.
13.4 Structure-Based Drug Design Using the Bacterial P Protein as Target
A virtual screening was performed using the crystal structure of the P protein from B. subtilis (Stams et al. 1998). Two crystal structures and one NMR structure of bacterial P proteins have revealed a conserved three-dimensional architecture despite low primary sequence conservation (Stams et al. 1998; Spitzfaden et al. 2000; Kazantsev et al. 2003). This is in line with the observation that the vast majority of bacterial P proteins are capable of functionally replacing the B. subtilis P protein in vivo (Gößringer and Hartmann 2007). Based on the protein’s topology, the conserved central cleft known to interact with 5′-leader sequences of ptRNA substrates appeared to us as a promising binding region for drug-like molecules, because it is the most pronounced cavity of the molecule and supports an essential RNA-protein interaction. A potential drawback of this strategy may arise from the fact that small compound inhibitors will likely have to compete with the 5′-leaders of cellular ptRNAs for binding to the central cleft. Since structural information on drug-like molecules in complex with bacterial P proteins is not available, the drug-like subset of the ZINC database (Irwin and Shoichet 2005) was chosen as a good starting point to explore possible binding candidates. The docking program GOLD (Verdonk et al. 2003) was used to dock ∼8 million compounds into the central cleft motif of the protein. Default parameters were set as suggested for docking calculations with GOLD. Subsequently, a rescoring of all computed binding geometries was performed using the scoring function DrugScoreCSD (Velec et al. 2005).
13.5 Inhibitors of RNase P from Eukaryotic Pathogens
Several antibiotics and other small ligands have been analyzed for their effect on RNase P. As a putative model system for eukaryotic pathogens, many of these screenings have been performed on partially purified nuclear RNase P from the slime mold Dictyostelium discoideum.
Among tested aminoglycosides, NeoB inhibited D. discoideum RNase P most effectively, an observation also made for bacterial RNase P (Mikkelsen et al. 1999). The aminoglycosides were found to act as classical noncompetitive inhibitors of D. discoideum RNase P, with K i values (as defined by enzyme kinetics) of ca 140 μM, 730 μM, 1.1 mM, 1.4 mM and 1.9 mM for NeoB, gentamycin, tobramycin, paromomycin and kanamycin, respectively (Tekos et al. 2000, 2004). Furthermore, aminoglycosides competed with Mg2+ ions for the same binding sites on the enzyme, a feature that was also observed for nuclear RNase P from human epidermal keratinocytes (Tekos et al. 2003) as well as for bacterial RNase P (Mikkelsen et al. 1999).
Other peptidyl transferase inhibitors – chloramphenicol, spiramycin, lincomycin, acting on prokaryotic ribosomes, and anisomycin, acting on eukaryotic ribosomes – were tested in a similar way, but were without effect on RNase P activity in this test system (Stathopoulos et al. 2000).
Calcipotriol, a synthetic analog of vitamin D3 (Fig. 13.5b) widely used in the treatment of psoriasis and other keratinization disorders, had a dose-dependent effect on RNase P from D. discoideum: calcipotriol acitvated the RNase P activity of D. discoideum cell lysates at concentrations below 50 µM, but was inhibitory at higher concentrations (IC50 = 0.18 mM). Interestingly, neither the parent compounds cholesterol and 7-dehydrocholesterol, nor vitamin D3 analogs very similar to calcipotriol, such as 1α-hydroxycholecalciferol and 1α,25-dihydroxycholecalciferol, displayed any effect on RNase P activity. The precise mode of action is unknown, but clearly independent of vitamin D-receptor mediated transcriptional control (Papadimou et al. 2000a). Similar to calcipotriol, anthralin (dithranol = 1,8-dihydroxy-anthrone; Fig. 13.5b), a derivative of chrysarobine also used for psoriasis treatment, inhibited RNase P in a cell-free D. discoideum system. Again, the mode of action remains unclear, and binding to allosteric inhibition sites as well as a direct effect of anthralin, or of the free radicals it is known to produce, on the RNA component of RNase P has been discussed. Additive inhibitory effects of calcipotriol and anthralin when present simultaneously point to different sites or modes of inhibition (Papadimou et al. 2000b).
Retinoids (Fig. 13.5c) are vitamin A analogs with functions ranging from the role in vision to multiple regulatory processes mediated by nuclear retinoid receptors which act as transcription factors. Their major therapeutic application is in a number of skin disorders. Analyzed for their effect on RNase P of D. discoideum, IC50 values of 80 and 500 μM and K i values of 15 and 1500 µM were determined for all-trans retinoic acid and retinol, respectively. For the synthetic retinoids isotretinoin and acitretin, IC50 was 60 μM and 40 μM and K i = 20 and 8 µM, respectively, with dose response curves similar to those observed for the natural retinoids (Papadimou et al. 1998). Also some arotinoids, third generation retinoids with two aromatic rings, have been tested similarly. By far the most efficient inhibitor among these was Ro13-7410 (Fig. 13.5c) with a K i of 45 µM, and a drastic increase in K i when the terminal carboxyl group was removed or replaced (Papadimou et al. 2000c). Detailed kinetic analysis further revealed that all the above retinoids including Ro13-7410 acted as simple competitive inhibitors with only one retinoid molecule acting on each molecule of RNase P (Papadimou et al. 1998, 2000c). Some of the retinoids were also tested on partially purified nuclear RNase P from human keratinocytes, with K i’s mostly in the same range as for the D. discoideum enzyme (Papadimou et al. 2003).
In conclusion, a number of compounds currently used as therapeutic agents and known for their effect on ribosomes or on intracellular receptors, thus also affect tRNA processing by eukaryotic RNase P. On the one hand, this might contribute to the understanding of the drugs’ modes of action, and on the other hand, this opens up perspectives for the design of drugs specifically targeted at RNase P from eukaryotic pathogens. However, such endeavors require the effect of such drugs on the eukaryotic host to be negligible compared to that on the pathogen, and at present it is not fully clear how close in terms of druggability RNase P from D. discoideum is to that of humans or eukaryotic pathogens. Its usefulness as a model system for either type of RNase P, therefore, requires further investigation. Finally, as with all data obtained from in vitro experiments, verification in vivo is crucial, and here the outcome might substantially differ from that in vitro, particularly in view of the low solubility and high K i’s of some of the compounds.
13.6 Other Small Ligand Effectors
13.6.1 Synthetic Inhibitors Which Act by Binding to the Substrate
Again motivated by evidence for binding of cationic porphyrins to tRNA at sites important for tRNA tertiary structure formation and possibly P RNA binding, the effect of several porphines and porphyrins on the cleavage reaction by E. coli P RNA was analyzed (Hori et al. 2005).
With K i values of 0.8–4.1 µM, the porphyrins T4MPyP (meso-tetrakis(N-methyl-pyridyl)porphine), TMAP (meso-tetrakis[4-(trimethylammonio)phenyl] porphine) and PPIX (Protoporphyrin IX; Fig. 13.6b) are among the strongest small ligand inhibitors of the RNase P reaction described so far. Fluorescence data indicated that 1:1 complexes of these compounds form with E. coli P RNA. Since these compounds bind with similar affinity to P RNA (K d’s of around 50 nM) and ptRNAs or tRNAs (K d’s of 0.1–1.2 µM), inhibition may be based on compound binding to P RNA, to substrate, or to both. Mechanistically, displacement of crucial metal ions, blockage of substrate-P RNA interaction and induction of conformational changes within RNA have been discussed (Hori et al. 2005).
13.6.2 Macrolides as Activators of Bacterial RNase P
Several macrolides tested at concentrations of up to 1 mM did not show any effect on RNase P of D. discoideum (see above). In contrast, the macrolides spiramycin (Fig. 13.6c), erythromycin, tylosin and roxithromycin affected the E. coli holoenzyme and P RNA-alone reaction in the low micromolar range, where they acted as dose-dependent activators (Toumpeki et al. 2008). Detailed analysis of the activation by spiramycin revealed a mixed-type activation mode with, at saturating concentrations of spiramycin, k cat/K s increased 18-fold in the holoenzyme reaction and 12-fold in the RNA-alone reaction. The activation was shown to be independent of pH in the range from 5 to 9, possibly indicative of hydrophobic interactions involved in binding of the macrolide to P RNA. Also, varying the Mg2+ concentration between 10 and 100 mM did not affect activation of the RNA-alone or holoenzyme reaction. There are two lines of evidence regarding the precise mode of spiramycin-mediated activation of RNase P: (1) Kinetic measurements demonstrate that spiramycin affects the catalytic step of the reaction in the RNA-alone as well as the holoenzyme reaction; (2) according to probing data, binding of spiramycin to E. coli P RNA leads to a structural rearrangement of the P10/11 region which is known to be involved in substrate binding, with A124, conserved in bacterial P RNAs, becoming more exposed. This may lead to increased affinity for the substrate and/or facilitated product release (Toumpeki et al. 2008).
13.7 Final Remarks
Various approaches and classes of potential inhibitors have been tested on RNase P as target and vivid research is going on. For some compounds, knowledge on the mode of inhibition has emerged, and further mechanistic insight is expected to be propelled by the recent advancements in our understanding of RNase P structure and function. However, up to now, therapeutic applications for any of these agents are far from being tangible. A fundamental question that arises in this context is whether natural inhibitors of bacterial RNase P exist, but have as yet escaped identification. Since RNase P is essential, differs substantially in bacteria versus eukarya, and has architectural properties that offer many routes to interfere with, one is tempted to entertain the suspicion that natural RNase P inhibitors exist in nature. On the other hand, the highly dynamic and flexible character of the enzyme, and particularly its RNA subunit (Kirsebom 2007), illustrated by the difficulties to obtain well-diffracting P RNA crystals (not to mention crystals of a holoenzyme or enzyme-substrate complex), may have prevented the evolution of RNase P-specific inhibitors. From this point of view, the bacterial P protein may be a more promising target than the RNA subunit. However, the bacterial P protein has characteristics of “intrinsically unstructured” proteins (Henkels et al. 2001) and binds via induced fit to the P RNA (Guo et al. 2006), suggesting that different conformers of the protein coexist in the cell, which might complicate its targetability. In conclusion, RNase P, although a promising drug target, remains a challenging one.
- Bichenkova EV, Sadat-Ebrahimi SE, Wilton AN et al (1998) Strong, specific, reversible binding ligands for transfer RNA: comparison by fluorescence and NMR spectroscopies with distamycin binding for a new structural class of ligand. Nucleosides Nucleotides Nucleic Acids 17:1651–1665CrossRefGoogle Scholar
- Giordano T, Sturgess MA, Rao SJ (2006) Inhibitors of RNase P proteins as antibacterial compounds. Unites States Patent US 7,001,924 B2Google Scholar
- Nulf CJ, Corey D (2004) Intracellular inhibition of hepatitis C virus (HCV) internal ribosomal entry site (IRES)-dependent translation by peptide nucleic acids (PNAs) and locked nucleic acids (LNAs). Nucleic Acids Res 32:3792–3798 Erratum in: Nucleic Acids Res 2004, 32:4954CrossRefPubMedGoogle Scholar
- Schedl P, Primakoff P, Roberts J (1974) Processing of E. coli tRNA precursors. Brookhaven Symp Biol 26:53–76Google Scholar
- Singh SK, Wengel J (1998) Universality of LNA-mediated high-affinity nucleic acid recognition. Chem Commun 12:1247–1248Google Scholar