Features, processing states, and heterologous protein interactions in the modulation of the retroviral nucleocapsid protein function

Abstract

Retroviral nucleocapsid (NC) is central to viral replication. Nucleic acid chaperoning is a key function for NC through the action of its conserved basic amino acids and zinc-finger structures. NC manipulates genomic RNA from its packaging in the producer cell to reverse transcription into the infected host cell. This chaperone function, in conjunction with NC’s aggregating properties, is up-modulated by successive NC processing events, from the Gag precursor to the fully mature protein, resulting in the condensation of the nucleocapsid within the capsid shell. Reverse transcription also depends on NC processing, whereas this process provokes NC dissociation from double-stranded DNA, leading to a preintegration complex (PIC), competent for host chromosomal integration. In addition NC interacts with cellular proteins, some of which are involved in viral budding, and also with several viral proteins. All of these properties are reviewed here, focusing on HIV-1 as a paradigmatic reference and highlighting the plasticity of the nucleocapsid architecture.

Introduction: NC and Viral Replication—Scope of this Special Review

In this review, attention is given to retroviral nucleocapsid (NC) proteins. These remarkable nonspecific RNA/DNA binding polypeptides modulate the thermodynamic folding pathways of nucleic acids, thereby orchestrating the sophisticated RNA-DNA choreography involved in retroviral biology. Our goal is to focus on two aspects of NC in the context of both its nucleic acid chaperone, and intravirion architectural modulating functions that are critical throughout viral assembly and replication. The first aspect discusses the strong requirement for conserved biochemical and structural features, the basic character of NC and its zinc finger or zinc knuckle (ZF) domains. The second presents contrasting considerations on the chaperone status of the NC domain as member of the Gag precursor compared with processed forms, highlighting the consequences of viral protease (PR) processing in RNA-NC interactions.

Similar strategies are employed by NC proteins from different retroviruses, with the notable exception of the spumaretroviruses that maintain their NC in a precursor form throughout replication. Insights into these strategies have been remarkable from the published works over the past several years, with subtle variations in (i) NC sequences, (ii) processing and (iii) RNA/DNA chaperone properties. These insights become more obvious after a thorough comparison of the paradigmatic NC proteins studied in the human immunodeficiency virus type 1 (HIV-1), simian immunodeficiency virus (SIV), murine leukemia virus (MuLV), Rous sarcoma virus (RSV) and human T-cell leukemia virus type 1 (HTLV-1) retroviral systems. This review will largely be based on HIV-1, comparing and contrasting it with the other retroviruses where possible. Additional aspects regarding NC interactions with viral and cellular proteins, and alternative nucleic acid chaperones will be discussed, providing some perplexing aspects to be considered. Figure 1 illustrates the HIV-1 replicative cycle showing where NC is directly involved, from Gag-RNA assemblies until the completion of reverse transcription.

NC Proteins

Retroviral NC proteins were first observed as the major protein component of reverse transcription-competent ribonucleoprotein (RNP) complexes purified from virions of avian sarcoma-leukosis virus and MuLV.1 Once isolated, NC was characterized as a nucleic acid-binding protein with preference for single-stranded sequences within structured RNA domains.2 Subsequently, NC was shown to be a major driving force in genomic RNA selection, dimerization and packaging during MuLV, RSV and HIV-1 assembly.3–6 NC was also found to direct viral RNA dimerization in vitro and annealing of the primer tRNA to the genome's primer-binding site, both in vitro and in several retroviral systems.7,8 These observations led to the discovery that NC was both the major architectural component of the RNP core complex and a RNA-DNA chaperone protein necessary for bona fide proviral DNA synthesis. These capabilities are now explained by the ability of NC to accelerate the most thermodynamically stable base-paired nucleic acid structures, which provokes both nucleic acid condensation into a RNP complex (i.e., the nucleocapsid) and configuration of RNA tertiary and quaternary structures required for successful reverse transcription (reviewed by J. G. Levin and K. Musier-Forsyth in this special issue). Over the years, specific features of NC have been well characterized: the NC ZFs function to destabilize nucleic acids base pairs, and the basic amino acids aid in aggregating RNA or DNA strands. Several in-depth reviews have described work on the basic amino acid residues and ZF structures in nucleic acid chaperone function.9–12 Due to pleiotropic effects of mutations to NC in cell culture, the best methods for gaining insights into NC's nucleic acid chaperone function still rely on model in vitro systems, which employ a wide array of biochemical, molecular biology and biophysical techniques and technologies.

Structural features of NC.

The N-terminal basic domain. The nucleocapsid proteins from the Retroviridae (http://www.ictvonline.org/virusTaxonomy.asp?version=2009) have a feature in common: high densities of basic amino acids that facilitate interactions with nucleic acids. Both the Orthoretrovirinae and Spumaretrovirinae have this feature. In general, the basic amino acids, Lys and Arg are involved in aggregating and annealing of nucleic acids.13,14 This strand aggregation ability is essential for facilitating the bimolecular nucleation step of strand annealing. The nature of this effect is widely discussed in the literature and is essentially electrostatic, due to the charge complementarities between nucleic acid backbones and the positively charged residues. In HIV-1, the basic residues of the N-terminal domain form a 3₁₀ helix, which optimizes interaction with the grooves of hairpin structures.15,16 Together with the N-terminal tail, the basic residues in the linker region between the HIV-1 NC ZFs contribute to robust (high affinity) but dynamic (rapid kinetics) binding/dissociation interactions with nucleic acids, which ensures nucleic acid aggregation and RNP energy minimization.17–20 Both effects are particularly apparent during minus-strand-transfer, from the accelerated annealing of minus-strand strong-stop DNA (containing a copy of the 5′-R region) to the 3′-R-RNA region of the genome.21–23 In agreement, deletion of basic residues or elevation of the buffer ionic strength significantly decrease the affinity of the protein toward nucleic acids.24 Interestingly, a recent molecular dynamics simulation study noted that the 3₁₀ helix provides a highly basic surface that could represent a nucleic acid dimerization site,25 which may explain the observation that HIV-1 NC has multiple nucleic acid-binding sites that allow intra- and intermolecular interactions.26 Most mature retroviral NC proteins have isoelectric points in the 10–11 range except for HTLV-1 and -2, which are neutral due to a patch of acidic amino acid residues in the C-terminal half of the proteins. NC proteins typically are highly flexible27–29 except for the conserved globular zinc-binding structure discussed next.

The ZF domain. A major distinction between the Ortho- and Spuma-retrovirinae subfamilies is that only the former contains a highly conserved element, a mini-globular zinc binding domain, commonly referred to as a ZF.30,31 Divided further, the Gammaretroviruses and Epsilonretroviruses (e.g., MuLV [GenBank accession no.: J02255] and walleye dermal sarcoma virus [GenBank accession no.: L41838], respectively) contain a single ZF in their NC proteins, with all other Orthoretrovirinae containing two of these zinc binding motifs. These structures are remarkably conserved with a general amino acid sequence of: -Cys-X₂-Cys-X₄-His-X₄-Cys- (CCHC), where X are variable amino acids. Typically there are one or two aromatic amino acids within the N-terminal loop, a Gly residue just before the conserved His residue and a carbonyl-containing residue, just prior to the C-terminal Cys residue30,31 in retroviral NC ZFs. More details about NC structure are available in Y. Mely's review in this issue.

NC chaperone function.

As shown in Figure 2 for HIV-1, RSV and MuLV NC, the primary function of this mini-globular metal binding domain is to bind specifically to exposed (unpaired) nucleobases of RNA or DNA. Folding into knuckle structures exposes a hydrophobic pocket centered on aromatic and several non polar residues, forming a cavity, which allows the insertion of a nucleobase (preferentially a guanosine). In free solution, there are transient interactions between these structures in HIV-1 NC, mediated through aromatic residues.29,32 The capability of HIV-1 NC to destabilize nucleic acid duplexes relies on its preferential binding to unpaired nucleic acids (single-stranded regions of RNA or DNA) via the capture of a guanosine by a cavity formed by two aromatic residues located on each ZF.33–35 The complex is stabilized by a network of hydrogen bonds, π-cation and π-π interactions with residues forming the cavity. When bound to nucleic acids, nucleobase capture is critical for destabilization, a process that is necessary for the many nucleic acid rearrangements that take place during viral replication. The culmination of the destabilization and annealing activities constitutes NC's nucleic acid chaperone activity.10–12 By this mechanism, melting by NC, of short helical segments bordered by mismatches or bulges provides a single-stranded nucleation site for the formation of a new, double-stranded helix with a neighboring single-stranded region. In addition, NC-induced duplex destabilization accelerates various processes of strand switching and strand transfers.10 Studies on other NC proteins and precursors remarkably support this mechanism. For example, RSV NC is an even better aggregating agent and exhibits rapid binding kinetics, but is a relatively poor duplex destabilizer. On the other hand, MuLV NC aggregates and destabilizes duplexes effectively; however, due to its slow dissociation kinetics, the protein is a less effective chaperone.19 As illustrated in Figure 2, for each NC, these characteristics are subtly tuned by variations in the nucleobase binding module (i.e., the ZF) and short, semi-structured basic residue-rich regions.

Most of the early work examining the role of NC has been performed comparing mutant and wild-type HIV-1 NC (NCp7 [55 amino acid form] or NCp9 [71 or 72 amino acid form; NCp7+SP2]) or MuLV NCp10 in in vitro assays. In addition, several virus-based assays have been performed to examine the properties of NC at various points in the viral replication cycle. Kafaie and coworkers examined an extensive collection of NC mutations and NC precursor forms on genome dimer maturation,36,37 a process that relies on NC's RNA chaperone function.3,38,39 It was demonstrated in this virus-based system that basic residues in the N-terminal and the linker regions, as well as the two ZFs were necessary for efficient genome dimerization. However there were a number of the mutations that also affected Gag processing and it was shown that processing per se can affect dimer maturation.36 Thus it is difficult to determine whether the mutations themselves or Gag processing defects were responsible for the effect on RNA chaperone-driven dimer maturation.

Recent in vitro studies have started to compare and contrast the nucleic acid chaperone properties of NC proteins from several retroviral Genera including HIV-1 NCp7, MuLV NCp10, RSV NCp12 and HTLV-1 NCp15. The properties of these NC's were characterized using a number of different analyses. The effectiveness of annealing falls into the rank order of HIV-1 and RSV > MuLV >> HTLV-1, which was assessed by gel-shift assays of a mini-TAR DNA/RNA system that mimics the minus-strand transfer process.19 Additionally, HTLV-1 NCp15 was shown by FRET analysis to facilitate annealing of a complementary DNA very poorly.40 Aggregation of nucleic acids, demonstrated by sedimentation showed that the HIV-1, RSV and MuLV NC proteins were equally effective at saturating protein levels, however HTLV-1 NC was a poor aggregator, the reasons for which will become apparent below.19

Effective destabilization afforded by these proteins typically relates to their higher relative binding affinities to single-stranded versus double-stranded nucleic acids. Typically single-stranded bulge regions in nucleic acid stems are starting points for destabilization as mentioned above.41 All four NC proteins have similar destabilization free energies, all favoring binding to single-stranded nucleic acids; fluorescence anisotropy measurements of NCs from various viruses showed that HTLV-1 > RSV ≈ MuLV > HIV-1 in this regard. Thus even though HTLV-1 NC is ineffective at aggregating nucleic acids, it appears to be the best destabilizer. Rapid binding/dissociation rates also contribute to the balance between annealing and destabilization, and thus overall chaperone effectiveness. Single-molecule DNA stretching studies provide excellent insights into the kinetics of binding/dissociation as well as information regarding destabilization; this is discussed in detail in the review of M.C. Williams and I. Rouzina in this issue. HIV-1 and RSV NCs both have rapid on/off rates from single-stranded nucleic acids, MuLV NC's dissociation is much slower accounting for its reduced annealing capacity; HTLV-1 NC has the slowest dissociation rate from single-stranded DNA (ssDNA), which contributes to its poor chaperone function.19

HTLV-1 NCp15 is a particularly poor nucleic acid chaperone due, in part, to its overall neutral charge at physiological pH and its inability to aggregate nucleic acids. However if the C-terminal acidic tail is removed (ΔC29), chaperone activity can be restored to levels seen with the other retroviral NC proteins. Additionally, binding/dissociation kinetics became much more rapid. Curiously, increasing the ionic strength also improved chaperone function of the native protein, leading to the hypothesis that an ionic interaction between the N-terminal basic and C-terminal acidic domains was occurring. ΔC29 HTLV-1 NC was examined for its annealing activity, which was reduced with increasing amounts of added C-terminal peptide; when the ionic strength was increased, annealing was again restored. Additional evidence for the detrimental effects of the C-terminal tail on chaperone function came from examination of an HIV-1 NCp7/HTLV-1 C-terminal tail chimera that had significantly reduced annealing activity.42

Spumaretroviruses as an exception.

Since their discovery, the spumaretroviruses, such as the prototype foamy retrovirus (PFV) have gained the attention of retrovirologists as they retain unique properties during replication, especially at the level of the budding and entry cycle.43 Characterization of the PFV NC-like domain also revealed unique properties.44 It is logically found at the C-terminus of Gag, but is not processed as a separated protein during Gag maturation, and there are no ZFs to contribute to RNA binding or destabilization. The nucleic acid binding properties of the PFV NC-like domain have been briefly described, with non-specific nucleic acid binding being attributed to arginine-rich regions (Gly-Arg motifs, which are also found in RSV NC). These motifs were shown to be necessary for PFV RNA and Pol packaging.45 The RNA chaperone activity of this particular NC has not been assessed, but should be absolutely necessary for assistance with reverse transcription, which is similar to the HIV-1 reverse transcription process although its spatiotemporal regulation is different.46 PFV reverse transcription is indeed performed predominantly prior to budding, resulting in a DNA-containing virus like HBV.47 Moreover, the critical Gag cleavage, which separates the N-terminal moiety, required for viral assembly and egress, from the C-terminus, has been shown to occur after reverse transcription and cell entry, providing for a unique PFV capsid uncoating mechanism.48

NC in Retroviral Replication

The number of NC domain-containing Gag precursors is largely dependent on virion size, but there is considerable debate as to the actual quantity. However, the quantity of NC domains associated with the genome dimer in an immature HIV-1 virion has been estimated to be around 3,000 and approximately 2,000 processed NC molecules are associated with the RNA in the mature particle.49 The NC:RNA molar ratio is therefore not far from 1 for ∼10 nucleotides, a value that is very similar to the ratio required in vitro for optimal efficiency in all assays where NC affords effective assistance.50 This ratio also typically drives efficient coaggregation of NC with nucleic acids18,51 and one could speculate that the NC domain aggregates with its genomic RNA partner in a macromolecular crowding context, from Gag assembly, formation of the mature nucleocapsid complex, to the reverse transcription complex (RTC). This aspect is common among both RNA chaperones and nucleic-acid architectural proteins.52

NC from most retroviruses functions in both precursor and fully processed forms throughout the replication cycle. Typically, in precursor forms (i.e., Pr55^Gag for HIV-1 or Pr65^Gag for MuLV), NC is involved in assembly processes, specifically recognizing and binding genomes as RNA dimers.53,54 The precursors are also involved in placement of the tRNA primer onto genomes.38,55 With processing of Gag by PR, successively smaller precursors containing the NC domain are generated until fully processed NC proteins are liberated, as illustrated in the cases of HIV-1 and MuLV (Fig. 3). The functions and characteristics of these precursors will be discussed further below.

Assembly functions of NC as an internal domain of Gag.

The HIV-1 Gag polyprotein (Fig. 3C) acts essentially as an assembly machine performing simultaneous membrane binding, RNA packaging and cofactor recruitment (for details, see reviews from A. Rein, D. Muriaux and J. L. Darlix, M. C. Williams and I. Rouzina and J. G. Levin and K. Musier-Forsyth in this issue). The NC domain of Gag plays a direct role by selecting two copies of viral RNA from a large pool of cellular RNA molecules. This is accomplished through the specific binding of several NC domains of Gag, with the four RNA stem loops (termed Ψ) that dimerize through self-complementarity. Structures of NC in complex with two of the isolated stem loops from HIV-1 Ψ show that RNA recognition is supported by specific contacts between the basic residues and ZFs, with exposed loop nucleotides (Fig. 2). For MuLV, an elegant mechanism combines genome dimerization and packaging in which the high-affinity binding site for NC becomes exposed upon dimerization.56–58 In HIV-1, the binding of several NC domains to the Ψ RNA region is thought to initiate particle assembly by congregating Gag on the RNA dimer, which then acts as a nucleation point for aggregation of other Gag units through Gag-Gag interactions and NC/RNA binding.59–61 Loading of Gag and Gag-Pol on the RNA, which functions as a scaffold for self-assembly of the particle, presumably relies on specific and nonspecific interactions between NC domains and RNA stems and the backbone.15 Although NC is believed to be the major determinant for binding to nucleic acids during retroviral assembly, participation of a basic amino acid patch in matrix (MA) has also been implicated in the case of HIV-1,62–64 and deltaretroviruses.65 Accordingly, the DNA binding constant of recombinant GagΔp6 is 10-fold higher than either NCp15, NCp9 or NCp7, as it is stabilized by additional Gag-Gag interactions,17,66 facilitated through the capsid (CA) domain.67 Gag has not been shown to facilitate more complex nucleic acid rearrangements, other than tRNA^lys3 annealing to the primer binding site (PBS). In contrast, NCp7 is solely capable of facilitating minus-strand transfer,21,68 although in vitro, at low concentrations, Gag (with p6 deleted) can do this as well.67

From Gag processing to nucleocapsid condensation.

Maturation of the Gag/gRNA nucleoprotein complex results in condensation and establishment of a RNP architecture that is able to support bona fide reverse transcriptase (RT)-catalyzed viral DNA synthesis, with subsequent conversion to a pre-integration complex. Immediately after budding, the C-terminal region of CA, the adjacent spacer peptide (SP1) and NC appear to form a continuous assembly domain, essential for formation of the immature virion.69 In this context, the NC domain of Gag and RNA delineate the innermost layer of the immature virion (Fig. 3D). Separation of the RNA-associated NC is the first step in Gag polyprotein cleavage, which neutralizes this assembly domain, separating the CA and NC domains and allowing condensation of the RNP complex that may subsequently function as a substrate for CA condensation.69 The sequential processing of the NC polypeptide by PR then correlates with the improvement of its chaperone character, presumably by increasing the flexibility of the NC and consequently increasing its binding/dissociation dynamics with nucleic acids.17,70 Remarkably, RNA bound to Gag or the various NC intermediates strongly stimulates PR activity, ensuring complete processing of the precursors.71–73 Destabilization of folded complementary sequences, activation of intermolecular base-pairing, sliding of NC, capture of multiple strands,26,74 and neutralization of RNA backbone repulsions by the polycationic character of NC then drives aggregation/condensation of the RNP complex towards the most thermodynamically stable architecture.17,51,75 Simultaneously, RNP condensation drives gRNA dimerization36,37 and presumably configures the proper RT pre-initiation complex.69,76,77

Several recent studies have shown that an essential step of this aggregative process is the separation of the p6 moiety, as NCp15 does not aggregate with either ssDNA51,71 or RNA (Lyonnais S, unpublished result) and dissociates more slowly than NCp7 from ssDNA.17,70 In contrast, NCp9 binds and dissociates much more rapidly, with moderate cooperativity and promotes extensive RNA and ssDNA aggregation.18,20,51 Finally, NCp7 appears to fulfill the optimal functionality, i.e., high mobility, to induce both nonspecific aggregation and facilitate the diffusional search of complementary nucleic acid regions.

The impact of maintaining NC in its precursor forms, NCp9 and NCp15, has also been studied. In virions, where NC is prevented from progressing past the NCp15 precursor, genome dimerization is significantly disrupted.36 Reverse transcription processes in this mutant, from minus-strand strong-stop syntheses to the plus-strand transfer event occur at near wild-type levels. However, conversion of plus-strand transfer products to 2-LTR circles is significantly reduced, due either to inefficient strand-displacement synthesis or NCp15 steric obstruction of the viral DNA ends to nuclear ligases; there are severe defects in integration, thus the mutant is replication defective.78 In mutant virions that maintain the NCp9 form, genome dimer formation and reverse transcription are not inhibited significantly, thus some insights into NC's nucleic acid chaperone function have been obtained from these HIV-1 based studies. As indicated above, due to the pleiotropic effects of mutations to NC, it is often difficult to interpret results in virus-based systems. On the other hand, clinical studies have shown that NC-p1 and p1–p6 regions are preferred sites, outside of PR, that acquire compensatory mutations associated with HIV-1 protease inhibitor resistance,79 the reasons for which require further investigation.

Reverse transcription.

RTCs evolve from within the nucleocapsid after introduction into the newly infected cell.80 Due to its nucleic acid chaperone activity, HIV-1 NCp7 efficiently assists RT in defined critical events such as tRNA^Lys3-directed initiation, the two obligate strand transfers and central DNA flap termination.10,12 Details regarding NCp7's function as a nucleic acid chaperone in reverse transcription and the implications of this activity are reviewed in this issue by J.G. Levin and K. Musier-Forsyth.

NC processing and reverse transcription. The loss of HIV-1 infectivity, observed under suboptimal concentrations of protease inhibitors has shown a predominant effect on reverse transcription, suggesting that CA and NC must be efficiently separated by PR.81 The same effects are observed with mutations that suppress Gag cleavage sites: trans-complementation of wild-type virus with a low level of a Gag mutant that does not complete NC maturation strongly affects wild-type reverse transcription.81 Taken together, these results suggests that at a minimum, NC maturation generating a species with the proper N-terminus (i.e., NCp15) is required in order to optimize the reverse transcription process, with two obvious consequences: (i) to favor the RNA/DNA chaperone activities required for the critical reverse transcription steps and (ii) to promote the RT-driven displacement of bound NC along the template. Non-processed Gag bound to a template would most likely be difficult to displace, similarly to that already shown for bound APOBEC3G.82 In support of this displacement hypothesis, a recent study showed that precursor forms of Gag, containing NC, not separated from CA could prevent reverse transcription at a threshold concentration much lower than peptides without an attached CA domain.67 In contrast, as mentioned above, mutations that impede C-terminal NC processing seem to not substantially affect reverse transcription, however preventing cleavage of the p1–p6 site strongly inhibits single-cycle viral infectivity at or near the completion of reverse transcription.78

Reverse transcription and NC dissociation. Within purified HIV-1 RTCs and PICs, CA and NCp7 appear predominantly to dissociate from the full-length linear viral double-stranded DNA (dsDNA) extracted from infected cells, in contrast to the efficient retention of integrase (IN) and viral protein R (Vpr).83–85 Capsid uncoating has been shown to be critical for the generation of an active PIC.86 The apparent loss of NCp7 also suggests that nucleocapsid disassembly and PIC assembly are directly interconnected during the reverse transcription process.

Changes in the HIV-1 nucleocapsid architecture have not been observed during reverse transcription processes in infected cells due to the difficulty in examining such a highly dynamic process. The mechanism demonstrating the loss of NCp7 arose from in vitro experiments: during DNA synthesis, RT sliding provokes NCp7 displacement from the template, whereas re-association with dsDNA appears severely limited, most likely due to a reduced affinity of NCp7 for dsDNA compared to single-stranded nucleic acids under ionic conditions for optimal RT activity, in vitro.51,71 It is interesting to note that parallel studies indicate that MuLV NCp10 strongly binds to both ss- and dsDNA, whereas it is not displaced by, and is also an inhibitor of HIV-1 RT,71 suggesting that RT and NC are tailored to one another, to manage NC dismantling from the nucleocapsid. Moreover, strategies of NC dismantling seem to be quite variable depending upon the class of retrovirus. In the case of HIV-1, this dismantling may favor IN and Vpr sequestration on dsDNA and the recruitment of suitable cellular dsDNA binding proteins.84 Finally, NCp7 seems easier to remove than NCp9 from nucleoprotein complexes in vitro,71 suggesting that p1 removal prior to initiation of reverse transcription would favor a more efficient RTC to PIC transition.

Thus, two alternative modes have been proposed to disrupt the HIV-1 nucleocapsid architecture in the context of reverse transcription within the cell:71 (i) dismantling of NC from RTCs, restricted to the plus-strand synthesis step resulting in the generation of dsDNA or (ii) progressive displacement of NC associated with RT translocation throughout the process. The first mode is supported by the in vitro aggregation of NCp7 with the ssRNA or ssDNA products observed during minus-strand synthesis,77,87 suggesting that NCp7 dissociation requires dsDNA. In contrast, the second mode could be supported by an additional dispersal force through the action of the RT-associated RNAse H, generating small RNA fragments during minus-strand DNA synthesis, causing competitive NCp7 dissociation from RTCs.88,89 Strands displaced during the course of plus-strand termination would be the last regions to retain NCp7, especially the central DNA flap,71 while the HIV-1 LTR dsDNA ends should promote active IN assemblies.90

Premature reverse transcription. A peculiar observation was made recently by several laboratories, where mutations to HIV-1 NC affected the timing of reverse transcription. Houzet and coworkers observed that when either or both zinc binding motifs were deleted from the NC in HIV-1, reverse transcription initiation and progression shifted from a point after introduction of the RNP complex into newly infected cells, to a step prior to virion budding. This resulted in virions with elevated levels of intravirion DNA, both from unspliced and multiply spliced viral RNAs.91 The study of Thomas and coworkers examined the kinetics of reverse transcription in cells infected with either NC mutant or wild-type HIV-1. Reverse transcription proceeded as expected with wild-type virions, with the increases in viral DNA that were maximal at about 8 h postinfection, whereas maximal levels of reverse transcripts were observed in the NC_H23C and NC_H44C mutants at the beginning of the time course.92 It was discovered in these particular mutants as well, that reverse transcription occurred before particles were released from cells.12,93 In both cases reverse transcriptase was required for the observed premature reverse transcription phenotype since RT inhibitor-treatment or active site RT mutations effectively reduced the levels of intravirion DNA. The mechanism of premature reverse transcription is not known at this time, although a similar phenotype was observed in HIV-1 with the p6, late domain mutation (PTAP to LIRL) that reduces virion budding.93

Other aspects of NC.

NC binding with cellular proteins. NC has been shown to engage in interactions with different cellular proteins, in addition to the viral RT. Relationships have not been established between these interactions and effects on RNA binding/chaperone activity: it is still not clear whether NC can bind simultaneously to RNA and another protein, or if these interactions are more exclusive in the context of the nucleocapsid architecture. Three examples will be highlighted here, with the main idea being that several thousands of copies of NC are able to accommodate many interactions with only a short sequence of amino acids, emphasizing its architectural adaptability within virions.

One of the first cellular proteins shown to be a HIV-1 NC ligand was actin. It binds to both the NC domain in Gag and fully processed NCp7.94,95 Additionally the N-terminal domain of HIV-1 NC has been implicated in this binding.96 More recently, interactions between actin and the NC domains of EIAV, MuLV or HIV-1 Gag have been also shown to positively impact viral budding.97,98 Nevertheless, more studies are expected to better identify these NC-actin interactions, their effects on NC binding to RNA and how they support viral replication.

Another protein that was shown to interact with NC was a cellular restriction factor, APOBEC3G that is incorporated into HIV-1 viral infectivity factor-negative (Vif^(-)) virions and inhibits replication.99 Its recruitment involves RNA and both the N-terminal and spacer domains of NC, again in the context of Gag precursor.100,101 Its antiviral action is due to its direct inhibition of reverse transcription plus its cytidine deaminase activity on the HIV-1 minus-strand DNA during reverse transcription.102 Even though this enzyme is incorporated with a low copy number in the absence of Vif (less than 10 copies), one study showed significant inhibition of proper NCp7-chaperoned annealing of tRNA^Lys3 on the PBS site,103 whereas two other studies showed no effect on annealing;82,104 APOBEC3G strongly affected the elongation steps of reverse transcription due to the inability of RT or NCp7 to effectively displace it.82,104

Recently, HIV-1 NC was shown to interact with ALIX and, as a domain in Gag, is necessary for ALIX recruitment together with the Gag late domain, p6.105 This cellular protein strongly enhances viral budding. With its boomerang-shaped Bro1 protein domain, it is involved in providing the correct membrane curvature required for microvesicular budding that is associated with the endosomal sorting complex required for transport (ESCRT). NC's ZFs have been shown to play a critical role in this binding interaction, whereas RNA appears dispensable. Moreover, NC can bind to a large class of proteins carrying Bro1 domains with an enhancement of virion particle release.106

The implications of NC sequences interacting with cellular proteins highlighted here indicate that these interactions are both required or can be a detriment to viral replication.

NC-Vif trans-interactions. Besides NC, it has been reported that two other HIV-1 proteins, trans-activator of transcription (Tat) and Vif, have some nucleic acid chaperone activity in vitro.107,108 The implication of the activities of these auxiliary proteins in vivo is not readily clear, however both proteins have been shown to perform other critical functions, due in large part to protein-protein interactions with other cellular factors.109,110 Little information regarding Tat interactions with NC is available, however Vif may be involved in some indirect modulation of NC function, while mutations in NC may conversely impact these interactions.

Along with its effectiveness in excluding APOBEC3G, Vif has been show to stabilize the condensed nucleocapsid assembly within virions111 (this may also be a consequence of APOBEC3G exclusion) and to interact with Gag at the MA-CA, SP1-NC and NC-SP2 boundaries. In conjunction with these interactions, it has been proposed that Vif may downregulate viral PR processing at these Gag cleavage sites.112–115 In vitro studies have shown that Vif acts as a negative regulator of NC-assisted RNA dimerization and tRNA^lys3 annealing.107 In spite of this, Vif has also been shown to interact with RT116 and to bind cooperatively with the HIV-1 5′-UTR RNA,117 supporting efficient reverse transcription. However, the influence of Vif within viral particles has resulted in intense debate due to its limited amount in the virion:107,118 Vif binding to Gag, Pol and RNA should promote its incorporation, which may be useful for proper timing of both maturation and reverse transcription. Taking into account that high expression of Vif negatively affects HIV-1 replication,115 cooperative Gag assembly followed by aggregation of mature NC with RNA may be a driving force for the exclusion of Vif from viral particles.

NC-p6 cis-interactions. HIV-1 NC binding to nucleic acids has been shown to be modulated by its C-terminal p6 domain within Gag.17,51,71 Interestingly, HTLV-1 NC contains a region that is similar to the HIV-1 p6 domain that also modulates HTLV-1 NC interactions with nucleic acids as mentioned above.42

This C-terminal domain, which is generally lacking in most other retroviruses, confers an optimal coordination for HIV-1 assembly and release with its late domain, containing the PTAP motif that facilitates interactions with the cellular MVB machinery.119 The NC and p6 domains appear to cooperate in their interactions with Alix, NC being bound to Alix Bro1 domain and the p6 LYPx_nL motif to Alix V-domain.105,106

The p6 domain is also necessary for efficient GagPol incorporation120 and is required for Vpr packaging into virions,121 incorporating both elements in close proximity to the NC domain that is bound to RNA. As NC interactions that have been described also include RT76 and Vpr,122 any impact of p6 on these interactions should be considered.

Cleavage of the SP2-p6 site results in p6 separation from NC and proteolytic cleavage is greatly enhanced when bound to RNA as mentioned above. This cleavage event leads to NC's efficient separation from within the capsid of the immature virion, whereupon NC and bound RNA undergo a vigorous condensation step to form the nucleocapsid found within the mature virion. More studies are needed to better understand the transient cis-actions between NC and p6 and its mediation by RNA.

Conclusion: Regulation of NC RNA Chaperone and the Spatiotemporal Control of Retroviral Plasticity

As presented in this review, a more complete view of NC behavior is actually emerging, in the context of HIV-1 replication. Becoming clearer is the critical role of the N-terminal domain and the two zinc fingers of this small protein in manipulating the viral RNA genome, from its capture to reverse transcription. As we mentioned above (section III.C.3.), a simple, but remarkable effect underlies this critical role: a simple substitution of the His by a Cys in either of the two conserved zinc knuckles provokes premature reverse transcription. On the other hand, the HIV-1 NC domain is remarkably conserved in its amino acid sequence, as confirmed by an analysis of all the HIV-1 sequences from the HIV sequence database (http://www.hiv.lanl.gov/content/sequence/HIV/mainpage.html): half of its residues (especially within the ZFs) are more than 99% conserved, demonstrating an exceptionally high genetic pressure to greatly maintain optimal NC function.

Another point that is developed here is modulation of the RNA chaperone activity of HIV-1 NC based on its processing state (Gag<NCp15<NCp9<NCp7). In HIV-1 and related lentiviruses, the C-terminal domain, p6, as well as two spacer peptides, SP1 and SP2 that flank the NC domain (Fig. 3) are a hallmark of the Gag polyprotein. These retroviruses have developed a complex strategy for the control of NC function, resulting in the viral NC-bound RNA playing a pivotal role in the activation of the three step, C-terminal processing events that direct nucleocapsid condensation (Figs. 1 and 3). Thus considering the state of NC processing provides a better understanding of the spatiotemporal regulation of intravirion architecture: optimal RNA chaperone function does not immediately commence after nucleocapsid separation within the capsid, but becomes so after p6 removal. In close relation to this change, NC's architectural properties, linked to RNA aggregation, are also greatly influenced by processing once NC is separated from CA-SP1 (i.e., separation of nucleocapsid from capsid): the first species, NCp15, is poorly aggregative; the second, NCp9, is highly aggregative; and the final, NCp7, is still aggregative but more mobile in its interaction with RNA (see the review of M. C. Williams and I. Rouzina in this issue). Rapid mobility has been proposed to be a favorable characteristic for NCp7 to assist in reverse transcription and the subsequent formation of PICs. Finally, we propose here that regulating both the RNA chaperone and architectural properties of the NC domain is a remarkable property of HIV-1 that confers an optimal architectural plasticity within the viral core, for its transfer from the producer to the target cell.

The HIV-1 NC domain and mature NCp7 have also been shown to engage in a diversified set of protein-protein interactions, either directly or mediated through the viral RNA. A better understanding of these interactions is now needed in order to determine their influence on the nucleic acid chaperone activity of HIV-1 NC and more generally in the plasticity of the viral architecture from states that allow escape from a producing cell to states that allow efficient introduction of the nucleoprotein complex into a newly infected cell. Such a scheme implies that NC must now be considered a nucleic acid architectural protein that directly coordinates the spatiotemporal progression of the HIV-1 genome from a messenger RNA to a double-stranded nuclear viral DNA poised for integration, with the support of two HIV-1 auxiliary proteins: Vif (escape steps) and Vpr (incoming steps). Such progression, finely tuned by PR, initiates with Gag assembly and the coordinated interactions of the NC and p6 domains in the binding of critical cellular proteins engaged in viral budding or ESCRT machinery. After budding and maturation, NC then packages a nucleocapsid complex associated with RT, Vpr and IN that is competent within a newly infected cell to mediate RTC, then PIC formation, with concomitant capsid uncoating and cytoskeleton-linked transport to the nucleus.

If we now take into account that the number of NC copies within one virion is close to 2000, it is tempting to speculate that division of labor among intravirion NC sub-populations should be a convenient model that can accommodate all of the NC properties described in this review. Thus a simple question can be formulated here: do all NC domains and mature NC proteins need to be bound to RNA within a virion? If so, is this binding exclusive? Development of future HIV-1 NC studies will hopefully clarify this point.

Finally, after examining NC properties, this first glimpse highlights that the general concept of the nucleocapsid is maintained in HIV-1 and other Orthoretroviruses. Nevertheless, significant differences are indeed found in structural features, processing from the Gag precursors and efficiencies of their RNA chaperone activities of NC's from different viruses. As presented in the Figure 2, NC structural organization can be quite different, with a single ZF and two long flanking domains in the case of MuLV or two ZFs with an extended spacer in the case of RSV. For HTLV-1 NC, a unique acidic C-terminal domain is present that appears, at first glance to be similar to unprocessed HIV-1 NCp15. At the level of processing, most retroviruses contain neither the spacer peptides SP1, SP2 nor p6-like peptides. In any event, NC proteins from these different retroviruses have clearly adopted unique nucleocapsid-based strategies to manage their RNA genomes and to promote their progression from mRNA to dsDNA, when traversing from the producer cell to the newly infected cell. Taking into account that various sub-families of retrovirus are implicated in human cancers (e.g., HTLV-1 and MuLV-related XMRV), we might expect more extensive studies on these retroviruses to improve our knowledge of retroviral nucleocapsids, to better identify their unique biological properties that may contribute to the discovery of new anti-cancer pharmacological strategies.

Abbreviations

CA	=	capsid
ds	=	double-stranded
ESCRT	=	endosomal sorting complex required for transport
HIV-1	=	human immunodeficiency virus type 1
HTLV-1	=	human T-cell leukemia virus type 1
IN	=	integrase
MuLV	=	murine leukemia virus
MA	=	matrix
NC	=	nucleocapsid
PFV	=	prototype foamy retrovirus
PIC	=	preintegration complex
PR	=	viral protease
RNP	=	ribonucleoprotein
RSV	=	Rous sarcoma virus
RT	=	reverse transcriptase
RTC	=	reverse transcription complex
SIV	=	simian immunodeficiency virus
ss	=	single-stranded
Tat	=	trans-activator of transcription
Vif	=	viral infectivity factor
Vpr	=	viral protein R
ZF	=	zinc finger or zinc knuckle

Figures and Tables

Figure 1 NC's involvements in HIV-1 replicative cycle. (1) Gag proteins are first produced and migrate through the cytoplasm towards the vicinity of the cellular membrane. N-terminal MA domains of Gag bind to the lipid raft domains of the cellular membrane while the C-termini project into the cytoplasm and bind viral RNA through their NC domains. Gag binding initiates at the Ψ RNA region ensuring viral RNA dimerization, which then acts as a nucleation point to aggregate other Gag and Gag-Pol units though Gag-Gag interactions and NC/RNA binding (MA/RNA binding has also been proposed62–64). Viral RNA capture competes with the translational and/or mRNA docking apparati whereas other viral components are loaded onto Gag such as Vpr. (2) RNA scaffolding and Gag-Gag interactions mediate assembly of the particle which subsequently buds from the cell as an immature and imperfect particle. (3) Contacts between Gag-Pol precursors drive PR dimerization/autoactivation, which initiates virus maturation. (4) Sequential proteolytic events of Gag and Gag-Pol by PR and the self-assembly capabilities of the cleavage products (MA, CA, NC) reorganize the viral core, generating a mature particle. Maturation results in NC/RNA complex (i.e., the nucleocapsid) condensation at the center of the core where NC chaperone activity modulates RNA tertiary and quaternary structures. (5) After entry of the HIV-1 particle into the cytoplasm of a newly infected cell, RT catalyses viral DNA synthesis in the context of a ribonucleoproteic complex where NC proteins accelerate the strand transfers required for (−) and (+) strand synthesis. Viral dsDNA synthesis proceeds with a complete remodeling of the capsid and its contents, which are ultimately converted into a pre-integration complex, depleted of most of the NC initially bound to the viral RNA.

Figure 2 Structural features of HIV-1, RSV and MuLV nucleocapsid proteins—Ψ RNA complexes. The figure focuses on the binding of key guanosine residues (red), by a motif formed by a pocket derived from the side chains of hydrophobic residues (green). Zinc atoms bound by the ZFs are grey spheres. Location of basic amino-acids (Lys and Arg), available for electrostatic binding with RNA, are highlighted in yellow. (A) HIV-1 NC complexed with SL3 Ψ RNA (PDB ID:1A1T). A guanosine residue fits between the two adjacent ZFs which form a hydrophobic pocket around Phe16 and Trp37. The N-terminal basic tail adopts a 3₁₀ helix, which binds to the major groove of the SL3 stem.16 (B) RSV NC and µΨ RNA structure (PDB ID:2IHX), where the guanosine base fits into a hydrophobic pocket defined by residues Tyr22, Tyr30, Leu20 and Gln31. Note that the RSV protein contains two ZF domains that adopt canonical folds observed for all other structurally characterized retroviral ZF, Tyr residues are used by RSV ZF1 instead of Trp and Phe residues used by HIV-1 NC. The C-terminal ZF is unusual as it does not contain aromatic residues nor a well-defined hydrophobic pocket. This finger packs against an adenine nucleobase though hydrogen bonds and salt bridges.123 (C) Complex of MuLV NC and the mΨ RNA core encapsidation signal (PDB ID:1U6P) in which the single CCHC ZF of NC interacts with the UCUG sequence (indicated in pink, with G in red). The guanosine base fits deeply into a pocket defined by the side chain of Leu21, Ala27, Trp35 and Ala36. The first U and C pack against the side chain of Tyr28 and the second U packs against the side chains of Ala27 and Leu 21. The binding also appears to be promoted by several interactions between the basic residues and the phosphodiester backbone.56

Figure 3 Proteolytic processing of HIV-1 and MuLV Gag by PR. The different protein species that exist during the steps of processing are indicated. (A) In HIV-1 Gag the initial cleavage occurs between SP1 and the NC domain, secondary cleavages occur at approximately 10-fold lower rates than the initial cut. The tertiary cleavages occur approximately 400-fold slower than the primary cleavage.72 During proteolysis, HIV-1 NC exists in two intermediate forms, NCp15 (partial cleavage product containing NC/SP2/p6) and NCp9 (partial cleavage product containing NC/SP2) and the fully processed form, NCp7. All three of these proteins exhibit nucleic acid chaperone activities.17 (B) Primary cleavage of MuLV Gag occurs between p12 and CA with slower secondary cleavages that yield the mature proteins MAp15, p12, CAp30 and NCp10.124 (C) Extended structural model of an HIV-1 Gag polypeptide assembled from high resolution structures of isolated domains (PDB IDs:1UPH, 3GV2, 1A1T, 2C55 respectively for MA, CA, NCp7 and p6). Unavailable and/or unstructured domains are represented by dashed lines. PR cleavage site are indicated by the arrowheads. (D) Schematic models of viral core maturation. SP1-NC cleavage by PR rapidly separates the MA-CA shell and the nucleocapsid complex formed between RNA, NCp15, RT and IN. NCp15 processing by PR into NCp9 and finally NCp7 leads to NC/RNA coaggregation/condensation within the confines of the capsid cone.

Acknowledgements

G.M. and S.L. are greatly indebted to Jose Maria Gatell and Laura Zamora for their kind hosting and support at the Infectious Disease and AIDS Unit. We wish to thank Sophie Brouillet, Joel Pothier (ABI, UPMC, Paris) and Anne Voinet (MTI-Université Paris Diderot, Paris) for the alignment of HIV-1 NC sequences. We also wish to thank James A. Thomas (NCI-Frederick) and Judith G. Levin (NICHD, NIH) for critical reading of this manuscript. S.L. is funded by a Marie Curie Intra-European Fellowship IEF N° 237738. (R.J.G.) this project has been funded in whole or in part with federal funds from the National Cancer Institute, National Institutes of Health, under contracts N01-CO-12400 and HHSN261200800001E. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products or organizations imply endorsement by the US Government.