Structural insights into the Nipah virus nucleocapsid assembly

Abstract

Nipah virus (NiV), a lethal zoonotic henipavirus, poses a major public health threat. Its nucleoprotein (N) safeguards the genome by forming a nucleocapsid, which, together with the RNA-dependent RNA polymerase (L) and phosphoprotein (P), constitutes the ribonucleoprotein complex. Here, we report a 2.96 Å cryo-EM structure of the NiV nucleocapsid co-expressed with L and P. The structure, resolved under C1 symmetry without imposing helical or other symmetry restraints, reveals irregular helical packing and axis bending that likely reflect the physiological nucleocapsid. Detailed analysis defines N–N and N–RNA interfaces, including critical salt bridges and hydrogen bonds, and uncovers a previously unrecognized loop-mediated contact bridging adjacent turns. Functional validation by site-directed mutagenesis and minigenome assays establishes their potential role of these interactions in nucleocapsid stability. Collectively, these integrated structural and functional insights refine our understanding of NiV nucleocapsid assembly and highlight structural vulnerabilities that could be targeted for antiviral development.

Cryo-EM structure of the Nipah virus nucleocapsid reveals flexible helical packing, key N–RNA and N–N interfaces, and loop-mediated inter-turn contacts important for viral RNA synthesis.

Introduction

The Nipah virus (NiV) and Hendra virus belong to the Henipavirus genus within the Paramyxoviridae family. NiV was first identified during outbreaks in Malaysia and Singapore in 1998 and 1999 and has since caused periodic epidemics in India, Philippines, and Bangladesh¹. As an emerging zoonotic RNA virus, NiV infects a range of animals, with fruit bats serving as its natural reservoir and pigs as important intermediate hosts. In humans, infection typically begins with fever and respiratory illness but can rapidly progress to encephalitis, convulsions, and coma². Transmission is often zoonotic, for example, through consumption of contaminated palm sap, but human-to-human spread has also been documented^3,4. Over the past 20 years, ~750 human cases have been reported, with more than 400 fatalities⁵. Due to its high lethality and epidemic potential, NiV is classified as a WHO priority pathogen, alongside Zika and Ebola viruses. Despite ongoing research, the mechanisms underlying NiV pathogenesis remain poorly understood, and no vaccines or approved antiviral therapies are currently available⁶.

NiV is a non-segmented, single-stranded, negative-sense RNA virus (nsNSV) with six genes encoding nine proteins. Six are structural: the nucleoprotein (N), which encapsidates viral RNA to form the nucleocapsid^7,8; the matrix protein (M), which surrounds the nucleocapsid and mediates viral particle formation⁹; the phosphoprotein (P) and the large RNA-dependent RNA polymerase (L), which are responsible for RNA transcription and replication and form the ribonucleoprotein (RNP) complex with the nucleocapsid⁷; and the fusion protein (F) and attachment glycoprotein (G), which mediate host cell receptor binding and viral entry^10–13. The replication cycle of NiV follows the canonical nsNSV pathway of entry, genome replication, transcription, morphogenesis, and budding¹⁴. The non-structural proteins V, W, and C facilitate immune evasion¹⁵ and promote budding¹⁶.

The NiV genome, like that of other paramyxoviruses, follows the “rule of six,” requiring genome lengths that are multiple of six for efficient replication¹⁷. Early in infection, abundant translation of viral proteins supports genome replication¹⁸, with the N protein playing a central role by encapsidating RNA into a helical nucleocapsid that protects the genome during synthesis. Structurally, the N protein comprises four domains: the N-terminal arm (NT-arm), N-terminal Ncore (NT-Ncore), C-terminal Ncore (CT-Ncore), and C-terminal arm (CT-arm) (Fig. 1A). As the most abundant and highly immunogenic viral protein, its structure, assembly, and function are critical for antiviral development, serology, and diagnostics¹⁹. Although N proteins structures have been determined from other Mononegavirales viruses^7,20–25, their overall sequence identities with NiV are relatively low ( ~ 20–32%), highlighting the need for NiV-specific structural studies to inform antiviral development. Recently, the structure of the NiV nucleocapsid assembly was resolved using cryo-electron microscopy (cryo-EM)²⁶, shedding light on its architecture.

Fig. 1. Overall representation of NiV nucleocapsid assembly.

A Schematic domain composition of the NiV N protein. The NT-arm, NT-Ncore, CT-Ncore, and CT-arm are colored in yellow, turquoise blue, pink, and frog green, respectively. B Cryo-EM map and model of the NiV nucleocapsid assembly. Domains are colored as described in panel A. C Surface representation of side and top projections of the pseudo-helical assembly. The indicated helical parameters (p, pitch; Δz, rise; Δφ, twist) are mean values derived from the 13 successive rigid-body transforms between adjacent N protomers (n → n + 1) in the C1 atomic model, with pitch computed from the mean rise and twist. Inner and outer diameters are also indicated. D Topology of the N protein. E Cartoon representation of the N protein monomer bound to ssRNA, fitted in the EM map. RNA is shown as sticks and balls.

In this study, we co-expressed L, P and N proteins in insect cells and determined a 2.96 Å cryo-EM structure of the NiV nucleocapsid complex under C1 symmetry, without imposing helical restraints. The reconstruction yielded an atomic model of a 14-N protomer RNA-bound helix, with irregular packing and axis bending, suggesting that the nucleocapsid may adopt more flexible conformations in vivo in the presence of L and P proteins than in assemblies formed by N alone. At this resolution, we mapped previously undefined N–N and N–RNA interactions, identifying key salt bridges, hydrogen bonds, and a loop-mediated interface bridging neighboring turns. Mutagenesis and minigenome assays showed that disrupting these elements severely impairs polymerase activity, confirming their potential role in nucleocapsid assembly and function. Together, these findings refine our understanding of NiV nucleocapsid architecture and highlight structural features that may represent therapeutic vulnerabilities.

Results

Overall structure of the NiV nucleocapsid

Full-length NiV N and P genes, as well as the L and P genes, were individually subcloned into the pFastBac Dual vector. Recombinant bacmids containing either N and P or L and P were generated and used for co-expression via the Bac-to-Bac system. The resulting protein complexes were purified and biochemical characterized. Size-exclusion chromatography and SDS-PAGE confirmed that the L, P, and N proteins have the expected molecular weights of ~250, 100, and 58 kDa, respectively (Supplementary Fig. 1A). Negative-stain EM verified the homogeneity of the N complex (Supplementary Fig. 1B), and its structural integrity was further assessed by cryo-EM. Samples from peaks A2 and A9 obtained from size-exclusion chromatography were analyzed. The A9 sample exhibited results consistent with our previous L–P–only sample²⁷, lacking any 2D classes corresponding to helical nucleocapsid particles, indicating that N likely exists as a monomer bound to the L–P complex within this peak.

Notably, cryo-EM revealed particles from peak A2 likely corresponding to the L–P–N RNP complex (Supplementary Fig. 1C), but subsequent data processing did not resolve this assembly, probably due to flexible interactions. The cryo-EM structure of the NiV N protein bound to native Trichoplusia ni RNA was determined at 2.96 Å resolution using C1 symmetry without applying helical reconstruction parameters (Fig. 1B, Supplementary Figs. 2 and 3, Table 1). The N protein forms a left-handed pseudo helical nucleocapsid (Fig. 1C), and the cryo-EM map allows to clearly accommodate 14 RNA-bound protomers, with distinct domains: NT-arm (residues 3–31, α1–η1), NT-Ncore, (residues 32–260, β1–α11), CT-Ncore (residues 261–369, α12–α15), and CT-arm (residues 370–400, α16) (Fig. 1D, E); the disordered C-terminal region (residues 401–532), including the Nα MoRE domain, was unresolved. Each N protomer binds seven nucleotides, with five nucleotides engaging a single protomer and one nucleotide at each end interacting with adjacent protomers (Fig. 2A). The helix exhibits a ~ 58 Å pitch, ~4.6 Å rise, ~28° twist, and inner/outer diameters of ~60 Å and ~205 Å, respectively (Fig. 1C). Due to potential helical irregularity, these parameters are approximate and may vary by several degrees or angstroms, as derived from de novo modeling of a full helical turn.

Table 1.

Cryo-EM data collection, refinement, and validation statistics

	NiV nucleocapsid complex (EMD-49803) (PDB 9NUF)
Data collection and processing
Magnification	130,000
Voltage (kV)	300
Electron exposure (e–/Å2)	50.5
Defocus range (μm)	−1.0 ~ −2.0
Pixel size (Å)	0.664
Symmetry imposed	C1
Initial particle images (no.)	1,326,637
Final particle images (no.)	138,389
Map resolution (Å)	2.96
FSC threshold	0.143
Map resolution range (Å)	2.4–6.4
Refinement
Initial model used (PDB code)	N/A
Model resolution (Å)	2.98
FSC threshold	0.5
Map sharpening B factor (Å2)	−60.0
Model composition
Non-hydrogen atoms	44996
Protein residues	5572
Nucleotide	84
B factors (Å2)
Protein	103.65
Nucleotide	81.15
R.m.s. deviations
Bond lengths (Å)	0.004
Bond angles (°)	0.881
Validation
MolProbity score	1.27
Clashscore	3.02
Poor rotamers (%)	0.00
Ramachandran plot
Favored (%)	96.93
Allowed (%)	3.07
Disallowed (%)	0.00

Fig. 2. Interactions mediating RNA encapsidation.

A The detailed interactions between N protein and RNA. The left panel shows an overall view of two protomers interacting with ssRNA. One protomer is represented as a cartoon, and the other as surface representation. The middle panel is a zoom-in view of the RNA binding cleft, highlighting essential residues involved in RNA binding. The right panel is a schematic representation of the RNA chain and the interacting residues. B The representation of surface electrostatic potential. The panel shows the protomers bound to ssRNA with the zoom-in views of the RNA binding cleft region. Residues Y258 and K178 involved in RNA chain flipping are indicated, along with a side view of the RNA-binding channel, showing size details.

During cryo-EM data processing, we used two parallel strategies. First, three box sizes, 480, 720, and 960 pixels, were tested with C1 symmetry. The 720-pixel dataset produced the best balance of resolution and coverage (Supplementary Figs. 2 and 4). The 480-pixel box gave a slightly better resolution and failed to capture a full helical turn, while the 960-pixel box resolved four full turns but had poorer local resolution and revealed bending of the helical axis, indicating irregular assembly (Supplementary Fig. 4). In the second approach, helical parameters were extracted for refinement, but this yielded only low-resolution (~ 8 Å) maps, likely due to the irregularity and curvature of the helix. Interestingly, unlike previous studies^22,26, only monodirectional helical structures were observed, with no clam-shaped particles detected (Supplementary Fig. 2).

Interactions mediating RNA encapsidation

The overall RNA conformation along the helix reveals a combination of triplets rotated inwards and outwards of the RNA-binding cleft (Fig. 2B)²⁰. Each N protomer binds seven nucleotides, of which five engage a single protomer, while one nucleotide at each end is shared with adjacent protomers (Fig. 2A). The key residues, including R352, A265, K178, R192, R193, and Y258, contact the phosphate backbone. In particular, K178 and Y258 stabilize phosphate groups that facilitate RNA chain rotation, and these residues are highly conserved across the Paramyxoviridae family. Notably, even when the sequences are not perfectly aligned, other members of the same order possess amino acids with similar properties that occupy analogous positions and likely serve similar functions as K178 and Y258.

Additional residues, including N349, S344, Q199, and Q200, interact with the RNA bases, while K196 engages both the base and sugar, potentially contributing to sequence-specific recognition²⁸. N349 and S344 are located within a topologically conserved loop, whereas K196, Q199, and Q200, from helix 8, are sequence-conserved across the Mononegavirales order (Supplementary Fig. 5). The RNA channel measures ~9.3 Å in height and ~16 Å in width. R352 interacts with the same nucleotide as Q199 from protomer n-1, while Q199 binds to the same nucleotide as R352 from protomer n + 1. Another residue, Q319, previously reported as important²⁶, lies near the RNA but likely forms only weak hydrogen bonds, as it is more than 3.5 Å from the nearest nucleotide. Although Y354 was also reported in a previous study²⁶ as being near the RNA, it is not positioned close enough in our structure to form direct contacts. Similarly, potential hydrophobic interactions involving P324, M345, and L348 are not observed in our structure due to the side chains being too far apart (Fig. 2A). On the other hand, Y258 forms a π-π hydrophobic interaction with an RNA base, which may facilitate RNA chain flip, potentially in coordination with Y354 (Fig. 2A). Comparison with the monomeric, RNA-free structure of the NiV N protein (PDB: 4CO6)²⁹ reveals several conformational changes upon RNA binding. Specifically, helix α15 rotates outward by approximately 31°, and the loop between α7 and α8 shifts away from the RNA-binding cleft, allowing additional interactions to form (Supplementary Fig. 6). In addition, compared with the previously published structure (PDB: 7NT5), the current structure resolves a pseudo-helical assembly with different helical parameters. This results in shifts of key regions involved in protein–protein interactions. Helices α1, α5, and α16 are displaced by up to ~5 Å over half of a helical turn when the two structures are aligned based on the RNA chain (Supplementary Fig. 7).

Interactions between N protomers

The buried surface area between adjacent N protomers is 3235 Ų, reflecting the extensive contacts required for stable oligomerization. The interface is reinforced by four salt bridges (R272–D290, K295–D3, R343–E356, and E398–K365), 14 hydrogen bonds (Fig. 3), and over 200 non-bonded interactions (Supplementary Fig. 8). Central to this interface is the NT-arm (α1) of one protomer, which inserts into a hydrophobic groove formed by helices α12 and α13 of the neighboring protomer (Fig. 3). This arm-groove interaction generates a hydrophobic core involving F11, F267, F268, and Y301, which acts like a “molecular lock” to stabilize the oligomer and rigidify the CT-Ncore²⁰.

Fig. 3. Protomer-protomer interactions.

The interactions between two protomers are displayed. Left: overall view of two RNA-bound protomers, Chain K (blue) and Chain L (olive). A key hydrophobic core, which stabilizes the oligomeric state, is indicated by a transparent magenta oval, formed between the NT-arm helix 1 of one protomer (L) and the CT-Ncore helices 12 and 13 of another protomer (K). Right: zoom-in views of the interaction regions, highlighting important residues involved in hydrogen bonds and salt bridges, which are labeled and depicted as sticks.

In the RNA-free NiV N monomer (PDB: 4CO6), this hydrophobic groove is instead occupied by a segment of the phosphoprotein P, which competitively blocks NT-arm insertion. This mutually exclusive binding mechanism prevents premature oligomerization by maintaining N in a monomeric, assembly-incompetent state²⁹. Thus, the switch between P binding and NT-arm insertion provides a structural basis for regulating the transition from RNA-free N to the assembled nucleocapsid.

Comparisons within the Mononegavirales order

We compared N protein structures across four Mononegavirales families. Sequence identity ranges from 25% to 90% (Supplementary Fig. 5)^{7,21–26,30,31}, with the highest similarity in the Paramyxoviridae family (up to 90%). In contrast, the Rhabdoviridae family shows the lowest sequence and structural similarity to the other three families. Despite this divergence, Paramyxoviridae, Pneumoviridae, and Filoviridae display strong topological conservation, particularly around the RNA-binding cleft (Fig. 4). The most pronounced deviations occur in peripheral regions of the NT-Ncore. In Pneumoviridae, helix α5 rotates ~45° counterclockwise, while in Filoviridae the same helix rotates ~80° (Fig. 4A). By comparison, deviations within Paramyxoviridae are minimal. Variation is also evident in the CT-Ncore, particularly in helices α15, and the CT-arm (α16). In Filoviridae, these helices are shifted, whereas in Pneumoviridae, helix α15 appears extended and more ordered (Fig. 4B). The RNA-binding groove is the most conserved region across families. Key RNA-contacting residues, including K178, R192, R193, and Y258, are highly conserved. An exception is Q199, which in Pneumoviridae is displaced due to a sharp turn in helix α8, moving both R198 and the Q199-equivalent residue away from the RNA backbone (Fig. 4C). Nevertheless, the topological equivalents of K178 and Y258 are universally maintained, underscoring their critical role in stabilizing the RNA chain flip (Fig. 4C).

Fig. 4. Structural comparisons within the Mononegavirales order.

Superposition of NiV N structure with 10 N protein structures from three families within the Mononegavirales order: Paramyxoviridae family - SeV (PDB: 6M7D), MeV (PDB: 4UFT), HPIV3 (PDB: 7EV8), HPIV5 (PDB: 4XJN), MuV (PDB: 7EWQ), NDV (PDB: 6JC3), HEV (PDB: 8C4H); Filoviridae family - Ebola (PDB: 6C54); Pneumoviridae family - RSV (PDB: 8IUO), and HMPV (PDB: 8PDS). A Structural differences in the positioning of helix 5 of NT-Ncore. B Differences within the CT-Ncore and CT-arm regions. C RNA binding cleft, with important conserved residue positions indicated based on the NiV protomer model from this study. Key residues, including K178 and Y258, are crucial for RNA chain flipping. Additionally, Q199, R192, and R193 are conserved within the family and topologically conserved across the order (indicated by a gray circle). An exception is observed in the Pneumoviridae family, where helix 8 is bent, and the corresponding R198 is distant from the RNA.

Functional validation of various interfaces

In light of our 2.96 Å cryo-EM structure of the NiV nucleocapsid, we mapped the intermolecular interactions essential for viral assembly and replication. The structure resolves N–N protomer interfaces, including four prominent salt bridges (D3–K295, R272–D290, R343–E356, and K365–E398), and N–RNA contacts primarily mediated by residues K178, S344, and R352 (Fig. 5A). Additionally, maps generated with 720- and 960-pixel box sizes revealed extra density between protomers in adjacent helical turns. At lower contour thresholds, density connecting the α5–β2 loop (residues 116–119) and the α5–α6 loop (residues 140–142) suggested a previously unrecognized inter-turn interface, likely stabilized through hydrophobic or electrostatic complementarity (Fig. 5B). Moreover, residues 116-119 are also unresolved in a previously published monomeric NiV N crystal structure (PDB: 4CO6)²⁹. Another interesting observation is that a recently published structure of the RSV nucleocapsid shows that inter-turn contacts are mediated by the CT-arm, suggesting that packing mechanisms may differ among members of the Mononegavirales³².

Fig. 5. Functional validation of N-N, N-RNA and two-turn interfaces.

A Overall view of two RNA-bound protomers with key residues mediating N–N (black) and N–RNA (red) interactions. B Low-pass filtered map (at 6 Å) showing density for potential inter-layer contacts. Loops involved in two-turn interactions are highlighted in red. C–E NanoLuc minigenome assays of NiV N mutants versus wild-type (WT). Data are shown as mean ± SD (n = 4). NC: negative control. Statistical significance was assessed by two-tailed unpaired t-test (*P ≤  0.05, ***P ≤  0.001, ****P  ≤  0.0001; ns, not significant). Source data are provided as Supplementary Data 1. F Focused multiple sequence alignment of Orthoparamyxoviridae N proteins, centered on the inter-turn loops. Sequences were obtained from UniProtKB: NiV, Nipah virus (Q9IK92); HeV, Hendra virus (O89339); CedV, Cedar virus (J7H328); GhV, Ghana virus (I0E088); LayV, Langya virus (A0AAX3C958); MojV, Mojiang virus (W8SIW1); MeV, Measles virus (P26029); HPIV3, Human parainfluenza virus type 3 (P06159); and SeV, Sendai virus (O57286).

To test the functional relevance of these interfaces, we performed systematic alanine mutagenesis (Supplementary Table 1) combined with quantitative minigenome assays. Western blotting confirmed comparable expression levels for all N mutants (Supplementary Fig. 9). Three key results emerged. First, disruption of the R343–E356 salt bridge markedly impaired polymerase activity (Fig. 5C), consistent with our EM results and underscoring its essential role in oligomerization, whereas substitutions at D3–K295, R272–D290, and K365–E398 had minimal impact, suggesting redundancy or context-specific roles. Second, mutations at S344 and R352 dramatically reduced transcription and replication (Fig. 5D), validating their central role in RNA binding. By contrast, K178A unexpectedly enhanced polymerase activity, hinting at a potential regulatory function rather than a strictly structural one. Third, multi-site alanine substitutions within inter-turn loops severely impaired activity: a triple mutant in the 116–119 loop (R116A/R117A/D119A, with G118 unchanged) and a double mutant in the 140–142 loop (K140A/K142A, with G141 unchanged) nearly abolished RNA synthesis (Fig. 5E), supporting their role as potential stabilizers of the helical assembly. Although residues in the 116–119 and 140–142 loops are not uniformly conserved across Mononegavirales, an Orthoparamyxoviridae-focused alignment reveals clear clade-restricted conservation (Supplementary Fig. 5 and Fig. 5F). Specifically, Arg117 is highly conserved within Orthoparamyxoviridae, Gly118 is conserved across most orthoparamyxoviruses, and Arg116 is conserved in henipaviruses, whereas Lys140/Lys142 positions are less strictly conserved but often retain basic residues (Lys/Arg) in henipaviruses (Fig. 5F). These patterns suggest that this region may contribute to nucleocapsid assembly through both residue-specific interactions and local positive charge.

To assess whether the observed functional effects were associated with altered nucleocapsid assembly, we examined the purified mutants by negative-stain EM under matched conditions (Supplementary Fig. 10). Most mutants formed nucleocapsid-like assemblies comparable to wild type, whereas E356A failed to adopt a regular helical conformation, indicating a mutation-specific defect in N assembly under our experimental conditions.

Together, these functional assays establish a refined model of NiV nucleocapsid assembly: canonical salt bridges and hydrogen bonds form the core of N–N and N–RNA interactions, while dynamic loop-mediated contacts provide additional possible stabilization across helical turns, collectively ensuring nucleocapsid integrity.

Discussion

At 2.96 Å resolution, we resolved 14 N protomers forming a pseudo left-handed helix with a ~ 58 Å pitch, ~4.6 Å rise, and ~28° twist, comprising 13 monomers per turn. All helices were monodirectional, and we did not observe the clam-shaped junctions reported previously^26,30,31. Both low-salt ( < 150 mM)³³ and high-salt (500 mM)²⁶ conditions have been used to assemble nucleocapsids with clam-shaped junctions. Our purification conditions closely match those previously reported²⁶, including high ionic strength (500 mM NaCl), making it unlikely that buffer composition alone accounts for the absence of clam-shaped junctions in our NiV N assemblies. Moreover, available evidence argues against expression host as the primary determinant of clam-junction formation. Recombinant HRSV N purified from insect cells forms double-headed nucleocapsids with paramyxovirus-like “spiral clams”³², and Sendai virus N forms similar junctions when expressed in either bacterial or mammalian cells³⁰. Taken together, these observations suggest that the presence of L–P, even if not resolved, more likely modulates the assembly landscape.

Despite relatively low sequence identity, the N protein of Mononegavirales, excluding the more distantly related Rhabdoviridae, exhibits a high degree of topological similarity. The most noticeable deviations occur around helix 5 and the helix pair 15-16 (Fig. 4A, B). Helix 5 and its surrounding loops (residues 106-148) likely contribute to stabilizing intra-layer/turn interactions. Structural fluctuations in this region may account for variations in helical parameters and drive the dynamic “breathing” of the nucleocapsid helix, leading to topological heterogeneity. This suggests that assembly does not adhere to strictly fixed parameters, such as 13.4 subunits per helical turn, but instead can accommodate 12-14 subunits^20,26,31, depending on environmental conditions and replication stage. Such variability could be masked during cryo-EM analysis when helical symmetry is imposed. Notably, helix 5 has also been identified as a key epitope in serological and immunological studies³⁴. Helices 15 and 16 undergo conformational rearrangements upon RNA binding, with the extent of movement differing across virus families and binding modes²⁴. Collectively, all these highlight the importance of probing nucleocapsid helical conformations under C1 symmetry, without applying predefined helical parameters.

Understanding NiV RNP function will ultimately require resolving the architecture of the complete L–P–N assembly. In our preparation, L and P co-purified with N (Supplementary Fig. 1A), and a minor fraction of particles appeared larger during cryo-EM screening (Supplementary Fig. 1C). However, density attributable to L or P was not observed in the final reconstruction, which therefore represents the helical N–RNA nucleocapsid alone. The lack of resolvable L–P features is consistent with sub-stoichiometric, flexible, and/or transient engagement of the polymerase cofactor with the nucleocapsid under our conditions, limiting structural conclusions about an intact L–P–N complex from this dataset. Accordingly, discussion of putative L–P–N contacts below is based on prior studies rather than direct visualization in our work. Previous structural work on the NiV L–P complex²⁷ has shown that P engages N at two distinct sites. The first interaction, captured in the crystal structure of RNA-free N bound to a 50–amino acid N-terminal fragment of P²⁹, involves the P NTD binding the CT-Ncore of N. This region overlaps with the NT-arm binding site from a neighboring N protomer, thereby blocking nucleocapsid oligomerization. The second interaction is mediated by the C-terminal XD domain of P and the intrinsically disordered Nα MoRE domain of N³⁵, which was not resolved in our reconstruction due to its flexibility.

Based on these established interactions, we built a composite model of the NiV RNP containing L, P, and N (Supplementary Fig. 11) by integrating four available structures: the L–P complex (PDB: 9CGI)²⁷, the Ntail–PXD complex (PDB: 7PNO)³⁵, the nucleocapsid assembly from this study (PDB: 9NUF), and the RNA-free N–P NTD complex (PDB: 4CO6)²⁹. Together, these structures provide a working framework for how the viral replication machinery may coordinate within the RNP.

In this model, the L–P complex functions as a dynamic scaffold recruited to the 3′ end of the RNA during genome replication. The N-terminal region of P engages multiple N monomers, while its XD domain binds the disordered C-terminal tails of N within the nucleocapsid. This coordinated engagement positions L–P relative to RNA-free N monomers and the existing nucleocapsid. We propose that RNP assembly begins when the L–P complex binds a flanking N monomer, displacing the NT-arm interaction and potentially exposing the RNA 3′ end to initiate replication. As transcription or replication proceeds, nascent RNA may be released adjacent to N monomers tethered by P, promoting displacement of the P N-terminus and triggering encapsidation by N. This sequential mechanism, in which L–P both guides and coordinates N loading, echoes findings from cryo-electron tomography of Ebola virus, where the L–P complex was observed simultaneously engaging the nucleocapsid at two distinct sites³⁶.

Overall, our findings provide structural and mechanistic insights into how the NiV nucleocapsid achieves both stability and flexibility to support viral replication. The irregular helical packing and axis bending observed in our C1 reconstruction suggest that structural plasticity may be an intrinsic property of the nucleocapsid, potentially facilitating dynamic transitions during genome transcription and replication. The identification of loop-mediated inter-turn contacts suggests a previously unrecognized layer of regulation, highlighting that nucleocapsid stability may be shaped not only by canonical N–N and N–RNA interactions but also by flexible interfaces that bridge helical turns. Together, these findings refine the current model of henipavirus nucleocapsid assembly and expose structural vulnerabilities within key protomer-protomer, protomer-RNA, and inter-turn interfaces, offering potential targets for antiviral strategies.

Methods

Gene cloning and protein expression

Codon-optimized gene sequences encoding Nipah virus (NiV) N (UniProtKB: Q9IK92), P (UniProtKB: Q9IK91), and L proteins (UniProtKB: Q997F0) were synthesized for optimal expression in insect cells. The NiV N gene as well as N mutant constructs, engineered to include an N-terminal hemagglutinin (HA) epitope tag, was cloned into the pFastBac Dual vector (Addgene plasmid #135584) under the control of the polyhedrin promoter. Concurrently, the NiV P gene, fused to a C-terminal 6×His tag, was inserted into the same vector downstream of the p10 promoter. For co-expression of the NiV L and P proteins, the NiV L gene was tagged at the N-terminus with a Twin-Strep tag and cloned downstream of the polyhedrin promoter in the pFastBac Dual vector, while the NiV P gene, also tagged with a C-terminal 6×His tag, was positioned downstream of the p10 promoter in the same construct. Recombinant bacmids encoding either NiV N and P or NiV L and P were generated following the Bac-to-Bac baculovirus expression protocol (Invitrogen). These bacmids were transfected into Spodoptera frugiperda (Sf9) cells to produce recombinant baculoviruses. Recombinant baculovirus stocks encoding NiV N–P and NiV L–P were amplified, mixed at a 1:2 (N–P:L–P) ratio, and used to infect 1 L cultures of Trichoplusia ni (Tni) cells at a multiplicity of infection (MOI) of 1; cells were maintained at 27 °C and harvested 72 h post-infection for protein expression and purification.

Protein purification

Infected Trichoplusia ni cells were harvested by centrifugation at 3470 × g for 30 minutes at 4 °C and resuspended in lysis buffer containing 50 mM Tris-HCl (pH 8.0), 500 mM NaCl, 10% (v/v) glycerol, 1 mM tris(2-carboxyethyl)phosphine (TCEP), 0.01% (v/v) Tween-20, 4 mM MgCl₂, and supplemented with Xpert Protease Inhibitor Cocktail (GenDEPOT). The cell suspension was lysed by sonication, and the lysate was clarified by ultracentrifugation at 100,650 × g for 2 hours at 4 °C. The supernatant was filtered through a 0.22-μm membrane and incubated with Strep-Tactin XT resin (Iba) for 30 minutes at 4 °C with gentle agitation to capture Twin-Strep-tagged proteins. The resin was washed four times with a buffer composed of 50 mM Tris-HCl (pH 8.0), 500 mM NaCl, 5% (v/v) glycerol, 1 mM TCEP, and 4 mM MgCl₂, and bound proteins were eluted with buffer containing 50 mM Tris-HCl (pH 8.0), 500 mM NaCl, 5% (v/v) glycerol, 1 mM TCEP, 4 mM MgCl₂, and 50 mM biotin. The eluted protein complexes were further purified via size-exclusion chromatography using a Superose 6 Increase 10/300 GL column (GE Healthcare) equilibrated with buffer containing 50 mM Tris-HCl (pH 8.0), 500 mM NaCl, 5% (v/v) glycerol, 1 mM TCEP, and 4 mM MgCl₂. Fractions corresponding to the major elution peak were analyzed by SDS-PAGE to assess protein composition and purity. Fractions containing co-eluting L, P, and N were pooled, flash-frozen in liquid nitrogen, and stored at −80 °C for subsequent analysis.

Negative staining EM

Negative staining electron microscopy (EM) was performed to assess the homogeneity of the NiV nucleocapsid (Supplementary Fig. 1B). Four microliters of the NiV nucleocapsid fraction ( ~ 0.2 μM) from the size-exclusion column were applied to freshly glow-discharged, 200-mesh carbon-coated copper grids (EM Sciences) and incubated for 1 minute, followed by blotting to remove excess liquid. The grids were washed three times with water and stained with a 0.75% uranyl formate solution for 30 seconds. Excess liquid was gently blotted from the side of the grids using filter paper, and the grids were air-dried before imaging. Imaging was conducted using a Biotwin Tecnai Spirit 120 kV electron microscope (ThermoFisher Scientific) equipped with a Gatan 4 K × 4 K CCD camera, at a defocus of −2 μm and a nominal magnification of 30,000× or 98,000×.

Cryo-EM sample preparation and data collection

Four microliters of 2 µM fresh NiV protein sample from the harvested gel filtration peak were applied to a freshly glow-discharged Quantifoil R1.2/1.3 300-mesh copper grid (EM Sciences) and blotted for 4 seconds at 4 °C under 100% chamber humidity, then plunge-frozen in liquid ethane using a Vitrobot Mark IV (FEI). Images were collected using the K3 Summit detector (Gatan) in super-resolution mode with binning 2 and CDS, along with a Gatan BioContinuum GIF energy filter (slit width of 20 eV) at the Hormel Institute, University of Minnesota³⁷. Data collection was performed using the EPU 2.8.0.1256REL software package (Thermo Fisher Scientific) with a pixel size of 0.664 Å (nominal magnification of 130,000x) and a nominal defocus value between −1.0 and −2.0 μm. Each image consists of 40 dose-framed fractions and was recorded with a total dose of 50.5 e⁻/Ų. Cryo-EM data collection statistics are summarized in Table 1.

Cryo-EM data processing

Cryo-EM data were processed using cryoSPARC v4.6.2³⁸, with procedures outlined in Supplementary Fig. 2. Briefly, dose-fractionated movies were subjected to Patch motion correction using MotionCor2³⁹ and Patch CTF estimation using CTFFIND-4.1.13⁴⁰. Images with defocus values between −0.5 and −3.0 μm and CTF fit resolutions better than 5 Å were retained for further processing, followed by optical grouping using Exposure Group Utilities. Particles were then picked using the Blob and Template pickers in cryoSPARC v4.6.2 and subjected to the Remove Duplicate Particles Tool. A total of 1,326,637 particles were extracted to a box size of 480 pixels and Fourier-cropped to a 360-pixel box size from the 6059 movies kept. Junk particles were removed through two rounds of 2D classifications. A total of 401,660 particles from the good 2D classes were used for Ab-initio reconstruction of four maps. Heterogeneous refinement produced 231,079 particles from the good class, which were then subjected to further non-uniform and CTF refinements to generate a final map of 2.86 Å resolution. Meanwhile, the good particles from 2D classifications were re-extracted using a box size of 720 or 960 pixels, Fourier-cropped to a 540-pixel box size, and then used for Ab-initio reconstruction of three maps, followed by further heterogeneous refinements and additional non-uniform and CTF refinements. The final maps with resolutions of 2.96 Å and 3.18 Å, respectively, were obtained. The maps were then sharpened by DeepEMhancer⁴¹. Map resolution was determined by gold-standard Fourier shell correlation (FSC) at 0.143 between the two half-maps. Local resolution variation was estimated from the two half-maps in cryoSPARC v4.6.2.

Model building and refinement

The maps with three different box sizes reveal the nucleocapsid spirals in various lengths. However, the map with a 720-pixel box size shows more protomers with clearer density and, therefore, was used for model building. The NiV nucleocapsid model was initially automatically built using ModelAngelo⁴² and also predicted by trRosetta⁴³, then manually rebuilt in COOT-0.8.9⁴⁴ until the final reliable models were obtained. The cryo-EM densities were sufficient to define the region of residues 3-400 of NiV N protein. The nucleotide sequence is unknown but was modeled using poly-U based on the density. A total of 14 protomers and 84-nt RNA were built into the map. Real-space refinements in Phenix-1.16⁴⁵ were performed to obtain the final models. In the real-space refinement, minimization global, local grid search, and adp were performed with the secondary structure, rotamer, and Ramachandran restraints applied throughout the entire refinement. The stereochemistry of all structural models was evaluated using MolProbity⁴⁶ in Phenix. Model and map statistics are summarized in Table 1. Figures were generated using UCSF Chimera X v1.8⁴⁷.

Minigenome luciferase reporter assay

293 T cells were seeded in 24-well plates and maintained in Dulbecco’s modified Eagle’s medium (DMEM, Gibco, 11965092) supplemented with 10% fetal bovine serum (FBS, Gibco, A5256701) and 1× penicillin-streptomycin (Gibco, 10378016) at 37 °C with 5% CO₂. A NiV replication minigenome encoding NanoLuc, containing a T7 promoter and the 5′- and 3′-terminal untranslated regions of the NiV genome, was constructed^48,49. Minigenome system plasmids were co-transfected into 293 T cells using Lipofectamine 3000 (Thermo Fisher Scientific, L3000001) at the following amounts per well: 50 ng pCAGGS-T7, 100 ng pCAGGS-minigenome, 450 ng pCAGGS-L, 50 ng pCAGGS-P, 50 ng pCAGGS-N or N mutants, and 1 ng pGL4.54-luc plasmid carrying firefly luciferase, with pCAGGS empty vector added to equalize total DNA across conditions. N mutations were introduced via site-directed mutagenesis, with primers listed in Supplementary Table 1. At 48 h post-transfection, cells were lysed with passive lysis buffer (Promega, E1941) for 15 min at room temperature, and 40 µL of lysate were used to measure luciferase activity using the Nano-Glo Dual-Luciferase Reporter Assay (Promega, N1610) according to the manufacturer’s instructions; reporter activity was expressed as the NanoLuc/Firefly luciferase ratio. All experiments were performed in quadruplicate wells.

Western blot analysis

293 T cells were lysed in buffer containing 25 mM HEPES (pH 7.8), 500 mM NaCl, 5% glycerol, 4 mM MgCl₂, 1 mM TCEP, and 0.01% Tween 20, supplemented with Xpert Protease Inhibitor Cocktail (GenDEPOT, P3100), and centrifuged at 13,000 × g for 10 min at 4 °C. Supernatants were mixed with 2× sample loading buffer, boiled at 100 °C for 5 min, and resolved by SDS-PAGE alongside a color-coded prestained protein marker (Cell Signaling Technology, 74124S). Proteins were transferred to PVDF membranes (Millipore, IPVH00010), blocked with 5% skim milk in TBST for 1 h, and incubated overnight at 4 °C with primary antibodies, rabbit anti-HA (1:2000, Cell Signaling Technology, C29F4) and mouse anti-GAPDH antibody (1:3000, Cell Signaling Technology, D4C6R). Detection was performed with HRP-conjugated secondary antibodies, goat anti-rabbit (7074S) and rabbit anti-mouse (58802S), each at 1:5000, and blots were imaged using a ChemiDoc MP system (Bio-Rad).

Alignment

Sequence alignments of N proteins were conducted using the online tools Multalin⁵⁰ and ESPript 3.0⁵¹. Conserved residues appear as white on a red background, highly similar residues as red framed in blue, with secondary structure annotations above and consensus residues below. The protein-protein and protein-RNA interfaces in the nucleocapsid complex were analyzed using PDBsum services⁵² from the European Bioinformatics Institute.

Statistics and reproducibility

Statistical analysis was performed using student’s t-test for comparisons between two groups and one-way analysis of variance (ANOVA) for comparisons among multiple groups. All data were analyzed and generated using GraphPad Prism 10.1 software. Graphs display the means of the experiments, with error bars representing one standard deviation. Significance is indicated as follows: ns, not significant; *P ≤ 0.05, **P ≤ 0.001, ***P ≤ 0.0001.

Author contributions

B.L. and G.Y. conceived the project. G.Y. and D.W. prepared the proteins. G.Y. performed functional assays. B.L. and G.Y. conducted cryo-EM data collection, cryo-EM image processing, atomic model building and refinement. D.K., G.Y., and B.L. wrote the manuscript.

Peer review

Peer review information

Communications Biology thanks Benoît Arragain, Marc Jamin and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editors: Dr. Nilanjan Banerjee. A peer review file is available.

Funding

This work was supported by funding awarded to B.L. from the National Institutes of Health (R35GM162600) and by the Hormel Institute, University of Minnesota.

Data availability

The atomic coordinate of the NiV nucleocapsid complex has been deposited in PDB with accession number 9NUF. The corresponding cryo-EM density map has been deposited in the Electron Microscopy Data Bank with the accession number EMD-49803. Source data are provided with this paper as Supplementary Data 1. Uncropped full-length western blot images corresponding to all western blotting results presented in this paper are provided in the Supplementary Information file. All other data supporting the findings of this study are available from the corresponding author upon reasonable request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-026-10379-6.