Self‐Assembly of Measles Virus Nucleocapsid‐like Particles: Kinetics and RNA Sequence Dependence

Abstract

Measles virus RNA genomes are packaged into helical nucleocapsids (NCs), comprising thousands of nucleo‐proteins (N) that bind the entire genome. N‐RNA provides the template for replication and transcription by the viral polymerase and is a promising target for viral inhibition. Elucidation of mechanisms regulating this process has been severely hampered by the inability to controllably assemble NCs. Here, we demonstrate self‐organization of N into NC‐like particles in vitro upon addition of RNA, providing a simple and versatile tool for investigating assembly. Real‐time NMR and fluorescence spectroscopy reveals biphasic assembly kinetics. Remarkably, assembly depends strongly on the RNA‐sequence, with the genomic 5′ end and poly‐Adenine sequences assembling efficiently, while sequences such as poly‐Uracil are incompetent for NC formation. This observation has important consequences for understanding the assembly process.

Paramyxoviruses are non‐segmented negative‐strand RNA viruses that express their own machinery for transcription and replication.1, 2, 3 Paramyxoviridae comprise a large number of human pathogens, including measles (MeV), but also the emergent Nipah and Hendra viruses, that present greater than 50 % fatality in humans and for which no successful treatment exists. Their genomes are packaged into large helical assemblies, termed nucleocapsids (NC), consisting of a multitude of nucleoprotein (N) copies that bind the entire sequence of the genome.4, 5, 6 The N‐RNA complex provides the template for replication and transcription by the viral polymerase complex composed of the phospho‐ (P) and large (L) proteins.7 Elucidation of the mechanisms that regulate transcription and replication holds great promise for the development of targeted drugs against paramyxoviral diseases.

MeV N comprises a folded RNA‐binding domain (N_CORE), as well as an intrinsically disordered C‐terminal domain (N_TAIL; Figure 1 a; Supporting Information, Figure S1).8, 9 A recent cryo‐electron microscopy (EM) structure10 revealed that assembly of NCs is mediated by the NTD_ARM and CTD_ARM sub‐domains, at the N‐ and C‐termini of N_CORE, that interact with neighboring N molecules in the NC (Figure 1 b), a mode of stabilization also observed in related viruses.11, 12, 13, 14

Figure 1.

a) Domain organization of MeV N showing the position of the N‐ and C‐terminal arms (NTD_ARM and CTD_ARM), N_CORE and N_TAIL. b) Structure of the folded domain of measles virus N in complex with RNA obtained from cryo‐electron microscopy on helical NCs (PDB: 4UFT). N_TAIL is not shown in this Figure.

The MeV P protein is tetrameric, also containing folded and unfolded domains,9, 15, 16, 17, 18 and interacts with N_CORE through a 300 amino acid, disordered N‐terminal domain, apparently to chaperone RNA‐free N prior to NC assembly on the RNA genome,19 in the so‐called N⁰P complex. Crystal structures of the N⁰P complex of paramyxoviruses20, 21 have been determined by engineering deletion mutants of N_CORE devoid of both NTD_ARM and CTD_ARM sub‐domains (as well as N_TAIL), to avoid assembly into NC particles. These structures revealed how the N‐terminus of P binds to N_CORE in a helix–kink–helix conformation. Comparison with the MeV NC10 revealed that the binding sites of the NTD_ARM and CTD_ARM in the NCs overlap with binding sites of two α‐helices of P in the N⁰P complex (Figure 1 b), providing a rational explanation of the chaperone activity of P.

Despite this insight into the structural basis of both N⁰P and assembled NCs, further understanding of the molecular mechanisms of NC assembly has been hindered by the inability to isolate intact N⁰P and subsequently assemble supramolecular NC particles in vitro. Indeed, assembled NCs from paramyxoviruses could only be purified by over‐expressing N in recombinant systems,4 for example in E. coli, where N spontaneously assembles on cellular RNA.

In this study, we engineer a soluble heterodimeric N⁰P complex of MeV including the two sub‐domains NTD_ARM and CTD_ARM, and show that the N⁰P complex can be triggered to initiate assembly of nucleocapsid‐like particles (NCLPs) simply upon addition of RNA. We use nuclear magnetic resonance (NMR), fluorescence spectroscopy, and EM to observe the kinetics of self‐assembly of NCLPs and to investigate the molecular basis of NC formation. Remarkably, assembly of MeV NCLPs exhibits strong RNA‐sequence dependence. While the 5′ genomic and poly‐Adenine sequences form very long NCLPs, certain sequences, such as poly‐Uracil, are incompetent for assembly.

An N⁰P complex of MeV that is capable of forming NCLPs in vitro was engineered using a fusion construct of MeV P peptide and N, including a TEV cleavage site separating N and P. The minimal peptide required to form the complex was identified using NMR spectroscopy as P_1‐50 (Figure S2), resulting in a high‐yielding, soluble, and stable N⁰P complex. This region of P resembles peptides co‐crystallized with the N_CORE of Nipah virus and MeV (with NTD_ARM, CTD_ARM and N_TAIL deleted).20, 21 RNA‐free heterodimeric N⁰P complexes of full length N (P₅₀N₅₂₅) and N_TAIL‐deleted N (P₅₀N₄₀₅) were purified using Ni‐affinity chromatography followed by TEV cleavage and gel filtration (Figure S3). MeV NCs obey the “rule of six”,1, 2, 22 with each N interacting with six nucleotides of the genome. Remarkably, upon addition of P₅₀N₄₀₅ to six‐nucleotide RNA strands (corresponding to the 5′ end of the viral genome, 5′‐RNA₆ ACCAGA), N spontaneously assembles into NCLPs whose negative‐staining EM resembles NCs purified directly by over‐expression of N in E. coli (Figure 2), demonstrating that continuous RNA is not a prerequisite for formation of NCLPs.

Figure 2.

Comparison of a) NCLPs assembled in vitro upon addition of 5′‐RNA₆ to P₅₀N₄₀₅ with b) NCLPs obtained from expression of N in E. coli and cleavage of N_TAIL by trypsin digestion post‐purification.

Disordered and flexible regions of P₅₀N₄₀₅ could be observed using solution NMR, corresponding to almost the entire CTD_ARM (residues 380–405 of N), the C‐terminal end of the P₅₀ peptide (residues 44–50 of P), and a few residues belonging to the NTD_ARM (residues 26–29; Figure S4). NMR thus can be used to follow NCLP assembly in real time. By adding a molar excess of RNA to P₅₀N₄₀₅ or P₅₀N₅₂₅, and observing the evolution of NMR signals using SOFAST HMQC,23 a time‐resolution of approximately four minutes can be achieved. Resonances of P₅₀ appeared as the peptide detaches from the surface of N_CORE upon addition of RNA, while flexible resonances of NTD_ARM and CTD_ARM disappear over time as the NCLPs assemble (Figures 3, S5, S6).

Figure 3.

a) Cartoon illustrating the NMR signals that appear or disappear upon assembly of NCLPs. In the N⁰P complex, both NTD_ARM and CTD_ARM of N are flexible, giving rise to observable backbone resonances in ¹H‐¹⁵ N spectra. P₅₀ is bound to N so that peaks are too weak to be observed. As N assembles, NTD_ARM and CTD_ARM stabilize the NCLPs so that these peaks disappear, while resonances from P₅₀ grow as more N proteins associate. b) Part of the ¹H‐¹⁵  n spectrum of P₅₀N₄₀₅ (red), P₅₀ (green), and P₅₀N₄₀₅ after incubation with 5′‐RNA₆₀ for 24 hours at 25°C (blue). Assignments are shown for the P₅₀N₄₀₅ and P₅₀ resonances. Subscript T refers to residues associated with cleavage or affinity tags. c) Assembly of NCLPs from P₅₀N₄₀₅ with 5′‐RNA₆ was initiated from 209 μm P₅₀N₄₀₅, adding 5′‐RNA₆ to 471 μm. Circles: intensities measured in SOFAST HMQC spectra and summed over appearing P₅₀ (left) or disappearing N₄₀₅ (right) resonances (P₅₀: 4, 5, 6, 22, 25, 38, N₄₀₅: 28, 385, 386, 387, 388, 389, 404, 405), red lines: simultaneous fit to double‐exponential with common assembly rates.

The assembly kinetics of P₅₀N₄₀₅ mixed with 5′‐RNA₆ were investigated (Figure 3 c). Time traces of N and P were simultaneously fitted using a double exponential function (statistical significance p<0.0001 compared to a single‐exponential fit), giving apparent rates 2.0×10⁻³ s⁻¹ and 3.6×10⁻⁴ s⁻¹ for 5′‐RNA₆ (Figure 3 c). Under similar conditions, assembly on 5′‐RNA₆₀ (corresponding to the first 60 nucleo‐tides of the 5′ end of the viral genome) was found to be slightly slower (Figures 3 b and S6 b). EM was used to confirm assembly of NCLPs by adding 5′‐RNA₆ to P₅₀N₄₀₅ at time 0 and visualizing this sample using EM at increasing time after mixing, showing NCLPs of increasing length (Figure S7).

The absolute assembly rates were determined by measuring time‐resolved fluorescence anisotropy of 10‐mer RNA with a MeV genome 5′ sequence, labelled at the 3′ terminus with a fluorescein derivative (5′‐RNA₁₀‐FAM; a longer RNA was used to allow the fluorescein to exit the RNA binding groove and to reduce the potential impact on assembly). The dye attached to free RNA undergoes fast rotation, exhibiting low fluorescence anisotropy, while upon binding to N, NCLP formation and subsequent elongation, the rotational freedom is significantly hindered. Addition of increasing concentrations of P₅₀N₄₀₅ again showed biphasic kinetics (Figure 4) with assembly rates of (2.2±0.2)×10³ and (3.37±0.04)×10²  m ⁻¹ s⁻¹. NMR and EM confirmed that the dye did not interfere with NCLP formation (Figure S8).

Figure 4.

Kinetics of NCLP assembly. a) Parallel and perpendicular polarized fluorescence light was acquired upon addition of 500 nm RNA₁₀‐FAM to P₅₀N₄₀₅ (top to bottom: 53, 42, 23, 7.5, 3.8 μm). Fluorescence anisotropy (r) was calculated for every time point. All of the curves were fit with a global fit (black lines) using two exponential rates, and imposing linearity between rates and protein concentration (see the Supporting Information). b) Fluorescence spectra at the end of kinetics (dotted lines: perpendicular‐polarized; solid line: parallel‐polarized fluorescence light) illustrate increased r and light scattering at higher protein concentration (black: 7.5 μm; grey: 53 μm P₅₀N₄₀₅). c) Kinetic rates of NCLP formation with P₅₀N₄₀₅ and RNA₁₀‐FAM corresponding to (a).

Both fluorescence and NMR data suggest that an initial reaction dependent on RNA binding is well‐separated from the second, slower component, possibly corresponding to assembly of N‐coated RNA strands into regular NCLPs. N is thought to bind the RNA immediately as it emerges from the viral polymerase,3, 24 so that RNA₆ may be physiologically relevant. We note that the observed rates of NCLP formation are compatible with the measured elongation speed of the viral polymerase in vivo.25

We have previously demonstrated that N_TAIL exfiltrates from the inside to the outside of NCLPs through the interstitial space between the N_CORE domains where it can interact with the polymerase complex.8, 26, 27 The role of N_TAIL in assembly was investigated here by measuring NMR, fluorescence, and EM after mixing P₅₀N₅₂₅ and RNA. In addition to backbone resonances that are visible for P₅₀N₄₀₅, P₅₀N₅₂₅ exhibit resonances from the entire N_TAIL domain, demonstrating that these residues remain flexible in full‐length N (Figure S9). The formation of NCLPs from 5′‐RNA₆ and P₅₀N₅₂₅ showed that N_TAIL does not interfere with particle assembly in vitro, although the NCLPs are shorter than those produced with P₅₀N₄₀₅ (Figure S10). Notably, a third, significantly slower kinetic rate (80±2 m ⁻¹ s⁻¹) is required to describe the assembly of P₅₀N₅₂₅ compared to P₅₀N₄₀₅, suggesting an additional step in the assembly process involving N_TAIL.

Different RNA sequences were tested for their ability to facilitate NCLP assembly in vitro from P₅₀N₄₀₅. Homopolymers comprising purines or pyrimidines alone, UUUUUU (polyU‐RNA₆) and AAAAAA (polyA‐RNA₆), were compared, as well as genomic 5′‐RNA₆. Remarkably, the efficiency of assembly is strongly sequence‐dependent (Figures 5 and S11). No assembly is observed for the polyU‐RNA₆ samples, while regular NCLPs are obtained for 5′‐RNA₆ and polyA‐RNA₆ with lengths reaching 1–2 μm. This differential ability to form NCLPs was verified using NMR and fluorescence. No change in fluorescence anisotropy (Figure 5) or displacement of P₅₀ peptide over a period of 24 hours at room temperature (Figure S12) were observed upon addition of polyU‐RNA₆. RNA‐sequence‐dependence is surprising considering the apparent necessity of N to encapsidate the entire genome, irrespective of sequence, suggesting the role of an initial specific nucleation site for successful encapsidation. In this respect, we note that the 5′ sequences of both the genome and anti‐genome of paramyxovirinae show strong conservation of ACCA in the first four nucleotides, followed by positions 5 and 6 occupied mainly by A/G. Indeed, the 5′ sequence of the MeV genome leads to formation of NCs with high efficiency. No assembly is observed on DNA (Figure S13). This specificity may be related to the observed N‐RNA interaction in RSV and rabies that is mediated by the 2′ ribose OH group with the carbonyls of N.12, 14

Figure 5.

Dependence of NCLP assembly on RNA sequence. (a–c) P₅₀N₄₀₅ was incubated with different RNAs for 24 hours at room temperature and visualized by negative‐stain electron microscopy. Assembly of NCLPs is observed for a) 5′‐RNA₆, b) polyA‐RNA₆, while no assembly is observed for c) polyU‐RNA₆ (See Figure S11 for additional representative EMs.) d–f) Fluorescence anisotropy during 2000 seconds after addition of 500 nm, d) 5′‐RNA₁₀‐FAM, e) polyA₁₀‐FAM, and f) polyU₁₀‐FAM to 23 μm P₅₀N₄₀₅.

In all of the structures of paramyxoviral NC particles determined so far, the RNA bases were modelled as generic base‐types (uracil), to account for signal averaging over different nucleotides from random cellular RNA. We are therefore unable to determine the molecular mechanisms by which N⁰P alone can discriminate between different RNA sequences. In the NC structure determined from cryo‐EM,10 three of the six bases associated with each N stack sequentially and face the surface of the protein, while the following three stack and point into the solvent. The only significant interactions between base moieties and the protein surface occur through a stacking of Y260, and the first of the three stacked‐in bases (position 2), with base‐specific hydrogen bonding identified between Q202 and the uracil modelled in position 1. Both amino acids are conserved in paramyxo‐virinae. An atomic‐resolution structural description of protein–RNA interactions is necessary to fully understand the molecular basis of the observed sequence discrimination.

Based on the rational design of N⁰P chaperone mimics, we were able to demonstrate self‐assembly of highly regular NCLPs of MeV in vitro, uniquely from protein and RNA. In addition to providing a quantitative determination of assembly rates using real‐time NMR and fluorescence spectroscopy, revealing biphasic assembly kinetics, we have demonstrated that assembly occurs both in the presence and absence of the 125 aa disordered N_TAIL domain. Importantly, we also identify a strong, and unexpected, dependence of NCLP assembly on the RNA sequence. While the genomic 5′ end and polyA sequences assemble with high efficiency, certain RNA sequences are incompetent for NC formation under our experimental conditions. We note that the ability to controllably assemble stable, protein‐based superstructures, whose regularly spaced intrinsically disordered domains can be decorated with sequence‐modifiable functionalities, may find diverse biotechnological applications both in vitro and in vivo. Finally, this fluorescence‐anisotropy‐based MeV NC assembly assay provides a powerful tool for the development of small molecule inhibitors of this essential step in viral replication.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

S. Milles, M. R. Jensen, G. Communie, D. Maurin, G. Schoehn, R. W. H. Ruigrok, M. Blackledge, Angew. Chem. Int. Ed. 2016, 55, 9356.