Beal Research Group
Department of Chemistry, University of Utah
Overview
Research in the Beal Group has focused primarily on the study of members of a fascinating and important family of RNA-binding proteins; those containing the double stranded RNA-binding motif (dsRBM). Members of this protein family bind double helical RNA structures and are critical players in such important processes as the cleavage, editing and trafficking of cellular RNAs. These proteins are also components of the RNA-interference and interferon signaling pathways. Although dsRBM proteins are of obvious functional importance, relatively little is known about their mechanism of action. For instance, the basis for the selectivity observed for certain RNAs is poorly understood as is the interplay between catalytic and RNA-binding domains for enzymes in this family. Furthermore, in contrast to the case of DNA-binding proteins, small molecule reagents capable of binding selectively to duplex RNA and inhibiting dsRBM-containing proteins do not currently exist. These compounds are extremely valuable in further defining the role dsRBM proteins play in the cell as well as allowing for the temporal control of their function. My group is currently studying two different dsRBM-containing proteins; ADAR2, an RNA-editing adenosine deaminase and PKR, the RNA-dependent protein kinase responsible for inhibition of translation during viral infection. These two projects are independently funded by research grants from the National Institutes of Health and are discussed in more detail below. In a third federally funded project described below, we have designed, synthesized and screened libraries of potential inhibitors of RNA function, including inhibitors of dsRBM-RNA binding. Our research program is unique in that we are using chemical approaches to define the mechanism of action for members of this important class of RNA-binding proteins, as well as developing synthetic compounds to control their activity.
ADAR2: An RNA-editing Adenosine Deaminase
ADARs are adenosine deaminases that act on RNA and are responsible for RNA-editing reactions that occur in eukaryotic mRNAs, including the mRNAs of glutamate and serotonin receptors. The deamination of adenosine in the mRNA results in inosine at that position. Since inosine is translated as guanosine, the ADAR reaction can lead to codon changes in the mRNA. (Figure 1) This results in the synthesis of proteins that are structurally and functionally distinct from those encoded in the genome. Our understanding of the molecular basis for these editing reactions is surprisingly limited. For instance, how specific adenosines in the mRNA are targeted for deamination is poorly understood at this time. The goal of this project is to define the molecular mechanism of the RNA-editing adenosine deamination reaction of ADAR2. Ultimately, we seek a complete understanding of how substrate specificity arises and how the chemical transformation proceeds for this important and interesting process.
Initially, we identified RNA substrates for ADAR2 that are accessible by chemical synthesis (1). In this paper, we reported that short duplexes derived from the pre-mRNA encoding a subunit of a glutamate-gated ion channel (GluR-B pre-mRNA) could be prepared using chemical synthesis and that these duplexes were recognized and processed by recombinant ADAR2. This was a critical early result in our project as it allowed us to use phosphoramidite chemistry to make subtle structural changes in the editing substrate. We have since used this approach to define the functional group requirements in the substrate for efficient editing (1 and 8), to introduce a fluorescent nucleotide into the RNA to monitor conformational changes during the reaction (7 and 12) and to prepare RNA with a potential inhibitor of adenosine deamination (11)(Figure 2).
ADAR2 catalyzes a reaction that is similar to that of the well-known nucleoside modifying enzyme adenosine deaminase (ADA). To define further the mechanistic relationship between these two enzymes, we studied the reaction of adenosine analogs of known ADA reactivity in the context of a natural editing site (8). Results of these studies indicate that ADAR2 likely uses a catalytic mechanism similar to that of ADA (i.e. activation of a water molecule for nucleophilic attack on the purine ring). However, the nature of adenosine recognition in the ADAR2 active site is distinct from that of ADA. Therefore, the ADA reaction is an imperfect model for the events occurring in the ADAR2 active site.
Our first generation substrate analogs were prepared with the editing site adenosine at the end of the duplex. Although this makes the enzyme assay more convenient, we found that the deamination reaction of ADAR2 proceeds at a higher rate when the editing site is flanked by duplex RNA (7). Furthermore, we could measure a steady state rate of deamination with this more rapidly processed substrate and show that the steady state and single turnover rate constants were similar in magnitude for this substrate (7). The rate of the ADAR2-catalyzed adenosine deamination was also shown to be sensitive to the sequence context of the adenosine, with a natural editing site favored by nearly two orders of magnitude over a different adenosine in this model substrate. By introducing the fluorescent base 2-aminopurine (2-AP) into the substrate at the editing site and monitoring changes in fluorescence in the presence of enzyme, we showed that the binding of ADAR2 causes a local unstacking of this nucleotide, consistent with a base flipping mechanism (7). The change in 2-AP fluorescence requires the catalytic domain of the enzyme and is localized to the editing site (12). We found that the isolated RNA-binding domain (RBD) of ADAR2 (which contains two dsRBMs) binds a model RNA substrate with high affinity and selectivity (12). Binding of this protein also causes a hypersensitivity to hydrolytic degradation near the editing site, suggesting that the RBD alters the conformational flexibility of the RNA at this site (12).
During these experiments, it became apparent that a variety of purine analogs with various substituents at C6 would be advantageous and that a simple method for generating a versatile intermediate in their synthesis would be preferred. To this end, we developed a new method for the preparation of 6-bromopurine ribonucleoside protected as the triacetyl derivative (2). This compound is a substrate for simple substitution reactions with various nucleophiles and can be used in metal-mediated cross couplings (13) ( Figure 3). We have used this compound in the synthesis of several new phosphoramidites for use in the study of ADAR2 (see for two examples 8 and 11). Furthermore, during this work we realized there was a need for practical methods for the synthesis of N6 and N2 substituted purine ribonucleoside analogs, particularly on solid phase. The development of this synthetic methodology is currently underway in our labs.
We have continued the analysis of substrate analogues to include 8-azaadenosine. Aza substitution at the 8-position is known to facilitate nucleophilic attack on the purine at C6. Indeed, we find that this substitution increases the rate of deamination of our model substrate by 20-fold, suggesting that nucleophilic attack on the purine ring is rate limiting for this substrate. To facilitate structural studies on the ADAR2-RNA complex (in collaboration with the Bass and Hill laboratories, Biochemistry Department, U. of Utah), we are synthesizing and analyzing RNAs modified at the editing site with active site directed inhibitor structures. We recently completed the synthesis of an 8-azapurine ribonucleoside phosphoramidite. Given the observation reported above, 8-aza-substitution of the purine should facilitate the initial attack by an activated water molecule. However, since 8-azapurine does not have a good leaving group at C6, the reaction should stall at this point, allowing for the analysis of the resulting complex. Other potential inhibitor structures are also being studied in the context of duplex RNA (e.g. 6-trifluoromethylpurine and 6-thioinosine).
Our laboratory has been quite successful using EDTA-Fe modification to study the binding selectivity of dsRBM-containing proteins (9, 14, 15, and 17). We are now using this approach to study the binding selectivity of the isolated RNA-binding domain of ADAR2. The results of these experiments are allowing us to determine the extent to which the editing selectivity observed is determined by the intrinsic binding selectivity of the RNA-binding domain.
PKR: An RNA-regulated, Antiviral Protein Kinase
The RNA-dependent protein kinase (PKR) is a component of the interferon signaling system, a collection of pathways that lead to growth inhibition in a number of different cell lines in response to viral infection. PKR is composed of an N-terminal RNA-binding domain with two dsRBMs and a C-terminal protein kinase domain. In vitro, PKR is activated by binding to RNA molecules with extensive duplex secondary structure. In vivo, the enzyme is believed to be activated by viral double stranded RNA (dsRNA) or viral replicative intermediates comprising dsRNA. Activated PKR can phosphorylate several protein substrates, including the alpha subunit of the heterotrimeric eukaryotic translation initiation factor 2 (eIF2a). Phosphorylation of eIF2a has the effect of inhibiting continued initiation of protein synthesis by the eIF2 complex (Figure 4). For the efficient synthesis of its proteins, a virus must inhibit the activity of PKR. Several strategies for viral inhibition of PKR are known, including virally encoded RNA molecules that bind to PKR's RNA-binding domain and block activation. Our efforts in this project have been directed at understanding the binding selectivity of PKR's RNA-binding domain, the differences between activating and inhibiting RNA ligands, the steps in the RNA activation mechanism and ways to control the activity of PKR using synthetic compounds.
By site-specific modification of the RNA binding domain of this enzyme, we introduced diagnostic functional groups at different positions to help define features of PKR-RNA complexes. These reagents allowed us to show that dsRBMI of PKR can bind RNA stem loops with an orientation preference and that PKR has an autophosphorylation site near the protein/RNA interface (3 and 9). This latter discovery led us to investigate the role of autophosphorylation in PKR's RNA binding activity (10). We found that, once autophosphorylated, PKR binds weakly to RNA. This result explained the previously puzzling observation that once activated by autophosphorylation, PKR was no longer regulated by RNA. During these experiments, we noted that PKR, isolated as a highly phosphorylated protein from a bacterial expression system, could be dephosphorylated by treatment with the catalytic subunit of protein phosphatase 1. This generates a form of the enzyme suitable for study of the RNA-activation mechanism (10). We have also shown that triple helix forming RNAs can block the binding of PKR to RNA regulators and thus control the activation state of this important kinase (5).
We have extended our studies of the RNA-binding domain of PKR to include the identification of its binding site on kinase-activating and kinase-inhibiting RNA ligands (14) as well as on an RNA generated by Epstein-Barr virus (the causative agent for mononucleosis) (15). Furthermore, using our binding assay as a screen, we identified ligands in a small combinatorial library of compounds that inhibit PKR-RNA binding (17)(Figure 5).
Inspired by the complex regulation of PKR's enzymatic activity by RNAs and ligands that bind the RNA regulators, we designed and executed a series of experiments to determine if RNA regulators of other enzymes could be discovered that were themselves regulated by small molecule ligands. We showed that RNA molecules can be discovered which bind both a protein and a small molecule in a mutually exclusive fashion, such that the RNA's binding to the protein can be regulated by the small molecule (18). Thus, the small molecule functions to reverse the inhibitory effect of the RNA and induce the protein's activity. We envision ligand-regulated RNA inhibitors as generally applicable tools for the study of protein function, including that of PKR. This approach will be particularly suited for the study of proteins involved in biological phenomena where the timing of molecular events is critical (e.g. the cell cycle), since addition of the inducer gives one precise temporal control of the protein's activity.
Our RNA cleavage data led us to propose molecular models for the complexes formed by the binding of PKR to various RNA ligands. We are carrying out a variety of experiments to test these models. These include the site-specific chemical modification of the RNA at positions predicted by the models to be in contact with PKR. We are also pursuing diffraction quality crystals of PKR-RNA complexes for high resolution structural analysis in collaboration with Prof. Martin Horvath in the Department of Biology.
Several published reports indicate that PKR undergoes a conformational change in the ATP-binding site during activation. To identify reagents that could be used to further study these changes as well as to identify a cell permeable inhibitor of PKR, we have initiated a collaboration with Glaxo/SmithKline to screen a subset of their ATP-binding site directed protein kinase inhibitors for those that potently inhibit PKR. The preliminary results are quite promising and interesting, suggesting that a low molecular weight compound that targets PKR's ATP binding site will be identified. This compound could be modified with a fluorophore to generate a probe of structural changes at the ATP-binding site during the activation process. This approach has precedence in the study of PKC activation using a fluorescently labeled derivative of staurosporine. In addition, a cell permeable, selective inhibitor of PKR would be valuable in the study of signal transduction via the PKR pathway. To this end, we are also preparing libraries of ATP analogues to screen for inhibitors of PKR's kinase activity.
4,9-Disubstituted Acridines: Intercalators with Structurally Diverse Substituents for RNA-specific Targeting
Although we found that triple helix forming oligonucleotides could inhibit the binding of PKR to RNA, the sequence requirements of triple helix formation limits this as a general approach to the inhibition of dsRBM-RNA binding. We are currently investigating a different class of nucleic acid binding molecule for the ability to block RNA function, including the binding of dsRBM proteins. The design of these molecules allows for an acridine unit to direct binding to base paired regions via intercalation. Substituents at the acridine 4- and 9-positions can then make stabilizing contacts to the RNA duplex grooves, adjacent loops or mismatched/bulged nucleotides, giving the molecule the desired selectivity (Figure 7). Without knowing the appropriate substituent to use for a given RNA a priori, we chose to prepare libraries of molecules with this design and select for the functionally interesting structures. Through this work, we are identifying substituents for the acridine 4- and 9-positions that lead to high affinity and selective binding to functionally important RNA structures.
We developed solid phase synthesis strategies to prepare disubstituted acridines (4 and 6)(Figure 8). Ours is the first solid phase synthesis of 4,9-disubstituted acridines where the structure of the 4- and 9-substituents can be varied by split and pool peptide synthesis. We have shown that a 9-anilinoacridine derivative can be used to cap the amino terminus of a peptide, linking the peptide to the acridine through a 4-carboxamide linkage (4). Furthermore, these conjugates can be analyzed by tandem mass spectrometry and their structures readily determined with small amounts of material. We also prepared an acridine-containing amino acid that allows for the solid phase synthesis of peptides bearing an acridine unit as part of the peptide backbone (6). These molecules are designed to bind duplexes via threading intercalation with the N-terminus of the peptide lying in one groove and the C-terminus in the other.
We have carried out selections from libraries of the peptides described above for inhibitors of the function of various RNAs. In one such selection, mixtures of soluble peptides were tested for their ability to inhibit the binding of PKR's RNA-binding domain to a viral inhibiting RNA (VAI RNA from adenovirus) (17)(Figure 9). The most potent mixture was deconvoluted and an inhibitor with an IC50 = 10 mM was identified. Although this compound displays modest potency and poor selectivity, we have shown that its potency is derived from the peptide appendage attached at the acridine 4-position. Thus, the structure of the molecule that is varied by combinatorial synthesis can control the potency of the inhibitor.
To allow for the analysis of more complex libraries, we are screening spatially arrayed library components. However, before this work could be done, we needed to determine if immobilizing the peptide-acridine conjugates affected their binding to RNA. We developed procedures to immobilized these molecules through either their N- or C-termini and an assay to measure RNA binding to the tethered molecules (16). We found that the sequence of the peptide and the point of attachment together control the binding of these molecules.
In development for this project is a screen in which compounds are immobilized on solid phase synthesis beads and fluorescently labeled RNA is used to identify beads bearing high affinity ligands (Figure 10). We have shown that we can identify highly fluorescent beads in such a screen and are currently confirming the "hits" in independent assays.
We are also screening these libraries for inhibitors of the splicing reaction of the group I intron ribozyme found in the large subunit rRNA of the fungal pathogen Candida albicans in collaboration with Douglas Turner's group at the University of Rochester. Approximately 150 acridine-peptides have been screened and several inhibitors have been identified. Toxicity to the fungus is currently being measured for the most potent inhibitors and second generation molecules are being prepared.
In addition to identifying substituents for the acridine 4- and 9-positions that will lead to high affinity binding to specific, predetermined RNA targets, we are carrying out selection experiments with sequence random RNA libraries to identify RNA motifs that bind specifically to fixed sequence acridine-peptides. These experiments will allow us to define the types of RNA structures that can be targeted by this class of molecules. This aspect of our peptide-acridine work was recently funding through a Camille Dreyfus Teacher-Scholar award to PAB.