Lifetime of ground conformational state determines the activity of structured RNA

Abstract

Biomolecules continually sample alternative conformations. Consequently, even the most energetically favored ground conformational state has a finite lifetime. Here, we show that, in addition to the three-dimensional (3D) structure, the lifetime of a ground conformational state determines its biological activity. Using hydrogen-deuterium exchange nuclear magnetic resonance spectroscopy, we found that Zika virus exoribonuclease-resistant RNA (xrRNA) encodes a ground conformational state with a lifetime that is ~10⁵-10⁷ longer than that of canonical base pairs. Mutations that shorten the apparent lifetime of the ground state without affecting its 3D structure decreased exoribonuclease resistance in vitro and impaired virus replication in cells. Additionally, we observed this exceptionally long-lived ground state in xrRNAs from diverse infectious mosquito-borne flaviviruses. These results demonstrate the biological importance of the lifetime of a preorganized ground state and further suggest that elucidating the lifetimes of dominant 3D structures of biomolecules may be crucial for understanding their behaviors and functions.

Extended Data Fig. 1 |. Zika virus replication assays.

a, Schematic illustration of sfRNA1 and sfRNA2 generation via xrRNA-dependent Xrn1 inhibition. b, Secondary structure of WT xrRNA1 and xrRNA2 in H/PF/2013 ZIKV used in the replication assays. Highlighted in orange is the long-range pseudoknot interaction in xrRNA1. c, Full images of the wells of infected Vero cells presented in Fig. 1d. Scale bars are not presented as the original exported images did not incorporate scale bars. Shown above each image is the name and sequence of xrRNA1 long-range pseudoknot interaction in WT and mutant ZIKV used in the assays.

Extended Data Fig. 2 |. Representative electron density of ZIKV xrRNA1 C55/57U mutant.

a, Electron density map, 2mFo-DFc, shown at 1σ on full-length C55/57U crystallography construct. b-d, Highlighted regions in Fig. 2b with electron density maps shown at 0.75σ (b, c) and 1σ (d).

Extended Data Fig. 3 |. Solution NMR and UV melting characterizations of the long-range pseudoknot interaction in ZIKV xrRNA1 mutants.

a, Comparison of the imino-proton region of ¹H-¹H NOESY spectra at 15 °C. The presence of G33-U55 wobble in C55/C7U mutant is supported from the observed cross peaks between G33-H1 and U55-H3 imino protons, which are not present in the WT construct. b, UV melting profiles of ZIKV xrRNA1^WT NMR construct (left), ZIKV xrRNA1^C55/57U NMR construct (center), and ZIKV xrRNA1^ΔPK NMR construct (right). The elevated baseline observed in xrRNA1^C55/57U construct is subject to future investigations. Thin vertical lines correspond to fitted melting temperatures.

Extended Data Fig. 4 |. Solution NMR characterization and identification of G7 as the highly stable residue.

a-d, Secondary structures and imino proton regions of NMR ¹H-¹H NOESY spectra for NMR constructs of ZIKV xrRNA1^WT (a), xrRNA1^C55/57U (b), xrRNA1^ΔPK (c), and xrRNA1^{ΔPK, G8A/C42U} (d). Shown above are 1D ¹H NMR spectra of imino proton region in H₂O and D₂O, where G7 imino protons are filled-in in the 1D ¹H spectra recorded in D₂O. The xrRNA1^{ΔPK, G8A/C42U} NMR construct (d) is designed to swap G8-C42 base pair, which resides above G7-C43 base pair, with A8-U42 base pair to confirm the resonance assignment of G7 imino proton. e-h, 1D ¹H NMR spectra of imino proton region in H₂O and upon dissolving in D₂O for xrRNA1^WT (e) and xrRNA1^ΔPK (f) in the absence of Mg²⁺ and xrRNA1^C22G (g) and xrRNA1^{ΔPK, C22G} (h) in the presence of Mg²⁺. The C22G mutation is designed to abolish xrRNA folding. For better visualizing residual signals, the spectra in D₂O are scaled up five times relative to those in H₂O.

Extended Data Fig. 5 |. Structural dynamics and functional characterization of ZIKV xrRNA1 and mutants.

a-d, Secondary structures and imino proton regions of NMR ¹H-¹H NOESY spectra for NMR constructs of ZIKV xrRNA1^C57U (a), xrRNA1^C56/57U (b), xrRNA1^C54/56U (c), and xrRNA1^C54/55U (d) at 15 °C. e, 1D ¹H NMR spectra of the imino proton for ZIKV xrRNA1^WT and xrRNA1^ΔPK NMR constructs upon dissolving in D₂O at 37 °C and their NMR ¹H-¹H NOESY spectra in H₂O at 37 °C, which display similar pattern as their corresponding spectra at 15 °C. f, 1D ¹H NMR spectra of the imino proton for ZIKV xrRNA1 mutant constructs upon dissolving in D₂O at 37 °C. g, Representative Xrn1 digestion assays from three replicates (n = 3) of full-length ZIKV xrRNA1^C57U, xrRNA1^C56/57U, xrRNA1^C54/56U, and xrRNA1^C54/55U assay constructs with P4 stem. h, Duplicates of NMR HDX profiles of G7 imino proton for six ZIKV NMR constructs measured at 37 °C. i, NMR HDX profile of G7 imino proton for ZIKV xrRNA1^ΔPK NMR construct where two separate HDX experiments with a total of six measurements at 37 °C were used.

Extended Data Fig. 6 |. Structural dynamics and functional characterization of other Mosquito-borne flaviviruses xrRNAs.

a-d, Secondary structures of dengue virus 1 (DENV-1) xrRNA1 (a), Japanese encephalitis virus (JEV) xrRNA1 (b), Saint Louis encephalitis virus (SLEV) xrRNA1 (c), and ZIKV xrRNA2 (d). e, 1D ¹H NMR spectra of the imino proton for NMR constructs of DENV-1 xrRNA1, JEV xrRNA1, SLEV xrRNA1, and ZIKV xrRNA2 without P4 upon dissolving in D₂O at 37 °C. f, Representative Xrn1 digestion assays from three replicates (n = 3) of the full-length DENV-1 xrRNA1, JEV xrRNA1, SLEV xrRNA1, and ZIKV xrRNA2 constructs. g, Duplicates of NMR HDX profiles of G7 imino proton for NMR constructs of DENV-1 xrRNA1, JEV xrRNA1, SLEV xrRNA1, and ZIKV xrRNA2 without P4 measured at 37 °C.

Extended Data Fig. 7 |. Constant-force molecular dynamics simulations of ZIKV xrRNA1s.

a, Ratio of pulled-through trajectories between C55/57U and WT xrRNA1 as a function of applied force in CFMD simulations. A pulled-through event is depicted as disrupted G7–C48–C22 triple. Colored in red are forces selected in large-scale CFMD simulations. b, Receiver operating characteristic (ROC) analyses of the size of the ring-like architecture and pulled-through event from WT (orange), C55/57U (blue), and combined (black) trajectories, where the high area under the curve (AUC) values suggests the larger size of the ring-like architecture contributes to pulled-through events. c, Representative ROC analyses of the opening of G34-C54 base pair and the opening of the ring-like architecture from WT (orange), C55/57U (blue), and combined (black) trajectories, where the high AUC values suggest the larger base pair opening contributes to the larger size of the ring-like architecture. d, Receiver operating characteristic (ROC) analysis of the pseudoknot base-pair motions in CFMD simulations. ROC plots for various base pair motions being used to predict the opening of the ring-like architecture. Indicated on each plot is the base pair motion type (stagger, shear, buckle, propeller, stretch, and opening) and base pairs in the pseudoknot region (G34-C54, G33-C55, G32-C56, and G31-C57) with corresponding area under the curve (AUC) values. Highlighted in red are curves with high AUC values that suggest stretching and opening motions of pseudoknot capping and stem base pairs mostly contribute to opening of the ring-like architecture.

Extended Data Fig. 8 |. Structural dynamics and functional characterization of ZIKV xrRNA1 mutants of the ring-like architecture.

a, Secondary structures of ZIKV xrRNA1 and mutants. b, Representative Xrn1 digestion assays of the full-length ZIKV xrRNA1^A37C, xrRNA1^A+, and xrRNA1^AA+ constructs. Plotted are average percentages of Xrn1 resistance from three replicates (n = 3) with standard deviations as error bars. c, Duplicates of NMR HDX profiles of G7 imino proton for NMR constructs of xrRNA1^A37C, xrRNA1^A+, and xrRNA1^AA+ without P4 measured at 37 °C. d, Diagram of Xrn1 resistances and the apparent lifetimes of G7 imino proton of xrRNAs. ZIKV xrRNA1 pseudoknot mutants and other wild-type MBFV xrRNAs in Fig. 4e are shown in gray for clarity. Plotted are average Xrn1 resistances and apparent lifetimes with s.d. of xrRNAs as reported in panels b and c and Fig. 4e.

Fig. 1 |. Pseudoknot interaction modulates Xrn1 resistance of ZIKV xrRNA1.

a, Secondary structure of ZIKV xrRNA1 and pseudoknot sequences of xrRNA1^WT and xrRNA1^C55/57U and xrRNA1^ΔPK mutants. b, In vitro Xrn1 digestion assays of ZIKV xrRNA1^WT, xrRNA1^C55/57U and xrRNA1^ΔPK with ³²P body labeling. Plotted are the average percentages of Xrn1 resistance from three replicates with the s.d. as error bars. c, Quantification of viral replication of WT and mutant ZIKV in Vero cells. Viral titers were quantified using focus-forming assays and reported as focus-forming units (FFU) per ml. Plotted are the average values from n = 15 replicates from five independent experiments for WT and C55/57U ZIKVs and n = 18 replicates from six independent experiments for ΔPK ZIKV with the s.e.m. as error bars. *P < 0.05 and ***P < 0.001 for statistical differences between WT and mutant ZIKV. P values were calculated using a two-way analysis of variance adjusted for multiple comparisons, resulting in P = 0.0010 between WT and C55/57U and P < 0.0001 between WT and ΔPK ZIKVs. Northern blots of sfRNAs from ZIKV-infected Vero cells are shown together with U6 RNA as an internal control. d, Quantification of cell-to-cell spread of WT and mutant ZIKV in Vero cells. Plotted are the mean areas of infectious foci from nine wells of one representative plate. NS, not significant (P > 0.05); ****P < 0.0001 for statistical differences between WT and mutant ZIKV. Scale bars are not presented as the original exported images did not incorporate scale bars. P values were calculated using an unpaired two-tailed t-test, resulting in P < 0.0001 between WT and C55/57U and P < 0.0001 between WT and ΔPK ZIKVs.

Fig. 2 |. Structure of ZIKV xrRNA1s.

a, Secondary structure of the ZIKV xrRNA1^C55/57U construct in a crystallography study. b, Electron density map, 2mF_o − DF_c, shown at 1σ on the ZIKV xrRNA1^C55/57U construct and surface representation of the crystal structure, highlighting the ring architecture. c, Structural comparison between ZIKV xrRNA1^C55/57U (rainbow) and xrRNA1^WT (orange; PDB 5TPY).

Fig. 3 |. Conformational dynamics of ZIKV xrRNA1s by HDX NMR.

a, HDX NMR of WT xrRNA1 NMR construct at 17 °C. Shown are the ¹H NMR spectra of the imino proton region in H₂O and D₂O. Time-dependent peak intensities of G7 imino proton upon dissolving in D₂O are fit to a monoexponential decay. Reported is the average apparent lifetime with s.d. estimated from fitting n = 241 data points. b, Tris-dependent HDX NMR of xrRNA1^WT NMR construct at 17 °C. Shown is a schematic diagram that depicts the HDX process of the G7 imino proton. Time-dependent peak intensities of G7 imino proton upon dissolving in D₂O are plotted as a function of Tris concentration. Individual average apparent lifetimes with s.d. estimated from n = 49–100 data points are fit linearly to 1/[Tris] to extract the intrinsic lifetime of the G7 imino proton. c,d, HDX NMR of xrRNA1^C55/57U (c) and xrRNA1^ΔPK (d) NMR constructs at 17 °C. Shown are the ¹H NMR spectra of the imino proton region in H₂O and D₂O. Time-dependent peak intensities of G7 imino proton upon dissolving in D₂O are fit to a monoexponential decay. Reported are the average apparent lifetimes with s.d. estimated from fitting n = 215 and 23 data points for xrRNA1^C55/57U and xrRNA1^ΔPK, respectively.

Fig. 4 |. Conformational dynamics and Xrn1 resistance of MBFV xrRNAs.

a, Secondary structures of WT and mutant ZIKV xrRNA1s. b, Dynamic and functional analyses of WT and mutant ZIKV xrRNA1s at 37 °C. Shown are the secondary structure of ZIKV xrRNA1 and pseudoknot sequences of WT and mutants. Plotted are average percentages of Xrn1 resistance from three replicates of in vitro Xrn1 digestion assay with the s.d. as error bars. Time-dependent peak intensities of G7 imino protons upon dissolving in D₂O are plotted for each construct and fit to monoexponential decays. Reported are the average apparent lifetimes with s.d. estimated from fitting two independent measurements with a total of n = 59, 40, 27, 92, 94, 27 and 4 data points for xrRNA1^WT, xrRNA1^C57U, xrRNA1^C56/57U, xrRNA1^C54/56U, xrRNA1^C54/55U, xrRNA1^C55/57U and xrRNA1^ΔPK, respectively. c, Secondary structures of other MBFV xrRNAs. d, Dynamic and functional analyses of other MBFV xrRNAs at 37 °C. Plotted are the average Xrn1 resistances of DENV-1 xrRNA1, JEV xrRNA1, SLEV xrRNA1 and ZIKV xrRNA2 (Z2) from three replicates of in vitro Xrn1 digestion assay with the s.d. as error bars. Time-dependent peak intensities of G7 imino protons upon dissolving in D₂O are plotted for each construct and fit to monoexponential decays. Reported are the average apparent lifetimes with the s.d. estimated from fitting two independent measurements with a total of n = 82, 66, 65 and 74 data points for DENV-1, JEV, SLEV and Z2, respectively. e, Diagram of Xrn1 resistances and the apparent lifetimes of G7 imino protons of xrRNAs. Plotted are the average Xrn1 resistances and apparent lifetimes with the s.d. of xrRNAs as reported in b and d.

Fig. 5 |. Xrn1 resistance of ZIKV xrRNA1s through constant-force simulations.

a, Representative trajectories of ZIKV xrRNA1^WT, xrRNA1^C55/57U and xrRNA1^ΔPK from CFMD simulations at 400 pN. Shown are the states of xrRNA1s captured at the same time steps along trajectories. The conserved G7–C48–C22 triple is highlighted in red. b, Accumulated pulled-through fractions of WT and C55/57U xrRNA1s as a function of time steps in CFMD simulations. The schematic diagram depicts a pulled-through event as a disrupted G7–C48–C22 triple. c, Distributions of the sizes of the ring architecture as the radius of gyration and the opening degrees of the pseudoknot base pair G34–C54 of xrRNA1^WT and xrRNA1^C55/57U in CFMD simulations. Violin plots show the full distribution of data points for xrRNA1^WT (n = 3,200) and xrRNA1^C55/57U (n = 3,195) through kernel density estimation, with width proportional to the number of observations at each value. Box plots overlaid within show the median (center line) with the bounds as the 25th and 75th percentiles. The whiskers extend to 1.5× the interquartile range. P values were calculated using a two-sided t-test with the Holm method, resulting in P = 1.92 × 10⁻¹¹⁰ and P = 1.27 × 10⁻²¹⁶ for ring size and base opening between xrRNA1^WT and xrRNA1^C55/57U, respectively.

Fig. 6 |. Model for xrRNA-dependent Xrn1 resistance.

xrRNA acts as a molecular wall to inhibit Xrn1 digestion toward generating sfRNAs to promote viral replication. The strength of exonuclease resistance is influenced by the lifetime of the ground conformational state of the closed G7–C48 base pair anchored on the ring architecture, which is modulated through the long-range pseudoknot interaction. As shown in the central panel, a longer ground-state lifetime of the xrRNA makes it less susceptible to Xrn1 digestion, which results in higher levels of accumulation of the sfRNAs. GS, ground state.