Molecular insights into the interaction between a disordered protein and a folded RNA

Intrinsically disordered protein regions (IDRs) are well-established as contributors to intermolecular interactions and the formation of biomolecular condensates. In particular, RNA-binding proteins (RBPs) often harbor IDRs in addition to folded RNA-binding domains that contribute to RBP function. To understand the dynamic interactions of an IDR-RNA complex, we characterized the RNA-binding features of a small (68 residues), positively charged IDR-containing protein, SERF. At high concentrations, SERF and RNA undergo charge-driven associative phase separation to form a protein- and RNA-rich dense phase. A key advantage of this model system is that this threshold for demixing is sufficiently high that we could use solution-state biophysical methods to interrogate the stoichiometric complexes of SERF with RNA in the one-phase regime. Herein, we describe our comprehensive characterization of SERF alone and in complex with a small fragment of the HIV-1 TAR RNA (TAR) with complementary biophysical methods and molecular simulations. We find that this binding event is not accompanied by the acquisition of structure by either molecule; however, we see evidence for a modest global compaction of the SERF ensemble when bound to RNA. This behavior likely reflects attenuated charge repulsion within SERF via binding to the polyanionic RNA and provides a rationale for the higher-order assembly of SERF in the context of RNA. We envision that the SERF-RNA system will lower the barrier to accessing the details that support IDR-RNA interactions and likewise deepen our understanding of the role of IDR-RNA contacts in complex formation and liquid-liquid phase separation.


Fig. S1: Local structural features of the SERF ensemble from NMR spectroscopy and allatom simulations. (A)
H, 15 N-HSQC spectrum with SERF assignments (see also Table S1).(B) 13 C, 15 N-CON spectrum of SERF with assignments (see also Table S1).

Fig. S2: Global SERF ensemble characterization by SAXS and SV-AUC. (A)
Raw scattering curves for two concentrations of SERF shown with scaled SAXS profile calculated from all-atom simulations using FoXS (5).(B) Guinier transformation of SAXS data at two concentrations with linear fit to approximate radius of gyration.The small spread of residuals suggests adequate linear fitting in both cases.(C) Raw scattering curves for each SERF (black) and TAR (green) alone, the sum of their scattering profiles (pink), and the scattering curve measured for the complex (blue).The lines correspond to the fit for a 1:1 complex with an apparent Kd = 260 µM.Notably, this apparent Kd differs from the Kd obtained from the binding isotherm based on fluorescence anisotropy.The discrepancy in the two Kd values might be due to the difference in protein and RNA concentrations present in the the two experiment types and possibly due to SERF-SERF interactions at high concentrations.Although the apparent Kd value determined by NMR titrations is not directly used in this study, it highlights that the TAR:SERF complex is far from being saturated at a 1:1 ratio.(C) RDC reproduction of TAR assuming two independent ideal A-form helices.The associated alignment tensors are given in Table S1.*the RG calculations for SERF in complex were performed on the protein chain only; the RNA chain was ignored.N/A -not applicable.
Acronyms: AFRC -analytical Flory random coil(1); ALBATROSS -deep learning-based predictor(2); EOM -ensemble optimization method(3); MFF -molecular form factor(4); SEstandard error; SD -standard deviation.‡ From experiments, the reported error is the standard deviation of the fit from the covariance matrix; the error from simulations is the standard error of the mean calculated from KD values across independent replicates.
Fig. S1: Local structural features of the SERF ensemble from NMR spectroscopy and allatom simulations.(A) 1 H, 15 N-HSQC spectrum with SERF assignments (see also TableS1).(B)13  C, 15 N-CON spectrum of SERF with assignments (see also TableS1).(C) Plot of SERF secondary structure from side chain chemical shifts (positive values indicate α-helix; negative values indicate β-strand) (left axis).The DSSP scores for α-helical character averaged over all simulation frames are shown as a dashed line (right axis).The yellow rectangle shows the C-SERF helical region of interest.(D) Zoomed-in representation of DSSP scores for C-SERF helix from simulations.The breakpoint in the helix is centered around a stretch of charged residues (DARREAEK).(E) Plot of T1 relaxation times per residue of SERF from 15 N spin relaxation.

Fig. S3 :
Fig. S3: Ion-mobility mass spectrometry of SERF-TAR complexes.(A) Plot of drift time as a function mass/charge ratio used to determine collision cross-section distributions.Each feature is assigned with the biomolecule(s) and ionization state it represents using the shapes given in the legend.The molecular weights for different species are given in the legend.(B) Plot of collision cross-sections from different carrier gases.The dashed line is the function y = x to guide the eye.

Fig. S4 :
Fig. S4: Monitoring 'SERF'-'TAR' binding by coarse-grained simulations using Mpipi.(A) Trace of distances between each 'SERF' and 'TAR' over the course of the simulation measured from the center of mass (COM) of each molecule.(B) COM-COM distances from (A) presented as a distribution and fit with a 2-Gaussian model.The green (low COM-COM distance) and grey (high COM-COM distance) shading reflects 'bound' and 'unbound' frame assignments, respectively.The grey vertical line at 52.5 Å is the 'cutoff' distance (see Methods) that minimizes the overlap of the two sub-distributions.(C) Distribution of KD values across five replicates for each included TAR conformation ranked by average affinity.The vertical grey dashed line and shaded region represent the average and standard deviation of dissociation constants over all listed conformers.(D) Cartoon depiction of all 20 TAR conformations from PDB ID 1ANR aligned for visualization.Five colored TAR conformations that were omitted from average KD calculations are shown in the overlay and separately.See also Table S#.

Fig. S5 :
Fig. S5: NMR spectroscopy of SERF bound to TAR RNA.(A) 13 C, 15 N-CON spectra of unbound SERF (grey, from Fig. S2) and TAR-bound SERF (green).Significant linebroadening of resonances for residues that directly contact RNA precluded use of the CON for analysis of chemical shift perturbations upon RNA binding.Resonance assignments are shown for unbound SERF.(B) 1 H, 15 N-HSQC spectra of SERF alone (grey) and bound to TAR (green).The peak assignments shown are those for the bound state, which were used to generate the plots in (C) and (D).(C) Plot of chemical shift perturbations (CSP) on SERF for binding to TAR RNA (green circles) or rU30 (magenta squares) on the same axis set.The yellow highlighting along the x-axis denotes the region of SERF that is examined more closely in (D).(D) Re-scaled representation of CSP plot from (C) to reflect the differences in CSP magnitude for the different RNA binding partners.(E) Plot of heteronuclear NOEs per residue for the unbound (red) and RNA-bound SERF (blue).

Fig. S6 :
Fig. S6: NMR spectroscopy of SERF-TAR interaction.(A) (top) Imino and (bottom) C1' TROSY-HSQC spectra of isotopically enriched TAR showing chemical shift changes upon titrating with unlabeled SERF (molar ratio color coded according to the insert scale).(D) Chemical shift perturbations (in ppm) of the residue C24 of TAR with increasing [SERF]/[TAR] from C5H5 (red), C6H6 (green), and C1'H1' (blue).The lines correspond to the fit for a 1:1 complex with an apparent Kd = 260 µM.Notably, this apparent Kd differs from the Kd obtained from the binding isotherm based on fluorescence anisotropy.The discrepancy in the two Kd values might be due to the difference in protein and RNA concentrations present in the the two experiment types and possibly due to SERF-SERF interactions at high concentrations.Although the apparent Kd value determined by NMR titrations is not directly used in this study, it highlights that the TAR:SERF complex is far from being saturated at a 1:1 ratio.(C) RDC reproduction of TAR assuming two independent ideal A-form helices.The associated alignment tensors are given in TableS1.

Fig. S7 :
Fig. S7: NMR titrations to probe the SERF-TAR interface.(A) Chemical shift perturbations and (B) intensity ratios of TAR for N1H1, N3H3, C2H2, C6H6, C8H8, C5H5, and C1'H1' spin pairs along the titration (Fig 5 and S6).Molar ratio color coded according to the insert scale.The star (*) indicates a higher uncertainty for signal highly overlapping in the TAR alone spectrum.

Fig. S8 :
Fig. S8: SERF and TAR undergo concentration-dependent assembly and phase separation.(A) Time-resolved DIC images of SERF-TAR droplets (in 10% PEG buffer) demonstrate their liquid-like behavior via droplet fusion.(B) DIC image of SERF-TAR droplets composed of 125 μM TAR and 312.5 μM SERF in 0% PEG.(C) SDS-PAGE showing DMTMM-crosslinked SERF species in the presence or absence of TAR RNA.Higher-order SERF species are observed only at high concentrations of SERF and TAR.(D) DIC images of SERF-TAR droplets (in 10% PEG buffer) showing sensitivity to NaCl concentrations.

Table S1 : Chemical shift assignments for SERF from 13 C direct-detect NMR experiments.
Atoms with missing assignments denoted by --are those for which we do not expect to observe a resonance (i.e., we cannot detect the amide nitrogen for N-terminal residue in 13 C-detected experiments; and Gly residues do not have Cβ atoms).Aside from these, only one atom is missing a chemical shift assignment (Hα from Asp20).n.d.-not determined.

Table S3 : Dissociation constants from experiments and simulations.
Dissociation constants from coarse-grained simulations are not absolute values; a correction factor of 10 was introduced into the analysis workflow for easier comparison between experiments and simulations. †