Measles virus is a human virus that, although the vaccine is very efficient, still kills about 100,000 people per year (Laksono et al., 2018; WHO, 2019). Measles virus is a single-stranded negative-sense RNA virus belonging to the Paramyxoviridae family. The viral RNA, 15,894 nucleotides long, is bound by thousands of copies of the viral nucleoprotein, forming the helical nucleocapsid (NC). The viral polymerase complex, comprising the viral L protein (L, for Large protein) and phosphoprotein, uses the NC as the template for transcription and replication. The only known structure of L of the viruses of Mononegavirales order, to which also paramyxoviruses belong, is the one from vesicular stomatitis virus (VSV) (Rahmeh et al., 2010; Liang et al., 2015; Qiu et al., 2016), and several reviews have used this structure as a model for the function and structure of measles virus L (Sourimant and Plemper, 2016; Fearns and Plemper, 2017). Figure 1 shows the genes of the measles virus, coding for eight proteins within six genes: nucleoprotein (N, blue), phosphoprotein (P, red), matrix (M), the fusion protein (F), and haemagglutinin (H) (the spike proteins), and the polymerase (L, yellow). The gene of P also encodes the mRNAs for the C and V proteins. These proteins have been found in many other Paramyxoviruses and speculated to be involved in regulation of host cell immune response through different pathways. In addition, it should be mentioned that V of measles virus share the first 231 amino acids with P and thus can also maintain N chaperoning.

This review describes recent results on the phosphoprotein (P) and nucleoprotein (N) of measles virus. The classical paradigm that a protein must have a defined and stable structure and that the functional mechanisms of the protein in the cell can be derived from this structure, is not applicable to P and N. This is because most of P is disordered, as is the C-terminal tail of N (Tompa, 2002; Uversky, 2002; Karlin et al., 2003; Longhi et al., 2003; Bourhis et al., 2004; Gerard et al., 2009; Habchi et al., 2014). As shown in Figure 1, 120 residues of N and 265 + 79 residues of P are intrinsically disordered in their functional state. Together this makes up to 1392 RNA bases, almost 9% of the full genome.

P and N have multi-domain structures with alternating ordered and disordered regions (Figure 1). P starts with a long disordered tail (304 amino acids) followed by a tetrameric domain of four long, parallel helices between residues 304 and 377 (Figure 1). Tetramer domains of similar length are found in Sendai, Nipah (Tarbouriech et al., 2000; Communie et al., 2013a; Bruhn et al., 2014). P of mumps also forms a tetramer of the same length but with two anti-parallel dimers (Cox et al., 2013; Pickar et al., 2015). For Sendai, Nipah and measles viruses, the polymerase binds to the tetramerization domain (Bowman et al., 1999; Bloyet et al., 2019; Bruhn et al., 2019). At the C-terminus of P there is another folded domain (XD) which is linked to the tetramerization domain via the unfolded P_loop (residues 377–457).

N comprises two distinct parts: Folded N_core and N_tail, an intrinsically disordered domain, containing a linear motif with α-helical propensity (residues 484–502; Figures 1, 2A). In solution this sequence folds into several helices of different length that exchange with each other and with a purely unfolded form on timescales faster than microseconds (Jensen et al., 2008, 2011; Communie et al., 2013b). This motif binds to the C-terminal domain of P, the X-domain (XD) (Figure 1; Johansson et al., 2003; Longhi et al., 2003; Kingston et al., 2004a,b; Bourhis et al., 2005; Houben et al., 2007; Schneider et al., 2015). Orthomyxoviruses like influenza virus have several separate RNA molecules, each bound to the viral polymerase. Because the nucleocapsids form circular complexes, with the 5′ and 3′ ends bound in close proximity on the polymerase (Ruigrok et al., 2011; Reich et al., 2014), the polymerase stays attached to the same gene while making new mRNAs. In paramyxoviruses such as measles, the polymerase first transcribes the leader RNA and then the mRNAs of N, P, M, F, H, and L. After each mRNA is terminated, L has to move from the end of the gene to the start of the next gene. The binding of N_tail and XD of P may assist in this process such that L remains in contact with the template (Brunel et al., 2014; Bloyet et al., 2016; Cox et al., 2017; Thakkar et al., 2018). The K_d of binding between N_tail and XD in paramyxoviruses is rather weak, between 3 and 8 μM (Schneider et al., 2015; Bloyet et al., 2016). If the affinity is artificially decreased further through mutations in the interacting protein domains, L has been shown to leave the template before transcribing all the genes (Bloyet et al., 2016).

N_core is flanked by N and C-arms which are disordered in the RNA-free form of the protein, and become more structured upon binding to the neighboring N_core subunits in nucleocapsids. During NC formation the N_arm and C_arm bind to the neighboring N protomers (Figure 2), with part of the N_arm forming a helix (residues 2–14) bound to the surface of the neighbor i-1 and the C_arm folding onto neighbor i+1. N can bind RNA, either viral RNA (vRNA) during infection, or cellular RNA during expression in cellular expression systems, making recombinant nucleocapsid-like structures. In order to avoid non-specific binding to cellular RNA, N is chaperoned by the N-terminal residues of P forming an N⁰P complex (Figures 1, 2A; Masters and Banerjee, 1988; Peluso and Moyer, 1988; Curran et al., 1993, 1995; Mavrakis et al., 2003). When the N_arm and C_arm from the N⁰P complex are removed, the complex can be crystallized and the structure can be determined through X-ray diffraction. In this way, the N⁰P structures of VSV, Nipah, measles, PIV5, HMPV, Marburg and Ebola viruses were determined (Leyrat et al., 2011; Yabukarski et al., 2014; Guryanov et al., 2015; Kirchdoerfer et al., 2015; Leung et al., 2015; Renner et al., 2016; Zhu et al., 2017; Aggarwal et al., 2018).

Up to now, there are no atomic resolution structures of the viral NCs, but NC-like particles (NCLPs) made from measles, Marburg, rabies and VSV viruses, for example, have similar structure as the viral nucleocapsids (Spehner et al., 1991; Fooks et al., 1993; Iseni et al., 1998; Mavrakis et al., 2002; Bhella et al., 2004; Schoehn et al., 2004). Some of the NCLPs assemble into rings rather than helical particles and the first two atomic resolution structures were determined from rings of VSV (10 protomers) and rabies (11 protomers per ring) with the RNA inside the ring (Albertini et al., 2006; Green et al., 2006). The N of RSV (respiratory syncytial virus) makes a ring with 10 to 11 protomers with the RNA on the outside of the ring (Tawar et al., 2009), the ring of PIV5 with 13 protomers with the RNA on the outside the ring (Alayyoubi et al., 2015) and HMPV with 10 protomers as with RSV (Renner et al., 2016). Most of the atomic-resolution structures were determined by crystallization and X-ray diffraction. Other NCLPs form helices of which the structures can be determined by Cryo-EM and single particle-based helical analysis. The first such structure was of measles virus NCLPs comprising 12.3 protomers (Gutsche et al., 2015) and Ebola virus with 24.4 protomers per turn of the helix and the RNA on the outside of the helix (Sugita et al., 2018), as with the rings of RSV and PIV5. Other structures of Ebola virus NCLPs with lower resolution show the same fold of the protein although the structure of the RNA was not visible (Wan et al., 2017; Su et al., 2018). A structure of the NCLP of mumps virus was also determined at a lower resolution (Cox et al., 2014; Severin et al., 2016). The comparison of N⁰P and the NCLPs structures reveals that, upon RNA binding the relative orientation of the lower and the upped lobes change by ~20^o (Green et al., 2006; Alayyoubi et al., 2015; Gutsche et al., 2015; Sugita et al., 2018) and became more closed than the N in the N⁰P from VSV, measles, PIV5 and Ebola viruses (Leyrat et al., 2011; Guryanov et al., 2015; Kirchdoerfer et al., 2015; Leung et al., 2015; Renner et al., 2016; Aggarwal et al., 2018) suggesting that in the N⁰P structures the RNA binding groove is primed for binding the RNA.

The structures of nucleoproteins of measles, PIV5, HMPV and Ebola viruses were determined in the absence of N_tail. The NCs of measles virus are very flexible when N_tail is present, but form tighter helices when N_tail is removed, and these NCs can be used for higher resolution image analysis (Figure 3; Heggeness et al., 1980; Bhella et al., 2002, 2004; Schoehn et al., 2004; Renner et al., 2016). The structure of the NCLP of measles virus shows that the C_arm of N is close to the center of the helix (Desfosses et al., 2011; Gutsche et al., 2015) indicating that N_tail starts inside the helix. Solution state NMR showed that the first 50 residues of the tail are conformationally restricted before the chain escapes to the outside of the helix (Figure 3; Jensen et al., 2008, 2011). The disordered chain could interfere with stacking of turns, resulting in loosening of the helix, facilitating the functions of the transcription and replication from L. The helical linear motif is part of the C-terminal end of N_tail although N has been shown to function also when this helical motif was moved to different sites within N (Cox et al., 2017).

All existing paramyxoviral NCLPs structures were determined from recombinant protein bound to RNA from the cells in which N was expressed. This means that the RNA is probably of random sequence, and the structures have only been able to reveal binding of the phosphates and 2' oxygen of the ribose but not interactions involving sequence-specific bases. Milles and coworkers constructed a fusion N⁰P complex where the P peptide was fused on the N_arm with a TEV cleavage site (Milles et al., 2016). The purified complex was then cleaved to make the P peptide (residues 1–50) and either the full N (residues 1–525) or N up to the N_arm (residues 1–405), with no RNA bound (Figures 2B, 4). During incubation of this complex with synthesized RNA, the RNA binds, the P peptide comes off and N associates into NCLPs (Figure 2B). The kinetics of the assembly process can be studied by nuclear magnetic resonance (NMR) spectroscopy because the signal of P becomes narrow (peaks of P appear in the spectrum) when the assembly process is initiated and P peptide dissociates from N. The signal for the residues of N_arm and C_arm become broader (the peaks disappear) as they bind to the neighboring N particles in the NC. In addition, using RNA labeled with a fluorescent probe, the kinetics of the assembly can be followed by measuring the changes in fluorescence anisotropy. Free RNA tumbles rapidly, however once it is bound to N and then assembles into NCs, the anisotropy increases. The resulting assembly can then be visualized with negative staining EM and NC growth can be followed during several hours (Milles et al., 2016; Figure 4). The nucleoprotein from Sendai and measles viruses bind 6 RNA bases per protomer (Calain and Roux, 1993; Pelet et al., 1996) and micrometer long NCLPs could be made with RNA molecules of only 6 bases in length; either with the 5′ viral genomic RNA (HO-ACCAGA-OH) or a polyA (HO-AAAAAA-OH) sequence and the resulting NCLPs structures were determined with Cryo-EM and image analysis (Desfosses et al., 2019). The structures show the RNA with a gap between the hexamers because the RNA molecules do not contain a 5′ phosphate (Figure 5A). Whereas, the previous structure (Gutsche et al., 2015) mainly showed only N residues binding the sugar and the phosphate of the RNA (residues K180, T183, R194, R195, and R354) the new structures also show residues binding the RNA bases; R195, Q202, Y260, E263, N351 (Figure 5B; Desfosses et al., 2019).

The structures of VSV and rabies virus NCs show that each N protomer can bind 9 bases and RSV can bind 7 bases. The N protomer of measles, PIV5 and Ebola viruses bind 6 bases. In these nucleocapsids the RNA form 2 short A-helices connected by a conformational flip of the backbone RNA; all structures show that 3 bases turn to the nucleoprotein (inward) and the others turn outward (Figure 6). If the RNA followed a long A-helix rather than 2 short helices, then there would be no room for the nucleoprotein and, probably, the new RNA strand made by L during transcription could easily form a low energy double helix RNA structure. These viruses do not code for a helicase; so the nucleoprotein functions as a helicase (Iseni et al., 2000). For VSV/rabies and RSV, NCLP structures show that for each protomer, the last 3′ base stacks with the first 5′ base stabilizing the structure of the nucleocapsids between the two protomers (Albertini et al., 2006; Green et al., 2006; Tawar et al., 2009; Ruigrok et al., 2011). The structure of the NCLP of PIV5 suggests that nucleotides 4–6 turn inward and nucleotides 1–3 outward (3′-in-in-in-out-out-out-5′) and would mean that bases 6 and 1 do not stack (Figure 6A). However, the chemical modification of nucleotide bases in the Sendai viral nucleocapsids (Iseni et al., 2002) showed that bases 1 and 6 are both more open for modification than the other bases. The structure of the NCLP of measles virus (Gutsche et al., 2015) suggested that nucleotides turn 3′-out-in-in-in-out-out-5′. Nucleotides 6 and 1 stack just like the NCLP of rabies virus, VSV, and RSV (Ruigrok et al., 2011; Figure 6B). The structure of the synthesized NCLPs of measles virus with 6 nucleotides (Figure 6C) shows how the 6 nucleotides bind on the protomer and these structures show the same orientation of nucleotides as suggested for the previous structure, with stacking for nucleotides 6 and 1.

With the help of NCs formed in vitro, we can identify both the binding of the 5′ end of the RNA to the first N protomer of the NCs, and the binding of the 3′ end on the final protomer of the NCs (Figure 5B): The lower part of N is still bound to the RNA but the upper part of N does not cover the terminal bases and only residue Q202 in the lower edge of the groove binds the last base at the 3′ end. Q202 is important for replication initiation in PIV2 because the mutant Q202A shows increased replication (Matsumoto et al., 2017, 2018). The fact that the 3′ end of the RNA is rather “open” may help the binding of L at the 3′ end of the vRNA.

NMR was used to investigate all the 304 disordered residues of the N-terminal domain of P, P_tail, identify the structural behavior of each residue, and characterize the relative flexibility of P_tail (Jensen et al., 2011, 2014; Ozenne et al., 2012; Abyzov et al., 2016; Milles et al., 2016, 2018). Figure 7A shows the rigidity for the residues, with a thick red ribbon for the flexible regions and in thin blue ribbon for the more rigid regions, and 4 regions with a propensity for making helices. The first two helices represent the transient helices α_1/2 that bind as helices in the N⁰P structure, residues 1–37, a short region around residue 110, α₃, the STAT binding site (Devaux et al., 2007), and helix α₄, around residues 189-198 (Milles et al., 2018). N⁰P with N⁰ and the 304 residues of P shows that α-helices 1 and 2 bind on N⁰ but also helix 4 and a short hydrophobic region just before helix 4, together named δα₄ (delta-alpha-4). This region δα₄ binds residues 96–127 on N⁰ (Figure 7B). This binding site and the helix α₄ seem to be conserved in many paramyxoviral P proteins. If the “HELL” residues are changed to AAAA in a measles virus transcription/replication assay, the activity is suppressed (Milles et al., 2018). This means that the functional core of N⁰P is not only N⁰ plus helices 1 and 2 (Yabukarski et al., 2014; Guryanov et al., 2015) but also up to the δα₄ region (Figure 7B). Between δα₄ and α₃ there is an acidic sequence with many Asp/Glu residues. As N⁰ cannot be purified in the absence of α_1/2, it is complicated to measure the K_d for their binding to N⁰, but it is estimated to be in the nanomolar range as the complex can be purified on a size-exclusion column. The binding of δα₄ to N⁰ is very weak, the K_d is around 0.6 mM, and we could show by NMR that δα₄ binds and dissociates very rapidly while the N-terminal helices α_1/2 remain bound, possibly so that the acidic loop blocks the cellular RNA binding to N⁰P.

The measles virus polymerase stutters on the template and adds extra G bases for making several mRNAs and therefore different proteins (Thomas et al., 1988; Cattaneo et al., 1989; Vidal et al., 1990). The V-protein uses the same P_tail as in the P-protein and changes sequence at residues 231 (the editing site, just after δα₄) up to residue 299, to form a zinc-binding domain of the V protein (Cattaneo et al., 1989; Liston and Briedis, 1994). For P we know what the domains do: Residues 1-200 bind N⁰ and the rest of P_tail in another register forms part of V, 304-377 for the tetramerization domain and 457-507 for XD. We do not know the function of the region 378–456 (P_loop) in the transcription and replication processes of measles virus. Additionally it should be mentioned that N and P are both known to be phosphorylated, which can be important for the viral cycle regulation, in particular phosphorylation is involved in the regulation of viral gene expression. However the mechanism of such regulation was not studied yet (Lamb et al., 1976; Sugita et al., 2018).

Cox and Plemper (2015) have suggested a model for the NCs in the presence of N, P and L for the paramyxoviruses. Similarly we have used recent results concerning the structures of P, N, and the nucleocapsids of measles virus (Figure 8) to develop a picture of the functioning polymerase complex. For this we used the structure of the NC with the 6 bases and for the structure of L we used the polymerase of VSV, to indicate the size of this component when bound to the tetramerization domain. In Figure 8 we show only one N_tail and only one P_tail. P could be bound to N with the XD-N_tail complex and with α₁ in the free α₁-site on N (Figure 2) and L could act on the first bases of the 3′ end of the vRNA (Figure 8A). For the last base at the 3′ end, the 2' of the ribose and the phosphate would be free for binding L. The very long P_tail and N_tail could keep L away from the RNA. However, if δα₄ region also binds to its binding site on N, then L could come closer to the vRNA (Figure 8B). To conclude, the current model illustrates the initial genomic RNA register and binding, and suggests a role for N-P interactions in positioning L during RNA processing.

RR and MB wrote the manuscript. SM and SG contributed to reviewing relevant literature, interpretation of studies, drafting the manuscript, and preparation of figures. All authors approved the article for publication.