Designing small molecules that target a cryptic RNA binding site via base displacement

Designing small molecules that target a cryptic RNA binding site via base displacement | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Designing small molecules that target a cryptic RNA binding site via base displacement Robert Batey, Lukasz Olenginski, Aleksandra Wierzba, Shawn Laursen This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5836924/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 29 Aug, 2025 Read the published version in Nature Chemical Biology → Version 1 posted You are reading this latest preprint version Abstract Most RNA-binding small molecules have limited solubility, weak affinity, and/or lack of specificity, restricting the medicinal chemistry often required for lead compound discovery. We reasoned that conjugation of these unfavorable ligands to a suitable “host” molecule can solubilize the “guest” and deliver it site-specifically to an RNA of interest to resolve these issues. Using this framework, we designed a small molecule library that was hosted by cobalamin (Cbl) to interact with the Cbl riboswitch through a common base displacement mechanism. Combining in vitro binding, cell-based assays, chemoinformatic modeling, and structure-based design, we unmasked a cryptic binding site within the riboswitch that was exploited to discover compounds that have affinity exceeding the native ligand, antagonize riboswitch function, or bear no resemblance to Cbl. These data demonstrate how a privileged biphenyl-like scaffold effectively targets RNA by optimizing π-stacking interactions within the binding pocket. Biological sciences/Biochemistry/RNA Biological sciences/Structural biology/X-ray crystallography Biological sciences/Drug discovery/Medicinal chemistry/Structure-based drug design RNA structure drug discovery small molecules structure-based design base displacement Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction The design and discovery of small molecules that selectively target RNA is a longstanding problem in chemical biology. This emerging field has the potential to develop chemical probes of RNA function and therapeutics to treat RNA-mediated disease. 1 – 3 Despite ongoing efforts, there is only one FDA-approved small molecule that targets RNA outside the ribosome, risdiplam. 4 This reinforces the notion that targeting RNA is difficult and often limited by the lack of druggable RNA targets and/or quantitative structure-activity relationship (QSAR) studies. 1 – 3 Compared to proteins, which are routinely targeted with small molecule therapeutics, 5 much less is known about the principles governing RNA-ligand interactions to guide discovery efforts. One approach to address this knowledge gap is the application of natural RNA-small molecule interactions such as those observed in riboswitches. These bacterial mRNA elements have evolved tertiary structures to selectively bind metabolites and control gene expression, 6 – 8 making them ideal models to explore RNA-ligand interactions. For example, BioRelix 9 , 10 and Merck 11 , 12 have reported promising lead compounds against the FMN riboswitch that demonstrate efficacy in animal models. Another system of therapeutic interest 13 is the cobalamin (Cbl) riboswitch, 14 which is broadly distributed across bacteria. 15 Different classes of Cbl riboswitches exhibit distinct binding preferences to forms of Cbl that differ only at their β-axial position (Fig. 1 a). 16 , 17 The biological forms of Cbl are 5′-deoxyadenosylCbl (AdoCbl) and methylCbl (MeCbl, 1 ), 13 but standard forms include the photolysis 18 product hydroxoCbl (OHCbl, 2 ) and the photostable 19 cyanoCbl (CNCbl, 3 ) (Fig. 1 a). We recently identified several β-axial derivatives 4 – 7 that bind the env8 MeCbl-selective riboswitch 20 and regulate in-cell function (Fig. 1 b,c). 21 This was a surprising result given that these derivatives host bulky β-axial moieties (Fig. 1 c) that present a significant steric problem within the binding pocket ( Supplementary Fig. 1a ). Chemical probing suggests that recognition of the higher affinity 4 and 7 involves the displacement of an adenosine (A20) from the RNA core ( Supplementary Fig. 1b ), which is likely replaced by the β-axial group. 21 Displacement of A20 would yield a cryptic binding site distinct from the displacement of an adjacent adenosine by the 5′-deoxyadenosyl moiety observed in AdoCbl-selective riboswitches ( Supplementary Fig. 1c ). 20 Our preliminary data suggest that the cryptic binding of 4 and 7 may explain their increased affinity and that chemical modifications to the β-axial group have robust effects on RNA binding. 21 In this work, we synthesized an expanded library of Cbl derivatives that host systematically varied β-axial moieties (Fig. 1 d,e) and employed a set of biochemical and cell-based assays to quantify their RNA binding affinities and regulatory activities. These data were used in a predictive modeling platform to determine the ligand properties associated with tight affinity and strong regulatory activity. As a complementary approach, we used structure-based design to unmask the cryptic β-axial binding pocket and discover lead compounds with affinity exceeding the native ligand. This structure-informed approach also enabled the identification of novel cryptic binders that are chemically distinct from Cbl. Collectively, our work outlines the molecular determinants of specific and high affinity base displacement RNA binding modes, which can guide future efforts to target these common RNA interactions. 22 – 25 Results Small molecule library design and synthesis Our Cbl derivative library was synthesized using a previously established reduction-free method 26 , 27 where the cyano group at the β-axial position of CNCbl is replaced by the R-group of a variable alkyne (Fig. 1 d). While this approach enabled the synthesis of a diverse assortment of photostable β-axial modified Cbls 8 – 44 (Fig. 1 e), reactions with alkynes containing amines or protonated nitrogen atoms were unsuccessful ( Supplementary Fig. 2 ). The logic of our library construction was to diversify around lead compounds 4 and 7 . This strategy enabled specific medicinal chemistry questions to be addressed with QSAR studies. Within this framework, we can determine which chemical features modulate RNA binding and function. Given that the majority of RNA-binding ligands are aromatic systems capable of π-stacking, 28 – 30 many RNA-targeting lead compounds suffer from limited solubility, low affinity, and/or lack of specificity, limiting the scope and utility of the downstream medicinal chemistry that is required for successful hit-to-lead discovery campaigns. 1 – 3 In our approach ( Supplementary Fig. 3a ), Cbl functions as a soluble “host” of a variable chemical “guest” that interacts with the env8 RNA target in the minor groove. This target-guest interaction is anchored by contacts from the host corrin ring and α-axial group (Fig. 1 a) to project the variable β-axial moiety to the same spatial location within the cryptic site. Thus, our model system enables a detailed and systematic approach to determine which chemical features confer high affinity base displacement RNA binding modes, which is common throughout RNA biology 22 – 25 ( Supplementary Fig. 3b ). Cbl derivatives productively bind env8 To quantify the binding of our library to env8 , we used a fluorophore-conjugated CNCbl probe (CNCbl-5×PEG-ATTO590 31 ) that undergoes fluorescence induction upon env8 binding and competed it off with ligands in our derivative library. 21 Titration of env8 into the probe gave a K D of 3.2 ± 0.1 nM (Fig. 2 a) and competitive titrations of MeCbl and Cbl 4 into the env8 -CNCbl-5×PEG-ATTO590 complex yielded K D values of 1.4 ± 0.8 nM and 10 ± 5 nM, respectively (Fig. 2 b). Using this fluorescence displacement assay, we measured the K D values for Cbls 1 – 44 . All ligands bound env8 with sub-micromolar affinity (Fig. 2 c). The tightest binding derivative was 29 (7 ± 7 nM) while 12 (800 ± 700 nM) was the weakest. Importantly, 29 functions as a binding lead with an affinity that exceeds 4 and 7 and is comparable ( P = 0.19) with the native MeCbl. QSAR analysis of RNA binding data To reveal chemical trends within our binding data, we employed a QSAR analysis. We quantified the “distance from the alkyne” of our ortho -, meta -, and para -substituted “phenyl-F” ( 9 – 11 ) and “phenyl-CH 3 ” ( 12 – 14 ) derivatives and compared this parameter against their K D values. We observed a strong relationship (R 2 = 0.97) for the smaller substituents and an even stronger relationship (R 2 = 0.99) for the larger groups ( Extended Data Fig. 1 a,b), suggesting steric limitations within the cryptic site. We also quantified the electron donating and withdrawing strength or Van der Waals (VdW) volume of the chemical groups in the para -substituted phenyls ( 4 , 5 , 8 , 11 , 14 , and 17 – 20 ) and compared these parameters against their K D values. This analysis revealed that phenyl ring electronics (R 2 = 0.10) and para -position sterics (R 2 = 0.01) both fail to explain env8 binding ( Extended Data Fig. 1 c-e). Additional QSAR trends emerge from our heterocyclic Cbls ( 21 – 26 ) ( Extended Data Fig. 1 f). Comparing the affinities of phenyl 8 (70 ± 20 nM) and pyrimidine 24 (70 ± 20 nM) suggest that heterocycles have no advantage over their hydrocarbon counterpart. Data from thiophene 21 (40 ± 30 nM), thiazole 22 (50 ± 40 nM), and furan 23 (50 ± 10 nM) demonstrate that five-membered heterocycles bind just as well as six-membered ones. Finally, the increased affinity for our two-ring heterocycles 25 (16 ± 8 nM) and 26 (12 ± 5 nM) suggests that two rings are preferred, which is consistent with our hypothesis that the β-axial group π-stacks within the cryptic site. This notion is supported by the binding data of our two-ring containing Cbls 27 – 29 , which all have affinities under 20 nM ( Extended Data Fig. 1 g). Even the three- ( 31 , 30 ± 20 nM) and four-ring ( 32 , 30 ± 10 nM) containing derivatives bind with high affinity ( Extended Data Fig. 1 g). To analyze these chemical trends in greater detail, we employed a multivariate analysis. Binding data were clustered into tight, moderate, and weak binders ( Extended Data Fig. 2 a), and 20 standard chemoinformatic parameters 32 – 35 ( Supplementary Table 1 ) were calculated for the β-axial group of Cbls 4 , 5 , and 7 – 44 . These descriptors were used in a linear discriminant analysis (LDA) 34 , 35 with our binding data. When viewed along the first two principal components (PC1 and PC2), each binding cluster occupies unique regions in chemical space ( Extended Data Fig. 2 b). The LDA loading plot provides qualitative trends for the molecular determinants of each cluster. For example, tight binders have more aromatic rings, moderate binders have more hydrogen bond donors, and weak binders have more accessible surface area ( Extended Data Fig. 2 c). QSAR model-based screening of potential binding leads Motivated by the LDA-identified trends, we sought to determine if the chemical identity of the β-axial group could be used to predict env8 binding affinity. An expanded set of 347 physiochemical descriptors were calculated for the β-axial group of Cbls 4 , 5 , and 7 – 44 and modeled against their natural log-transformed K D data using a least absolute shrinkage and selection operator (lasso)-multiple linear regression (MLR) strategy. 36 When using all data in training, we obtained a baseline model that predicted our binding data with good accuracy (R 2 = 0.71) using six descriptors ( Extended Data Fig. 3 a,b), which are listed along with their physical meaning in Supplementary Table 2 . When using a stratified data split (Fig. 2 d and Extended Data Fig. 3 c) and implementing our lasso-MLR strategy, we constructed new models that prioritized either the predictive ability on the test set (Q 2 -focused) or the overall fit of the training set (R 2 -focused). The Q 2 -focused (R 2 Training = 0.48, Q 2 Test = 0.81) and R 2 -focused (R 2 Training = 0.65, Q 2 Test = 0.60) models showed a similar level of performance (Fig. 2 e and Extended Data Fig. 3 d). We then used our binding-based models to screen for high affinity β-axial groups among a 513-compound alkyne library, which included modified-phenyl (“phenyl-X”) and aliphatic (“amide-X”) alkynes (Fig. 2 f) that are compatible with our Cbl synthesis (Fig. 1 d) and representative of our initial library (Fig. 1 e). Among these compounds, 184 were predicted to have tighter affinities than our current binding lead 29 ( Extended Data Fig. 3 e). However, the majority of these small molecules were synthetically inaccessible using the non-reductive method 26 , 27 ( Extended Data Fig. 3 f). These considerations left us with four remaining leads, which were synthesized into the corresponding Cbls 45 – 48 along with compounds that were predicted to be moderate ( 49 and 50 ) and weak ( 51 – 53 ) binders (Fig. 2 g). We then used our fluorescence displacement assay to quantify the binding of these derivatives to env8 . The experimentally derived ln [K D ] data agreed well with our model-based predictions for the moderate and weak binders but deviated significantly for the predicted leads 45 and 46 (Fig. 2 h). One explanation for this observation is the sparseness of tight binders in our dataset (8 of 44, 18%). Synthetic routes that access chemical groups facilitating high affinity RNA binding interactions (e.g., amino or other positively charged groups 28 – 30 ) could address this issue. Cbl derivatives promote RNA regulatory activity Given that all the derivatives in our library bind env8 , we wanted to determine whether they drive RNA regulatory activity in a cellular environment. We employed a previously established 37 cell-based assay in which env8 was placed upstream of GFPuv, whose expression is repressed by Cbl-dependent occlusion of the ribosome-binding site. In the absence of Cbl, we detected high fluorescence, indicative of GFPuv expression (Fig. 3 a). Conversely, in the presence of 10 nM of MeCbl or Cbl 4 , we observed attenuated fluorescence, demonstrating that these ligands promote env8 repression of GFPuv expression in E. coli (Fig. 3 a). To quantify the extent to which each derivative regulates env8 function, we calculated the fold repression 37 for Cbls 1 – 44 . While all ligands were able to regulate env8 function to some extent, the repression was highly variable. The strongest repressors were 36 (8 ± 1) and 41 (7 ± 2) while 43 (1.4 ± 0.1) and 44 (2.0 ± 0.1) were the weakest (Fig. 3 b). Importantly, 36 and 41 are new functional leads with fold repression values that exceed 4 and 7 and are comparable ( 36 , P = 0.31 and 41 , P = 0.19) with the native MeCbl (9.0 ± 0.2). These data enable the exploration of the relationship between derivative binding and regulatory activity. In general, while many tight binding derivatives are also strong repressors and vice versa, this was not always true (R 2 = 0.34) (Fig. 3 c). This fact reflects the complex nature of promoting a ligand-induced regulatory response, which involves Cbl cellular import, β-axial repair by endogenous enzymes, and env8 -Cbl binding within the appropriate time scale of a co-transcriptional process. 37 To investigate this further, we carried out growth experiments in ΔMetE cells, which lack the Cbl-independent methionine synthase MetE. 38 In the absence of methionine, these cells must use Cbl for one carbon metabolism, making MeCbl essential for growth. 38 In the presence of 10 nM derivative, both weak (Cbl 44 ) and strong (Cbl 36 ) repressors support ΔMetE growth with the same doubling time as CNCbl, suggesting that these ligands are imported and repaired to some extent ( Extended Data Fig. 4 ). However, the fact that repression is variable among our derivatives in wild-type E. coli cells suggests that these compounds exist predominantly in their unrepaired, β-axial derivatized state. A confounding factor in comparing the binding and functional data is potentially variable levels of derivative metabolism within the cell. QSAR model-based screening of potential functional leads While individual QSAR trends were less pronounced within our functional data ( Supplementary Fig. 4 ), an LDA using the same set of chemoinformatic parameters 32 – 35 ( Supplementary Table 1 ) demonstrated that functional clusters ( Extended Data Fig. 5 a) occupy unique regions in chemical space ( Extended Data Fig. 5 b). The LDA loading plot suggests that strong repressors have more hydrogen bond donors, moderate repressors have more aromatic rings, and weak repressors have more sp 3 centers ( Extended Data Fig. 5 c). These trends motivated us to build QSAR models using the same lasso-MLR workflow 36 described above, except using our repression data as the dependent variable. We obtained baseline, Q 2 -focused, and R 2 -focused models that predict the functional data reasonably well with four to six descriptors (Fig. 3 d,e and Extended Data Fig. 6 a-d), which are listed in Supplementary Table 3 . We then used our function-based models to screen for strong repressive β-axial groups among the same alkyne library (Fig. 3 f). Although 151 small molecules were predicted to have stronger repression than our current functional lead 36 ( Extended Data Fig. 6 e), many of these leads were unavailable for the same reasons described above ( Extended Data Fig. 6 f). We identified four available predicted leads, which were synthesized into the corresponding Cbl 54 – 57 along with compounds that were predicted to be moderate ( 58 and 59 ) and weak ( 60 – 62 ) repressors (Fig. 3 g). The fold repression of 54 – 62 were then measured using our reporter assay. Much like our binding data, the experimentally derived repression data agreed well with our model-based predictions for the moderate and weak repressors but deviated significantly for the predicted functional leads 54 – 56 (Fig. 3 h), which is likely explained by the confounding role of derivative metabolism on regulatory activity. env8 -Cbl co-crystal structures unmask cryptic binding site As a complementary approach to QSAR modeling, we used structure-based design to investigate env8 -ligand interactions. We carried out three independent 1 µs molecular dynamics (MD) simulations of env8 -CNCbl and env8 -Cbl 4 and measured the A20(C6)-G19(C6) and A20(C6)-A68(N7) distances to assess A20 π-stacking with neighboring nucleotides (Fig. 4 a). Over the course of the MD trajectories, the A20(C6)-G19(C6) distances were identical among both complexes, while 4 did induce slight increases in A20(C6)-A68(N7) distances (Fig. 4 b). However, these changes were not a result of A20 displacement but were caused by the β-axial phenyl- p NO 2 group intercalating between A20 and A68 ( Supplementary Fig. 5 ), which is inconsistent with previous chemical probing data ( Supplementary Fig. 1b ). 21 The limitations of our MD analysis motivated us to employ X-ray crystallography. We determined co-crystal structures of the Cbl riboswitch aptamer domain in complex with CNCbl and 4 . However, we only obtained diffraction-quality crystals using the closely related env2 RNA, which shares 95% sequence identity with env8 ( Extended Data Fig. 7a ). Given that env2 shows near-identical ligand binding and regulatory activity as env8 ( Extended Data Fig. 7b,c ), all structural information can be translated to env8 . The env2 -CNCbl structure shows A20 π-stacked with G19 and A68 (Fig. 4 c) with strong agreement to the env8 -OHCbl structure 20 (RMSD = 0.45 Å) ( Extended Data Fig. 7d ). The env2 -Cbl 4 structure, on the other hand, reveals that A20 undergoes base displacement from the RNA core toward the major groove and that the β-axial phenyl ring π-stacks with A68 (Fig. 4 d), unmasking the cryptic binding pocket in agreement with our hypothesis. 21 We also determined the structure of env2 in the apo state, which shows a near-identical (RMSD = 0.39 Å) binding pocket to our env2 -CNCbl structure ( Extended Data Fig. 8a ), demonstrating that CNCbl interacts with a pre-organized binding pocket. Notably, the env2 -Cbl 4 structure reinforces the inability of the MD simulation to access base displacement of nucleotides. To gain further insight into cryptic β-axial pocket, we determined nine additional env2 -ligand co-crystal structures ( Extended Data Fig. 8b-j ). The structure of env2 -Cbl 32 with its bulky pyrene β-axial group showcases that base displacement provides a lot of room (> 340 Å 3 ) in the binding pocket (Fig. 4 e). With its four phenyl rings, pyrene π-stacks well with A68 but is unable to optimize π-stacking with G19 (Fig. 4 e). In contrast, the structure of env2 -Cbl 29 demonstrates how the biphenyl β-axial group maximizes π-stacking with both G19 and A68 via rotation of the two rings relative to one another (Fig. 4 f). Interestingly, our data support the fitting of two conformations of A20, with the more populated (55%) conformer partially engaged within the RNA core (Fig. 4 f), providing a means to offset the energetic penalty of A20 displacement. It is important to note that π-stacking is not the only binding mode available within the cryptic site. The structure of env2 -Cbl 42 suggests that two hydrogen bonds to A68 provide comparable affinity (50 ± 10 nM) with a π-stacking interaction (e.g., 8 , 70 ± 20 nM) (Fig. 4 g). Together, our structural data suggest a set of design principles to develop new lead compounds: (1) hydrogen-bonding to the phosphate backbone, (2) cation-π-stacking, and (3) engagement of A20 (Fig. 4 h). Structure-based design identifies Cbls with affinity exceeding the native ligand In order to explore design principles (1) and (2), we adopted new chemistry using a click reaction of Cbl 58 with a variable azide 26 to form a biphenyl-like scaffold where the second ring is an R-group harboring triazole (Fig. 5 a). We synthesized Cbls 63 – 65 to install amino-, guanidinium-, and hydroxyl-substituted triazoles, respectively (Fig. 5 b). We then used our fluorescence displacement assay to quantify the binding of these derivatives to env8 and discovered that 63 (1.0 ± 0.6 nM, P = 0.62) and 64 (1.3 ± 0.7 nM, P = 0.92) have comparable affinity to CNCbl (Fig. 5 c). However, these interactions approach the accuracy limit of our binding assay. To confirm these results, we used isothermal titration calorimetry (ITC) with the env8 aptamer domain, which lacks an element of RNA structure that interacts with the α-axial face of the Cbl and thereby lowering the observed affinity. These data revealed significantly tighter affinity to 63 (32 ± 6 nM, P = 0.02) and 64 (40 ± 3 nM, P = 0.02) as compared to CNCbl (260 ± 60 nM) (Fig. 5 d). To understand how these new lead compounds bind env8 , we determined co-crystal structures of env2 -Cbl 63 and env2 -Cbl 64 . While weakly supported by the electron density, both structures suggest that the amino (Fig. 5 e) or guanidinium (Fig. 5 f) groups likely electrostatically interact with backbone phosphate of G19 ( env2 -Cbl 64 ) or A68 ( env2 -Cbl 63 ). However, our env2 -Cbl 63 structure unambiguously demonstrates that the terminal triazole ring twists in order to π-stack with A20 (Fig. 5 e), again suggesting that rotatable bonds are beneficial for optimizing the π-stacking network. This is the first evidence of strong engagement of A20 in any of our structures. Notably, when π-stacked with the ligand, A20 adopts a perpendicular orientation relative to G19, a highly unusual mode of base-base interaction in nucleic acid. Despite their high affinity, when used in our reporter assay, Cbls 63 – 65 confer minimal repression (Fig. 5 g). To investigate this further, we repeated our growth experiments in ΔMetE cells. In the presence of 10 nM derivative, 63 – 65 support ΔMetE growth with the same doubling time as CNCbl, suggesting that these ligands are imported and repaired to some extent ( Extended Data Fig. 9a,b ). However, unlike 36 and 44 , elevated concentrations of 63 – 65 prevented growth of ΔMetE cells ( Extended Data Fig. 9c,d ), suggesting that these derivatives antagonize some essential aspect of Cbl metabolism in E. coli . 39 To fully assess how 63 – 65 affect regulatory activity, we repeated our reporter assay in wild-type E. coli , which grow normally in the presence of high Cbl concentrations ( Extended Data Fig. 9e ). Adding excess amounts of 63 – 65 to cells containing 10 nM CNCbl resulted in loss of CNCbl-induced repression, confirming that these derivatives are antagonists of env8 function (Fig. 5 g). To further explore design principle (2), we synthesized the pyridine Cbl 66 to introduce a conditional positive charge (Fig. 5 h). Under the conditions of our binding experiments (pH 8), 66 should behave exactly like its non-pyridine counterpart 29 . However, at a pH below its predicted pK a (~ 5.25), 66 should be charged and we would expect an increase in affinity that is absent in 29 . To test this hypothesis, we carried out ITC measurements at pH 8 and 5 with the env8 aptamer domain and CNCbl, 29 , and 66 . The non-titratable CNCbl and 29 showed a ~ two-fold reduction in binding affinity at reduced pH, whereas the affinity of 66 increased 1.7-fold (Fig. 5 i), in agreement with our hypothesis. These data provide proof-of-principle that installing pyridines, whose pK a can be lowered by interaction with RNA, is an important design principle that leverages the increased affinity of cation-π interactions without violating Lipinski’s rules. 40 , 41 Structure-informed screen identifies cryptic binders of env8 divorced from Cbl Our structural biology efforts unmasked the cryptic β-axial binding pocket within env8 , which allows us to explore whether small molecules divorced from a Cbl host can also target this site. To address this issue, we carried out a high-throughput computational screen using a 28,000-compound RNA-focused library. These compounds were docked against env2 in two conformations: one in which A20 was displaced toward the major groove (A20-out, e.g., env2 -Cbl 29 ) and one where A20 was engaged in the RNA core (A20-in, e.g., env2 -Cbl 63 ) ( Supplementary Fig. 6 ). To leverage our structural data, we specified that docking hits contain the biphenyl-like scaffold and identified eight potential lead compounds (Fig. 6 a). To survey the binding of these ligands to env8 , we adopted a thiazole orange (TO) displacement assay ( Supplementary Fig. 7a ). 42 Upon addition of E1 - E8 , E2 and E5 led to sufficient (> 25%) displacement of TO indicative of binding (Fig. 6 b). This conclusion was confirmed with microscale thermophoresis (MST) (Fig. 6 c). Competitive titrations of CNCbl, E2 , and E5 into the env8 -TO complex yielded K D values of 38 ± 5 nM, 34 ± 9 µM, and 50 ± 10 µM, respectively (Fig. 6 d). The affinity of CNCbl agrees well with data obtained from ITC (38 ± 8 nM) ( Supplementary Fig. 7b,c ), supporting the use of this method to quantify ligand binding to env8 . To verify that E2 and E5 target the Cbl binding pocket, we employed an electrophoresis mobility shift assay (EMSA) that monitors the ability of env8 to adopt a kissing-loop conformation in the presence of Cbl, 37 which migrates faster than the unbound RNA (Fig. 6 e). Given that the kissing-loop is facilitated by interactions with the α-axial face of the Cbl corrin ring, 37 E2 and E5 are not expected to induce this conformational change. Thus, if these ligands competitively bind the same site within env8 , then adding E2 or E5 to env8 in the presence of CNCbl should revert the RNA back to the slower migrating species. When excess amounts of E2 or E5 were added to an env8 -CNCbl complex both ligands prevented kissing-loop formation, suggestive of competitive binding (Fig. 6 f). While this phenomenon was unambiguous for E5 , addition of E2 reproducibly led to weaker gel staining (Fig. 6 f). We therefore confirmed the competitive binding of both ligands with MST (Fig. 6 g). These data support the top-ranked docking poses of E2 and E5 which show their biphenyl-like scaffolds maximizing π-stacking to G19 and A68 within the cryptic β-axial binding pocket and making hydrogen bonds to the phosphate backbone (Fig. 6 h). Discussion Targeting RNA with small molecules is an emerging field hindered by an incomplete understanding of the basic principles governing RNA-ligand interactions. 1 – 3 To address this knowledge gap, we explored the chemical features promoting specific and high affinity base displacement RNA binding interactions using the Cbl riboswitch. Despite limitations of our QSAR chemoinformatic modeling, structure-based design successfully identified lead compounds with affinities exceeding the native ligand. All leads share a biphenyl-like structure, which has been previously identified as a privileged RNA-binding scaffold. 28 – 30 , 43 Our data demonstrate that twisting around a rotatable bond confers the ability to optimize the π-stacking network within the binding pocket. Thus, our results not only reinforce our understanding of RNA chemical space but yield new structural insights into how these scaffolds interact with RNA. Across all newly reported RNA-ligand co-crystal structures, binding pocket nucleotides G19 and A68 are in near-identical positions, indicating that structural plasticity is conferred solely by the displacement of A20, providing a large (> 340 Å 3 ) cryptic site that accommodates the β-axial group of all derivatives. One mechanism to explain this phenomenon is ligand docking into a rigid RNA to induce A20 displacement; however, large, rigid β-axial groups likely prevent the corrin ring from productively engaging the RNA. A more likely explanation is that A20 undergoes transient excursions from the RNA core to form a binding-competent state, a motion that is well documented in nucleic acids and consistent with the time scale of ligand binding. 44 The energetic penalty of A20 displacement to expose the cryptic pocket is offset to varying degrees by each β-axial group, reflecting different derivative binding affinities. This illustrates that a consideration of local binding pocket dynamics is an essential feature to unmask additional RNA cryptic sites, 45 , 46 which will facilitate the development of new chemical probes of RNA function and therapeutics to treat RNA-mediated disease. However, our work cautions against using a purely computational approach, as MD simulations could not account for the nuances of base displacement within env8 . Advances in machine-learning 47 and artificial intelligence 48 methods may address these limitations in the future. Our work also showcases that bifunctional host-guest systems are a valuable approach to targeting RNA. Our soluble Cbl host enabled access to chemical space that is typically intractable to robust medicinal chemistry. Exploration of general host-small molecule conjugates is therefore of critical importance. Nucleic acid-based hosts are an attractive candidate because they can solubilize the attached ligand, deliver it to RNA in a sequence specific manner, and contribute binding energy that can turn modest and/or non-specific binders to ligands with high affinity and specificity. For example, TO has been conjugated to a peptide nucleic acid (PNA) host to monitor double-stranded RNA formation. 49 PNA hosts have also been shown to rescue the binding of small molecules that have no affinity for RNA by themselves. 50 This conceptual framework is similar to the widespread use of bifunctional/multivalent small molecules in protein targeting, 51 which has recently seen parallels in RNA-targeting strategies (e.g., RiboTAC 43 and RiboSNAP 52 ). We believe that a host-guest approach is generalizable and will enable detailed chemical and structural analysis of underexplored chemical space to better understand RNA-small molecule interactions. Methods Synthesis of Cbl derivatives Cbl derivatives 4 – 66 were synthesized according to previously described procedures. 26 , 27 See the Supplementary Information for additional details. RNA preparation All RNAs (full sequences shown in Supplementary Table 4 ) were prepared using DNA templates amplified by PCR, transcribed with T7 RNA polymerase, and purified using preparative denaturing polyacrylamide gel electrophoresis. 53 Purified RNA was buffer exchanged and concentrated into Milli-Q H 2 O using centrifugal concentrators (Millipore-Sigma). Final RNA concentrations were calculated using absorbance at 260 nm and extinction coefficients determined from the summation of the individual bases. Prior to all binding experiments, RNA was heated at 95°C for 2 minutes, incubated on ice for 10 minutes, and then allowed to equilibrate to room temperature. Cbl fluorescence binding assays Both direct CNCbl-5×PEG-ATTO590 fluorescence induction and displacement assays were performed as previously described. 21 The K D values for all Cbl derivatives to env8 can be found in Supplementary Table 5 . Cell-based reporter assay These were conducted as previously reported 37 with minor alterations. Firstly, plasmids harboring env8 -GFPuv (Addgene #99831) were transformed into E. coli BW25113 cells. Second, 1 µl of a saturated overnight culture was added to 1 ml of CSB media supplemented with 100 µg/ml carbenicillin and 10 nM Cbl (unless otherwise noted) and grown to mid-log phase at 37°C. Third, from each biological replicate, 200 µl of cells were added to wells in a 96-well plate (Costar). The fold repression values 37 for all Cbl derivatives can be found in Supplementary Table 6 . ΔMetE growth assays For growth experiments, 5 µl of a saturated overnight culture of ΔMetE E. coli cells (Keio collection, JW3805) was added to 5 ml of methionine-dropout CSB media supplemented with either 335 µM methionine or 10 nM Cbl and grown to mid-log phase at 37°C. From each biological replicate, 200 µl of cells were added to wells in a 96-well plate (Costar) and growth was monitored by measuring OD 600 using a microplate reader (Tecan). For growth rate experiments, the same procedure was used except the cells were grown for 24 hours and the OD 600 was measured at various time intervals. For titration experiments, the same procedure was used except variable amounts of methionine or Cbl was added to the media. Chemoinformatic linear discriminant (LDA) analysis A set 20 standard chemoinformatic parameters 32 – 35 ( Supplementary Table 1 ) were calculated for the β-axial group of Cbls 4 , 5 , and 7 – 44 as previously described. 34 , 35 All descriptors that were non-constant were then used in a LDA with our clustered binding or functional data (see Extended Data Figs. 2 and 5 ) using Discriminatory Analysis plugin from XLSTAT as previously described. 34 , 35 Excel files with the calculated descriptors are available upon request. QSAR modeling The β-axial group of Cbls 4 , 5 , and 7 – 44 were tuned to the correct protonation and tautomerization states using molecular operating environment (MOE, v2022.02, Chemical Computing Group) and each state was sent to conformational search as previously described. 36 A total of 347 descriptors (all but x3D and protein descriptors) were calculated for each conformation in MOE, averaged using the Boltzmann-weighted equation, and filtered to remove multicollinear features as previously described. 36 Then, a lasso-MLR strategy was used to model these descriptors (independent variable) against either the natural-log transformed K D or fold repression data (dependent variables) in matlab (v2019a). Lasso was implemented with five-fold cross-validation to find an optimal lambda value in descriptor selection. MLR exhaustively searched all possible models from the lasso-selected descriptors (with a maximum number of descriptors set by Topliss rule 54 ) with five-fold cross-validation to obtain the best model as defined by specified criteria. We constructed a baseline model without a data split that maximized the R 2 . We also used a stratified 80/20 data split to ensure that training and test sets had equivalent proportions of tight/strong, moderate, and weak binders/repressors and constructed models that prioritized either the predictive ability on the test set (Q 2 -focused) or the fit of the training set (R 2 -focused). Excel files with the calculated descriptors and matlab scripts for modeling are available upon request. Lead compound prediction We used our binding- and function-based models to screen for high affinity β-axial groups among a curated Enamine alkyne library. These small molecules were selected by a substructure search ( https://enaminestore.com/search ) of modified-phenyl (phenyl-X) and aliphatic (amide-X) alkynes (precise structures used for this search are shown in Fig. 2 f). This left us with a 513-compound library (469 phenyl-X and 44 amide-X). The binding- and function-based model-selected descriptors ( Extended Data Figs. 3 a and 6 a) were calculated for the 513-compound library in MOE as described above. The predicted ln [K D ] and fold repression values from baseline, Q 2 -focused, and R 2 -focused models were averaged and used to select tight, moderate, and weak binders (Cbls 45–53 ) (Fig. 2 g) and strong, moderate, and weak repressors (Cbls 54–62 ) (Fig. 3 g). Excel files with the calculated descriptors and SDF files from phenyl-X and alkyne-X substructure search are available upon request. Molecular dynamic (MD) simulations All-atom MD simulations of env8 -CNCbl and env8 -Cbl 4 complexes were performed using the PMEMD module 55 in AMBER22. 56 CNCbl was manually built from OHCbl (PDB 4FRG 20 ) in PyMol (v2.5.7, Schrödinger) and Cambridge Structural Database deposition 961228 27 was used for Cbl 4 . Ligands were docked into env8 by alignment to OHCbl from the env8 -OHCbl structure (PDB 4FRG 20 ). Each system was neutralized and solvated in a cubic simulation box with a 25 Å buffer of OPC water molecules 57 and 150 mM NaCl to mimic physiological conditions. RNA was parameterized using the OL3 force field, 58 incorporating the RNA.LJbb correction 59 for improved backbone accuracy. Force field parameters for each ligand were generated using Antechamber with the General Amber Force Field. The metal center, including the cobalt ion, was parameterized using MCPB.py following established guidelines. 60 Quantum mechanical calculations were performed with Gaussian16 to optimize the geometry and derive charges. The cobalt-carbon bond was explicitly defined in the MCPB.py input to ensure accurate representation of the ligand coordination environment. 60 To improve simulation efficiency, hydrogen mass repartitioning was applied, allowing a 4 fs time step. 61 Following system preparation, energy minimization, gradual heating to 300 K, and equilibration under constant pressure were conducted. Production simulations were performed in triplicate for each system under constant temperature (300 K) and pressure (1 atm) using the NPT ensemble, with each replicate running for at least 1 µs. Trajectory analysis and post-processing A20(C6)-G19(C6) and A20(C6)-A68(N7) distance calculations were carried out using Cpptraj. 62 Crystallization of env2- ligand complexes The env2 aptamer domain RNA in complex with various Cbl derivatives was crystallized at 30°C using hanging drop vapor diffusion. The RNA-ligand complexes were prepared as solutions containing 250 µM RNA and 375–500 µM ligand in 1X TE buffer. For crystallization, drops containing 2 µl of RNA-ligand solution and 2 µl precipitant solution (40 mM sodium cacodylate pH 7, 10% (w/v) 2-methyl-2,4-pentanediol (MPD), 12 mM spermine tetrahydrochloride, 80 mM KCl, and 20 mM MgCl 2 ) were suspended above a reservoir solution of 500 µl of 35% MPD and grown for 24–96 hours. For cryoprotection, the crystals were soaked in a cryoprotectant (precipitant solution with MPD increased to 25%) for 2 min and flash frozen in liquid nitrogen. Some data were collected on a home-source Rigaku MicroMax-003 X-ray source with a Dectris Pilatus 200K detector, and then indexed, integrated, and scaled with HKL3000 63 ( Supplementary Tables 7–10 ). Other data were collected at the Advanced Light Source on a Dectris on the ALS Beamline 8.2.2. with a Dectris Pilatus3 2M detector, and then indexed, integrated, and scaled with XDS 64 ( Supplementary Tables 7–10 ). The crystals that yielded the env2 apo structure were grown in the same solution described above but in the presence of compound E5 . Structure determination and model refinement For all structures, an initial electron density map was calculated using molecular replacement with Phaser 65 in PHENIX 66 using env8 as a starting model (PDB 4FRG 20 ) but with OHCbl and A20 removed to minimize model bias. All ligands were manually built from Cbl 4 (Cambridge Structural Database deposition 961228 27 ) in PyMol (v2.5.7, Schrödinger) and restraint files were generated with eLBOW. 67 After initial refinement, the ligand density was unambiguous, and ligands were added to the model with LigandFit. 68 After another round of refinement, A20 was manually built into the model. The map and model were refined, and the solvent was built until a maximum agreement between the map and the model was reached. During this process, to mitigate the effect of model bias, multiple rounds of combined high temperature (5,000 K) simulated annealing in torsion space and maximum-likelihood refinement were performed. 69 At the end of refinement, RNA geometry was corrected with ERRASER. 70 A subset of the RNA-ligand co-crystal structures had sufficient resolution to warrant simulated annealing in Cartesian space and individual ADP refinement. A simulated annealing 2F o -F c map where A20 and the ligand were omitted from the model was calculated for all structures to support the final placement of the ligand and A20. Models and associated structure factors have been deposited in the RCSB Protein Data Bank. Isothermal titration calorimetry (ITC) RNA was exchanged into a 1X RNA binding buffer (50 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) pH 8.0, 100 mM KCl, 10 mM NaCl, and 1 mM MgCl 2 ) by successive washing in centrifugal concentrators (Millipore-Sigma). RNA was diluted up to a final concentration of 10 µM and titrated with ligand (dissolved in the flow-through of the final wash from buffer exchange) at 100 µM. All titrations were performed at 25°C using a MicroCal ITC200 and collected in triplicate. The data were fit to a single-site binding model using Origin 7 ITC software (MicroCal). Computational docking The 28,000-RNA library from Enamine ( https://enamine.net/compound-libraries/targeted-libraries/rna ) was downloaded as an SDF file and imported into MOE (v2022.02, Chemical Computing Group) as a database file. The small molecules were tuned to the correct protonation and tautomerization states and each state was sent to conformational search 36 as described above. RNA receptor preparation was carried out using the default Quick Prep protocol in MOE. Our env2 -Cbl 29 (A20-out) and env2 -Cbl 63 (A20-in) structures were used as representative receptors where A20 was displaced toward the major groove or engaged in the RNA core, respectively. To ensure that A20 was not accessible for A20-out docking, all A20 atoms were removed from the RNA receptor in MOE. Rigid receptor docking was carried out in a templated manner, where ligand search space started from the biphenyl-like scaffold present in the co-crystal structures with 29 and 63 . For A20-in docking, the terminal triazole ring was rebuilt into a phenyl group, so that the scaffolds were identical in both RNA receptors. With this approach, only the 319 ligands with the required scaffold were docked and scored. Thiazole orange (TO) binding assay For fluorescence induction titration experiments, 60 µl reactions containing TO (5 nM), 1X RNA binding buffer, and increasing amounts of env8 were incubated in 384-well plates (Corning). Reactions were incubated for 15 minutes and TO fluorescence was monitored (490 ± 10 nm excitation, 530 ± 10 nm emission) using a CLARIOstarPLU microplate reader (BMG Labtech). For competitive fluorescence displacement assays, the same procedure was carried out except with a fixed amount of 2 µM RNA, 20 µM TO (in excess of its K D to ensure RNA saturation), and 2 mM competing ligand. Displacement titrations were carried out in the same manner except using increasing amounts of E2 or E5 . All titrations were performed at room temperature and collected in triplicate. The data were fit to a single-site binding model with a HillSlope parameter and the K D from the induction experiment was used to convert the IC 50 from the competitive fluorescence displacement assays to K D values, as previously described. 21 Electrophoresis mobility shift assay (EMSA) To validate the CNCbl-induced env8 conformational change, 20 µl reactions containing RNA (0.5 µM), 1X RNA binding buffer, and increasing amounts of CNCbl (0.1–10 µM) were incubated at 37°C for 15 minutes and then at 4°C for 15 additional minutes. Once equilibrated to 4°C, 15 µl of the samples were loaded onto a native 10% (19:1) polyacrylamide gel in 0.5X THE buffer and 1 mM MgCl 2 and run at constant 7 W for 3 hours at 4°C in 0.5X THE buffer with 1 mM MgCl 2 . For the competitive EMSA experiment, the same procedure was used except using a pre-bound env8 -CNCbl complex (0.5 µM CNCbl) and adding 5 mM E2 or E5 . All gels were stained with ethidium bromide, imaged on a UV illuminator (Alpha Innotech), and processed with AlphaView software (AlphaImagerHP). Unmodified images of the gels shown in Fig. 6 e,f and their replicates can be found in Supplementary Fig. 8 . Microscale thermophoresis (MST) For MST binding experiments, env8 was 3′-end labeled with pCp-AF488 (NU-1706-AF488, Jena Bioscience). The 50 µl labeling reaction comprised 200–500 nmol RNA, 1 mM ATP, 10% (w/v) DMSO, 1X T4 RNA ligase 1 buffer (#B0204S, New England Biolabs (NEB)), 15% (w/v) polyethylene glycol 8,000, 48 µM pCp-AF488, and 4 µl T4 RNA ligase 1 (#M0204L, NEB) and was incubated at 16°C for 18 hours. To monitor RNA-ligand binding, labeled env8 (300 nmol) was incubated in 1X RNA binding buffer and 250 µM ligand for 15 minutes and transferred to capillaries (#MO-K022, NanoTemper). Alexafluor488 fluorescence was read on a Monolith NT.115 (NanoTemper) instrument. Competitive MST experiments were carried out in the same manner except using a pre-bound env8 -CNCbl complex (300 nmol RNA and 10 µM ligand) and adding 1 mM E2 or E5 . Declarations Data availability Atomic coordinates and structure factors have been deposited in the Protein Data Bank (PDB) (https://www.rcsb.org/) under accession numbers 9MFH ( env2 apo), 9E5H ( env2 -CNCbl), 9E5I ( env2 -Cbl 4 ), 9E5J ( env2 -Cbl 5 ), 9E5K ( env2 -Cbl 13 ), 9E5L ( env2 -Cbl 26 ), 9E5M ( env2 -Cbl 29 ), 9E5O ( env2 -Cbl 32 ), 9E5P ( env2 -Cbl 33 ), 9E5Q ( env2 -Cbl 36 ), 9ELR ( env2 -Cbl 37 ), 9E5R ( env2 -Cbl 42 ), 9E5S ( env2 -Cbl 63 ), and 9E5T ( env2 -Cbl 64 ). Source data are provided with this paper. Acknowledgements This work is supported by the National Institutes of Health (R35 GM152029 to R.T.B. and R35 GM139644 to A.J.W.) and the Howard Hughes Medical Institute (S.P.L). We thank J. Nix and the staff of beamline 8.2.2. of the Advanced Light Source, Lawrence Berkeley National Laboratory for their support with remote crystallographic data collection. Beamline 8.2.2. of the Advanced Light Source, a DOE Office of Science User Facility under Contract No. DE-AC02-05CH11231, is supported in part by the ALS-ENABLE program funded by the National Institutes of Health (P30 GM124169-01). We thank the Macromolecular X-ray Crystallography Core (RRID:SCR_019310) at the University of Colorado Boulder for crystallographic data collection and the Shared Instruments Pool (RRID: SCR_018986) of the Department of Biochemistry at the University of Colorado Boulder for the use of the Monolith NT.115 NanoTemper instrument. We acknowledge A. Erbse for her assistance with the resources utilized at the University of Colorado Boulder. We thank F. Longshore-Neate and D. Patel for help with derivative synthesis as well as N. Zeps and T. Wolters for help with RNA-ligand co-crystallization. We are grateful to Q. Vicens for comments on the manuscript. Author information Shawn P. Laursen Present address: National Renewable Energy Laboratory, Golden Colorado, USA. Authors and Affiliations Department of Biochemistry, University of Colorado Boulder, Boulder, CO, USA Lukasz T. Olenginski, Aleksandra J. Wierzba, & Robert T. Batey. BioFrontiers Institute, University of Colorado, Boulder, CO, USA Aleksandra J. Wierzba Department of Molecular, Cellular, and Developmental Biology, University of Colorado Boulder, Boulder, CO, USA Shawn P. Laursen Howard Hughes Medical Institute, Chevy Chase, MD, USA Shawn P. Laursen Contributions R.T.B. conceptualized the project. L.T.O. designed the small molecule library and carried out its synthesis, purification, and characterization. A.J.W. helped supervise synthesis and characterization. S.P.L. conducted MD simulations. L.T.O. performed the in vitro binding and cell-based reporter assays, carried out the chemoinformatic modeling, conducted RNA crystallization and structure determination, implemented computational docking, and analyzed all data. L.T.O. wrote the manuscript with input from all authors. Corresponding authors Correspondence to Robert T. Batey . Ethics declaration Competing interests R.T.B. serves on the Scientific Advisory Boards of Expansion Therapeutics, SomaLogic, and MeiraGTx. References Childs-Disney, J. L. et al. Targeting RNA structures with small molecules. Nat. Rev. Drug Discov. 21, 736–762 (2022). Warner, K. D., Hajdin, C. E. & Weeks, K. M. Principles for targeting RNA with drug-like small molecules. Nat. Rev. Drug Discov. 17, 547–558 (2018). Umuhire Juru, A. & Hargrove, A. E. Frameworks for targeting RNA with small molecules. J. Biol. Chem. 296, 100191 (2021). Ratni, H. et al. Discovery of Risdiplam, a Selective Survival of Motor Neuron-2 (SMN2) Gene Splicing Modifier for the Treatment of Spinal Muscular Atrophy (SMA). J. Med. Chem. 61, 6501–6517 (2018). Santos, R. et al. A comprehensive map of molecular drug targets. Nat. Rev. Drug Discov. 16, 19–34 (2017). Olenginski, L. T., Spradlin, S. F. & Batey, R. T. Flipping the script: Understanding riboswitches from an alternative perspective. J. Biol. Chem. 300, (2024). Breaker, R. R. The Biochemical Landscape of Riboswitch Ligands. Biochemistry 61, 137–149 (2022). Serganov, A. & Nudler, E. A decade of riboswitches. Cell 152, 17–24 (2013). Blount, K. F. et al. Novel Riboswitch-Binding Flavin Analog That Protects Mice against Clostridium difficile Infection without Inhibiting Cecal Flora. Antimicrob. Agents Chemother. 59, 5736 (2015). Vicens, Q. et al. Structure-Activity Relationship of Flavin Analogues That Target the Flavin Mononucleotide Riboswitch. ACS Chem. Biol. 13, 2908–2919 (2018). Rizvi, N. F. et al. Discovery of Selective RNA-Binding Small Molecules by Affinity-Selection Mass Spectrometry. ACS Chem. Biol. 13, 820–831 (2018). Howe, J. A. et al. Selective small-molecule inhibition of an RNA structural element. Nature 526, 672–677 (2015). Kräutler, B. & EurJ, C. Antivitamins B12—Some Inaugural Milestones. Chem. – A Eur. J. 26, 15438–15445 (2020). Nahvi, A. et al. Genetic Control by a Metabolite Binding mRNA. Chem. Biol. 9, 1043–1049 (2002). Mccown, P. J., Corbino, K. A., Stav, S., Sherlock, M. E. & Breaker, R. R. Riboswitch diversity and distribution. RNA 23, 995–1011 (2017). Lennon, S. R. & Batey, R. T. Regulation of Gene Expression Through Effector-dependent Conformational Switching by Cobalamin Riboswitches. J. Mol. Biol. 434, 167585 (2022). Polaski, J. T., Webster, S. M., Johnson, J. E. & Batey, R. T. Cobalamin riboswitches exhibit a broad range of ability to discriminate between methylcobalamin and adenosylcobalamin. J. Biol. Chem. 292, 11650–11658 (2017). Demerre, L. J. & Wilson, C. Photolysis of vitamin B12. J. Am. Pharm. Assoc. 45, 129–134 (1956). Juzeniene, A. & Nizauskaite, Z. Photodegradation of cobalamins in aqueous solutions and in human blood. J. Photochem. Photobiol. B. 122, 7–14 (2013). Johnson, J. E., Reyes, F. E., Polaski, J. T. & Batey, R. T. B12 cofactors directly stabilize an mRNA regulatory switch. Nat. 2012 4927427 492, 133–137 (2012). Lennon, S. R. et al. Targeting Riboswitches with Beta-Axial-Substituted Cobalamins. ACS Chem. Biol. 18, 1136–1147 (2023). Peselis, A. & Serganov, A. Structural insights into ligand binding and gene expression control by an adenosylcobalamin riboswitch. Nat. Struct. Mol. Biol. 2012 1911 19, 1182–1184 (2012). Davila-Calderon, J. et al. IRES-targeting small molecule inhibits enterovirus 71 replication via allosteric stabilization of a ternary complex. Nat. Commun. 2020 111 11, 1–13 (2020). Chen, J. L. et al. Design, Optimization, and Study of Small Molecules That Target Tau Pre-mRNA and Affect Splicing. J. Am. Chem. Soc. 142, 8706–8727 (2020). Matyjasik, M. M., Hall, S. D. & Batey, R. T. High Affinity Binding of N2-Modified Guanine Derivatives Significantly Disrupts the Ligand Binding Pocket of the Guanine Riboswitch. Molecules 25, (2020). Chromiński, M., Lewalska, A., Karczewski, M. & Gryko, D. Vitamin B12 derivatives for orthogonal functionalization. J. Org. Chem. 79, 7532–7542 (2014). Chromiński, M., Lewalska, A. & Gryko, D. Reduction-free synthesis of stable acetylide cobalamins. Chem. Commun. 49, 11406–11408 (2013). Donlic, A. et al. R-BIND 2.0: An Updated Database of Bioactive RNA-Targeting Small Molecules and Associated RNA Secondary Structures. ACS Chem. Biol. 17, 1556–1566 (2022). Morgan, B. S. et al. R-BIND: An Interactive Database for Exploring and Developing RNA-Targeted Chemical Probes. ACS Chem. Biol. 14, 2691–2700 (2019). Morgan, B. S., Forte, J. E., Culver, R. N., Zhang, Y. & Hargrove, A. E. Discovery of Key Physicochemical, Structural, and Spatial Properties of RNA-Targeted Bioactive Ligands. Angew. Chemie Int. Ed. 56, 13498–13502 (2017). Braselmann, E. et al. A multicolor riboswitch-based platform for imaging of RNA in live mammalian cells. Nat. Chem. Biol. 2018 1410 14, 964–971 (2018). Morgan, B. S., Forte, J. E., Culver, R. N., Zhang, Y. & Hargrove, A. E. Discovery of Key Physicochemical, Structural, and Spatial Properties of RNA-Targeted Bioactive Ligands. Angew. Chem. Int. Ed. Engl. 56, 13498–13502 (2017). Stratton, C. F., Newman, D. J. & Tan, D. S. Cheminformatic comparison of approved drugs from natural product versus synthetic origins. Bioorg. Med. Chem. Lett. 25, 4802–4807 (2015). Patwardhan, N. N., Cai, Z., Umuhire Juru, A. & Hargrove, A. E. Driving factors in amiloride recognition of HIV RNA targets. Org. Biomol. Chem. 17, 9313–9320 (2019). Patwardhan, N. N. et al. Amiloride as a new RNA-binding scaffold with activity against HIV-1 TAR. Medchemcomm 8, 1022 (2017). Cai, Z., Zafferani, M., Akande, O. M. & Hargrove, A. E. Quantitative Structure-Activity Relationship (QSAR) Study Predicts Small-Molecule Binding to RNA Structure. J. Med. Chem. 65, 7262–7277 (2022). Polaski, J. T., Holmstrom, E. D., Nesbitt, D. J. & Batey, R. T. Mechanistic Insights into Cofactor-Dependent Coupling of RNA Folding and mRNA Transcription/Translation by a Cobalamin Riboswitch. Cell Rep. 15, 1100–1110 (2016). Neidhardt, F. C. Escherichia coli and Salmonella typhimurium - Cellular and Molecular Biology . Washington: American Society for Microbiology (Washington: American Society for Microbiology, 1987). Figge, R. M. Methionine Biosynthesis in Escherichia coli and Corynebacterium glutamicum. in Wendisch , V.F. (eds) Amino Acid Biosynthesis ~ Pathways, Regulation and Metabolic Engineering 163–193 (Springer, Berlin, Heidelberg, 2006). doi: 10.1007/7171_2006_059 . Lipinski, C. A. Lead- and drug-like compounds: the rule-of-five revolution. Drug Discov. Today. Technol. 1, 337–341 (2004). Lipinski, C. A., Lombardo, F., Dominy, B. W. & Feeney, P. J. Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv. Drug Deliv. Rev. 46, 3–26 (2001). Wicks, S. L. & Hargrove, A. E. Fluorescent indicator displacement assays to identify and characterize small molecule interactions with RNA. Methods 167, 3–14 (2019). Tong, Y. et al. Programming inactive RNA-binding small molecules into bioactive degraders. Nat. 2023 6187963 618, 169–179 (2023). Geng, A., Roy, R. & Al-Hashimi, H. M. Conformational penalties: New insights into nucleic acid recognition. Curr. Opin. Struct. Biol. 89, (2024). Stelzer, A. C. et al. Discovery of selective bioactive small molecules by targeting an RNA dynamic ensemble. Nat. Chem. Biol. 2011 78 7, 553–559 (2011). Panei, F. P., Gkeka, P. & Bonomi, M. Identifying small-molecules binding sites in RNA conformational ensembles with SHAMAN. Nat. Commun. 2024 151 15, 1–15 (2024). Zhou, Y. & Chen, S.-J. Advances in machine-learning approaches to RNA-targeted drug design. Artif. Intell. Chem. 2, 100053 (2024). Morishita, E. C. & Nakamura, S. Recent applications of artificial intelligence in RNA-targeted small molecule drug discovery. Expert Opin. Drug Discov. 19, 415–431 (2024). Sonar, M. V. et al. Fluorescence Detection of KRAS2 mRNA Hybridization in Lung Cancer Cells with PNA-Peptides Containing an Internal Thiazole Orange. Bioconjug. Chem. 25, 1697–1708 (2014). Wierzba, A. J., Richards, E. M., Lennon, S. R., Batey, R. T. & Palmer, A. E. Unveiling the promise of peptide nucleic acids as functional linkers for an RNA imaging platform. RSC Chem. Biol. (2024) doi: 10.1039/D4CB00274A . Corson, T. W., Aberle, N. & Crews, C. M. Design and applications of bifunctional small molecules: Why two heads are better than one. ACS Chem. Biol. 3, 677–692 (2008). Angelbello, A. J. et al. Precise small-molecule cleavage of an r(CUG) repeat expansion in a myotonic dystrophy mouse model. Proc. Natl. Acad. Sci. U. S. A. 116, 7799–7804 (2019). Edwards, A. L., Garst, A. D. & Batey, R. T. Determining structures of RNA aptamers and riboswitches by X-ray crystallography. Methods Mol. Biol. 535, 135–163 (2009). Topliss, J. G. & Edwards, R. P. Chance factors in studies of quantitative structure-activity relationships. J. Med. Chem. 22, 1238–1244 (1979). Cheatham, T. E., Miller, J. L., Fox, T., Darden, T. A. & Kollman, P. A. Molecular Dynamics Simulations on Solvated Biomolecular Systems: The Particle Mesh Ewald Method Leads to Stable Trajectories of DNA, RNA, and Proteins. J. Am. Chem. Soc. 117, 4193–4194 (1995). Case, D. A. et al. The Amber biomolecular simulation programs. J. Comput. Chem. 26, 1668–1688 (2005). Izadi, S., Anandakrishnan, R. & Onufriev, A. V. Building water models: A different approach. J. Phys. Chem. Lett. 5, 3863–3871 (2014). Zgarbová, M. et al. Refinement of the Cornell et al. Nucleic acids force field based on reference quantum chemical calculations of glycosidic torsion profiles. J. Chem. Theory Comput. 7, 2886–2902 (2011). Steinbrecher, T., Latzer, J. & Case, D. A. Revised AMBER parameters for bioorganic phosphates. J. Chem. Theory Comput. 8, 4405–4412 (2012). Li, P. & Merz, K. M. MCPB.py: A Python Based Metal Center Parameter Builder. J. Chem. Inf. Model. 56, 599–604 (2016). Hopkins, C. W., Le Grand, S., Walker, R. C. & Roitberg, A. E. Long-time-step molecular dynamics through hydrogen mass repartitioning. J. Chem. Theory Comput. 11, 1864–1874 (2015). Roe, D. R. & Cheatham, T. E. PTRAJ and CPPTRAJ: Software for processing and analysis of molecular dynamics trajectory data. J. Chem. Theory Comput. 9, 3084–3095 (2013). Minor, W., Cymborowski, M., Otwinowski, Z. & Chruszcz, M. HKL-3000: the integration of data reduction and structure solution–from diffraction images to an initial model in minutes. Acta Crystallogr. D. Biol. Crystallogr. 62, 859–866 (2006). Kabsch, W. XDS. Acta Crystallogr. Sect. D Struct. Biol. 66, 125–132 (2010). Bunkóczi, G. et al. Phaser.MRage: automated molecular replacement. Acta Crystallogr. Sect. D Biol. Crystallogr. 69, 2276–2286 (2013). Adams, P. D. et al. The Phenix software for automated determination of macromolecular structures. Methods 55, 94–106 (2011). Moriarty, N. W., Grosse-Kunstleve, R. W. & Adams, P. D. electronic Ligand Builder and Optimization Workbench (eLBOW): a tool for ligand coordinate and restraint generation. Acta Crystallogr. Sect. D Biol. Crystallogr. 65, 1074–1080 (2009). Terwilliger, T. C., Klei, H., Adams, P. D., Moriarty, N. W. & Cohn, J. D. Automated ligand fitting by core-fragment fitting and extension into density. Acta Crystallogr. Sect. D Biol. Crystallogr. 62, 915–922 (2006). Adams, P. D., Pannu, N. S., Read, R. J. & Brunger, A. T. Extending the limits of molecular replacement through combined simulated annealing and maximum-likelihood refinement. Acta Crystallogr. D. Biol. Crystallogr. 55, 181–190 (1999). Chou, F. C., Sripakdeevong, P., Dibrov, S. M., Hermann, T. & Das, R. Correcting pervasive errors in RNA crystallography through enumerative structure prediction. Nat. Methods 2012 101 10, 74–76 (2013). Additional Declarations Yes there is potential Competing Interest. R.T.B. serves on the Scientific Advisory Boards of Expansion Therapeutics, SomaLogic, and MeiraGTx. Supplementary Files GA.png Graphical abstract 250113LTONCBCblpapersupplementalinformation.docx Designing small molecules that target a cryptic RNA binding site via base displacement Extendeddata.docx Cite Share Download PDF Status: Published Journal Publication published 29 Aug, 2025 Read the published version in Nature Chemical Biology → Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-5836924","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":406450013,"identity":"93d7ce64-7adf-4172-a8f0-8518af8583d8","order_by":0,"name":"Robert Batey","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABBklEQVRIiWNgGAWjYFACHjCZACY/MDDIgGgJIGZswKslAQjZgKpmQE0gXgszDzFazNt7Dz6u/MGQxz+/+Zi0TY0ND7/04Yc3PjDYyG44gF2LzJlzyYZnEhiKJY6xpUnnHEvjkexLM7acwZBmjEuLhESOmWRDAkNiwzEeM+nchsM8BmcYzKR5GA4n4tFi/hOkZf4x/m/SlmAt7N+AWv7j02LGCNKy4RgPmzQjWAsPyJYDuLXwnEuWbEiTKDY8BvRCD8gvPTzFljMMko1n4tLC3nvwY4ONTZ7cYWBA/aixkePnYd9440OFnWwfDi0wnSCCRQIhYIBXORwwfyBO3SgYBaNgFIw0AADqH1X0n47cTAAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0000-0002-1384-6625","institution":"University of Colorado Boulder","correspondingAuthor":true,"prefix":"","firstName":"Robert","middleName":"","lastName":"Batey","suffix":""},{"id":406450014,"identity":"1981033a-873e-44b6-891a-2a34b6af5ada","order_by":1,"name":"Lukasz Olenginski","email":"","orcid":"","institution":"University of Colorado Boulder","correspondingAuthor":false,"prefix":"","firstName":"Lukasz","middleName":"","lastName":"Olenginski","suffix":""},{"id":406450015,"identity":"d163e43c-6a66-4e6d-a630-e9240b925fbc","order_by":2,"name":"Aleksandra Wierzba","email":"","orcid":"https://orcid.org/0000-0001-8347-5498","institution":"University of Colorado Boulder","correspondingAuthor":false,"prefix":"","firstName":"Aleksandra","middleName":"","lastName":"Wierzba","suffix":""},{"id":406450016,"identity":"2d021fa8-95ef-4a4d-8b31-d626178e2fbe","order_by":3,"name":"Shawn Laursen","email":"","orcid":"","institution":"University of Colorado Boulder","correspondingAuthor":false,"prefix":"","firstName":"Shawn","middleName":"","lastName":"Laursen","suffix":""}],"badges":[],"createdAt":"2025-01-15 19:40:22","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5836924/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5836924/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41589-025-02018-8","type":"published","date":"2025-08-29T04:00:00+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":74988148,"identity":"3c17c049-3e25-4be4-a1fe-e37301f25292","added_by":"auto","created_at":"2025-01-29 06:31:57","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":765250,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMotivation and design of our expanded Cbl derivative library\u003c/strong\u003e.\u003cstrong\u003e a\u003c/strong\u003e, Chemical structure of standard Cbls with their corresponding β-axial groups shown. \u003cstrong\u003eb\u003c/strong\u003e, Co-crystal structure of the \u003cem\u003eenv8\u003c/em\u003e-OHCbl complex (PDB 4FRG\u003csup\u003e20\u003c/sup\u003e). Global RNA architecture (left) is displayed as cartoon, nucleotides critical for recognition of the variable β-axial group are shown in cyan (A68), green (A20), and yellow (G19), and OHCbl is represented by magenta Van der Waals spheres. The ligand binding pocket (right) consists of a π-stacking network between nucleotides G19-A20 and A20-A68 that is unimpeded by the small (-OH) β-axial group. \u003cstrong\u003ec\u003c/strong\u003e, Simplified chemical structures of previously characterized\u003csup\u003e21\u003c/sup\u003e β-axial-modified Cbls \u003cstrong\u003e4\u003c/strong\u003e-\u003cstrong\u003e7\u003c/strong\u003e tested against \u003cem\u003eenv8\u003c/em\u003e. \u003cstrong\u003ed\u003c/strong\u003e, Reduction-free synthetic route\u003csup\u003e26,27\u003c/sup\u003e to make Cbl derivatives from CNCbl using a variable alkyne. \u003cstrong\u003ee\u003c/strong\u003e, Overview of our expanded β-axial-modified Cbl library.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-5836924/v1/5ead9443e3e3a3d275fa988d.png"},{"id":74988150,"identity":"d3becc57-c74e-4038-9a93-28d5de87f4f1","added_by":"auto","created_at":"2025-01-29 06:31:57","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":354638,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMeasuring and predicting RNA binding affinity. a\u003c/strong\u003e, Binding curve of the fluorescence induction of CNCbl-5×PEG-ATTO590 binding to \u003cem\u003eenv8\u003c/em\u003e. \u003cstrong\u003eb\u003c/strong\u003e, Representative displacement binding curve for MeCbl and Cbl \u003cstrong\u003e4\u003c/strong\u003e. \u003cstrong\u003ec\u003c/strong\u003e, Log-transformed K\u003csub\u003eD\u003c/sub\u003e values for \u003cstrong\u003e1\u003c/strong\u003e-\u003cstrong\u003e44\u003c/strong\u003e. In \u003cstrong\u003ea\u003c/strong\u003e-\u003cstrong\u003ec\u003c/strong\u003e, mean and s.d. from independent experiments (\u003cem\u003en\u003c/em\u003e = 3-5) are shown. \u003cstrong\u003ed\u003c/strong\u003e, Locations of the training and test set from the Q\u003csup\u003e2\u003c/sup\u003e-focused modeling in two-dimensional chemical space constructed from PC1 and PC2 of the whole data set. \u003cstrong\u003ee\u003c/strong\u003e, Measured ln [K\u003csub\u003eD\u003c/sub\u003e] values plotted with the value predicted by the Q\u003csup\u003e2\u003c/sup\u003e-focused model. \u003cstrong\u003ef\u003c/strong\u003e, Overview of our lead compound screen from a 513-compound alkyne library using our binding-based models. \u003cstrong\u003eg\u003c/strong\u003e, Simplified chemical structures of representative derivatives that our binding-based models predict to be tight (\u003cstrong\u003e45\u003c/strong\u003e-\u003cstrong\u003e48\u003c/strong\u003e), moderate (\u003cstrong\u003e49\u003c/strong\u003e and \u003cstrong\u003e50\u003c/strong\u003e), and weak (\u003cstrong\u003e51\u003c/strong\u003e-\u003cstrong\u003e53\u003c/strong\u003e) binders. \u003cstrong\u003eh\u003c/strong\u003e, Measured ln [K\u003csub\u003eD\u003c/sub\u003e] values plotted with the value predicted by our binding-based models. The experimental data are represented as mean and s.d. from independent experiments (\u003cem\u003en\u003c/em\u003e = 3-5) and the predicted data are shown as the mean and s.d. from independent predictions (\u003cem\u003en\u003c/em\u003e = 3) from our three models (baseline, R\u003csup\u003e2\u003c/sup\u003e-focused, and Q\u003csup\u003e2\u003c/sup\u003e-focused).\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-5836924/v1/31b901db40ab0943905bb63c.png"},{"id":74988154,"identity":"ae575c1b-d25b-4a46-8811-195cad7f13c2","added_by":"auto","created_at":"2025-01-29 06:31:57","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":370780,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMeasuring and predicting RNA regulatory activity. a\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eBox-and-whisker plot of the OD\u003csub\u003e600\u003c/sub\u003e-corrected relative fluorescence units (RFU) of \u003cem\u003eE. coli\u003c/em\u003e cells with and without reporter plasmid (\u003cem\u003eenv8\u003c/em\u003e-GFPuv) and ligand (MeCbl and Cbl \u003cstrong\u003e4\u003c/strong\u003e) added. \u003cstrong\u003eb\u003c/strong\u003e, Fold repression (defined as the ratio of “(–)-Cbl RFU” and “(+)-Cbl RFU” values)\u003csup\u003e37\u003c/sup\u003e for our \u003cem\u003eenv8\u003c/em\u003e reporter system in the presence of Cbl \u003cstrong\u003e1\u003c/strong\u003e-\u003cstrong\u003e44\u003c/strong\u003e. Mean and s.d. from biological replicates (\u003cem\u003en\u003c/em\u003e = 4) are shown. \u003cstrong\u003ec\u003c/strong\u003e, Plot comparing the log-transformed fold repression and K\u003csub\u003eD\u003c/sub\u003e data. \u003cstrong\u003ed\u003c/strong\u003e, Locations of the training and test set from the Q\u003csup\u003e2\u003c/sup\u003e-focused modeling in two-dimensional chemical space constructed from PC1 and PC2 of the whole data set. \u003cstrong\u003ee\u003c/strong\u003e, Measured fold repression values plotted with the value predicted by the Q\u003csup\u003e2\u003c/sup\u003e-focused model. \u003cstrong\u003ef\u003c/strong\u003e, Overview of our lead compound screen from a 513-compound alkyne library using our function-based models. \u003cstrong\u003eg\u003c/strong\u003e, Simplified chemical structures of representative derivatives that our function-based models predict to be strong (\u003cstrong\u003e54\u003c/strong\u003e-\u003cstrong\u003e57\u003c/strong\u003e), moderate (\u003cstrong\u003e58\u003c/strong\u003e and \u003cstrong\u003e59\u003c/strong\u003e), and weak (\u003cstrong\u003e60\u003c/strong\u003e-\u003cstrong\u003e62\u003c/strong\u003e) repressors. \u003cstrong\u003eh\u003c/strong\u003e, Measured fold repression values plotted with the value predicted by our function-based models. The experimental data are represented as mean and s.d. from biological replicates (\u003cem\u003en\u003c/em\u003e = 4) and the predicted data are shown as the mean and s.d. from independent predictions (\u003cem\u003en\u003c/em\u003e = 3) from our three models (i.e., baseline, R\u003csup\u003e2\u003c/sup\u003e-focused, and Q\u003csup\u003e2\u003c/sup\u003e-focused).\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-5836924/v1/d6dedff9f15f4bc3628255d4.png"},{"id":74988770,"identity":"c70fb191-c123-4e80-a8f1-dc7cc2553b06","added_by":"auto","created_at":"2025-01-29 06:39:57","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":893292,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCo-crystal structures confirm A20 displacement and inform lead compound design. a\u003c/strong\u003e, Representative MD starting states of the \u003cem\u003eenv8\u003c/em\u003e-CNCbl and \u003cem\u003eenv8\u003c/em\u003e-Cbl \u003cstrong\u003e4\u003c/strong\u003e complexes from one (of three) independent 1 µs trajectory. The A20(C6)-G19(C6) and A20(C6)-A68(N7) distances are shown as double arrows. In all structural representations, the binding pocket nucleotides G19 (yellow), A20 (green), and A68 (cyan) are numbered in reference to full-length \u003cem\u003eenv8\u003c/em\u003e and colored, the β-axial group is shown in magenta, and dashed lines represent proposed hydrogen bonds. \u003cstrong\u003eb\u003c/strong\u003e, The average A20(C6)-G19(C6) (left) and A20(C6)-A68(N7) (right) distances measured over the course of three independent MD trajectories for both \u003cem\u003eenv8\u003c/em\u003e-CNCbl and \u003cem\u003eenv8\u003c/em\u003e-Cbl \u003cstrong\u003e4\u003c/strong\u003e complexes. \u003cstrong\u003ec\u003c/strong\u003e, Co-crystal structure of \u003cem\u003eenv2\u003c/em\u003e-CNCbl showing only the binding pocket nucleotides and the β-axial group. All mesh representations correspond to a simulated annealing 2F\u003csub\u003eo\u003c/sub\u003e-F\u003csub\u003ec \u003c/sub\u003emap where A20 and the ligand were omitted from the model and are shown at 1 σ contour. \u003cstrong\u003ed\u003c/strong\u003e, Same as in \u003cstrong\u003ec\u003c/strong\u003e but for \u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cstrong\u003e4\u003c/strong\u003e. \u003cstrong\u003ee\u003c/strong\u003e, Same as in \u003cstrong\u003ec\u003c/strong\u003e but for \u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cstrong\u003e32 \u003c/strong\u003e(left) and also showing a Van der Waal sphere representation of the binding pocket (right). \u003cstrong\u003ef\u003c/strong\u003e, Same as in \u003cstrong\u003ee \u003c/strong\u003ebut for \u003cem\u003eenv2\u003c/em\u003e-Cbl\u003cstrong\u003e 29\u003c/strong\u003e. \u003cstrong\u003eg\u003c/strong\u003e, Same as in \u003cstrong\u003ec \u003c/strong\u003ebut for \u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cstrong\u003e42\u003c/strong\u003e. \u003cstrong\u003eh\u003c/strong\u003e, Schematic representation of design principles that emerge from our 11 RNA-ligand co-crystal structures.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-5836924/v1/6eefc6ddb66cc9842772ec0c.png"},{"id":74988920,"identity":"6b1f42a3-1589-49f5-913d-81e961e190b0","added_by":"auto","created_at":"2025-01-29 06:47:57","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":423872,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eStructure-based design of lead compounds with affinity exceeding the native ligand. a\u003c/strong\u003e, Modified click reaction\u003csup\u003e26\u003c/sup\u003e to make Cbl derivatives from \u003cstrong\u003e58\u003c/strong\u003e using a variable azide. \u003cstrong\u003eb\u003c/strong\u003e, Simplified chemical structures of Cbls \u003cstrong\u003e63\u003c/strong\u003e-\u003cstrong\u003e65\u003c/strong\u003e that were made to systematically test our design principles. \u003cstrong\u003ec\u003c/strong\u003e, Fluorescence displacement binding curve for CNCbl and \u003cstrong\u003e63\u003c/strong\u003e-\u003cstrong\u003e65\u003c/strong\u003e. K\u003csub\u003eD\u003c/sub\u003e values are reported as mean and s.d. from independent experiments (\u003cem\u003en\u003c/em\u003e = 3-5). \u003cstrong\u003ed\u003c/strong\u003e, To verify the near-picomolar binding of\u003cstrong\u003e 63\u003c/strong\u003e and\u003cstrong\u003e 64\u003c/strong\u003e, we used ITC with the \u003cem\u003eenv8 \u003c/em\u003eaptamer domain. Given that this RNA shows weaker ligand binding than the full-length \u003cem\u003eenv8\u003c/em\u003e,\u003csup\u003e20\u003c/sup\u003e it is in the perfect regime for ITC. Representative ITC binding isotherm for the \u003cem\u003eenv8 \u003c/em\u003eaptamer domain with CNCbl, \u003cstrong\u003e63\u003c/strong\u003e, and \u003cstrong\u003e64\u003c/strong\u003e. \u003cstrong\u003ee\u003c/strong\u003e, Co-crystal structure of \u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cstrong\u003e63\u003c/strong\u003e showing only the binding pocket nucleotides and the β-axial group. Proposed hydrogen bonds, which are weakly supported by the electron density, are shown as double arrows. All mesh representations correspond to a simulated annealing 2F\u003csub\u003eo\u003c/sub\u003e-F\u003csub\u003ec \u003c/sub\u003emap where A20 and the ligand were omitted from the model and are shown at 1 σ contour. \u003cstrong\u003ef\u003c/strong\u003e, Same as in \u003cstrong\u003ee \u003c/strong\u003ebut for \u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cstrong\u003e64\u003c/strong\u003e. \u003cstrong\u003eg\u003c/strong\u003e, Fold repression for our reporter system in the presence of CNCbl and \u003cstrong\u003e63\u003c/strong\u003e-\u003cstrong\u003e65 \u003c/strong\u003eand for cells with CNCbl and excess amounts of \u003cstrong\u003e63\u003c/strong\u003e-\u003cstrong\u003e65 \u003c/strong\u003eadded. Mean and s.d. from biological replicates (\u003cem\u003en\u003c/em\u003e = 4) are shown.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003eh\u003c/strong\u003e, Simplified chemical structure of the titratable pyridine Cbl \u003cstrong\u003e66\u003c/strong\u003e. \u003cstrong\u003ei\u003c/strong\u003e, Representative ITC binding isotherms for \u003cem\u003eenv8 \u003c/em\u003eaptamer domain with CNCbl, \u003cstrong\u003e29\u003c/strong\u003e, and \u003cstrong\u003e66\u003c/strong\u003e at pH 8 and 5. For all ITC data, K\u003csub\u003eD\u003c/sub\u003e values are reported as mean and s.d. from independent experiments (\u003cem\u003en\u003c/em\u003e = 3).\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-5836924/v1/b5eb91ce7c46ea3c413a3a93.png"},{"id":74988152,"identity":"cf4c3636-d47b-4280-84c5-f6da9fbb5e8a","added_by":"auto","created_at":"2025-01-29 06:31:57","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":603566,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eStructure-informed screen identifies cryptic binders of \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003eenv8\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e divorced from Cbl. a\u003c/strong\u003e, Chemical structure of the eight hits that emerged from our structure-informed docking, with their biphenyl scaffold highlighted in gray. \u003cstrong\u003eb\u003c/strong\u003e, TO-based displacement assay to identify \u003cem\u003eenv8\u003c/em\u003e binding compounds that induce fluorescence attenuation. This was used as an orthogonal binding experiment because we observed ligand-induced fluorescence induction of CNCbl-5×PEG-ATTO590. \u003cstrong\u003ec\u003c/strong\u003e, Representative MST trace of fluorescently labeled \u003cem\u003eenv8\u003c/em\u003e alone and in the presence of CNCbl, \u003cstrong\u003eE2\u003c/strong\u003e, and \u003cstrong\u003eE5\u003c/strong\u003e. All ligands induced an observable change to the MST trace of \u003cem\u003eenv8\u003c/em\u003e suggestive of binding. \u003cstrong\u003ed\u003c/strong\u003e, TO-displacement binding curve for CNCbl, \u003cstrong\u003eE2\u003c/strong\u003e, and \u003cstrong\u003eE5\u003c/strong\u003e. K\u003csub\u003eD\u003c/sub\u003e values are reported as mean and s.d. from independent experiments (\u003cem\u003en\u003c/em\u003e = 3). \u003cstrong\u003ee\u003c/strong\u003e, Native polyacrylamide gel showing a gradual shift from a slower migrating, extended conformation to a faster migrating, kissing-loop conformation of \u003cem\u003eenv8\u003c/em\u003e upon titration of CNCbl (0.01-10 µM). \u003cstrong\u003ef\u003c/strong\u003e, Same as in \u003cstrong\u003ee\u003c/strong\u003e but after addition of CNCbl, \u003cstrong\u003eE2\u003c/strong\u003e, and/or \u003cstrong\u003eE5 \u003c/strong\u003eto demonstrate competitive binding of \u003cstrong\u003eE2\u003c/strong\u003e and \u003cstrong\u003eE5\u003c/strong\u003e. \u003cstrong\u003eg\u003c/strong\u003e, Representative MST trace from a competition experiment where \u003cem\u003eenv8\u003c/em\u003e-CNCbl was monitored with and without the addition of excess amounts of \u003cstrong\u003eE2\u003c/strong\u003e or \u003cstrong\u003eE5\u003c/strong\u003e. Both ligands induced an observable change to the MST trace of \u003cem\u003eenv8\u003c/em\u003e-CNCbl suggestive of competitive binding. \u003cstrong\u003eh\u003c/strong\u003e, Top-ranked docking poses of \u003cstrong\u003eE2\u003c/strong\u003e\u003cem\u003e \u003c/em\u003eand \u003cstrong\u003eE5\u003c/strong\u003e\u003cem\u003e \u003c/em\u003eto \u003cem\u003eenv2 \u003c/em\u003ein the\u003cem\u003e \u003c/em\u003eA20-out and A20-in state.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-5836924/v1/66956d51801afd68990483cc.png"},{"id":90216939,"identity":"d1d63703-55fc-47ca-be2d-dd56074913e3","added_by":"auto","created_at":"2025-08-30 07:07:07","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":5785034,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5836924/v1/ea443801-6719-4ad1-b850-092bb277cdea.pdf"},{"id":74988151,"identity":"c26b7885-d81a-4e57-bdf7-cba432c42999","added_by":"auto","created_at":"2025-01-29 06:31:57","extension":"png","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":361304,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGraphical abstract\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"GA.png","url":"https://assets-eu.researchsquare.com/files/rs-5836924/v1/defcc0afb8520050408bbcba.png"},{"id":74988186,"identity":"370df375-db3d-40a4-ac62-a89e74b3cd7f","added_by":"auto","created_at":"2025-01-29 06:31:58","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":109444751,"visible":true,"origin":"","legend":"\u003cp\u003eDesigning small molecules that target a cryptic RNA binding site via base displacement\u003c/p\u003e","description":"","filename":"250113LTONCBCblpapersupplementalinformation.docx","url":"https://assets-eu.researchsquare.com/files/rs-5836924/v1/5f87f03cc3c980cbc97a61c4.docx"},{"id":74988166,"identity":"3fdf3693-44c8-4328-9c3f-c62f55f5f77a","added_by":"auto","created_at":"2025-01-29 06:31:57","extension":"docx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":2535944,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cbr\u003e\u003c/p\u003e","description":"","filename":"Extendeddata.docx","url":"https://assets-eu.researchsquare.com/files/rs-5836924/v1/6707f9cd1fe55d34e7cdadfa.docx"}],"financialInterests":"\u003cb\u003eYes\u003c/b\u003e there is potential Competing Interest.\nR.T.B. serves on the Scientific Advisory Boards of Expansion Therapeutics, SomaLogic, and MeiraGTx.","formattedTitle":"Designing small molecules that target a cryptic RNA binding site via base displacement","fulltext":[{"header":"Introduction","content":"\u003cp\u003eThe design and discovery of small molecules that selectively target RNA is a longstanding problem in chemical biology. This emerging field has the potential to develop chemical probes of RNA function and therapeutics to treat RNA-mediated disease.\u003csup\u003e\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e Despite ongoing efforts, there is only one FDA-approved small molecule that targets RNA outside the ribosome, risdiplam.\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e This reinforces the notion that targeting RNA is difficult and often limited by the lack of druggable RNA targets and/or quantitative structure-activity relationship (QSAR) studies.\u003csup\u003e\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e Compared to proteins, which are routinely targeted with small molecule therapeutics,\u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e much less is known about the principles governing RNA-ligand interactions to guide discovery efforts.\u003c/p\u003e \u003cp\u003eOne approach to address this knowledge gap is the application of natural RNA-small molecule interactions such as those observed in riboswitches. These bacterial mRNA elements have evolved tertiary structures to selectively bind metabolites and control gene expression,\u003csup\u003e\u003cspan additionalcitationids=\"CR7\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e making them ideal models to explore RNA-ligand interactions. For example, BioRelix\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e and Merck\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e,\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e have reported promising lead compounds against the FMN riboswitch that demonstrate efficacy in animal models. Another system of therapeutic interest\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e is the cobalamin (Cbl) riboswitch,\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e which is broadly distributed across bacteria.\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e Different classes of Cbl riboswitches exhibit distinct binding preferences to forms of Cbl that differ only at their β-axial position (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea).\u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e,\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e The biological forms of Cbl are 5\u0026prime;-deoxyadenosylCbl (AdoCbl) and methylCbl (MeCbl, \u003cb\u003e1\u003c/b\u003e),\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e but standard forms include the photolysis\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e product hydroxoCbl (OHCbl, \u003cb\u003e2\u003c/b\u003e) and the photostable\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e cyanoCbl (CNCbl, \u003cb\u003e3\u003c/b\u003e) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea).\u003c/p\u003e \u003cp\u003eWe recently identified several β-axial derivatives \u003cb\u003e4\u003c/b\u003e\u0026ndash;\u003cb\u003e7\u003c/b\u003e that bind the \u003cem\u003eenv8\u003c/em\u003e MeCbl-selective riboswitch\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e and regulate in-cell function (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb,c).\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e This was a surprising result given that these derivatives host bulky β-axial moieties (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec) that present a significant steric problem within the binding pocket (\u003cb\u003eSupplementary Fig.\u0026nbsp;1a\u003c/b\u003e). Chemical probing suggests that recognition of the higher affinity \u003cb\u003e4\u003c/b\u003e and \u003cb\u003e7\u003c/b\u003e involves the displacement of an adenosine (A20) from the RNA core (\u003cb\u003eSupplementary Fig.\u0026nbsp;1b\u003c/b\u003e), which is likely replaced by the β-axial group.\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e Displacement of A20 would yield a cryptic binding site distinct from the displacement of an adjacent adenosine by the 5\u0026prime;-deoxyadenosyl moiety observed in AdoCbl-selective riboswitches (\u003cb\u003eSupplementary Fig.\u0026nbsp;1c\u003c/b\u003e).\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e Our preliminary data suggest that the cryptic binding of \u003cb\u003e4\u003c/b\u003e and \u003cb\u003e7\u003c/b\u003e may explain their increased affinity and that chemical modifications to the β-axial group have robust effects on RNA binding.\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eIn this work, we synthesized an expanded library of Cbl derivatives that host systematically varied β-axial moieties (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ed,e) and employed a set of biochemical and cell-based assays to quantify their RNA binding affinities and regulatory activities. These data were used in a predictive modeling platform to determine the ligand properties associated with tight affinity and strong regulatory activity. As a complementary approach, we used structure-based design to unmask the cryptic β-axial binding pocket and discover lead compounds with affinity exceeding the native ligand. This structure-informed approach also enabled the identification of novel cryptic binders that are chemically distinct from Cbl. Collectively, our work outlines the molecular determinants of specific and high affinity base displacement RNA binding modes, which can guide future efforts to target these common RNA interactions.\u003csup\u003e\u003cspan additionalcitationids=\"CR23 CR24\" citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eSmall molecule library design and synthesis\u003c/h2\u003e \u003cp\u003eOur Cbl derivative library was synthesized using a previously established reduction-free method\u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e,\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e where the cyano group at the β-axial position of CNCbl is replaced by the R-group of a variable alkyne (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ed). While this approach enabled the synthesis of a diverse assortment of photostable β-axial modified Cbls \u003cb\u003e8\u003c/b\u003e\u0026ndash;\u003cb\u003e44\u003c/b\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ee), reactions with alkynes containing amines or protonated nitrogen atoms were unsuccessful (\u003cb\u003eSupplementary Fig.\u0026nbsp;2\u003c/b\u003e). The logic of our library construction was to diversify around lead compounds \u003cb\u003e4\u003c/b\u003e and \u003cb\u003e7\u003c/b\u003e. This strategy enabled specific medicinal chemistry questions to be addressed with QSAR studies. Within this framework, we can determine which chemical features modulate RNA binding and function.\u003c/p\u003e \u003cp\u003eGiven that the majority of RNA-binding ligands are aromatic systems capable of π-stacking,\u003csup\u003e\u003cspan additionalcitationids=\"CR29\" citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e many RNA-targeting lead compounds suffer from limited solubility, low affinity, and/or lack of specificity, limiting the scope and utility of the downstream medicinal chemistry that is required for successful hit-to-lead discovery campaigns.\u003csup\u003e\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e In our approach (\u003cb\u003eSupplementary Fig.\u0026nbsp;3a\u003c/b\u003e), Cbl functions as a soluble \u0026ldquo;host\u0026rdquo; of a variable chemical \u0026ldquo;guest\u0026rdquo; that interacts with the \u003cem\u003eenv8\u003c/em\u003e RNA target in the minor groove. This target-guest interaction is anchored by contacts from the host corrin ring and α-axial group (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea) to project the variable β-axial moiety to the same spatial location within the cryptic site. Thus, our model system enables a detailed and systematic approach to determine which chemical features confer high affinity base displacement RNA binding modes, which is common throughout RNA biology\u003csup\u003e\u003cspan additionalcitationids=\"CR23 CR24\" citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e (\u003cb\u003eSupplementary Fig.\u0026nbsp;3b\u003c/b\u003e).\u003c/p\u003e \u003cp\u003e \u003cb\u003eCbl derivatives productively bind\u003c/b\u003e \u003cb\u003eenv8\u003c/b\u003e\u003c/p\u003e \u003cp\u003eTo quantify the binding of our library to \u003cem\u003eenv8\u003c/em\u003e, we used a fluorophore-conjugated CNCbl probe (CNCbl-5\u0026times;PEG-ATTO590\u003csup\u003e31\u003c/sup\u003e) that undergoes fluorescence induction upon \u003cem\u003eenv8\u003c/em\u003e binding and competed it off with ligands in our derivative library.\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e Titration of \u003cem\u003eenv8\u003c/em\u003e into the probe gave a K\u003csub\u003eD\u003c/sub\u003e of 3.2\u0026thinsp;\u0026plusmn;\u0026thinsp;0.1 nM (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea) and competitive titrations of MeCbl and Cbl \u003cb\u003e4\u003c/b\u003e into the \u003cem\u003eenv8\u003c/em\u003e-CNCbl-5\u0026times;PEG-ATTO590 complex yielded K\u003csub\u003eD\u003c/sub\u003e values of 1.4\u0026thinsp;\u0026plusmn;\u0026thinsp;0.8 nM and 10\u0026thinsp;\u0026plusmn;\u0026thinsp;5 nM, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb). Using this fluorescence displacement assay, we measured the K\u003csub\u003eD\u003c/sub\u003e values for Cbls \u003cb\u003e1\u003c/b\u003e\u0026ndash;\u003cb\u003e44\u003c/b\u003e. All ligands bound \u003cem\u003eenv8\u003c/em\u003e with sub-micromolar affinity (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec). The tightest binding derivative was \u003cb\u003e29\u003c/b\u003e (7\u0026thinsp;\u0026plusmn;\u0026thinsp;7 nM) while \u003cb\u003e12\u003c/b\u003e (800\u0026thinsp;\u0026plusmn;\u0026thinsp;700 nM) was the weakest. Importantly, \u003cb\u003e29\u003c/b\u003e functions as a binding lead with an affinity that exceeds \u003cb\u003e4\u003c/b\u003e and \u003cb\u003e7\u003c/b\u003e and is comparable (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.19) with the native MeCbl.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eQSAR analysis of RNA binding data\u003c/h3\u003e\n\u003cp\u003eTo reveal chemical trends within our binding data, we employed a QSAR analysis. We quantified the \u0026ldquo;distance from the alkyne\u0026rdquo; of our \u003cem\u003eortho\u003c/em\u003e-, \u003cem\u003emeta\u003c/em\u003e-, and \u003cem\u003epara\u003c/em\u003e-substituted \u0026ldquo;phenyl-F\u0026rdquo; (\u003cb\u003e9\u003c/b\u003e\u0026ndash;\u003cb\u003e11\u003c/b\u003e) and \u0026ldquo;phenyl-CH\u003csub\u003e3\u003c/sub\u003e\u0026rdquo; (\u003cb\u003e12\u003c/b\u003e\u0026ndash;\u003cb\u003e14\u003c/b\u003e) derivatives and compared this parameter against their K\u003csub\u003eD\u003c/sub\u003e values. We observed a strong relationship (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.97) for the smaller substituents and an even stronger relationship (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.99) for the larger groups (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea,b), suggesting steric limitations within the cryptic site. We also quantified the electron donating and withdrawing strength or Van der Waals (VdW) volume of the chemical groups in the \u003cem\u003epara\u003c/em\u003e-substituted phenyls (\u003cb\u003e4\u003c/b\u003e, \u003cb\u003e5\u003c/b\u003e, \u003cb\u003e8\u003c/b\u003e, \u003cb\u003e11\u003c/b\u003e, \u003cb\u003e14\u003c/b\u003e, and \u003cb\u003e17\u003c/b\u003e\u0026ndash;\u003cb\u003e20\u003c/b\u003e) and compared these parameters against their K\u003csub\u003eD\u003c/sub\u003e values. This analysis revealed that phenyl ring electronics (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.10) and \u003cem\u003epara\u003c/em\u003e-position sterics (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.01) both fail to explain \u003cem\u003eenv8\u003c/em\u003e binding (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec-e).\u003c/p\u003e \u003cp\u003eAdditional QSAR trends emerge from our heterocyclic Cbls (\u003cb\u003e21\u003c/b\u003e\u0026ndash;\u003cb\u003e26\u003c/b\u003e) (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ef). Comparing the affinities of phenyl \u003cb\u003e8\u003c/b\u003e (70\u0026thinsp;\u0026plusmn;\u0026thinsp;20 nM) and pyrimidine \u003cb\u003e24\u003c/b\u003e (70\u0026thinsp;\u0026plusmn;\u0026thinsp;20 nM) suggest that heterocycles have no advantage over their hydrocarbon counterpart. Data from thiophene \u003cb\u003e21\u003c/b\u003e (40\u0026thinsp;\u0026plusmn;\u0026thinsp;30 nM), thiazole \u003cb\u003e22\u003c/b\u003e (50\u0026thinsp;\u0026plusmn;\u0026thinsp;40 nM), and furan \u003cb\u003e23\u003c/b\u003e (50\u0026thinsp;\u0026plusmn;\u0026thinsp;10 nM) demonstrate that five-membered heterocycles bind just as well as six-membered ones. Finally, the increased affinity for our two-ring heterocycles \u003cb\u003e25\u003c/b\u003e (16\u0026thinsp;\u0026plusmn;\u0026thinsp;8 nM) and \u003cb\u003e26\u003c/b\u003e (12\u0026thinsp;\u0026plusmn;\u0026thinsp;5 nM) suggests that two rings are preferred, which is consistent with our hypothesis that the β-axial group π-stacks within the cryptic site. This notion is supported by the binding data of our two-ring containing Cbls \u003cb\u003e27\u003c/b\u003e\u0026ndash;\u003cb\u003e29\u003c/b\u003e, which all have affinities under 20 nM (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eg). Even the three- (\u003cb\u003e31\u003c/b\u003e, 30\u0026thinsp;\u0026plusmn;\u0026thinsp;20 nM) and four-ring (\u003cb\u003e32\u003c/b\u003e, 30\u0026thinsp;\u0026plusmn;\u0026thinsp;10 nM) containing derivatives bind with high affinity (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eg).\u003c/p\u003e \u003cp\u003eTo analyze these chemical trends in greater detail, we employed a multivariate analysis. Binding data were clustered into tight, moderate, and weak binders (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea), and 20 standard chemoinformatic parameters\u003csup\u003e\u003cspan additionalcitationids=\"CR33 CR34\" citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e (\u003cb\u003eSupplementary Table\u0026nbsp;1\u003c/b\u003e) were calculated for the β-axial group of Cbls \u003cb\u003e4\u003c/b\u003e, \u003cb\u003e5\u003c/b\u003e, and \u003cb\u003e7\u003c/b\u003e\u0026ndash;\u003cb\u003e44\u003c/b\u003e. These descriptors were used in a linear discriminant analysis (LDA)\u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e,\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e with our binding data. When viewed along the first two principal components (PC1 and PC2), each binding cluster occupies unique regions in chemical space (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb). The LDA loading plot provides qualitative trends for the molecular determinants of each cluster. For example, tight binders have more aromatic rings, moderate binders have more hydrogen bond donors, and weak binders have more accessible surface area (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec).\u003c/p\u003e\n\u003ch3\u003eQSAR model-based screening of potential binding leads\u003c/h3\u003e\n\u003cp\u003eMotivated by the LDA-identified trends, we sought to determine if the chemical identity of the β-axial group could be used to predict \u003cem\u003eenv8\u003c/em\u003e binding affinity. An expanded set of 347 physiochemical descriptors were calculated for the β-axial group of Cbls \u003cb\u003e4\u003c/b\u003e, \u003cb\u003e5\u003c/b\u003e, and \u003cb\u003e7\u003c/b\u003e\u0026ndash;\u003cb\u003e44\u003c/b\u003e and modeled against their natural log-transformed K\u003csub\u003eD\u003c/sub\u003e data using a least absolute shrinkage and selection operator (lasso)-multiple linear regression (MLR) strategy.\u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e When using all data in training, we obtained a baseline model that predicted our binding data with good accuracy (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.71) using six descriptors (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea,b), which are listed along with their physical meaning in \u003cb\u003eSupplementary Table\u0026nbsp;2\u003c/b\u003e. When using a stratified data split (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ed and \u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec) and implementing our lasso-MLR strategy, we constructed new models that prioritized either the predictive ability on the test set (Q\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e-focused) or the overall fit of the training set (R\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e-focused). The Q\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e-focused (R\u003csup\u003e2\u003c/sup\u003e\u003csub\u003eTraining\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;0.48, Q\u003csup\u003e2\u003c/sup\u003e\u003csub\u003eTest\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;0.81) and R\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e-focused (R\u003csup\u003e2\u003c/sup\u003e\u003csub\u003eTraining\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;0.65, Q\u003csup\u003e2\u003c/sup\u003e\u003csub\u003eTest\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;0.60) models showed a similar level of performance (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ee and \u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ed).\u003c/p\u003e \u003cp\u003eWe then used our binding-based models to screen for high affinity β-axial groups among a 513-compound alkyne library, which included modified-phenyl (\u0026ldquo;phenyl-X\u0026rdquo;) and aliphatic (\u0026ldquo;amide-X\u0026rdquo;) alkynes (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ef) that are compatible with our Cbl synthesis (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ed) and representative of our initial library (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ee). Among these compounds, 184 were predicted to have tighter affinities than our current binding lead \u003cb\u003e29\u003c/b\u003e (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ee). However, the majority of these small molecules were synthetically inaccessible using the non-reductive method\u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e,\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ef). These considerations left us with four remaining leads, which were synthesized into the corresponding Cbls \u003cb\u003e45\u003c/b\u003e\u0026ndash;\u003cb\u003e48\u003c/b\u003e along with compounds that were predicted to be moderate (\u003cb\u003e49\u003c/b\u003e and \u003cb\u003e50\u003c/b\u003e) and weak (\u003cb\u003e51\u003c/b\u003e\u0026ndash;\u003cb\u003e53\u003c/b\u003e) binders (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eg). We then used our fluorescence displacement assay to quantify the binding of these derivatives to \u003cem\u003eenv8\u003c/em\u003e. The experimentally derived ln [K\u003csub\u003eD\u003c/sub\u003e] data agreed well with our model-based predictions for the moderate and weak binders but deviated significantly for the predicted leads \u003cb\u003e45\u003c/b\u003e and \u003cb\u003e46\u003c/b\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eh). One explanation for this observation is the sparseness of tight binders in our dataset (8 of 44, 18%). Synthetic routes that access chemical groups facilitating high affinity RNA binding interactions (e.g., amino or other positively charged groups\u003csup\u003e\u003cspan additionalcitationids=\"CR29\" citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e) could address this issue.\u003c/p\u003e\n\u003ch3\u003eCbl derivatives promote RNA regulatory activity\u003c/h3\u003e\n\u003cp\u003eGiven that all the derivatives in our library bind \u003cem\u003eenv8\u003c/em\u003e, we wanted to determine whether they drive RNA regulatory activity in a cellular environment. We employed a previously established\u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e cell-based assay in which \u003cem\u003eenv8\u003c/em\u003e was placed upstream of GFPuv, whose expression is repressed by Cbl-dependent occlusion of the ribosome-binding site. In the absence of Cbl, we detected high fluorescence, indicative of GFPuv expression (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea). Conversely, in the presence of 10 nM of MeCbl or Cbl \u003cb\u003e4\u003c/b\u003e, we observed attenuated fluorescence, demonstrating that these ligands promote \u003cem\u003eenv8\u003c/em\u003e repression of GFPuv expression in \u003cem\u003eE. coli\u003c/em\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea).\u003c/p\u003e \u003cp\u003eTo quantify the extent to which each derivative regulates \u003cem\u003eenv8\u003c/em\u003e function, we calculated the fold repression\u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e for Cbls \u003cb\u003e1\u003c/b\u003e\u0026ndash;\u003cb\u003e44\u003c/b\u003e. While all ligands were able to regulate \u003cem\u003eenv8\u003c/em\u003e function to some extent, the repression was highly variable. The strongest repressors were \u003cb\u003e36\u003c/b\u003e (8\u0026thinsp;\u0026plusmn;\u0026thinsp;1) and \u003cb\u003e41\u003c/b\u003e (7\u0026thinsp;\u0026plusmn;\u0026thinsp;2) while \u003cb\u003e43\u003c/b\u003e (1.4\u0026thinsp;\u0026plusmn;\u0026thinsp;0.1) and \u003cb\u003e44\u003c/b\u003e (2.0\u0026thinsp;\u0026plusmn;\u0026thinsp;0.1) were the weakest (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb). Importantly, \u003cb\u003e36\u003c/b\u003e and \u003cb\u003e41\u003c/b\u003e are new functional leads with fold repression values that exceed \u003cb\u003e4\u003c/b\u003e and \u003cb\u003e7\u003c/b\u003e and are comparable (\u003cb\u003e36\u003c/b\u003e, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.31 and \u003cb\u003e41\u003c/b\u003e, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.19) with the native MeCbl (9.0\u0026thinsp;\u0026plusmn;\u0026thinsp;0.2).\u003c/p\u003e \u003cp\u003eThese data enable the exploration of the relationship between derivative binding and regulatory activity. In general, while many tight binding derivatives are also strong repressors and vice versa, this was not always true (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.34) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec). This fact reflects the complex nature of promoting a ligand-induced regulatory response, which involves Cbl cellular import, β-axial repair by endogenous enzymes, and \u003cem\u003eenv8\u003c/em\u003e-Cbl binding within the appropriate time scale of a co-transcriptional process.\u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e To investigate this further, we carried out growth experiments in ΔMetE cells, which lack the Cbl-independent methionine synthase MetE.\u003csup\u003e\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e In the absence of methionine, these cells must use Cbl for one carbon metabolism, making MeCbl essential for growth.\u003csup\u003e\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e In the presence of 10 nM derivative, both weak (Cbl \u003cb\u003e44\u003c/b\u003e) and strong (Cbl \u003cb\u003e36\u003c/b\u003e) repressors support ΔMetE growth with the same doubling time as CNCbl, suggesting that these ligands are imported and repaired to some extent (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). However, the fact that repression is variable among our derivatives in wild-type \u003cem\u003eE. coli\u003c/em\u003e cells suggests that these compounds exist predominantly in their unrepaired, β-axial derivatized state. A confounding factor in comparing the binding and functional data is potentially variable levels of derivative metabolism within the cell.\u003c/p\u003e\n\u003ch3\u003eQSAR model-based screening of potential functional leads\u003c/h3\u003e\n\u003cp\u003eWhile individual QSAR trends were less pronounced within our functional data (\u003cb\u003eSupplementary Fig.\u0026nbsp;4\u003c/b\u003e), an LDA using the same set of chemoinformatic parameters\u003csup\u003e\u003cspan additionalcitationids=\"CR33 CR34\" citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e (\u003cb\u003eSupplementary Table\u0026nbsp;1\u003c/b\u003e) demonstrated that functional clusters (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ea) occupy unique regions in chemical space (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eb). The LDA loading plot suggests that strong repressors have more hydrogen bond donors, moderate repressors have more aromatic rings, and weak repressors have more sp\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e centers (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ec). These trends motivated us to build QSAR models using the same lasso-MLR workflow\u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e described above, except using our repression data as the dependent variable. We obtained baseline, Q\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e-focused, and R\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e-focused models that predict the functional data reasonably well with four to six descriptors (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ed,e and \u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ea-d), which are listed in \u003cb\u003eSupplementary Table\u0026nbsp;3\u003c/b\u003e.\u003c/p\u003e \u003cp\u003eWe then used our function-based models to screen for strong repressive β-axial groups among the same alkyne library (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ef). Although 151 small molecules were predicted to have stronger repression than our current functional lead \u003cb\u003e36\u003c/b\u003e (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ee), many of these leads were unavailable for the same reasons described above (\u003cb\u003eExtended Data\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ef). We identified four available predicted leads, which were synthesized into the corresponding Cbl \u003cb\u003e54\u003c/b\u003e\u0026ndash;\u003cb\u003e57\u003c/b\u003e along with compounds that were predicted to be moderate (\u003cb\u003e58\u003c/b\u003e and \u003cb\u003e59\u003c/b\u003e) and weak (\u003cb\u003e60\u003c/b\u003e\u0026ndash;\u003cb\u003e62\u003c/b\u003e) repressors (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eg). The fold repression of \u003cb\u003e54\u003c/b\u003e\u0026ndash;\u003cb\u003e62\u003c/b\u003e were then measured using our reporter assay. Much like our binding data, the experimentally derived repression data agreed well with our model-based predictions for the moderate and weak repressors but deviated significantly for the predicted functional leads \u003cb\u003e54\u003c/b\u003e\u0026ndash;\u003cb\u003e56\u003c/b\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eh), which is likely explained by the confounding role of derivative metabolism on regulatory activity.\u003c/p\u003e \u003cp\u003e \u003cb\u003eenv8\u003c/b\u003e \u003cb\u003e-Cbl co-crystal structures unmask cryptic binding site\u003c/b\u003e \u003c/p\u003e \u003cp\u003eAs a complementary approach to QSAR modeling, we used structure-based design to investigate \u003cem\u003eenv8\u003c/em\u003e-ligand interactions. We carried out three independent 1 \u0026micro;s molecular dynamics (MD) simulations of \u003cem\u003eenv8\u003c/em\u003e-CNCbl and \u003cem\u003eenv8\u003c/em\u003e-Cbl \u003cb\u003e4\u003c/b\u003e and measured the A20(C6)-G19(C6) and A20(C6)-A68(N7) distances to assess A20 π-stacking with neighboring nucleotides (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea). Over the course of the MD trajectories, the A20(C6)-G19(C6) distances were identical among both complexes, while \u003cb\u003e4\u003c/b\u003e did induce slight increases in A20(C6)-A68(N7) distances (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb). However, these changes were not a result of A20 displacement but were caused by the β-axial phenyl-\u003cem\u003ep\u003c/em\u003eNO\u003csub\u003e2\u003c/sub\u003e group intercalating between A20 and A68 (\u003cb\u003eSupplementary Fig.\u0026nbsp;5\u003c/b\u003e), which is inconsistent with previous chemical probing data (\u003cb\u003eSupplementary Fig.\u0026nbsp;1b\u003c/b\u003e).\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eThe limitations of our MD analysis motivated us to employ X-ray crystallography. We determined co-crystal structures of the Cbl riboswitch aptamer domain in complex with CNCbl and \u003cb\u003e4\u003c/b\u003e. However, we only obtained diffraction-quality crystals using the closely related \u003cem\u003eenv2\u003c/em\u003e RNA, which shares 95% sequence identity with \u003cem\u003eenv8\u003c/em\u003e (\u003cb\u003eExtended Data Fig.\u0026nbsp;7a\u003c/b\u003e). Given that \u003cem\u003eenv2\u003c/em\u003e shows near-identical ligand binding and regulatory activity as \u003cem\u003eenv8\u003c/em\u003e (\u003cb\u003eExtended Data Fig.\u0026nbsp;7b,c\u003c/b\u003e), all structural information can be translated to \u003cem\u003eenv8\u003c/em\u003e. The \u003cem\u003eenv2\u003c/em\u003e-CNCbl structure shows A20 π-stacked with G19 and A68 (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ec) with strong agreement to the \u003cem\u003eenv8\u003c/em\u003e-OHCbl structure\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e (RMSD\u0026thinsp;=\u0026thinsp;0.45 \u0026Aring;) (\u003cb\u003eExtended Data Fig.\u0026nbsp;7d\u003c/b\u003e). The \u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cb\u003e4\u003c/b\u003e structure, on the other hand, reveals that A20 undergoes base displacement from the RNA core toward the major groove and that the β-axial phenyl ring π-stacks with A68 (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ed), unmasking the cryptic binding pocket in agreement with our hypothesis.\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e We also determined the structure of \u003cem\u003eenv2\u003c/em\u003e in the apo state, which shows a near-identical (RMSD\u0026thinsp;=\u0026thinsp;0.39 \u0026Aring;) binding pocket to our \u003cem\u003eenv2\u003c/em\u003e-CNCbl structure (\u003cb\u003eExtended Data Fig.\u0026nbsp;8a\u003c/b\u003e), demonstrating that CNCbl interacts with a pre-organized binding pocket. Notably, the \u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cb\u003e4\u003c/b\u003e structure reinforces the inability of the MD simulation to access base displacement of nucleotides.\u003c/p\u003e \u003cp\u003eTo gain further insight into cryptic β-axial pocket, we determined nine additional \u003cem\u003eenv2\u003c/em\u003e-ligand co-crystal structures (\u003cb\u003eExtended Data Fig.\u0026nbsp;8b-j\u003c/b\u003e). The structure of \u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cb\u003e32\u003c/b\u003e with its bulky pyrene β-axial group showcases that base displacement provides a lot of room (\u0026gt;\u0026thinsp;340 \u0026Aring;\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e) in the binding pocket (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ee). With its four phenyl rings, pyrene π-stacks well with A68 but is unable to optimize π-stacking with G19 (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ee). In contrast, the structure of \u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cb\u003e29\u003c/b\u003e demonstrates how the biphenyl β-axial group maximizes π-stacking with both G19 and A68 via rotation of the two rings relative to one another (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ef). Interestingly, our data support the fitting of two conformations of A20, with the more populated (55%) conformer partially engaged within the RNA core (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ef), providing a means to offset the energetic penalty of A20 displacement. It is important to note that π-stacking is not the only binding mode available within the cryptic site. The structure of \u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cb\u003e42\u003c/b\u003e suggests that two hydrogen bonds to A68 provide comparable affinity (50\u0026thinsp;\u0026plusmn;\u0026thinsp;10 nM) with a π-stacking interaction (e.g., \u003cb\u003e8\u003c/b\u003e, 70\u0026thinsp;\u0026plusmn;\u0026thinsp;20 nM) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eg). Together, our structural data suggest a set of design principles to develop new lead compounds: (1) hydrogen-bonding to the phosphate backbone, (2) cation-π-stacking, and (3) engagement of A20 (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eh).\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eStructure-based design identifies Cbls with affinity exceeding the native ligand\u003c/h2\u003e \u003cp\u003eIn order to explore design principles (1) and (2), we adopted new chemistry using a click reaction of Cbl \u003cb\u003e58\u003c/b\u003e with a variable azide\u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e to form a biphenyl-like scaffold where the second ring is an R-group harboring triazole (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ea). We synthesized Cbls \u003cb\u003e63\u003c/b\u003e\u0026ndash;\u003cb\u003e65\u003c/b\u003e to install amino-, guanidinium-, and hydroxyl-substituted triazoles, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eb). We then used our fluorescence displacement assay to quantify the binding of these derivatives to \u003cem\u003eenv8\u003c/em\u003e and discovered that \u003cb\u003e63\u003c/b\u003e (1.0\u0026thinsp;\u0026plusmn;\u0026thinsp;0.6 nM, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.62) and \u003cb\u003e64\u003c/b\u003e (1.3\u0026thinsp;\u0026plusmn;\u0026thinsp;0.7 nM, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.92) have comparable affinity to CNCbl (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ec). However, these interactions approach the accuracy limit of our binding assay. To confirm these results, we used isothermal titration calorimetry (ITC) with the \u003cem\u003eenv8\u003c/em\u003e aptamer domain, which lacks an element of RNA structure that interacts with the α-axial face of the Cbl and thereby lowering the observed affinity. These data revealed significantly tighter affinity to \u003cb\u003e63\u003c/b\u003e (32\u0026thinsp;\u0026plusmn;\u0026thinsp;6 nM, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.02) and \u003cb\u003e64\u003c/b\u003e (40\u0026thinsp;\u0026plusmn;\u0026thinsp;3 nM, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.02) as compared to CNCbl (260\u0026thinsp;\u0026plusmn;\u0026thinsp;60 nM) (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ed).\u003c/p\u003e \u003cp\u003eTo understand how these new lead compounds bind \u003cem\u003eenv8\u003c/em\u003e, we determined co-crystal structures of \u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cb\u003e63\u003c/b\u003e and \u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cb\u003e64\u003c/b\u003e. While weakly supported by the electron density, both structures suggest that the amino (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ee) or guanidinium (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ef) groups likely electrostatically interact with backbone phosphate of G19 (\u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cb\u003e64\u003c/b\u003e) or A68 (\u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cb\u003e63\u003c/b\u003e). However, our \u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cb\u003e63\u003c/b\u003e structure unambiguously demonstrates that the terminal triazole ring twists in order to π-stack with A20 (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ee), again suggesting that rotatable bonds are beneficial for optimizing the π-stacking network. This is the first evidence of strong engagement of A20 in any of our structures. Notably, when π-stacked with the ligand, A20 adopts a perpendicular orientation relative to G19, a highly unusual mode of base-base interaction in nucleic acid.\u003c/p\u003e \u003cp\u003eDespite their high affinity, when used in our reporter assay, Cbls \u003cb\u003e63\u003c/b\u003e\u0026ndash;\u003cb\u003e65\u003c/b\u003e confer minimal repression (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eg). To investigate this further, we repeated our growth experiments in ΔMetE cells. In the presence of 10 nM derivative, \u003cb\u003e63\u003c/b\u003e\u0026ndash;\u003cb\u003e65\u003c/b\u003e support ΔMetE growth with the same doubling time as CNCbl, suggesting that these ligands are imported and repaired to some extent (\u003cb\u003eExtended Data Fig.\u0026nbsp;9a,b\u003c/b\u003e). However, unlike \u003cb\u003e36\u003c/b\u003e and \u003cb\u003e44\u003c/b\u003e, elevated concentrations of \u003cb\u003e63\u003c/b\u003e\u0026ndash;\u003cb\u003e65\u003c/b\u003e prevented growth of ΔMetE cells (\u003cb\u003eExtended Data Fig.\u0026nbsp;9c,d\u003c/b\u003e), suggesting that these derivatives antagonize some essential aspect of Cbl metabolism in \u003cem\u003eE. coli\u003c/em\u003e.\u003csup\u003e\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e To fully assess how \u003cb\u003e63\u003c/b\u003e\u0026ndash;\u003cb\u003e65\u003c/b\u003e affect regulatory activity, we repeated our reporter assay in wild-type \u003cem\u003eE. coli\u003c/em\u003e, which grow normally in the presence of high Cbl concentrations (\u003cb\u003eExtended Data Fig.\u0026nbsp;9e\u003c/b\u003e). Adding excess amounts of \u003cb\u003e63\u003c/b\u003e\u0026ndash;\u003cb\u003e65\u003c/b\u003e to cells containing 10 nM CNCbl resulted in loss of CNCbl-induced repression, confirming that these derivatives are antagonists of \u003cem\u003eenv8\u003c/em\u003e function (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eg).\u003c/p\u003e \u003cp\u003eTo further explore design principle (2), we synthesized the pyridine Cbl \u003cb\u003e66\u003c/b\u003e to introduce a conditional positive charge (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eh). Under the conditions of our binding experiments (pH 8), \u003cb\u003e66\u003c/b\u003e should behave exactly like its non-pyridine counterpart \u003cb\u003e29\u003c/b\u003e. However, at a pH below its predicted pK\u003csub\u003ea\u003c/sub\u003e (~\u0026thinsp;5.25), \u003cb\u003e66\u003c/b\u003e should be charged and we would expect an increase in affinity that is absent in \u003cb\u003e29\u003c/b\u003e. To test this hypothesis, we carried out ITC measurements at pH 8 and 5 with the \u003cem\u003eenv8\u003c/em\u003e aptamer domain and CNCbl, \u003cb\u003e29\u003c/b\u003e, and \u003cb\u003e66\u003c/b\u003e. The non-titratable CNCbl and \u003cb\u003e29\u003c/b\u003e showed a\u0026thinsp;~\u0026thinsp;two-fold reduction in binding affinity at reduced pH, whereas the affinity of \u003cb\u003e66\u003c/b\u003e increased 1.7-fold (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ei), in agreement with our hypothesis. These data provide proof-of-principle that installing pyridines, whose pK\u003csub\u003ea\u003c/sub\u003e can be lowered by interaction with RNA, is an important design principle that leverages the increased affinity of cation-π interactions without violating Lipinski\u0026rsquo;s rules.\u003csup\u003e\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e,\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003e \u003cb\u003eStructure-informed screen identifies cryptic binders of\u003c/b\u003e \u003cb\u003eenv8\u003c/b\u003e \u003cb\u003edivorced from Cbl\u003c/b\u003e\u003c/p\u003e \u003cp\u003eOur structural biology efforts unmasked the cryptic β-axial binding pocket within \u003cem\u003eenv8\u003c/em\u003e, which allows us to explore whether small molecules divorced from a Cbl host can also target this site. To address this issue, we carried out a high-throughput computational screen using a 28,000-compound RNA-focused library. These compounds were docked against \u003cem\u003eenv2\u003c/em\u003e in two conformations: one in which A20 was displaced toward the major groove (A20-out, e.g., \u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cb\u003e29\u003c/b\u003e) and one where A20 was engaged in the RNA core (A20-in, e.g., \u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cb\u003e63\u003c/b\u003e) (\u003cb\u003eSupplementary Fig.\u0026nbsp;6\u003c/b\u003e). To leverage our structural data, we specified that docking hits contain the biphenyl-like scaffold and identified eight potential lead compounds (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ea).\u003c/p\u003e \u003cp\u003eTo survey the binding of these ligands to \u003cem\u003eenv8\u003c/em\u003e, we adopted a thiazole orange (TO) displacement assay (\u003cb\u003eSupplementary Fig.\u0026nbsp;7a\u003c/b\u003e).\u003csup\u003e\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e Upon addition of \u003cb\u003eE1\u003c/b\u003e-\u003cb\u003eE8\u003c/b\u003e, \u003cb\u003eE2\u003c/b\u003e and \u003cb\u003eE5\u003c/b\u003e led to sufficient (\u0026gt;\u0026thinsp;25%) displacement of TO indicative of binding (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eb). This conclusion was confirmed with microscale thermophoresis (MST) (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ec). Competitive titrations of CNCbl, \u003cb\u003eE2\u003c/b\u003e, and \u003cb\u003eE5\u003c/b\u003e into the \u003cem\u003eenv8\u003c/em\u003e-TO complex yielded K\u003csub\u003eD\u003c/sub\u003e values of 38\u0026thinsp;\u0026plusmn;\u0026thinsp;5 nM, 34\u0026thinsp;\u0026plusmn;\u0026thinsp;9 \u0026micro;M, and 50\u0026thinsp;\u0026plusmn;\u0026thinsp;10 \u0026micro;M, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ed). The affinity of CNCbl agrees well with data obtained from ITC (38\u0026thinsp;\u0026plusmn;\u0026thinsp;8 nM) (\u003cb\u003eSupplementary Fig.\u0026nbsp;7b,c\u003c/b\u003e), supporting the use of this method to quantify ligand binding to \u003cem\u003eenv8\u003c/em\u003e.\u003c/p\u003e \u003cp\u003eTo verify that \u003cb\u003eE2\u003c/b\u003e and \u003cb\u003eE5\u003c/b\u003e target the Cbl binding pocket, we employed an electrophoresis mobility shift assay (EMSA) that monitors the ability of \u003cem\u003eenv8\u003c/em\u003e to adopt a kissing-loop conformation in the presence of Cbl,\u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e which migrates faster than the unbound RNA (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ee). Given that the kissing-loop is facilitated by interactions with the α-axial face of the Cbl corrin ring,\u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e \u003cb\u003eE2\u003c/b\u003e and \u003cb\u003eE5\u003c/b\u003e are not expected to induce this conformational change. Thus, if these ligands competitively bind the same site within \u003cem\u003eenv8\u003c/em\u003e, then adding \u003cb\u003eE2\u003c/b\u003e or \u003cb\u003eE5\u003c/b\u003e to \u003cem\u003eenv8\u003c/em\u003e in the presence of CNCbl should revert the RNA back to the slower migrating species. When excess amounts of \u003cb\u003eE2\u003c/b\u003e or \u003cb\u003eE5\u003c/b\u003e were added to an \u003cem\u003eenv8\u003c/em\u003e-CNCbl complex both ligands prevented kissing-loop formation, suggestive of competitive binding (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ef). While this phenomenon was unambiguous for \u003cb\u003eE5\u003c/b\u003e, addition of \u003cb\u003eE2\u003c/b\u003e reproducibly led to weaker gel staining (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ef). We therefore confirmed the competitive binding of both ligands with MST (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eg). These data support the top-ranked docking poses of \u003cb\u003eE2\u003c/b\u003e and \u003cb\u003eE5\u003c/b\u003e which show their biphenyl-like scaffolds maximizing π-stacking to G19 and A68 within the cryptic β-axial binding pocket and making hydrogen bonds to the phosphate backbone (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eh).\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eTargeting RNA with small molecules is an emerging field hindered by an incomplete understanding of the basic principles governing RNA-ligand interactions.\u003csup\u003e\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e To address this knowledge gap, we explored the chemical features promoting specific and high affinity base displacement RNA binding interactions using the Cbl riboswitch. Despite limitations of our QSAR chemoinformatic modeling, structure-based design successfully identified lead compounds with affinities exceeding the native ligand. All leads share a biphenyl-like structure, which has been previously identified as a privileged RNA-binding scaffold.\u003csup\u003e\u003cspan additionalcitationids=\"CR29\" citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e,\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e Our data demonstrate that twisting around a rotatable bond confers the ability to optimize the π-stacking network within the binding pocket. Thus, our results not only reinforce our understanding of RNA chemical space but yield new structural insights into how these scaffolds interact with RNA.\u003c/p\u003e \u003cp\u003eAcross all newly reported RNA-ligand co-crystal structures, binding pocket nucleotides G19 and A68 are in near-identical positions, indicating that structural plasticity is conferred solely by the displacement of A20, providing a large (\u0026gt;\u0026thinsp;340 \u0026Aring;\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e) cryptic site that accommodates the β-axial group of all derivatives. One mechanism to explain this phenomenon is ligand docking into a rigid RNA to induce A20 displacement; however, large, rigid β-axial groups likely prevent the corrin ring from productively engaging the RNA. A more likely explanation is that A20 undergoes transient excursions from the RNA core to form a binding-competent state, a motion that is well documented in nucleic acids and consistent with the time scale of ligand binding.\u003csup\u003e\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e\u003c/sup\u003e The energetic penalty of A20 displacement to expose the cryptic pocket is offset to varying degrees by each β-axial group, reflecting different derivative binding affinities. This illustrates that a consideration of local binding pocket dynamics is an essential feature to unmask additional RNA cryptic sites,\u003csup\u003e\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e,\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e\u003c/sup\u003e which will facilitate the development of new chemical probes of RNA function and therapeutics to treat RNA-mediated disease. However, our work cautions against using a purely computational approach, as MD simulations could not account for the nuances of base displacement within \u003cem\u003eenv8\u003c/em\u003e. Advances in machine-learning\u003csup\u003e\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e\u003c/sup\u003e and artificial intelligence\u003csup\u003e\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e\u003c/sup\u003e methods may address these limitations in the future.\u003c/p\u003e \u003cp\u003eOur work also showcases that bifunctional host-guest systems are a valuable approach to targeting RNA. Our soluble Cbl host enabled access to chemical space that is typically intractable to robust medicinal chemistry. Exploration of general host-small molecule conjugates is therefore of critical importance. Nucleic acid-based hosts are an attractive candidate because they can solubilize the attached ligand, deliver it to RNA in a sequence specific manner, and contribute binding energy that can turn modest and/or non-specific binders to ligands with high affinity and specificity. For example, TO has been conjugated to a peptide nucleic acid (PNA) host to monitor double-stranded RNA formation.\u003csup\u003e\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e\u003c/sup\u003e PNA hosts have also been shown to rescue the binding of small molecules that have no affinity for RNA by themselves.\u003csup\u003e\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e\u003c/sup\u003e This conceptual framework is similar to the widespread use of bifunctional/multivalent small molecules in protein targeting,\u003csup\u003e\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e\u003c/sup\u003e which has recently seen parallels in RNA-targeting strategies (e.g., RiboTAC\u003csup\u003e\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e and RiboSNAP\u003csup\u003e\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e\u003c/sup\u003e). We believe that a host-guest approach is generalizable and will enable detailed chemical and structural analysis of underexplored chemical space to better understand RNA-small molecule interactions.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\n\u003ch2\u003eSynthesis of Cbl derivatives\u003c/h2\u003e\n\u003cdiv id=\"Sec12\" class=\"Section3\"\u003e\n\u003cp\u003eCbl derivatives \u003cstrong\u003e4\u003c/strong\u003e\u0026ndash;\u003cstrong\u003e66\u003c/strong\u003e were synthesized according to previously described procedures.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e26\u003c/span\u003e,\u003cspan class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e See the \u003cstrong\u003eSupplementary Information\u003c/strong\u003e for additional details.\u003c/p\u003e\n\u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\n\u003ch2\u003eRNA preparation\u003c/h2\u003e\n\u003cp\u003eAll RNAs (full sequences shown in \u003cstrong\u003eSupplementary Table\u0026nbsp;4\u003c/strong\u003e) were prepared using DNA templates amplified by PCR, transcribed with T7 RNA polymerase, and purified using preparative denaturing polyacrylamide gel electrophoresis.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e53\u003c/span\u003e\u003c/sup\u003e Purified RNA was buffer exchanged and concentrated into Milli-Q H\u003csub\u003e2\u003c/sub\u003eO using centrifugal concentrators (Millipore-Sigma). Final RNA concentrations were calculated using absorbance at 260 nm and extinction coefficients determined from the summation of the individual bases. Prior to all binding experiments, RNA was heated at 95\u0026deg;C for 2 minutes, incubated on ice for 10 minutes, and then allowed to equilibrate to room temperature.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\n\u003ch2\u003eCbl fluorescence binding assays\u003c/h2\u003e\n\u003cp\u003eBoth direct CNCbl-5\u0026times;PEG-ATTO590 fluorescence induction and displacement assays were performed as previously described.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e The K\u003csub\u003eD\u003c/sub\u003e values for all Cbl derivatives to \u003cem\u003eenv8\u003c/em\u003e can be found in \u003cstrong\u003eSupplementary Table\u0026nbsp;5\u003c/strong\u003e.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\n\u003ch2\u003eCell-based reporter assay\u003c/h2\u003e\n\u003cp\u003eThese were conducted as previously reported\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e with minor alterations. Firstly, plasmids harboring \u003cem\u003eenv8\u003c/em\u003e-GFPuv (Addgene #99831) were transformed into \u003cem\u003eE. coli\u003c/em\u003e BW25113 cells. Second, 1 \u0026micro;l of a saturated overnight culture was added to 1 ml of CSB media supplemented with 100 \u0026micro;g/ml carbenicillin and 10 nM Cbl (unless otherwise noted) and grown to mid-log phase at 37\u0026deg;C. Third, from each biological replicate, 200 \u0026micro;l of cells were added to wells in a 96-well plate (Costar). The fold repression values\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e for all Cbl derivatives can be found in \u003cstrong\u003eSupplementary Table\u0026nbsp;6\u003c/strong\u003e.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e\n\u003ch2\u003e\u0026Delta;MetE growth assays\u003c/h2\u003e\n\u003cp\u003eFor growth experiments, 5 \u0026micro;l of a saturated overnight culture of \u0026Delta;MetE \u003cem\u003eE. coli\u003c/em\u003e cells (Keio collection, JW3805) was added to 5 ml of methionine-dropout CSB media supplemented with either 335 \u0026micro;M methionine or 10 nM Cbl and grown to mid-log phase at 37\u0026deg;C. From each biological replicate, 200 \u0026micro;l of cells were added to wells in a 96-well plate (Costar) and growth was monitored by measuring OD\u003csub\u003e600\u003c/sub\u003e using a microplate reader (Tecan). For growth rate experiments, the same procedure was used except the cells were grown for 24 hours and the OD\u003csub\u003e600\u003c/sub\u003e was measured at various time intervals. For titration experiments, the same procedure was used except variable amounts of methionine or Cbl was added to the media.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec17\" class=\"Section2\"\u003e\n\u003ch2\u003eChemoinformatic linear discriminant (LDA) analysis\u003c/h2\u003e\n\u003cp\u003eA set 20 standard chemoinformatic parameters\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e32\u003c/span\u003e\u0026ndash;\u003cspan class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e (\u003cstrong\u003eSupplementary Table\u0026nbsp;1\u003c/strong\u003e) were calculated for the \u0026beta;-axial group of Cbls \u003cstrong\u003e4\u003c/strong\u003e, \u003cstrong\u003e5\u003c/strong\u003e, and \u003cstrong\u003e7\u003c/strong\u003e\u0026ndash;\u003cstrong\u003e44\u003c/strong\u003e as previously described.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e34\u003c/span\u003e,\u003cspan class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e All descriptors that were non-constant were then used in a LDA with our clustered binding or functional data (see \u003cstrong\u003eExtended Data\u003c/strong\u003e Figs.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e and \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e) using Discriminatory Analysis plugin from XLSTAT as previously described.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e34\u003c/span\u003e,\u003cspan class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e Excel files with the calculated descriptors are available upon request.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec18\" class=\"Section2\"\u003e\n\u003ch2\u003eQSAR modeling\u003c/h2\u003e\n\u003cp\u003eThe \u0026beta;-axial group of Cbls \u003cstrong\u003e4\u003c/strong\u003e, \u003cstrong\u003e5\u003c/strong\u003e, and \u003cstrong\u003e7\u003c/strong\u003e\u0026ndash;\u003cstrong\u003e44\u003c/strong\u003e were tuned to the correct protonation and tautomerization states using molecular operating environment (MOE, v2022.02, Chemical Computing Group) and each state was sent to conformational search as previously described.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e A total of 347 descriptors (all but x3D and protein descriptors) were calculated for each conformation in MOE, averaged using the Boltzmann-weighted equation, and filtered to remove multicollinear features as previously described.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e Then, a lasso-MLR strategy was used to model these descriptors (independent variable) against either the natural-log transformed K\u003csub\u003eD\u003c/sub\u003e or fold repression data (dependent variables) in matlab (v2019a). Lasso was implemented with five-fold cross-validation to find an optimal lambda value in descriptor selection. MLR exhaustively searched all possible models from the lasso-selected descriptors (with a maximum number of descriptors set by Topliss rule\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e54\u003c/span\u003e\u003c/sup\u003e) with five-fold cross-validation to obtain the best model as defined by specified criteria. We constructed a baseline model without a data split that maximized the R\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. We also used a stratified 80/20 data split to ensure that training and test sets had equivalent proportions of tight/strong, moderate, and weak binders/repressors and constructed models that prioritized either the predictive ability on the test set (Q\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e-focused) or the fit of the training set (R\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e-focused). Excel files with the calculated descriptors and matlab scripts for modeling are available upon request.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec19\" class=\"Section2\"\u003e\n\u003ch2\u003eLead compound prediction\u003c/h2\u003e\n\u003cp\u003eWe used our binding- and function-based models to screen for high affinity \u0026beta;-axial groups among a curated Enamine alkyne library. These small molecules were selected by a substructure search (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://enaminestore.com/search\u003c/span\u003e\u003c/span\u003e) of modified-phenyl (phenyl-X) and aliphatic (amide-X) alkynes (precise structures used for this search are shown in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003ef). This left us with a 513-compound library (469 phenyl-X and 44 amide-X). The binding- and function-based model-selected descriptors (\u003cstrong\u003eExtended Data\u003c/strong\u003e Figs.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003ea and \u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003ea) were calculated for the 513-compound library in MOE as described above. The predicted ln [K\u003csub\u003eD\u003c/sub\u003e] and fold repression values from baseline, Q\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e-focused, and R\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e-focused models were averaged and used to select tight, moderate, and weak binders (Cbls \u003cstrong\u003e45\u0026ndash;53\u003c/strong\u003e) (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eg) and strong, moderate, and weak repressors (Cbls \u003cstrong\u003e54\u0026ndash;62\u003c/strong\u003e) (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eg). Excel files with the calculated descriptors and SDF files from phenyl-X and alkyne-X substructure search are available upon request.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec20\" class=\"Section2\"\u003e\n\u003ch2\u003eMolecular dynamic (MD) simulations\u003c/h2\u003e\n\u003cp\u003eAll-atom MD simulations of \u003cem\u003eenv8\u003c/em\u003e-CNCbl and \u003cem\u003eenv8\u003c/em\u003e-Cbl \u003cstrong\u003e4\u003c/strong\u003e complexes were performed using the PMEMD module\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e55\u003c/span\u003e\u003c/sup\u003e in AMBER22.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e56\u003c/span\u003e\u003c/sup\u003e CNCbl was manually built from OHCbl (PDB 4FRG\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e) in PyMol (v2.5.7, Schr\u0026ouml;dinger) and Cambridge Structural Database deposition 961228\u003csup\u003e27\u003c/sup\u003e was used for Cbl \u003cstrong\u003e4\u003c/strong\u003e. Ligands were docked into \u003cem\u003eenv8\u003c/em\u003e by alignment to OHCbl from the \u003cem\u003eenv8\u003c/em\u003e-OHCbl structure (PDB 4FRG\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e). Each system was neutralized and solvated in a cubic simulation box with a 25 \u0026Aring; buffer of OPC water molecules\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e57\u003c/span\u003e\u003c/sup\u003e and 150 mM NaCl to mimic physiological conditions. RNA was parameterized using the OL3 force field,\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e58\u003c/span\u003e\u003c/sup\u003e incorporating the RNA.LJbb correction\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e59\u003c/span\u003e\u003c/sup\u003e for improved backbone accuracy. Force field parameters for each ligand were generated using Antechamber with the General Amber Force Field. The metal center, including the cobalt ion, was parameterized using MCPB.py following established guidelines.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e60\u003c/span\u003e\u003c/sup\u003e Quantum mechanical calculations were performed with Gaussian16 to optimize the geometry and derive charges. The cobalt-carbon bond was explicitly defined in the MCPB.py input to ensure accurate representation of the ligand coordination environment.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e60\u003c/span\u003e\u003c/sup\u003e To improve simulation efficiency, hydrogen mass repartitioning was applied, allowing a 4 fs time step.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e61\u003c/span\u003e\u003c/sup\u003e Following system preparation, energy minimization, gradual heating to 300 K, and equilibration under constant pressure were conducted. Production simulations were performed in triplicate for each system under constant temperature (300 K) and pressure (1 atm) using the NPT ensemble, with each replicate running for at least 1 \u0026micro;s. Trajectory analysis and post-processing A20(C6)-G19(C6) and A20(C6)-A68(N7) distance calculations were carried out using Cpptraj.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e62\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCrystallization of\u003c/strong\u003e \u003cstrong\u003eenv2-\u003c/strong\u003e\u003cstrong\u003eligand complexes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe \u003cem\u003eenv2\u003c/em\u003e aptamer domain RNA in complex with various Cbl derivatives was crystallized at 30\u0026deg;C using hanging drop vapor diffusion. The RNA-ligand complexes were prepared as solutions containing 250 \u0026micro;M RNA and 375\u0026ndash;500 \u0026micro;M ligand in 1X TE buffer. For crystallization, drops containing 2 \u0026micro;l of RNA-ligand solution and 2 \u0026micro;l precipitant solution (40 mM sodium cacodylate pH 7, 10% (w/v) 2-methyl-2,4-pentanediol (MPD), 12 mM spermine tetrahydrochloride, 80 mM KCl, and 20 mM MgCl\u003csub\u003e2\u003c/sub\u003e) were suspended above a reservoir solution of 500 \u0026micro;l of 35% MPD and grown for 24\u0026ndash;96 hours. For cryoprotection, the crystals were soaked in a cryoprotectant (precipitant solution with MPD increased to 25%) for 2 min and flash frozen in liquid nitrogen. Some data were collected on a home-source Rigaku MicroMax-003 X-ray source with a Dectris Pilatus 200K detector, and then indexed, integrated, and scaled with HKL3000\u003csup\u003e63\u003c/sup\u003e (\u003cstrong\u003eSupplementary Tables\u0026nbsp;7\u0026ndash;10\u003c/strong\u003e). Other data were collected at the Advanced Light Source on a Dectris on the ALS Beamline 8.2.2. with a Dectris Pilatus3 2M detector, and then indexed, integrated, and scaled with XDS\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e64\u003c/span\u003e\u003c/sup\u003e (\u003cstrong\u003eSupplementary Tables\u0026nbsp;7\u0026ndash;10\u003c/strong\u003e). The crystals that yielded the \u003cem\u003eenv2\u003c/em\u003e apo structure were grown in the same solution described above but in the presence of compound \u003cstrong\u003eE5\u003c/strong\u003e.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec21\" class=\"Section2\"\u003e\n\u003ch2\u003eStructure determination and model refinement\u003c/h2\u003e\n\u003cp\u003eFor all structures, an initial electron density map was calculated using molecular replacement with Phaser\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e65\u003c/span\u003e\u003c/sup\u003e in PHENIX\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e66\u003c/span\u003e\u003c/sup\u003e using \u003cem\u003eenv8\u003c/em\u003e as a starting model (PDB 4FRG\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e) but with OHCbl and A20 removed to minimize model bias. All ligands were manually built from Cbl \u003cstrong\u003e4\u003c/strong\u003e (Cambridge Structural Database deposition 961228\u003csup\u003e27\u003c/sup\u003e) in PyMol (v2.5.7, Schr\u0026ouml;dinger) and restraint files were generated with eLBOW.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e67\u003c/span\u003e\u003c/sup\u003e After initial refinement, the ligand density was unambiguous, and ligands were added to the model with LigandFit.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e68\u003c/span\u003e\u003c/sup\u003e After another round of refinement, A20 was manually built into the model. The map and model were refined, and the solvent was built until a maximum agreement between the map and the model was reached. During this process, to mitigate the effect of model bias, multiple rounds of combined high temperature (5,000 K) simulated annealing in torsion space and maximum-likelihood refinement were performed.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e69\u003c/span\u003e\u003c/sup\u003e At the end of refinement, RNA geometry was corrected with ERRASER.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e70\u003c/span\u003e\u003c/sup\u003e A subset of the RNA-ligand co-crystal structures had sufficient resolution to warrant simulated annealing in Cartesian space and individual ADP refinement. A simulated annealing 2F\u003csub\u003eo\u003c/sub\u003e-F\u003csub\u003ec\u003c/sub\u003e map where A20 and the ligand were omitted from the model was calculated for all structures to support the final placement of the ligand and A20. Models and associated structure factors have been deposited in the RCSB Protein Data Bank.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec22\" class=\"Section2\"\u003e\n\u003ch2\u003eIsothermal titration calorimetry (ITC)\u003c/h2\u003e\n\u003cp\u003eRNA was exchanged into a 1X RNA binding buffer (50 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) pH 8.0, 100 mM KCl, 10 mM NaCl, and 1 mM MgCl\u003csub\u003e2\u003c/sub\u003e) by successive washing in centrifugal concentrators (Millipore-Sigma). RNA was diluted up to a final concentration of 10 \u0026micro;M and titrated with ligand (dissolved in the flow-through of the final wash from buffer exchange) at 100 \u0026micro;M. All titrations were performed at 25\u0026deg;C using a MicroCal ITC200 and collected in triplicate. The data were fit to a single-site binding model using Origin 7 ITC software (MicroCal).\u003c/p\u003e\n\u003cdiv id=\"Sec23\" class=\"Section3\"\u003e\n\u003ch2\u003eComputational docking\u003c/h2\u003e\n\u003cp\u003eThe 28,000-RNA library from Enamine (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://enamine.net/compound-libraries/targeted-libraries/rna\u003c/span\u003e\u003c/span\u003e) was downloaded as an SDF file and imported into MOE (v2022.02, Chemical Computing Group) as a database file. The small molecules were tuned to the correct protonation and tautomerization states and each state was sent to conformational search\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e as described above. RNA receptor preparation was carried out using the default Quick Prep protocol in MOE. Our \u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cstrong\u003e29\u003c/strong\u003e (A20-out) and \u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cstrong\u003e63\u003c/strong\u003e (A20-in) structures were used as representative receptors where A20 was displaced toward the major groove or engaged in the RNA core, respectively. To ensure that A20 was not accessible for A20-out docking, all A20 atoms were removed from the RNA receptor in MOE. Rigid receptor docking was carried out in a templated manner, where ligand search space started from the biphenyl-like scaffold present in the co-crystal structures with \u003cstrong\u003e29\u003c/strong\u003e and \u003cstrong\u003e63\u003c/strong\u003e. For A20-in docking, the terminal triazole ring was rebuilt into a phenyl group, so that the scaffolds were identical in both RNA receptors. With this approach, only the 319 ligands with the required scaffold were docked and scored.\u003c/p\u003e\n\u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec24\" class=\"Section2\"\u003e\n\u003ch2\u003eThiazole orange (TO) binding assay\u003c/h2\u003e\n\u003cp\u003eFor fluorescence induction titration experiments, 60 \u0026micro;l reactions containing TO (5 nM), 1X RNA binding buffer, and increasing amounts of \u003cem\u003eenv8\u003c/em\u003e were incubated in 384-well plates (Corning). Reactions were incubated for 15 minutes and TO fluorescence was monitored (490\u0026thinsp;\u0026plusmn;\u0026thinsp;10 nm excitation, 530\u0026thinsp;\u0026plusmn;\u0026thinsp;10 nm emission) using a CLARIOstarPLU microplate reader (BMG Labtech). For competitive fluorescence displacement assays, the same procedure was carried out except with a fixed amount of 2 \u0026micro;M RNA, 20 \u0026micro;M TO (in excess of its K\u003csub\u003eD\u003c/sub\u003e to ensure RNA saturation), and 2 mM competing ligand. Displacement titrations were carried out in the same manner except using increasing amounts of \u003cstrong\u003eE2\u003c/strong\u003e or \u003cstrong\u003eE5\u003c/strong\u003e. All titrations were performed at room temperature and collected in triplicate. The data were fit to a single-site binding model with a HillSlope parameter and the K\u003csub\u003eD\u003c/sub\u003e from the induction experiment was used to convert the IC\u003csub\u003e50\u003c/sub\u003e from the competitive fluorescence displacement assays to K\u003csub\u003eD\u003c/sub\u003e values, as previously described.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\n\u003cdiv id=\"Sec25\" class=\"Section3\"\u003e\n\u003ch2\u003eElectrophoresis mobility shift assay (EMSA)\u003c/h2\u003e\n\u003cp\u003eTo validate the CNCbl-induced \u003cem\u003eenv8\u003c/em\u003e conformational change, 20 \u0026micro;l reactions containing RNA (0.5 \u0026micro;M), 1X RNA binding buffer, and increasing amounts of CNCbl (0.1\u0026ndash;10 \u0026micro;M) were incubated at 37\u0026deg;C for 15 minutes and then at 4\u0026deg;C for 15 additional minutes. Once equilibrated to 4\u0026deg;C, 15 \u0026micro;l of the samples were loaded onto a native 10% (19:1) polyacrylamide gel in 0.5X THE buffer and 1 mM MgCl\u003csub\u003e2\u003c/sub\u003e and run at constant 7 W for 3 hours at 4\u0026deg;C in 0.5X THE buffer with 1 mM MgCl\u003csub\u003e2\u003c/sub\u003e. For the competitive EMSA experiment, the same procedure was used except using a pre-bound \u003cem\u003eenv8\u003c/em\u003e-CNCbl complex (0.5 \u0026micro;M CNCbl) and adding 5 mM \u003cstrong\u003eE2\u003c/strong\u003e or \u003cstrong\u003eE5\u003c/strong\u003e. All gels were stained with ethidium bromide, imaged on a UV illuminator (Alpha Innotech), and processed with AlphaView software (AlphaImagerHP). Unmodified images of the gels shown in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003ee,f and their replicates can be found in \u003cstrong\u003eSupplementary Fig.\u0026nbsp;8\u003c/strong\u003e.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec26\" class=\"Section3\"\u003e\n\u003ch2\u003eMicroscale thermophoresis (MST)\u003c/h2\u003e\n\u003cp\u003eFor MST binding experiments, \u003cem\u003eenv8\u003c/em\u003e was 3\u0026prime;-end labeled with pCp-AF488 (NU-1706-AF488, Jena Bioscience). The 50 \u0026micro;l labeling reaction comprised 200\u0026ndash;500 nmol RNA, 1 mM ATP, 10% (w/v) DMSO, 1X T4 RNA ligase 1 buffer (#B0204S, New England Biolabs (NEB)), 15% (w/v) polyethylene glycol 8,000, 48 \u0026micro;M pCp-AF488, and 4 \u0026micro;l T4 RNA ligase 1 (#M0204L, NEB) and was incubated at 16\u0026deg;C for 18 hours. To monitor RNA-ligand binding, labeled \u003cem\u003eenv8\u003c/em\u003e (300 nmol) was incubated in 1X RNA binding buffer and 250 \u0026micro;M ligand for 15 minutes and transferred to capillaries (#MO-K022, NanoTemper). Alexafluor488 fluorescence was read on a Monolith NT.115 (NanoTemper) instrument. Competitive MST experiments were carried out in the same manner except using a pre-bound \u003cem\u003eenv8\u003c/em\u003e-CNCbl complex (300 nmol RNA and 10 \u0026micro;M ligand) and adding 1 mM \u003cstrong\u003eE2\u003c/strong\u003e or \u003cstrong\u003eE5\u003c/strong\u003e.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec27\" class=\"Section3\"\u003e\u0026nbsp;\u003c/div\u003e\n\u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAtomic coordinates and structure factors have been deposited in the Protein Data Bank (PDB) (https://www.rcsb.org/) under accession numbers 9MFH (\u003cem\u003eenv2\u003c/em\u003e apo), 9E5H (\u003cem\u003eenv2\u003c/em\u003e-CNCbl), 9E5I (\u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cstrong\u003e4\u003c/strong\u003e), 9E5J (\u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cstrong\u003e5\u003c/strong\u003e), 9E5K (\u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cstrong\u003e13\u003c/strong\u003e), 9E5L (\u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cstrong\u003e26\u003c/strong\u003e), 9E5M (\u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cstrong\u003e29\u003c/strong\u003e), 9E5O (\u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cstrong\u003e32\u003c/strong\u003e), 9E5P (\u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cstrong\u003e33\u003c/strong\u003e), 9E5Q (\u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cstrong\u003e36\u003c/strong\u003e), 9ELR (\u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cstrong\u003e37\u003c/strong\u003e), 9E5R (\u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cstrong\u003e42\u003c/strong\u003e), 9E5S (\u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cstrong\u003e63\u003c/strong\u003e), and 9E5T (\u003cem\u003eenv2\u003c/em\u003e-Cbl \u003cstrong\u003e64\u003c/strong\u003e). Source data are provided with this paper.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work is supported by the National Institutes of Health (R35 GM152029 to R.T.B. and R35 GM139644 to A.J.W.) and the Howard Hughes Medical Institute (S.P.L). We thank J. Nix and the staff of beamline 8.2.2. of the Advanced Light Source, Lawrence Berkeley National Laboratory for their support with remote crystallographic data collection. Beamline 8.2.2. of the Advanced Light Source, a DOE Office of Science User Facility under Contract No. DE-AC02-05CH11231, is supported in part by the ALS-ENABLE program funded by the National Institutes of Health (P30 GM124169-01). We thank the Macromolecular X-ray Crystallography Core (RRID:SCR_019310) at the University of Colorado Boulder for crystallographic data collection and the Shared Instruments Pool (RRID: SCR_018986) of the Department of Biochemistry at the University of Colorado Boulder for the use of the Monolith NT.115 NanoTemper instrument. We acknowledge A. Erbse for her assistance with the resources utilized at the University of Colorado Boulder. We thank F. Longshore-Neate and D. Patel for help with derivative synthesis as well as N. Zeps and T. Wolters for help with RNA-ligand co-crystallization. We are grateful to Q. Vicens for comments on the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor information\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eShawn P. Laursen\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePresent address: National Renewable Energy Laboratory, Golden Colorado, USA.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors and Affiliations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDepartment of Biochemistry, University of Colorado Boulder, Boulder, CO, USA\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eLukasz T. Olenginski, Aleksandra J. Wierzba, \u0026amp; Robert T. Batey.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBioFrontiers Institute, University of Colorado, Boulder, CO, USA\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAleksandra J. Wierzba\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDepartment of Molecular, Cellular, and Developmental Biology, University of Colorado Boulder, Boulder, CO, USA\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eShawn P. Laursen\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eHoward Hughes Medical Institute, Chevy Chase, MD, USA\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eShawn P. Laursen\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eContributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eR.T.B. conceptualized the project. L.T.O. designed the small molecule library and carried out its synthesis, purification, and characterization. A.J.W. helped supervise synthesis and characterization. S.P.L. conducted MD simulations. L.T.O. performed the \u003cem\u003ein vitro\u003c/em\u003e binding and cell-based reporter assays, carried out the chemoinformatic modeling, conducted RNA crystallization and structure determination, implemented computational docking, and analyzed all data. L.T.O. wrote the manuscript with input from all authors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCorresponding authors\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCorrespondence to \u003cstrong\u003eRobert T. Batey\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics declaration\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCompeting interests\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eR.T.B. serves on the Scientific Advisory Boards of Expansion Therapeutics, SomaLogic, and MeiraGTx.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eChilds-Disney, J. L. \u003cem\u003eet al.\u003c/em\u003e Targeting RNA structures with small molecules. Nat. Rev. Drug Discov. 21, 736\u0026ndash;762 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWarner, K. D., Hajdin, C. E. \u0026amp; Weeks, K. M. Principles for targeting RNA with drug-like small molecules. Nat. Rev. Drug Discov. 17, 547\u0026ndash;558 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUmuhire Juru, A. \u0026amp; Hargrove, A. E. Frameworks for targeting RNA with small molecules. J. Biol. Chem. 296, 100191 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRatni, H. \u003cem\u003eet al.\u003c/em\u003e Discovery of Risdiplam, a Selective Survival of Motor Neuron-2 (SMN2) Gene Splicing Modifier for the Treatment of Spinal Muscular Atrophy (SMA). J. Med. Chem. 61, 6501\u0026ndash;6517 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSantos, R. \u003cem\u003eet al.\u003c/em\u003e A comprehensive map of molecular drug targets. Nat. Rev. Drug Discov. 16, 19\u0026ndash;34 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOlenginski, L. T., Spradlin, S. F. \u0026amp; Batey, R. T. Flipping the script: Understanding riboswitches from an alternative perspective. J. Biol. Chem. 300, (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBreaker, R. R. The Biochemical Landscape of Riboswitch Ligands. Biochemistry 61, 137\u0026ndash;149 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSerganov, A. \u0026amp; Nudler, E. A decade of riboswitches. Cell 152, 17\u0026ndash;24 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBlount, K. F. \u003cem\u003eet al.\u003c/em\u003e Novel Riboswitch-Binding Flavin Analog That Protects Mice against Clostridium difficile Infection without Inhibiting Cecal Flora. Antimicrob. Agents Chemother. 59, 5736 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVicens, Q. \u003cem\u003eet al.\u003c/em\u003e Structure-Activity Relationship of Flavin Analogues That Target the Flavin Mononucleotide Riboswitch. ACS Chem. Biol. 13, 2908\u0026ndash;2919 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRizvi, N. F. \u003cem\u003eet al.\u003c/em\u003e Discovery of Selective RNA-Binding Small Molecules by Affinity-Selection Mass Spectrometry. ACS Chem. Biol. 13, 820\u0026ndash;831 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHowe, J. A. \u003cem\u003eet al.\u003c/em\u003e Selective small-molecule inhibition of an RNA structural element. Nature 526, 672\u0026ndash;677 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKr\u0026auml;utler, B. \u0026amp; EurJ, C. Antivitamins B12\u0026mdash;Some Inaugural Milestones. Chem. \u0026ndash; A Eur. J. 26, 15438\u0026ndash;15445 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNahvi, A. \u003cem\u003eet al.\u003c/em\u003e Genetic Control by a Metabolite Binding mRNA. Chem. Biol. 9, 1043\u0026ndash;1049 (2002).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMccown, P. J., Corbino, K. A., Stav, S., Sherlock, M. E. \u0026amp; Breaker, R. R. Riboswitch diversity and distribution. RNA 23, 995\u0026ndash;1011 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLennon, S. R. \u0026amp; Batey, R. T. Regulation of Gene Expression Through Effector-dependent Conformational Switching by Cobalamin Riboswitches. J. Mol. Biol. 434, 167585 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePolaski, J. T., Webster, S. M., Johnson, J. E. \u0026amp; Batey, R. T. Cobalamin riboswitches exhibit a broad range of ability to discriminate between methylcobalamin and adenosylcobalamin. J. Biol. Chem. 292, 11650\u0026ndash;11658 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDemerre, L. J. \u0026amp; Wilson, C. Photolysis of vitamin B12. J. Am. Pharm. Assoc. 45, 129\u0026ndash;134 (1956).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJuzeniene, A. \u0026amp; Nizauskaite, Z. Photodegradation of cobalamins in aqueous solutions and in human blood. J. Photochem. Photobiol. B. 122, 7\u0026ndash;14 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJohnson, J. E., Reyes, F. E., Polaski, J. T. \u0026amp; Batey, R. T. B12 cofactors directly stabilize an mRNA regulatory switch. Nat. 2012 4927427 492, 133\u0026ndash;137 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLennon, S. R. \u003cem\u003eet al.\u003c/em\u003e Targeting Riboswitches with Beta-Axial-Substituted Cobalamins. ACS Chem. Biol. 18, 1136\u0026ndash;1147 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePeselis, A. \u0026amp; Serganov, A. Structural insights into ligand binding and gene expression control by an adenosylcobalamin riboswitch. Nat. Struct. Mol. Biol. 2012 1911 19, 1182\u0026ndash;1184 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDavila-Calderon, J. \u003cem\u003eet al.\u003c/em\u003e IRES-targeting small molecule inhibits enterovirus 71 replication via allosteric stabilization of a ternary complex. \u003cem\u003eNat. Commun.\u003c/em\u003e 2020 \u003cem\u003e111\u003c/em\u003e 11, 1\u0026ndash;13 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen, J. L. \u003cem\u003eet al.\u003c/em\u003e Design, Optimization, and Study of Small Molecules That Target Tau Pre-mRNA and Affect Splicing. J. Am. Chem. Soc. 142, 8706\u0026ndash;8727 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMatyjasik, M. M., Hall, S. D. \u0026amp; Batey, R. T. High Affinity Binding of N2-Modified Guanine Derivatives Significantly Disrupts the Ligand Binding Pocket of the Guanine Riboswitch. Molecules 25, (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChromiński, M., Lewalska, A., Karczewski, M. \u0026amp; Gryko, D. Vitamin B12 derivatives for orthogonal functionalization. J. Org. Chem. 79, 7532\u0026ndash;7542 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChromiński, M., Lewalska, A. \u0026amp; Gryko, D. Reduction-free synthesis of stable acetylide cobalamins. Chem. Commun. 49, 11406\u0026ndash;11408 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDonlic, A. \u003cem\u003eet al.\u003c/em\u003e R-BIND 2.0: An Updated Database of Bioactive RNA-Targeting Small Molecules and Associated RNA Secondary Structures. ACS Chem. Biol. 17, 1556\u0026ndash;1566 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMorgan, B. S. \u003cem\u003eet al.\u003c/em\u003e R-BIND: An Interactive Database for Exploring and Developing RNA-Targeted Chemical Probes. ACS Chem. Biol. 14, 2691\u0026ndash;2700 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMorgan, B. S., Forte, J. E., Culver, R. N., Zhang, Y. \u0026amp; Hargrove, A. E. Discovery of Key Physicochemical, Structural, and Spatial Properties of RNA-Targeted Bioactive Ligands. Angew. Chemie Int. Ed. 56, 13498\u0026ndash;13502 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBraselmann, E. \u003cem\u003eet al.\u003c/em\u003e A multicolor riboswitch-based platform for imaging of RNA in live mammalian cells. Nat. Chem. Biol. \u003cem\u003e2018 1410\u003c/em\u003e 14, 964\u0026ndash;971 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMorgan, B. S., Forte, J. E., Culver, R. N., Zhang, Y. \u0026amp; Hargrove, A. E. Discovery of Key Physicochemical, Structural, and Spatial Properties of RNA-Targeted Bioactive Ligands. Angew. Chem. Int. Ed. Engl. 56, 13498\u0026ndash;13502 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStratton, C. F., Newman, D. J. \u0026amp; Tan, D. S. Cheminformatic comparison of approved drugs from natural product versus synthetic origins. Bioorg. Med. Chem. Lett. 25, 4802\u0026ndash;4807 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePatwardhan, N. N., Cai, Z., Umuhire Juru, A. \u0026amp; Hargrove, A. E. Driving factors in amiloride recognition of HIV RNA targets. Org. Biomol. Chem. 17, 9313\u0026ndash;9320 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePatwardhan, N. N. \u003cem\u003eet al.\u003c/em\u003e Amiloride as a new RNA-binding scaffold with activity against HIV-1 TAR. Medchemcomm 8, 1022 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCai, Z., Zafferani, M., Akande, O. M. \u0026amp; Hargrove, A. E. Quantitative Structure-Activity Relationship (QSAR) Study Predicts Small-Molecule Binding to RNA Structure. J. Med. Chem. 65, 7262\u0026ndash;7277 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePolaski, J. T., Holmstrom, E. D., Nesbitt, D. J. \u0026amp; Batey, R. T. Mechanistic Insights into Cofactor-Dependent Coupling of RNA Folding and mRNA Transcription/Translation by a Cobalamin Riboswitch. Cell Rep. 15, 1100\u0026ndash;1110 (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNeidhardt, F. C. \u003cem\u003eEscherichia coli and Salmonella typhimurium - Cellular and Molecular Biology\u003c/em\u003e. \u003cem\u003eWashington: American Society for Microbiology\u003c/em\u003e (Washington: American Society for Microbiology, 1987).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFigge, R. M. Methionine Biosynthesis in Escherichia coli and Corynebacterium glutamicum. in \u003cem\u003eWendisch\u003c/em\u003e, \u003cem\u003eV.F. (eds) Amino Acid Biosynthesis\u0026thinsp;~\u0026thinsp;Pathways, Regulation and Metabolic Engineering\u003c/em\u003e 163\u0026ndash;193 (Springer, Berlin, Heidelberg, 2006). doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/7171_2006_059\u003c/span\u003e\u003cspan address=\"10.1007/7171_2006_059\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLipinski, C. A. Lead- and drug-like compounds: the rule-of-five revolution. Drug Discov. Today. Technol. 1, 337\u0026ndash;341 (2004).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLipinski, C. A., Lombardo, F., Dominy, B. W. \u0026amp; Feeney, P. J. Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv. Drug Deliv. Rev. 46, 3\u0026ndash;26 (2001).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWicks, S. L. \u0026amp; Hargrove, A. E. Fluorescent indicator displacement assays to identify and characterize small molecule interactions with RNA. Methods 167, 3\u0026ndash;14 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTong, Y. \u003cem\u003eet al.\u003c/em\u003e Programming inactive RNA-binding small molecules into bioactive degraders. Nat. 2023 6187963 618, 169\u0026ndash;179 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGeng, A., Roy, R. \u0026amp; Al-Hashimi, H. M. Conformational penalties: New insights into nucleic acid recognition. Curr. Opin. Struct. Biol. 89, (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStelzer, A. C. \u003cem\u003eet al.\u003c/em\u003e Discovery of selective bioactive small molecules by targeting an RNA dynamic ensemble. Nat. Chem. Biol. \u003cem\u003e2011 78\u003c/em\u003e 7, 553\u0026ndash;559 (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePanei, F. P., Gkeka, P. \u0026amp; Bonomi, M. Identifying small-molecules binding sites in RNA conformational ensembles with SHAMAN. Nat. Commun. 2024 \u003cem\u003e151\u003c/em\u003e 15, 1\u0026ndash;15 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhou, Y. \u0026amp; Chen, S.-J. Advances in machine-learning approaches to RNA-targeted drug design. Artif. Intell. Chem. 2, 100053 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMorishita, E. C. \u0026amp; Nakamura, S. Recent applications of artificial intelligence in RNA-targeted small molecule drug discovery. Expert Opin. Drug Discov. 19, 415\u0026ndash;431 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSonar, M. V. \u003cem\u003eet al.\u003c/em\u003e Fluorescence Detection of KRAS2 mRNA Hybridization in Lung Cancer Cells with PNA-Peptides Containing an Internal Thiazole Orange. Bioconjug. Chem. 25, 1697\u0026ndash;1708 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWierzba, A. J., Richards, E. M., Lennon, S. R., Batey, R. T. \u0026amp; Palmer, A. E. Unveiling the promise of peptide nucleic acids as functional linkers for an RNA imaging platform. RSC Chem. Biol. (2024) doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1039/D4CB00274A\u003c/span\u003e\u003cspan address=\"10.1039/D4CB00274A\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCorson, T. W., Aberle, N. \u0026amp; Crews, C. M. Design and applications of bifunctional small molecules: Why two heads are better than one. ACS Chem. Biol. 3, 677\u0026ndash;692 (2008).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAngelbello, A. J. \u003cem\u003eet al.\u003c/em\u003e Precise small-molecule cleavage of an r(CUG) repeat expansion in a myotonic dystrophy mouse model. \u003cem\u003eProc. Natl. Acad. Sci. U. S. A.\u003c/em\u003e 116, 7799\u0026ndash;7804 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEdwards, A. L., Garst, A. D. \u0026amp; Batey, R. T. Determining structures of RNA aptamers and riboswitches by X-ray crystallography. Methods Mol. Biol. 535, 135\u0026ndash;163 (2009).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTopliss, J. G. \u0026amp; Edwards, R. P. Chance factors in studies of quantitative structure-activity relationships. J. Med. Chem. 22, 1238\u0026ndash;1244 (1979).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCheatham, T. E., Miller, J. L., Fox, T., Darden, T. A. \u0026amp; Kollman, P. A. Molecular Dynamics Simulations on Solvated Biomolecular Systems: The Particle Mesh Ewald Method Leads to Stable Trajectories of DNA, RNA, and Proteins. J. Am. Chem. Soc. 117, 4193\u0026ndash;4194 (1995).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCase, D. A. \u003cem\u003eet al.\u003c/em\u003e The Amber biomolecular simulation programs. J. Comput. Chem. 26, 1668\u0026ndash;1688 (2005).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIzadi, S., Anandakrishnan, R. \u0026amp; Onufriev, A. V. Building water models: A different approach. J. Phys. Chem. Lett. 5, 3863\u0026ndash;3871 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZgarbov\u0026aacute;, M. \u003cem\u003eet al.\u003c/em\u003e Refinement of the Cornell et al. Nucleic acids force field based on reference quantum chemical calculations of glycosidic torsion profiles. \u003cem\u003eJ. Chem. Theory Comput.\u003c/em\u003e 7, 2886\u0026ndash;2902 (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSteinbrecher, T., Latzer, J. \u0026amp; Case, D. A. Revised AMBER parameters for bioorganic phosphates. J. Chem. Theory Comput. 8, 4405\u0026ndash;4412 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi, P. \u0026amp; Merz, K. M. MCPB.py: A Python Based Metal Center Parameter Builder. J. Chem. Inf. Model. 56, 599\u0026ndash;604 (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHopkins, C. W., Le Grand, S., Walker, R. C. \u0026amp; Roitberg, A. E. Long-time-step molecular dynamics through hydrogen mass repartitioning. J. Chem. Theory Comput. 11, 1864\u0026ndash;1874 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRoe, D. R. \u0026amp; Cheatham, T. E. PTRAJ and CPPTRAJ: Software for processing and analysis of molecular dynamics trajectory data. J. Chem. Theory Comput. 9, 3084\u0026ndash;3095 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMinor, W., Cymborowski, M., Otwinowski, Z. \u0026amp; Chruszcz, M. HKL-3000: the integration of data reduction and structure solution\u0026ndash;from diffraction images to an initial model in minutes. Acta Crystallogr. D. Biol. Crystallogr. 62, 859\u0026ndash;866 (2006).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKabsch, W. XDS. Acta Crystallogr. Sect. D Struct. Biol. 66, 125\u0026ndash;132 (2010).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBunk\u0026oacute;czi, G. \u003cem\u003eet al.\u003c/em\u003e Phaser.MRage: automated molecular replacement. Acta Crystallogr. Sect. D Biol. Crystallogr. 69, 2276\u0026ndash;2286 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAdams, P. D. \u003cem\u003eet al.\u003c/em\u003e The Phenix software for automated determination of macromolecular structures. Methods 55, 94\u0026ndash;106 (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMoriarty, N. W., Grosse-Kunstleve, R. W. \u0026amp; Adams, P. D. electronic Ligand Builder and Optimization Workbench (eLBOW): a tool for ligand coordinate and restraint generation. Acta Crystallogr. Sect. D Biol. Crystallogr. 65, 1074\u0026ndash;1080 (2009).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTerwilliger, T. C., Klei, H., Adams, P. D., Moriarty, N. W. \u0026amp; Cohn, J. D. Automated ligand fitting by core-fragment fitting and extension into density. Acta Crystallogr. Sect. D Biol. Crystallogr. 62, 915\u0026ndash;922 (2006).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAdams, P. D., Pannu, N. S., Read, R. J. \u0026amp; Brunger, A. T. Extending the limits of molecular replacement through combined simulated annealing and maximum-likelihood refinement. Acta Crystallogr. D. Biol. Crystallogr. 55, 181\u0026ndash;190 (1999).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChou, F. C., Sripakdeevong, P., Dibrov, S. M., Hermann, T. \u0026amp; Das, R. Correcting pervasive errors in RNA crystallography through enumerative structure prediction. Nat. Methods 2012 \u003cem\u003e101\u003c/em\u003e 10, 74\u0026ndash;76 (2013).\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"RNA structure, drug discovery, small molecules, structure-based design, base displacement","lastPublishedDoi":"10.21203/rs.3.rs-5836924/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5836924/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Most RNA-binding small molecules have limited solubility, weak affinity, and/or lack of specificity, restricting the medicinal chemistry often required for lead compound discovery. We reasoned that conjugation of these unfavorable ligands to a suitable “host” molecule can solubilize the “guest” and deliver it site-specifically to an RNA of interest to resolve these issues. Using this framework, we designed a small molecule library that was hosted by cobalamin (Cbl) to interact with the Cbl riboswitch through a common base displacement mechanism. Combining in vitro binding, cell-based assays, chemoinformatic modeling, and structure-based design, we unmasked a cryptic binding site within the riboswitch that was exploited to discover compounds that have affinity exceeding the native ligand, antagonize riboswitch function, or bear no resemblance to Cbl. These data demonstrate how a privileged biphenyl-like scaffold effectively targets RNA by optimizing π-stacking interactions within the binding pocket.","manuscriptTitle":"Designing small molecules that target a cryptic RNA binding site via base displacement","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-01-29 06:31:52","doi":"10.21203/rs.3.rs-5836924/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-chemical-biology","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"nchembio","sideBox":"Learn more about [Nature Chemical Biology](http://www.nature.com/nchembio/)","snPcode":"","submissionUrl":"","title":"Nature Chemical Biology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Research","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"a4615b35-90a1-42f5-a724-6f1adc4d9319","owner":[],"postedDate":"January 29th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":43343972,"name":"Biological sciences/Biochemistry/RNA"},{"id":43343973,"name":"Biological sciences/Structural biology/X-ray crystallography"},{"id":43343974,"name":"Biological sciences/Drug discovery/Medicinal chemistry/Structure-based drug design"}],"tags":[],"updatedAt":"2025-08-30T07:06:55+00:00","versionOfRecord":{"articleIdentity":"rs-5836924","link":"https://doi.org/10.1038/s41589-025-02018-8","journal":{"identity":"nature-chemical-biology","isVorOnly":false,"title":"Nature Chemical Biology"},"publishedOn":"2025-08-29 04:00:00","publishedOnDateReadable":"August 29th, 2025"},"versionCreatedAt":"2025-01-29 06:31:52","video":"","vorDoi":"10.1038/s41589-025-02018-8","vorDoiUrl":"https://doi.org/10.1038/s41589-025-02018-8","workflowStages":[]},"version":"v1","identity":"rs-5836924","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5836924","identity":"rs-5836924","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}