Minireview Approaching the Golden Fleece a Molecule at a Time: Biophysical Insights into Argonaute-Instructed Nucleic Acid Interactions Veronika A. Herzog1 and Stefan L. Ameres1,* 1Institute of Molecular Biotechnology of the Austrian Academy of Sciences (IMBA), Dr. Bohr-Gasse 3, 1030 Vienna, Austria *Correspondence: [email protected]
Argonaute proteins act at the core of nucleic acid-guided interference pathways that regulate gene expression and defend organisms against foreign genetic elements in all domains of life. Here, we review recent biophysical studies on how Argonaute proteins instruct oligonucleotides in the process of target finding, binding, cleavage, and release, as measured at high spatiotemporal resolution by single-molecule approaches. In the context of previous structural, biochemical, and computational studies, a model emerges for how Argonaute proteins manipulate the thermodynamic rules for nucleic acid hybridization to convey efficiency and specificity to RNA- and DNA-guided regulatory processes. Sequence-specific recognition of nucleic acids is of fundamental importance to the regulation of gene expression and the defense against foreign genetic elements. The versatility of short oligonucleotides to form specific interactions via base pairing provides an attractive means to attain such specificity. In that respect, the discovery of 20 to 30 nucleotide (nt) small RNAs that mediate the sequence-specific regulation of genes through pathways—collectively referred to as RNA interference (RNAi)—represents a landmark in modern molecular biology. It revolutionized our view on remarkably diverse biological processes, established new experimental tools, and prompted entirely novel therapeutic approaches. Several classes of small RNAs, as defined by their distinct biogenesis and function, have been described in plants, animals, fungi, and viruses (Ghildiyal and Zamore, 2009): for example in animals, microRNAs (miRNAs) act as key regulators of development, physiology, and disease through a pathway that typically involves translational repression and destabilization of mRNAs, triggered by binding of miRNAs via partial complementary; and small interfering RNAs (siRNAs) initiate the endonucleolytic cleavage of highly complementary sites to defend some organisms against foreign genetic elements, such as viruses and transposons. In a unifying theme across all RNAi pathways, small RNAs function through association with proteins of the Argonaute (AGO) family. Together, AGO and a single-stranded small RNA form the RNA-induced silencing complexes (RISCs). Biochemical, structural, and computational studies suggest that an AGO protein divides the small RNA into five domains with distinct functions (Bartel, 2009; Sasaki and Tomari, 2012; Wee et al., 2012). (1) The 50 anchor (nt 1) clings the small RNA to a specific binding pocket in the MID domain of AGO and facilitates small RNA loading. (2) The seed sequence (nt 2–8) initiates target binding and is the primary determinant of binding specificity; seedbinding is frequently sufficient to mediate silencing by miRNAs. (3) The central region (nt 9–12) needs to base pair in order to initiate siRNA-directed endonucleolytic cleavage catalyzed 4 Molecular Cell 59, July 2, 2015 ª2015 Elsevier Inc.
by the AGO protein itself. (4) The 30 supplementary region (nt 13–16) is thought to complement seed pairing for some miRNA targets. (5) The 30 tail region (nt 17 and onward) anchors the 30 end of the small RNA to AGO via the PAZ domain. While crystallographic studies have provided valuable insights into the structural organization and rearrangements of AGO proteins, frozen in specific states (i.e., empty and bound to guide strand, without and with target with increasing complementarity), the kinetic aspects of these transitions remained unknown. Biochemical studies have filled many gaps but merely report ensemble averages of a population of molecules. Such bulk studies are inherently prone to neglecting transient intermediate states, heterogeneity in population, and parallel reaction pathways. These problems can be overcome by the use of biophysical approaches that trace the motion of individual molecules at biologically relevant resolution (sub-nanometer) and timescale (microseconds to minutes) (Joo et al., 2008). Four recent studies employed fluorescently labeled guide and target nucleic acid molecules by total internal reflection fluorescence (TIRF) microscopy in order to monitor the process of nucleic acid hybridization in the context of mammalian, fly, and bacterial AGO proteins (for an overview of experimental setups, see Figure 1), providing remarkably detailed insights into the mechanisms of RNAi-related processes (Salomon et al., 2015; Chandradoss et al., 2015; Jo et al., 2015; Yao et al., 2015). Target Finding Finding complementary targets with high specificity and efficiency among the vast array of RNA molecules in a cell is a major challenge for RISC. Biochemical and structural studies indicated that RISC circumvents the kinetic and thermodynamic pitfalls of ‘‘naked’’ nucleic acid oligomers, which readily form promiscuous interactions, by providing a scaffold that exposes only the short seed sequence to the solvent while preventing residual nucleotides from pairing (Tomari and Zamore, 2005; Wang et al., 2009). The arrangement of the seed in an A-form-like helical geometry by AGO has been proposed to make productive
Figure 1. Overview of Experimental Strategies to Characterize the Biophysical Properties of RISC by Single-Molecule Approaches (A) In total internal reflection fluorescence (TIRF) microscopy (left), fluorophores directly adjacent (100 nm) to the glass-water interface are selectively illuminated by an evanescent excitation field. To monitor binding and dissociation of AGO proteins, target RNAs are tethered to a polymerized glass surface by biotinstreptavidin interactions. The free ends of the target RNA were labeled with a fluorophore (see also B). A second fluorophore in the guide oligonucleotide bound to an AGO protein labels RISC complexes. A representative microscopy snapshot (right) shows the localization of a RISC-free target RNA (in red) or the target RNA bound by RISC (in green) resulting in a yellow spot, while unbound RISC molecules in solution are not detected. Each spot is recorded over time to obtain binding and dissociation constants. (B) Overview of the experimental setup used in the indicated studies. In all cases, a labeled target RNA was tethered to the glass surface and RISC was labeled with a fluorophore in the guide RNA at the indicated position. Note that in some instances the fluorophore and the biotin group at the target were reversed to tether either the 50 or 30 end to the glass surface (Salomon et al., 2015; Yao et al., 2015) or the two fluorophores were exchanged between guide RNA and target RNA (Yao et al., 2015). Binding events were either detected by co-localization of the two fluorophores or by fluorescence resonance energy transfer (FRET). Length and composition of the targets, as well as identity and position of the fluorophores and source of the AGO proteins (m, mouse; Tt, Thermus thermophilus; Dm, Drosophila melanogaster; h, human) are indicated. (C) Exemplary illustration of a fluorescence intensity time track for a single target molecule. Changes in fluorescence states upon RISC binding, dissociation, target cleavage, and product release from RISC using a 50 -labeled target (red) and a 30 -labeled guide RNA (green) are indicated. Parallel recordings of several hundred events were quantified to calculate kinetic parameters.
collisions with targets more likely, because it reduces the entropic cost inherent to base pairing (Parker et al., 2009); but previous ensemble studies could only indirectly assess the rate of association (kon) (Deerberg et al., 2013; Wee et al., 2012). Single-molecule studies now reveal that AGO proteins accelerate target RNA binding by up to 250-fold, when compared to the kinetics of protein-free nucleic acids (Salomon et al., 2015;
Chandradoss et al., 2015; Jo et al., 2015). Target binding is initiated by interactions involving as little as three nucleotides within nt 2 to 5 of the guide, with increasing importance toward the 50 end of the guide (Salomon et al., 2015; Chandradoss et al., 2015). This indicates that the actual ‘‘seed’’ for target binding is smaller than previously annotated. Structural studies already anticipated a sub-division of the seed sequence because in Molecular Cell 59, July 2, 2015 ª2015 Elsevier Inc. 5
Minireview human AGO2 the guide nt 6 to 8 are prevented from pairing by a conserved alpha helix (helix 7 in human AGO2), which also disrupts the helical geometry of the seed beyond nt 5 (Schirle et al., 2014). By reducing the region involved in establishing the initial contact to targets, RISC is able to obtain on rates (up to 4 3 108 M1 s1) that approach the limits of macromolecular diffusion (109 M1 s1), suggesting that target finding is merely limited by the chance of two molecules to collide in solution (Salomon et al., 2015; Chandradoss et al., 2015; Jo et al., 2015). Chandradoss et al. also propose that transient subseed-interactions might enable lateral diffusion of RISC along RNA molecules, thereby temporarily reducing the search-space from three to one dimension (Chandradoss et al., 2015). In theory, such a scanning process could accelerate target finding by more than two orders of magnitude (Tafvizi et al., 2011). While this model finds some support in recent studies on the binding of transcription factors to DNA, it is challenged by the fact that RISC:target associations—even if transient—have not yet been observed to exceed the limits of macromolecular diffusion, as expected if lateral diffusion were to be involved. Nevertheless, the low affinity of sub-seed interactions is most certainly beneficial to the overall target search kinetics in vivo because it minimizes the time RISC spends on a plethora of unintended sites in the transcriptome (Chandradoss et al., 2015; Salomon et al., 2015). Target Binding Once identified, RISC needs to stay bound to its target long enough, without falling off, in order to initiate regulation. This is particularly relevant for miRNA-mediated gene silencing in animals, which involves the subsequent recruitment of co-factors (Huntzinger and Izaurralde, 2011). Single-molecule analyses highlight the importance of complete seed pairing (i.e., nt 2 to 8) for stable target-binding by miRNAs (Bartel, 2009; Wee et al., 2012): while RISC dissociated from sub-seed sites within seconds, pairing of nt 2 to 8 persisted for 5 min (Chandradoss et al., 2015; Salomon et al., 2015). In fact, when compared to naked RNA, miRNAs bind 4 million times tighter and dissociate 14,000 times slower from seed-matched sites in the context of AGO, enabling RISC to bind with a strength (KD 10 nM) and motif size (7 nt) that is similar to high-affinity RNA-binding proteins (Wee et al., 2012; Salomon et al., 2015). At the same time, mismatches within the seed impact binding strength more than predicted by thermodynamic rules, with greater effects toward the 50 end of the seed, enabling effective proofreading of seed interactions (Salomon et al., 2015; Chandradoss et al., 2015). The nt 6 to 8 contribute less to this control, presumably because their binding requires a conformational rearrangement of helix 7 in AGO (see above) (Schirle et al., 2014). Notably, also the extension of seed pairing by one or two nucleotides (involving nt 9 and 10) destabilized binding (Salomon et al., 2015), perhaps because it induces the opening of a central cleft in AGO (Schirle et al., 2014). For some animal miRNA-targets, base pairing to the 30 supplementary region (nt 13–16) of the guide productively augments seed binding: such sites tend to result in more regulation and increase the probability to predict true targets (Grimson et al., 2007). Biophysical studies now indicate that 30 supple6 Molecular Cell 59, July 2, 2015 ª2015 Elsevier Inc.
mentary binding appears to be most relevant to miRNAs with weak (i.e., AU-rich) seed pairing (e.g., miR-21), for which 30 supplementary binding increases affinity, while this was not observed for a miRNA with GC-rich seed (e.g., let-7) (Salomon et al., 2015). Most miRNA binding sites tend to act independently, but closely spaced sites result in more repression than expected from the independent contributions of each individual site (Bartel, 2009). Along these lines, Chandradoss et al. showed that neighboring sites for the same miRNA resulted in higher than expected residence times of RISC on the target, perhaps because RISC is able to shuttle between closely spaced sites (Chandradoss et al., 2015). But how miRNAs with different seed sequence can function cooperatively remains unclear. MicroRNA-mediated gene regulation in animals occurs in the cytoplasm, but several eukaryotic RNAi-related pathways rely on RISC to find their targets in the nucleus, triggering diverse processes such as transcriptional gene regulation and DNA elimination (Schraivogel and Meister, 2014). Although the prevailing models for targeting in these cases propose AGO-assisted interactions with nascent RNA transcripts, direct targeting of DNA has not been excluded experimentally. Interestingly, Salomon et al. report that mammalian RISC binds to single-stranded DNA at rates comparable to RNA, but rapidly falls off again, with dissociation kinetics that are >100-fold faster for DNA than for RNA (Salomon et al., 2015). This is in line with the proposed targeting specificity of animal AGOs for RNA. However, the bacterial AGO from Thermus thermophilus bound to and released RNA and DNA at similar rates (Salomon et al., 2015), consistent with its ability to target foreign DNA upon horizontal gene transfer in vivo (Vogel, 2014). The different targeting specificities might originate from sequence independent contacts of mammalian AGO to its bound target, employing shape complementarity to the minor grove of the duplex, a feature that is absent in TtAGO (Schirle et al., 2014; Wang et al., 2009). Target Cleavage and Release In contrast to miRNAs, siRNAs bind to highly complementary sites. Depending on the AGO protein, this conformation results in endonucleolytic cleavage of the target, because it induces structural rearrangements that position a catalytic tetrad, present in the RNaseH-like fold of AGO’s PIWI domain, close to the scissile phosphate of the target across the guide nt 10 and 11, where cleavage occurs (Wang et al., 2009; Nakanishi et al., 2012; Sheng et al., 2014; Elbashir et al., 2001). In this scenario RISC can act as a multiple turnover enzyme (Wee et al., 2012), but how the removal of the cleaved product in between each cleavage cycle is coordinated could so far not be addressed. Single-molecule studies now show that—in contrast to the directionality of siRNA binding, which always occurs from 50 to 30 (Chandradoss et al., 2015; Salomon et al., 2015; Jo et al., 2015; Yao et al., 2015)—the release of the products follows no strict order but largely depends on the relative base-pairing stability of the fragments (Yao et al., 2015; Salomon et al., 2015). Irrespective of the order, dissociation occurs faster than expected from the stability of the individual duplexes, and release of either cleavage products accelerates removal of the second, indicating that AGO generates an environment that actively promotes the
Minireview release of the cleavage products (Salomon et al., 2015; Jo et al., 2015).
Huntzinger, E., and Izaurralde, E. (2011). Gene silencing by microRNAs: contributions of translational repression and mRNA decay. Nat. Rev. Genet. 12, 99–110.
Conclusions Together, these single-molecule studies provide detailed insights into one of nature’s remarkable solutions to the thermodynamic pitfalls that proteins encounter when utilizing nucleic acids as specificity determinants. On a practical note, the systematic quantitative assessment of RISC’s biophysical properties will certainly add to existing de novo prediction algorithms for miRNA targets. While the unusual efficiency at which minimal RISC in isolation finds, binds, cleaves, and releases its targets questions the contribution of accessory factors to many of these processes, further studies in the more complex context of a living cell are required. From an evolutionary point of view, the sophisticated scaffold of AGO proteins appears remarkably successful since this protein family has been consulted for many biological problems across all domains of life (Swarts et al., 2014; Ghildiyal and Zamore, 2009). But other protein frameworks exist that also employ nucleic acids as specificity determinants, the bacterial CRISPR-Cas effector complexes being one of the most recently described additions (Heidrich and Vogel, 2013). Notably, also Cas9 appears to employ sub-seed-like interactions to interrogate target sites (Sternberg et al., 2014), perhaps hinting at a common theme in nucleic acid-guided processes. Finally, recent single-molecule studies have already advanced our molecular understanding of RNA silencing pathways beyond the function of AGO proteins, such as the biogenesis of small RNAs and their loading into AGO (Nguyen et al., 2015; Iwasaki et al., 2015) and will certainly reveal more insights in the near future.
Iwasaki, S., Sasaki, H.M., Sakaguchi, Y., Suzuki, T., Tadakuma, H., and Tomari, Y. (2015). Defining fundamental steps in the assembly of the Drosophila RNAi enzyme complex. Nature 521, 533–536.
ACKNOWLEDGMENTS Work in the S.L.A. lab is supported by the Austrian Science Fund FWF (Y-733B22 START and W1207- B09) and the European Research Council (ERC-StG338252 miRLIFE). REFERENCES Bartel, D.P. (2009). MicroRNAs: target recognition and regulatory functions. Cell 136, 215–233. Chandradoss, S.D., Schirle, N.T., Szczepaniak, M., MacRae, I.J., and Joo, C. (2015). A dynamic search process underlies microRNA targeting. Cell 162, http://dx.doi.org/10.1016/j.cell.2015.06.032. Deerberg, A., Willkomm, S., and Restle, T. (2013). Minimal mechanistic model of siRNA-dependent target RNA slicing by recombinant human Argonaute 2 protein. Proc. Natl. Acad. Sci. USA 110, 17850–17855. Elbashir, S.M., Martinez, J., Patkaniowska, A., Lendeckel, W., and Tuschl, T. (2001). Functional anatomy of siRNAs for mediating efficient RNAi in Drosophila melanogaster embryo lysate. EMBO J. 20, 6877–6888. Ghildiyal, M., and Zamore, P.D. (2009). Small silencing RNAs: an expanding universe. Nat. Rev. Genet. 10, 94–108. Grimson, A., Farh, K.K., Johnston, W.K., Garrett-Engele, P., Lim, L.P., and Bartel, D.P. (2007). MicroRNA targeting specificity in mammals: determinants beyond seed pairing. Mol. Cell 27, 91–105. Heidrich, N., and Vogel, J. (2013). Same same but different: new structural insight into CRISPR-Cas complexes. Mol. Cell 52, 4–7.
Jo, M.H., Shin, S., Jung, S.-R., Kim, E., Song, J.-J., and Hohng, S. (2015). Human Argonaute 2 has diverse reaction pathways on target RNAs. Mol. Cell 59, this issue, 117–124. Joo, C., Balci, H., Ishitsuka, Y., Buranachai, C., and Ha, T. (2008). Advances in single-molecule fluorescence methods for molecular biology. Annu. Rev. Biochem. 77, 51–76. Nakanishi, K., Weinberg, D.E., Bartel, D.P., and Patel, D.J. (2012). Structure of yeast Argonaute with guide RNA. Nature 486, 368–374. Nguyen, T.A., Jo, M.H., Choi, Y.-G., Park, J., Kwon, S.C., Hohng, S., Kim, V.N., and Woo, J.-S. (2015). Functional Anatomy of the Human Microprocessor. Cell 161, 1374–1387. Parker, J.S., Parizotto, E.A., Wang, M., Roe, S.M., and Barford, D. (2009). Enhancement of the seed-target recognition step in RNA silencing by a PIWI/MID domain protein. Mol. Cell 33, 204–214. Salomon, W.E., Samson, M.J., Moore, M.J., Zamore, P.D., and Serebrov, V. (2015). Single-molecule imaging reveals argonaute mechanisms in the miRNA and RNAi pathways. Cell 162, http://dx.doi.org/10.1016/j.cell.2015.06.029. Sasaki, H.M., and Tomari, Y. (2012). The true core of RNA silencing revealed. Nat. Struct. Mol. Biol. 19, 657–660. Schirle, N.T., Sheu-Gruttadauria, J., and MacRae, I.J. (2014). Structural basis for microRNA targeting. Science 346, 608–613. Schraivogel, D., and Meister, G. (2014). Import routes and nuclear functions of Argonaute and other small RNA-silencing proteins. Trends Biochem. Sci. 39, 420–431. Sheng, G., Zhao, H., Wang, J., Rao, Y., Tian, W., Swarts, D.C., van der Oost, J., Patel, D.J., and Wang, Y. (2014). Structure-based cleavage mechanism of Thermus thermophilus Argonaute DNA guide strand-mediated DNA target cleavage. Proc. Natl. Acad. Sci. USA 111, 652–657. Sternberg, S.H., Redding, S., Jinek, M., Greene, E.C., and Doudna, J.A. (2014). DNA interrogation by the CRISPR RNA-guided endonuclease Cas9. Nature 507, 62–67. Swarts, D.C., Makarova, K., Wang, Y., Nakanishi, K., Ketting, R.F., Koonin, E.V., Patel, D.J., and van der Oost, J. (2014). The evolutionary journey of Argonaute proteins. Nat. Struct. Mol. Biol. 21, 743–753. Tafvizi, A., Mirny, L.A., and van Oijen, A.M. (2011). Dancing on DNA: kinetic aspects of search processes on DNA. ChemPhysChem 12, 1481–1489. Tomari, Y., and Zamore, P.D. (2005). Perspective: machines for RNAi. Genes Dev. 19, 517–529. Vogel, J. (2014). Biochemistry. A bacterial seek-and-destroy system for foreign DNA. Science 344, 972–973. Wang, Y., Juranek, S., Li, H., Sheng, G., Wardle, G.S., Tuschl, T., and Patel, D.J. (2009). Nucleation, propagation and cleavage of target RNAs in Ago silencing complexes. Nature 461, 754–761. Wee, L.M., Flores-Jasso, C.F., Salomon, W.E., and Zamore, P.D. (2012). Argonaute divides its RNA guide into domains with distinct functions and RNA-binding properties. Cell 151, 1055–1067. Yao, C., Sasaki, H.M., Ueda, T., Tomari, Y., and Tadakuma, H. (2015). Singlemolecule analysis of the target cleavage reaction by Drosophila RNAi enzyme complex. Mol. Cell 59, this issue, 125–132.
Molecular Cell 59, July 2, 2015 ª2015 Elsevier Inc. 7