RNA-Binding Domains of Heterologous Viral Proteins Substituted for Basic Residues in the RSV Gag NC Domain Restore Specific Packaging of Genomic RNA.

The Rous sarcoma virus Gag polyprotein transiently traffics through the nucleus, which is required for efficient incorporation of the viral genomic RNA (gRNA) into virus particles. Packaging of gRNA is mediated by two zinc knuckles and basic residues located in the nucleocapsid (NC) domain in Gag. To further examine the role of basic residues located downstream of the zinc knuckles in gRNA encapsidation, we used a gain-of-function approach. We replaced a basic residue cluster essential for gRNA packaging with heterologous basic residue motif (BR) with RNA-binding activity from either the HIV-1 Rev protein (Rev BR) or the HSV ICP27 protein (ICP27 BR). Compared to wild-type Gag, the mutant ICP27 BR and Rev BR Gag proteins were much more strongly localized to the nucleus and released significantly lower levels of virus particles. Surprisingly, both the ICP27 BR and Rev BR mutants packaged normal levels of gRNA per virus particle when examined in the context of a proviral vector, yet both mutants were noninfectious. These results support the hypothesis that basic residues located in the C-terminal region of NC are required for selective gRNA packaging, potentially by binding non-specifically to RNA via electrostatic interactions.


Introduction
The retroviral Gag protein directs the assembly of new virus particles at the plasma membrane, specifically selecting the viral genomic RNA (gRNA) from the milieu of cellular and viral RNAs. The mechanism by which selective packaging occurs within the cell remains incompletely understood. The Gag NC (nucleocapsid) domain is essential for encapsidation of gRNA and also plays an important role in the subcellular trafficking of the Gag protein [1][2][3][4][5][6]. RSV Gag is initially synthesized in the cytoplasm and then undergoes transient nuclear trafficking, a step that is required for efficient gRNA packaging [7,8]. Mutants of Gag with reduced nuclear trafficking have lower levels of gRNA incorporation, whereas enhancing nuclear localization restores gRNA packaging [8]. After particle release, RSV Gag is cleaved into MA (matrix), p2, p10, CA (capsid), NC, and small spacer peptides. The mature NC protein plays important roles during the early stages of infection.

Confocal Imaging
For pNC.YFP constructs co-expressed with pFibrillarin.CFP, 0.2 × 10 6 cells were seeded onto 35-mm glass-bottomed dishes (MatTek Corporation, Ashland, MA, USA) and imaged using a Leica AOBS SP2 confocal microscope with a 63×/1.4 oil objective at a 4× zoom at 14 to 24 h post-transfection. Sequential scanning settings were used to differentiate CFP (excitation at 458 nm, emission at 465-490 nm, and 50% laser power) and YFP (excitation at 514 nm, emission at 530-600 nm, and 10% laser power) emission spectra. The Gag-YFP wild-type and mutant proteins were imaged on a Leica SP8 confocal microscope with a 63×/1.4 oil objective at a 4× zoom, with DAPI excited with the 405 nm UV laser at 20% laser power using a photomultiplier tube detector and YFP imaged using the white light laser with a laser line excitation of 514 nm using a hybrid detector.

Budding Analysis
Budding assays were performed as previously described in detail in [37]. Briefly, Gag expression within cell lysates was labelled for 5 min using 35 S-Met/Cys. Lysates were collected, then Gag was immunoprecipitated using an α-RSV antibody and was resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and quantified using a PhosphorImager (Bio-Rad, Hercules, CA, USA). After 35 S-Met/Cys labeling for 2.5 h, supernatants were clarified, and virus particles were pelleted and immunoprecipitated using an α-RSV antibody and separated by SDS-PAGE and imaged by phosphorimaging. Budding efficiency was calculated as a ratio of the CA present in the media divided by the total amount of Gag expressed in the lysates. The release of the wild-type RS.V8 was set at 100%, and all mutants were expressed as a percentage of the wild-type level of budding.

Ribonuclease Protection Assays
QT6 cells were transfected with either wild-type or mutant proviral DNA constructs. Culture media was collected after 48 h, cells were pelleted using a low speed spin, and the media was passed through a 0.2 µm filter. Virus particles were pelleted by ultracentrifugation at 126,000× g through a 25% sucrose cushion. After resuspension of the pellet, aliquots were removed for reverse transcriptase (RT) assays. The mean RT values were used to normalize the amount of virus particles for each sample as previously described [38]. Viral RNA was extracted using a QiaAMP viral RNA mini kit (Qiagen, Hilden, Germany). Ribonuclease protection assays were performed as previously described [38]. A 318-nucleotide antisense probe transcribed with T7 RNA polymerase with [ 32 P]CTP, spanning the splice acceptor site of the env gene, was used to detect both unspliced (263-nucleotide fragment) and spliced (183-nucleotide fragment) viral RNA as previously described in detail [39]. The unprotected fragments of the RNAs were digested with RNase and the samples were separated by gel electrophoresis and quantified using a PhosphorImager (BioRad, Hercules, CA, USA).

Viral Infectivity Assay
Infection assays were performed as previously described in detail [37]. Supernatants were collected from QT6 cells expressing either wild-type or mutant provirus after 48 h. Virus particles were concentrated by ultracentrifugation at 126,000× g through a 25% sucrose cushion. RT assays were performed, as described above, to normalize the amount of virus used for infection. Equivalent amounts of RT counts from each concentrated virus preparation were added to naive DF1 cells. Cells were then assayed for the presence of GFP, which is expressed from the RS.V8 provirus, by flow cytometry (FACSCanto, BD Biosciences, San Jose, CA, USA). The percentage of cells expressing GFP was measured every three days until all infectious virus constructs reached approximately 95% green cells, or after 21 days for all remaining viruses. A minimum of three infectivity assays was performed for each proviral construct from two separate transfections.

Substitution of the ICP27 or Rev Basic Residue Motifs(BR) for Basic Residues in NC Altered the Budding of Virus Particles
To examine whether the basic residues after the second Cys-His box in RSV Gag NC were important for virus assembly, we deleted residues 61-73 in the context of the proviral construct pRS.V8 (pRS.V8.∆61-73) ( Figure 1A), and a quantitative radioimmunoprecipitation assay was performed. A representative experiment showing Gag expression within the cell lysates after a 5-min labeling period is presented in Figure 1B. After detection of metabolically labeled Gag proteins released into the supernatant, quantitation of virus release was performed by dividing the amount of CA in the media by the amount of Gag in the cell lysates [37]. For RS.V8.∆61-73, we observed that budding was not significantly changed compared to wild-type Gag ( Figure 1C,D). To determine whether budding would be affected by replacing the basic residue-containing region between amino acids 61-73 with a stretch of basic amino acids from the viral RNA-binding proteins HIV-1 Rev and HSV-1 ICP27, these heterologous BRs were inserted into this region of NC [40,41], forming the chimeric proviral constructs pRS.V8.Rev and pRS.V8.ICP27 ( Figure 1A).
To determine whether insertion of the BRs from ICP27 and Rev altered virus particle production, budding efficiency for each mutant was compared to wild-type, which was set at 100% ( Figure 1D). RS.V8.∆61-73 was not significantly different from wild-type (147%; p > 0.5). RS.V8.Rev was reduced in budding (12%; p = 0.0013), as was RS.V8.ICP27 (41%; p = 0.0253). These results indicate that the BR from Rev and ICP27 negatively affects budding, but the complete removal of the BR in the C-terminal region of NC (RS.V8.∆61-73) does not. released into the supernatant, quantitation of virus release was performed by dividing the amount of CA in the media by the amount of Gag in the cell lysates [37]. For RS.V8.61-73, we observed that budding was not significantly changed compared to wild-type Gag ( Figure 1C,D). To determine whether budding would be affected by replacing the basic residue-containing region between amino acids 61-73 with a stretch of basic amino acids from the viral RNA-binding proteins HIV-1 Rev and HSV-1 ICP27, these heterologous BRs were inserted into this region of NC [40,41], forming the chimeric proviral constructs pRS.V8.Rev and pRS.V8.ICP27 ( Figure 1A).  CT, C-terminal region. Wild-type residues are shown for the C-terminal basic region. Dashed lines represent deleted residues (∆61-73). Bold residues depict sequences used to replace the deleted amino acids (ICP27 BR and Rev BR). (B) Gag expression within the cell lysates after a 5-min labeling period with 35 S-Met/Cys. After collection of cell lysates and immunoprecipitation using an α-RSV antibody, the proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and visualized by phosphorimaging. Differences in molecular weights (predicted masses listed in parentheses) were noticeable, with ∆61-73 mutant Gag (73 kD) running faster than wild-type (74.5 kD), and both Rev (75.3 kD) and ICP27 (74.7 kD) running slightly slower in the gel. The arrow indicates the position of the Gag band. (C) After radioactive labeling for 2.5 h, supernatants were clarified, and virus particles were pelleted and immunoprecipitated using an α-RSV antibody (media samples). Proteins were separated by SDS-PAGE and visualized by phosphorimaging. The arrow indicates CA (25.8 kD). (D) The average of four independent budding assays is presented within the bar graph, with the bars representing the standard error. * p-value = 0.0253, ** p-value = 0.0013 calculated by Student's t-test.

Subcellular Localization of the ICP27 BR and Rev BR Gag Mutants
The amino acid sequences derived from ICP27 and Rev are capable of general RNA binding as well as facilitating nucleolar localization [24][25][26][27][29][30][31]]. Therefore, we tested whether the addition of these heterologous viral RNA-binding domains, which also serve as nucleolar localization signals, would alter the normal trafficking patterns of RSV Gag. To investigate this possibility, we examined cells expressing wild-type or mutant proviral constructs by immunofluorescence staining using a polyclonal α-RSV antibody ( Figure 2A). The wild-type Gag protein expressed using a proviral vector exhibited a small amount of nuclear fluorescence, with the majority of the fluorescence signal in the cytoplasm, forming discrete foci at the plasma membrane ( Figure 2A, panel a). The ∆61-73 mutant showed similar localization, with fluorescent signal primarily in the cytoplasm and along the plasma membrane, and possibly even more strongly at the plasma membrane compared to wild-type ( Figure 2A, panel b). The Gag localization results correlate with the budding data from Figure 1, in that the average budding efficiency of ∆61-73 was greater than wild-type, even though not statistically significant. Unexpectedly, Viruses 2020, 12, 370 6 of 14 although both RS.V8 ICP27 and Rev mutants contain the NES in the Gag p10 domain, they both were concentrated in the nucleus and formed numerous foci in the nucleoplasm (Figure 2A, panels c and d, respectively). cells expressing wild-type or mutant proviral constructs by immunofluorescence staining using a polyclonal -RSV antibody (Figure 2A). The wild-type Gag protein expressed using a proviral vector exhibited a small amount of nuclear fluorescence, with the majority of the fluorescence signal in the cytoplasm, forming discrete foci at the plasma membrane ( Figure 2A, panel a). The 61-73 mutant showed similar localization, with fluorescent signal primarily in the cytoplasm and along the plasma membrane, and possibly even more strongly at the plasma membrane compared to wild-type ( Figure  2A, panel b). The Gag localization results correlate with the budding data from Figure 1, in that the average budding efficiency of Δ61-73 was greater than wild-type, even though not statistically significant. Unexpectedly, although both RS.V8 ICP27 and Rev mutants contain the NES in the Gag p10 domain, they both were concentrated in the nucleus and formed numerous foci in the nucleoplasm (Figure 2A, panels c and d, respectively). To determine whether viral nucleic acids or proteins expressed from the proviral vectors affected Gag localization, we expressed Gag fused to YFP using a CMV promoter ( Figure 2B). Gag.YFP was similar in appearance, with mostly cytoplasmic and plasma membrane fluorescence and a lower amount of diffuse and focal nuclear signals ( Figure 2B, panel a). The deletion mutant Gag.∆61-73 had similar localization compared to wild-type Gag ( Figure 2B, panel b). However, when either the Gag.ICP27.YFP or Gag.Rev.YFP mutants were expressed, they localized almost exclusively to the nucleus, forming numerous nucleoplasmic foci ( Figure 2B, panels c and d, respectively). Of note, full length ICP27 forms nucleoplasmic foci in HSV-infected cells [24][25][26], whereas HIV-1 Rev localizes to nucleoli under steady-state conditions [30].
When expressed by itself, wild-type NC localizes primarily to nucleoli [11] ( Figure 2C, panel a). By contrast, the NC.∆61-73 deletion mutant was nuclear-localized but excluded nucleoli ( Figure 2C,  panel b). Substitution of the ICP27 or Rev BRs for residues 61-73 restored nucleolar localization of the NC protein ( Figure 2C, panels c and d, respectively). These data indicate that replacement of amino acids 61-73 with either the ICP27 or Rev BR is sufficient to direct NC nucleolar localization.

Heterologous BRs Substituted in the Gag NC Domain Restore gRNA Packaging
The Cys-His boxes in RSV Gag specifically bind to the psi-packaging sequence [17,18,42], although basic residues within NC also contribute to gRNA packaging [21,22]. To investigate the Viruses 2020, 12, 370 7 of 14 importance of basic residues in the C-terminal region of the Gag for gRNA encapsidation, we measured the relative amount of gRNA in virus particles using a quantitative ribonuclease protection assay. The amount of gRNA isolated from wild-type and mutant viruses was normalized using reverse transcriptase activity, as described in Materials and Methods. We used a probe spanning the 3 splice acceptor site in env to quantitate the amount of spliced viral RNA and unspliced gRNA isolated from purified particles. Representative autoradiograms are shown ( Figure 3A), and the means of at least three independent experiments for each mutant were plotted ( Figure 3B). The amount of gRNA detected in wild-type virions was set to 100%, and each mutant was compared to wild-type. The Gag.∆61-73 mutant virus was significantly reduced in its ability to package gRNA (15% of the wild-type level, p = 0.0001). The RS.V8.Rev and RS.V8.ICP27 viruses packaged RSV gRNA much more efficiently, with levels of 73% and 78% compared to wild-type, respectively, although gRNA incorporation for RS.V8.Rev was statistically lower than wild-type (p = 0.0278). These results indicate that insertion of a heterologous BR restores incorporation of gRNA into virus particles at near wild-type levels. When the ratio of spliced:unspliced viral RNA was examined, both BR insertion mutants packaged nearly the same ratio as the wild-type virus. However, in the case of the ∆61-73 mutant virus, the level of gRNA packaging was drastically reduced (15%), and the ratio of spliced:unspliced viral RNA detected by the phosphorimager was increased when compared to the wild-type virus, demonstrating a reduction in the specificity of the type of viral RNA incorporated into particles. BR or Gag.Rev BR were transfected into QT6 cells. Virus particles were collected for 48 h and were normalized using reverse transcriptase assay. Equivalent amounts of virus particles were used to extract RNA, and ribonuclease protection assay was used to measure the levels of spliced and unspliced viral RNA present. After hybridization of the 318-nt 32 P-radiolabled antisense riboprobe, which spans the 3′-splice acceptor site in env, and digestion of the unprotected fragment with RNase treatment, the RNA was separated by gel electrophoresis and visualized using a phosphorimager. Black lines represent lanes that were removed and spaces between the gels represent results from independent experiments. Arrows to the left identify nucleotide lengths, and arrows to the right denote the species of RNA (free probe, unspliced gRNA, and spliced viral RNA). (B) The results of four independent experiments are shown on the graph, with the bars representing standard error of the mean. * p-value = 0.0278; ** p-value = 0.0001 when compared to wild-type levels by Student's ttest.

Viruses with Deletions of Basic Residues or Heterologous BR Substitutions in the NC Domain of Gag Are Noninfectious
To determine whether the basic residues in the C-terminal region of Gag are required for virus replication, we performed infectivity assays. Virus particles were collected from QT6 cells expressing either wild-type or mutant proviral constructs, the particles were normalized by reverse transcriptase activity, and equivalent amounts of virus were placed on uninfected DF1 cells. The RS.V8 provirus contains a GFP gene in the place of src, so the ability of mutant viruses to spread through the cell culture was monitored by measuring fluorescence using fluorescence-activated cells every three days BR or Gag.Rev BR were transfected into QT6 cells. Virus particles were collected for 48 h and were normalized using reverse transcriptase assay. Equivalent amounts of virus particles were used to extract RNA, and ribonuclease protection assay was used to measure the levels of spliced and unspliced viral RNA present. After hybridization of the 318-nt 32 P-radiolabled antisense riboprobe, which spans the 3 -splice acceptor site in env, and digestion of the unprotected fragment with RNase treatment, the RNA was separated by gel electrophoresis and visualized using a phosphorimager. Black lines represent lanes that were removed and spaces between the gels represent results from independent experiments. Arrows to the left identify nucleotide lengths, and arrows to the right denote the species of RNA (free probe, unspliced gRNA, and spliced viral RNA). (B) The results of four independent experiments are shown on the graph, with the bars representing standard error of the mean. * p-value = 0.0278; ** p-value = 0.0001 when compared to wild-type levels by Student's t-test.

Viruses with Deletions of Basic Residues or Heterologous BR Substitutions in the NC Domain of Gag Are Noninfectious
To determine whether the basic residues in the C-terminal region of Gag are required for virus replication, we performed infectivity assays. Virus particles were collected from QT6 cells expressing either wild-type or mutant proviral constructs, the particles were normalized by reverse transcriptase

Discussion
The NC domain of the RSV Gag protein is enriched in basic residues, which play a variety of roles in virus replication [20][21][22]. To date, the functions of basic residues in NC have been primarily studied using deletion mutagenesis, which produces negative or loss-of-function results. More informative are genetic gain-of-function experiments, in which an essential function is reconstituted using functional domains derived from heterologous proteins. In this report, we first deleted a region following the Cys-His boxes of RSV NC that contains five positively charged residues, finding that Gag localization and budding were not adversely affected, whereas gRNA packaging and infectivity were severely compromised, and nucleolar localization of the mature NC protein was abrogated. The data suggest that these basic residues do not affect general Gag-nucleic acid interactions that are needed for particle assembly [43][44][45], but appear to be involved in the specificity of binding viral gRNA.
In an attempt to separate the contributions of basic residues encompassing amino acids 61-73 in RSV NC, we substituted the sequences from two different viral RNA-binding proteins [27,46] that also contain nucleolar localization signals. Even though these sequences were derived from very different viral proteins, those of HSV-1 ICP27 and HIV-1 Rev, the resulting chimeric proteins behaved very similarly. The most striking feature was the re-localization of these chimeric Gag proteins to the nucleus, with accumulation in discrete nucleoplasmic foci that resembled ICP27 nuclear foci in HSV-1-infected cells [28,47,48]. In fact, quite surprisingly, the ICP27 and Rev Gag nuclear foci are very similar in appearance to RSV Gag treated with leptomycin B (an inhibitor of Crm1 nuclear export) or with the introduction of a point mutant in the p10 NES of Gag that blocks nuclear egress [14,15].
What accounts for the accumulation of Gag in nuclear foci with the addition of BRs from Rev and ICP27? The NC domain of Gag is a major site of Gag-Gag interactions mediated through proteinprotein or protein-RNA binding [23,49,50], so perhaps the BRs enhance Gag-Gag interactions. We previously showed that Gag-Gag intermolecular interactions in the nucleus depend on the NC domain and its RNA-binding ability [14]. In addition, the ICP27 and Rev RNA-binding domains may interact with host nuclear factors, either proteins and/or RNAs, that serve to tether Gag more strongly in the nucleus [28,47,[51][52][53][54]. We previously reported that RSV Gag colocalizes with splicing factors [15], as does ICP27 [28]. Whether Gag influences RNA splicing, as has been found for ICP27 [51], is not yet known. It is also possible that the insertions of ICP27 and Rev BRs in NC alter the conformation of the Gag protein, interfering with the function of the NES in p10, leading to nuclear The wild-type virus (RS.V8) was able to efficiently infect and replicate within cells, reaching approximately 95% of cells expressing the GFP protein by day 6 post-infection. The RS.V8.∆61-73 mutant was noninfectious, which was expected given the low incorporation of gRNA. However, even though gRNA incorporation was restored by insertion of the exogenous viral RNA-binding domains from ICP27 and Rev, these substitutions were unable to restore infectivity. We cannot distinguish between the possibilities that these mutant viruses are defective in establishment of infection and/or cell-to-cell spread.

Discussion
The NC domain of the RSV Gag protein is enriched in basic residues, which play a variety of roles in virus replication [20][21][22]. To date, the functions of basic residues in NC have been primarily studied using deletion mutagenesis, which produces negative or loss-of-function results. More informative are genetic gain-of-function experiments, in which an essential function is reconstituted using functional domains derived from heterologous proteins. In this report, we first deleted a region following the Cys-His boxes of RSV NC that contains five positively charged residues, finding that Gag localization and budding were not adversely affected, whereas gRNA packaging and infectivity were severely compromised, and nucleolar localization of the mature NC protein was abrogated. The data suggest that these basic residues do not affect general Gag-nucleic acid interactions that are needed for particle assembly [43][44][45], but appear to be involved in the specificity of binding viral gRNA.
In an attempt to separate the contributions of basic residues encompassing amino acids 61-73 in RSV NC, we substituted the sequences from two different viral RNA-binding proteins [27,46] that also contain nucleolar localization signals. Even though these sequences were derived from very different viral proteins, those of HSV-1 ICP27 and HIV-1 Rev, the resulting chimeric proteins behaved very similarly. The most striking feature was the re-localization of these chimeric Gag proteins to the nucleus, with accumulation in discrete nucleoplasmic foci that resembled ICP27 nuclear foci in HSV-1-infected cells [28,47,48]. In fact, quite surprisingly, the ICP27 and Rev Gag nuclear foci are very Viruses 2020, 12, 370 9 of 14 similar in appearance to RSV Gag treated with leptomycin B (an inhibitor of Crm1 nuclear export) or with the introduction of a point mutant in the p10 NES of Gag that blocks nuclear egress [14,15].
What accounts for the accumulation of Gag in nuclear foci with the addition of BRs from Rev and ICP27? The NC domain of Gag is a major site of Gag-Gag interactions mediated through protein-protein or protein-RNA binding [23,49,50], so perhaps the BRs enhance Gag-Gag interactions. We previously showed that Gag-Gag intermolecular interactions in the nucleus depend on the NC domain and its RNA-binding ability [14]. In addition, the ICP27 and Rev RNA-binding domains may interact with host nuclear factors, either proteins and/or RNAs, that serve to tether Gag more strongly in the nucleus [28,47,[51][52][53][54]. We previously reported that RSV Gag colocalizes with splicing factors [15], as does ICP27 [28]. Whether Gag influences RNA splicing, as has been found for ICP27 [51], is not yet known. It is also possible that the insertions of ICP27 and Rev BRs in NC alter the conformation of the Gag protein, interfering with the function of the NES in p10, leading to nuclear retention of the mutant Gag proteins.
The increased nuclear localization of the chimeric Gag proteins likely explains the budding defect of the RS.V8.ICP27 and RS.V8.Rev viruses. These results provide further evidence that nuclear trafficking of RSV Gag is intrinsic to the virus assembly pathway since enhancing nuclear localization was linked to an assembly defect of these mutants. In spite of the defect in particle production, the ICP27 and Rev chimeric viruses both restored gRNA packaging when compared to the mutant bearing a deletion of basic residues 61-73 of the NC domain. This restoration of gRNA packaging could be due to the RNA-binding capabilities of the RS.V8.Rev and RS.V8.ICP27 mutants, although it was somewhat unexpected that the RNA-binding domains from heterologous viruses would confer packaging of the RSV genome, maintaining the proper ratio of spliced:unspliced viral RNAs incorporated into virus particles.
One possible explanation for the effect of the BR insertions is that the presence of several non-specific RNA-binding domains may contribute to specific gRNA binding, although the mechanisms underlying specificity of RNA recognition are not fully understood [55]. For example, RNA-binding domains rich in RS or RG residues, such as the RGG box of ICP27, in combination with other basic sequences, have been shown to mediate both specific and non-specific interactions with RNA [56]. The properties of RNA-binding proteins have been extensively studied in serine/arginine-rich (SR) proteins involved in RNA splicing [56,57]. SR proteins generally contain one or two of the RNA recognition motifs at the N-terminus and an RS motif at the C-terminus. Typically, the RNA recognition motifs are involved in recognizing the specific target RNAs, while the RS motifs participate in indirect protein-protein and non-specific protein-RNA interactions that bring the SR proteins and the target RNAs within close proximity [56,57]. Other common sequences typically found in RNA-binding proteins are motifs enriched in R/K residues, in which four to eight residues in small patches form highly positive regions that mediate molecular interactions. They frequently flank globular domains, assisting in RNA binding [56]. Therefore, it is possible that the deletion of basic residues immediately after the second Cys-His box in RSV Gag eliminates specific binding of the psi sequence, as previously suggested by Lee et al. [22]. Introduction of additional basic residues in the ICP27 and Rev heterologous RNA-binding domains, which contain 8 and 10 basic residues, respectively, restores the interaction with RSV RNA via electrostatic interactions that bring Gag and gRNA in close proximity, allowing specific gRNA binding through the zinc knuckles in NC.
Electrostatic interactions play a role in the binding of positively charged RNA-binding proteins to negatively charged RNAs through both specific and non-specific mechanisms. We searched the literature for cellular RNA-binding proteins with mechanisms that explain how non-specific electrostatic interactions could contribute to specific binding to their cognate RNA molecules. We found that the spliceosomal protein, U1A, uses a non-specific "lure" step followed by a specific "lock" step to bind to its SL2 RNA partner (summarized in [58]). Short-range electrostatic interactions during the "lure" step are proposed to attract the SL2 RNA to U1A, followed by the "lock" step, in which long-range electrostatic interactions between the protein and RNA create specific binding. We hypothesize that RSV Gag could be utilizing a similar technique for binding to viral gRNA. The basic residues in the C-terminus of NC could be involved in the initial electrostatic "lure" step, bringing gRNA in closer proximity. This binding event would then allow the zinc knuckles in NC to specifically bind to the psi sequence on the viral RNA. Thus, whether selective gRNA packaging in the chimeric ICP27 and Rev viruses is due to the addition of these heterologous RNA-binding domains or simply the addition of additional basic amino acids is not clear. Further studies to examine this question would be needed, perhaps by inserting a random sequence of basic residue sequences versus RNA-binding domains that do not contain basic amino acids (KH domain or additional zinc fingers) to further dissect the mechanism.
Although insertion of heterologous BRs from ICP27 and Rev into NC rescued selective incorporation of gRNA into virus particles, these mutants were noninfectious. Because the mutations were inserted into NC, the most likely explanation is that functions of NC in other steps of infection were impaired, including gRNA dimerization [16,59,60], reverse transcription [61][62][63][64][65], transport of the preintegration complex [66], or chaperone function [67,68]. We can conclude that nucleolar localization of NC, which was restored in the ICP27 and Rev chimeras, is not sufficient for NC-mediated replication activities early in infection. Further investigation into the mechanisms underlying the replication defect of these chimeric viruses will be enlightening.

Conclusions
The work presented here demonstrates the importance of the basic residues after the Cys-His boxes in the C-terminus of RSV NC in gRNA packaging. When this sequence was deleted, virus particles were able to bud normally from cells, but were not infectious. However, when these residues were replaced with heterologous basic residues from other viral RNA-binding proteins, the subcellular localization of Gag became predominantly nuclear, and gRNA packaging was restored. Even though gRNA packaging in the chimeras was increased to near normal levels, these viruses remained noninfectious. These results suggest that the C-terminal basic residues in NC are important for facilitating gRNA binding. A deeper understanding of the mechanism by which retroviral Gag proteins selectively incorporate their genomes may be helpful in future antiviral and vaccine development [69][70][71][72][73].

Author Contributions:
Writing-original draft preparation, conceptualization, formal analysis, and investigation, B.L.R.; conceptualization, formal analysis, investigation, methodology, writing-original draft, and writing-review and editing, T.L.L.; conceptualization, funding acquisition, project supervision, resources, writing-review and editing, L.J.P. All authors have read and agreed to the published version of the manuscript.