Identification of Pr78Gag binding sites on the Mason-Pfizer monkey virus genomic RNA packaging determinants.

How retroviral Gag proteins recognize the packaging signals (Psi) on their genomic RNA (gRNA) is a key question that we addressed here using Mason-Pfizer monkey virus (MPMV) as a model system by combining band-shift assays and footprinting experiments. Our data show that Pr78Gag selects gRNA against spliced viral RNA by simultaneously binding to two single stranded loops on the MPMV Psi RNA: (1) a large purine loop (ssPurines), and (2) a loop which partially overlaps with a mostly base-paired purine repeat (bpPurines) and extends into a GU-rich binding motif. Importantly, this second Gag binding site is located immediately downstream of the major splice donor (mSD) and is thus absent from the spliced viral RNAs. Identifying elements crucial for MPMV gRNA packaging should help in understanding not only the mechanism of virion assembly by retroviruses, but also facilitate construction of safer retroviral vectors for human gene therapy.


Introduction
Specific selection of the retroviral genome is central to the process of virion assembly, during which a dimeric form of retroviral genomic RNA (gRNA) is selectively packaged into the nascently forming virions [1][2][3][4][5][6]. Despite the fact that the viral gRNA constitutes only ~1% of the total RNA in the cell milieu it is still specifically selected from a vast array of spliced viral and cellular RNAs [7][8][9][10][11][12][13]. This highly controlled and selective process is dependent on two important factors: (1) the presence of specific sequences or structures within the gRNA, and (2) the retroviral precursor polyprotein Gag and its ability to identify and bind to these unique sequences or structures [7,[9][10][11][12][13][14][15].
A key player in the process of specific gRNA packaging is the retroviral Gag polyprotein.
The three major domains of the polyprotein are matrix (MA), capsid (CA) and nucleocapsid (NC), found ubiquitously in all retroviruses [6,[12][13][14]. From the time of its synthesis in the cytoplasm to virion release and maturation, the various domains of Gag drive the assembly process and are responsible for one or more key events in the retroviral life cycle. One of these events is to identify and bind to specific sequences on the Psi RNA and bring about selective 4 packaging of its gRNA. Although Gag domains are most often studied independently of each other, their various functions are achieved within the native context of the polyprotein. These domains are liberated from one another only upon virus maturation post virion assembly and release from the infected cell. Among these domains, the NC is primarily responsible for the selective binding and packaging of gRNA in most retroviruses. It is a highly basic and hydrophobic protein, harboring two highly conserved zinc finger domains. These fingers consist of CCHC arrays (C-X 2 -C-X 4 -H-X 4 -C; where C = Cys, H = His, X n = n number of other amino acids) that sequester zinc ions required for specific gRNA binding [6,[12][13][14].
A majority of studies conducted to date to identify the specific high affinity binding sites of Gag on retroviral Psi RNA have been restricted to the NC domain alone, either in its immature or mature form [28][29][30][31][32][33][34][35][36][37]. Preliminary studies carried out on human immunodeficiency virus type 1 (HIV-1) indicated the high affinity binding site of its NC domain to be located on the apical loop of SL3 [28,30,37]. However, use of full-length HIV-1 Pr55 Gag has revealed the presence of a high affinity binding site on the internal (G//AGG) loop of SL1 of the HIV Psi RNA [9].
These findings suggest the importance of studying the selective packaging of retroviral gRNA in the context of the full length precursor polyprotein, Gag. However, since NC plays its role as part of full-length Gag, the role of domains other than NC cannot be excluded. In fact, a number of studies have implicated other domains of Gag in specific binding to gRNA, including the MA [38][39][40][41][42], CA [43], p1 [44], p2 [45,46] and p6 [47,48] domains of HIV-1. Results from these studies suggest that all three major domains of Gag (MA, CA, and NC) and the other domains are capable of binding to various structured motifs on the Psi RNA with varying affinities, emphasizing the importance of studying selective Psi RNA packaging in the native context of full-length Gag. 5 Mason Pfizer monkey virus (MPMV) is the most widely studied prototypic type D betaretrovirus. It was first isolated from the breast adenocarcinoma of a rhesus monkey (Macaca mulatta) and is known to cause fatal immunodeficiency in macaques [49,50]. MPMV serves as a potential candidate for the development of gene therapy vectors due to its phylogenetic distance from human retroviruses, such as HIV-1. It harbors promoters that are functional in human cells and also its constitutive transport element (CTE), analogous to the HIV-1 Rev responsive element (RRE), allows for efficient cytoplasmic transport of viral RNA independent of any viral protein [51][52][53][54]. MPMV has also been widely studied to decipher the assembly process of retroviruses and thus can be used as an experimental tool to investigate potential inhibitors of retroviral particle assembly [55]. It differs distinctly from C-type retroviruses such as HIV-1 by adopting an intracytoplasmic A-type morphology with spherical capsids.
The 5' end of MPMV has been extensively investigated to demarcate the boundaries of the minimal packaging sequences required for efficient incorporation of gRNA into viral particles [24,27,[56][57][58][59][60][61][62][63]. These sequences span from the 5' UTR into the gag open reading frame, similar to most other retroviruses [14]. Systematic mutational analyses carried out on this region has revealed a discontinuous or bipartite signal consisting of the first 50 nts of the 5' UTR, inclusive of the palindromic stem loop (Pal SL) that serves as the dimerization initiation site (DIS) for the gRNA, and the last 23 nts of the 5' UTR followed by the first 120 nts of gag, both of which are required for successful MPMV gRNA packaging [ Figure 1A; 59,60].
Consistent with other retroviruses, the Psi sequences on MPMV gRNA fold into a higher order structure comprising of various structural motifs [24,60]. Among these structural motifs, two purine-rich motifs, the single stranded purines (ssPurines; U 191 UAAAAGUGAAAGUAA 206 ) and the base paired purines (bpPurines; G 246 AAAGUAA 253 ), have been identified as unique regions 6 in their composition and positioning that may contribute to Gag binding [24,60]; Figure 1B).
Presence of these purine-rich sequences in the MPMV Psi RNA is consistent with the fact that a stretch of purines in the Psi of other retroviral gRNAs has been proposed to facilitate gRNA packaging by functioning as a potential NC binding site [3,9,15,17,18,[64][65][66][67][68][69].
The ssPurines present in MPMV Psi RNA by far forms the largest single-stranded purine rich region found in any widely studied retroviral Psi RNA ( Figure 1B). It consists of 16 singlestranded nucleotides with 75% purines, and more importantly, is located immediately downstream of the DIS which makes it a potential motif for Gag binding. This observation is consistent with HIV-1 where the primary Gag binding site (G//AGG) is located on the internal loop of SL1, whose apical loop also functions as DIS [9]. Interestingly, the latter half of the ssPurines sequence is found repeated downstream in a base paired manner to form the bpPurines ( Figure 1B; [60]. Employing genetic, biochemical, and structure-function approaches, a recent study has pointed towards these two purine-rich regions (ss-and bp-Purines) functioning as redundant packaging motifs and possible Gag binding sites during viral assembly [63]. Therefore, the current study was undertaken to establish whether these two purine-rich regions (ss-and bp-Purines) on MPMV Psi RNA truly function as MPMV Gag precursor polyprotein (Pr78 Gag ) binding sites ( Figure 1B). Our results indicate that the MPMV polyprotein, Pr78 Gag binds to two loops: 1) the ssPurines loop (U 191 UAAAAGUGAAAGUAA 206 ) and 2) a second loop (A 252 AGUGU 257 ) corresponding to the last two purines of the bpPurines and extending into a GU-rich region ( Figure 1B). Interestingly, this second binding site is located immediately downstream of the mSD and is thus absent from the spliced viral RNAs ( Figure 1B). Finally, we propose a model for the specific selection of full length unspliced MPMV RNA over cellular and viral spliced env RNA by Pr78 Gag .

His 6 -tag fusion protein) by dynamic light scattering (DLS)
To determine whether the two purine-rich regions (ss-and bp-Purines) important for MPMV gRNA packaging, act as binding sites for the Gag precursor polyprotein, large scale expression and purification of Pr78 Gag was performed and the protein characterized for its biological function, revealing that it could assemble in vitro to form virus-like particles (VLPs), and also form VLPs in bacteria, and that the VLPs produced in eukaryotic cells could encapsidate MPMV RNA containing the Psi region [62].
The bacterially expressed and full-length Pr78 Gag was characterized by DLS, which revealed that the purified protein did not contain any aggregates. The mean hydrodynamic radius (R h ) based on volume (percent) and number (percent) distribution was estimated to be 6.7 and 5.8 nm, respectively. This corresponded to a molecular weight of 288 and 206 kDa, respectively, indicative of Pr78 Gag trimers (Supplemental figure 1).
Pr78 Gag discriminates between full-length, unspliced sub-genomic Psi RNA and spliced env

RNA
We first attempted to establish if Pr78 Gag specifically binds to the full-length, unspliced gRNA (RCR001) over spliced env RNA (FN42; Figure 2A). The T7-based plasmids expressing these RNAs were designed in a way that they would express RNAs of the same length (549 nucleotides). As a first step, we performed band shift assays to determine the optimal concentration of Pr78 Gag at which a complete shift of dimeric gRNA would occur upon successful Gag-RNA complex formation. These band shift assays were performed with a constant amount of radiolabeled unspliced gRNA (50,000 cpm, ~5 nM) against progressively increasing concentrations of Pr78 Gag (from 0 to 2000 nM). Results of these band shift assays indicated a complete shift of the dimeric gRNA at a protein concentration of 500 nM ( Figure   2B). The band shift gel was then quantitated, and the data plotted to fit Hill's equation ( Figure   2C). The Hill's coefficient was estimated to be 1.216 ± 0.422 (mean ± SD), suggesting that binding of Pr78 Gag to gRNA is weakly cooperative or non-cooperative. The apparent K d obtained was 216.2 ± 76.10 nM (mean ± SD).
Having determined the optimal concentration of Pr78 Gag for a complete shift of the dimeric gRNA, we proceeded to perform competitive band shift assays to determine the differential ability of Pr78 Gag to bind to unspliced gRNA (RCR001) versus spliced env RNA  Figure 2D). This is further supported by the statistically significant difference observed for the percentage of bound RNA between the unspliced gRNA (RCR001) and the spliced env RNA (FN42). These results clearly indicate the preferential binding of Pr78 Gag to unspliced MPMV gRNA rather than to its spliced env RNA.
Pr78 Gag binds redundantly to both the ssPurines and bpPurines.
Next, we investigated where Pr78 Gag binds on the full-length MPMV gRNA. Our earlier genetic and biochemical analyses had identified two structural elements crucial for MPMV gRNA packaging, ssPurines and bpPurines [24,60]. Mutations introduced in these purine-rich regions had shown that both these motifs were responsible for efficient gRNA packaging in a redundant fashion [63]. Thus, competitive band shift assays were performed on a number of mutant clones in these motifs to determine whether binding of Pr78 Gag to these regions could be observed ( Figure 3A).
Using a similar experimental design as described for env RNA, the mutant RNAs were in vitro transcribed, and increasing concentrations of the mutant RNAs (from 0 to 400 nM; Figure   3A) were used as competitors against the radiolabeled unspliced gRNA (RCR001) to assess their ability to displace the radiolabeled unspliced gRNA from the protein-RNA complex.
Experiments for each mutant clone were performed in triplicates, the resultant gels were quantified, and the results depicted as percentage of bound radiolabeled RNA in the protein-RNA complex versus increasing competitor RNA ( Figure 3B & C). The data thus obtained was compared to that obtained from the WT unspliced gRNA (RCR001) to determine the competing (or non-competing) ability of these mutants to bind Pr78 Gag . Mutant FN26, containing a complete deletion of the ssPurines, demonstrated efficient competition against the labeled unspliced gRNA, revealing that this particular mutant RNA was able to efficiently displace the bound protein from the labeled RNA-protein complex ( Figure 3B). The mean percentage of bound RNA displaced by this mutant was within the range of that for the WT RNA (RCR001), indicating the presence of other Pr78 Gag binding site(s) within this mutant that were not affected by the absence of the ssPurines. Given the absence of a statistically significant reduction in Pr78 Gag binding to the FN26 mutant RNA, we proceeded to investigate the effect of protein binding to several bpPurines mutants (FN16, FN19, FN30). Briefly, these clones included deletion of bpPurines (FN16), deletion of the helix/stem of bpPurines only (FN19), and deletion of the complementary sequence of the bpPurines (FN30; Figure 3A). The competitive band shift data for these bpPurine mutants showed a similar pattern of binding as that of FN26 indicating that they competed well against the labeled WT RNA with no statistically significant difference between their percentage of bound RNA ( Figure 3B  Having observed no significant effect on the in vitro binding of Pr78 Gag to these individual ss-and bp-Purines mutants, we next investigated the combined effect of the deletion of both these purine-rich regions on in vitro Pr78 Gag binding. As expected for a redundant role of ssPurines and bpPurines, the competitive band shift data for FN15 (containing deletion of both the ssPurines and bpPurines; Figure 3A) revealed poor competition for Pr78 Gag binding against the labeled WT RNA with a statistically significant difference between the levels of their bound RNA ( Figure 3C).
The in vitro Pr78 Gag binding to these gRNA mutants correlated well with their in vivo packaging, since simultaneous deletion of ssPurines and bpPurines had a dramatic effect on packaging in contrast with deletion of either of these regions alone, which had minimal effects ( Figure 3A; [63]).

MPMV Pr78 Gag binds to the ssPurines, bpPurines, and a single-stranded GU-rich region located immediately downstream of bpPurines
To observe direct binding of Pr78 Gag to MPMV Psi RNA, we combined hSHAPE, a technique that allows structural investigation at single nucleotide resolution following structuredependent modification of the RNA, with footprinting assays. In these experiments, RNA- Thus, the WT MPMV packaging signal RNA (RCR001) was tested in the presence of four molar excess of competitor RNA, which in this case was MPMV spliced env RNA (FN42) and then incubated with either 6 µM of Pr78 Gag or no Pr78 Gag . The protein-RNA complex or RNA alone was then subjected to chemical modification using BzCN, followed by hSHAPE analysis to obtain reactivity data. Experimental triplicates were used to obtain the mean reactivity data for nucleotides in each case and the dataset without Pr78 Gag was used to obtain the secondary structure of the packaging signal RNA. Mean reactivity data from triplicate experiments with protein was applied onto the secondary structure of the MPMV packaging signal obtained in the absence of Pr78 Gag . Changes in reactivity data, with p values ≤ 0.05, between the samples treated with and without Pr78 Gag , were considered significant (Supplemental table 1). Any significant attenuation in reactivities of nucleotides were regarded as Pr78 Gag binding nucleotides or sites.
The secondary structure of the MPMV packaging signal RNA obtained via hSHAPE and footprinting in the absence of protein was identical to the previously published hSHAPE structure [24]; Figure 4A). In the presence of Pr78 Gag , a significant attenuation of reactivities was observed in two major regions of the Psi, including: (1)  In the case of bpPurines, only 2 out of 8 of the nucleotides indicated significant attenuation in reactivities and hence potential Gag binding. These two nucleotides are the unpaired adenine nucleotides (A252 and A253) located at the 3' end of the bpPurines ( Figure   4C-E). Quite unexpectedly, this Pr78 Gag footprinting extended into an immediately adjacent cluster of guanosine and uracil nucleotides and hence is referred to here as the "GU-rich region" 13 ( Figure 1B, 4C & D). This GU-rich region is also located within the region determined earlier to be indispensable for in vivo MPMV RNA packaging [59,60]. It comprises of the sequence G 254 UGUU 258 and along with the two 3' nucleotides of the bpPurines (A252 and A253) is part of a continuous single-stranded loop A 252 AGUGUU 258 ( Figure 4C & D). Our footprinting results show that three nucleotides (G254, G256, and U258) out of these five nucleotides (G 254 UGUU 258 ) showed a significant attenuation in hSHAPE reactivity and hence binding to Pr78 Gag ( Figure 4C-E). In addition, the nucleotide G259 located immediately after the GU-rich region showed a significant attenuation in hSHAPE reactivity ( Figure 4C-E). However, since this nucleotide is involved in a G-C base pair, it is unclear as to whether this is an attenuation due to Pr78 Gag binding or due to conformational changes in the structure of the RNA as a result of Pr78 Gag binding.

Discussion
This study aimed to determine the Gag-binding sites on MPMV gRNA important for the specific incorporation of its genome into the assembling virion. Packaging signals for MPMV has been mapped to the 5 'UTR and the first 120 nts of gag, a region that assumes a complex secondary RNA structure ( Figure 1B; [24,59]). Interestingly, while most of the structural motifs of MPMV Psi RNA contribute to its structural stability, two purine-rich regions, (1) ssPurines, located immediately downstream of the Pal SL (DIS), and (2) the bpPurines, have been proposed to function redundantly as packaging signals [60,63]. Thus, using purified full-length MPMV Gag protein in footprinting assays, our results provide direct evidence for Pr78 Gag  Additionally, as MPMV assembles in the cytoplasm, in vitro studies of its Gag-gRNA interaction are likely to have direct implications for the selection and packaging of the gRNA. Interestingly, some Gag mutations in MPMV can divert its assembly from the cytoplasm to the plasma membrane [70]. Therefore, it would be interesting to determine if re-directing MPMV assembly to the plasma membrane affects Gag-gRNA interactions and thereby RNA packaging.
Competitive band shift assays carried out on ss-and bp-Purines mutants provides a mechanistic rationale for the previously published in vivo packaging data for the mutants harboring the same mutations in these regions. Despite the deletion of either of these regions, a majority of the mutants (FN16, 19, 26, and 30; Figure 3A) were capable of competing well against the labeled WT unspliced gRNA ( Figure 3B & C), revealing the presence of other Pr78 Gag binding sites on these mutant RNAs. As expected, it was only the simultaneous deletion of both the ss-and bp-Purines in mutant FN15 that led to a drastic effect on Pr78 Gag binding ( Figure 3C). Since both these purine-rich regions contribute to the structural stability of the gRNA, their simultaneous deletion in FN15 resulted in the overall loss of the RNA structure [63]. This perhaps is responsible for the poor Pr78 Gag binding observed for this mutant. The redundant nature of Gag binding to purine residues have been implicated in other retroviruses.
For example, in the case of HIV-1 it has been reported that multiple G residues are important during in vivo RNA packaging [69]. When these G residues were substituted individually, they did not show a significant change in gRNA packaging, while substitution of these nucleotides collectively resulted in a drastic reduction in packaging, supporting the redundant role of these G residues during Gag binding and gRNA packaging, even though the structure of these HIV-1 mutants has not been studied [69].  [68,69,74]. Analysis of the Pr78 Gag binding data 20 with the Hill equation indicates that its binding is weakly or not cooperative ( Figure 2C). This is in contrast with the situation prevailing in MMTV, since similar experiments showed a high degree of cooperativity of MMTV Pr77 Gag to its cognate gRNA (Hill coefficient = 3; Chameettachal A. et al, to be published). In HIV-1, while binding of Gag∆p6 to gRNA is cooperative [41], binding of the full-length Pr55 Gag is not cooperative, but there are two classes of Pr55 Gag binding sites with different binding affinities [15]. Irrespective, our data suggest that  Figure 1B). The loss of these LRIs during splicing results in disruption of the native structure of the Psi RNA [61], including the native ssPurines conformation, preventing Pr78 Gag binding to it. As is evident from our footprinting data on the WT gRNA, Pr78 Gag binding to the ssPurines is limited to the nucleotides at the apical end of the loop ( Figure 4B-D), suggesting the requirement of a spatially-accessible loop structure. Secondary structural analysis of the spliced env RNA, performed using hSHAPE, indicates that this spatially-accessible loop is lost and instead attains a partially base-paired conformation in which nucleotides in the ssPurines loop are base paired ( Figure 6). This is a distinct alteration from its native single-stranded form which would otherwise be completely accessible for Pr78 Gag binding. Such a conformational change results in unpairing of 6 of the 9 nucleotides of the ssPurines which showed Gag footprints ( Figure 6). This suggests partial accessibility of Pr78 Gag to the spliced env RNA, which may explain the 30% bound RNA being observed upon competitive band shift assays performed for env spliced RNA ( Figure 2D). Furthermore, since spliced env RNA is not packaged, these results suggest that potential limited Gag binding to spliced env RNA may allow its initial capture from the distinctly non-specific cellular RNAs (Figure 7). However, such inadequate binding may not be enough for Gag multimerization on the spliced env RNA necessary for encapsidation. On the other hand, the full length unspliced RNA, having an additional Gag binding GU-rich region (G 254 UGUU 258 ) downstream of the mSD ( Figure 4B-D), may facilitate Gag multimerization onto the unspliced RNA, resulting in its preferential packaging over the spliced RNA (Figure 7). Thus, the overall conformational change of ssPurines in the spliced env RNA, the restriction exerted by the partially base paired nucleotides for Pr78 Gag binding, as well as the absence of the GU-rich region following splicing may explain the preferential selection of unspliced Psi RNAs over spliced env RNA in MPMV gRNA packaging.

Conclusion
In conclusion, our study demonstrates that Pr78 Gag binds to two single-stranded loops positioned oppositely on the unspliced MPMV Psi RNA consisting of the ssPurines, and the A 252 AGUGUU 258 loop which includes two nucleotides of the bpPurines and an adjacent GU-rich motif (G 254 UGUU 258 ; Figure 7). Based on structural differences, our findings also demonstrate that Pr78 Gag can effectively discriminate between unspliced MPMV Psi RNA from spliced env RNA, revealing how MPMV differentiates between the two RNA substrates. Thus, identification of structural elements crucial in MPMV gRNA packaging should help in understanding not only the mechanism of virion assembly by retroviruses, but also facilitate development of safe and efficient retroviral vectors for human gene therapy.

Nucleotide numbering system
Nucleotide numbers in this study refer to the MPMV genome with the Genbank accession number M12349.1 [76].

Expression and purification of Pr78 Gag
Pr78 Gag was expressed with a C-terminal hexa-histidine (His 6 ) tag and purified via immobilized metal affinity chromatography (IMAC) followed by size exclusion chromatography 23 (SEC). The purified protein was characterized using western blot. Methodology employed to express and purify the protein has already been described [62].

Physical characterization of Pr78 Gag by dynamic light scattering (DLS)
Prior to the in vitro assays, purified Pr78 Gag was characterized by DLS using a DynaPro Nanostar (100 mW He-Ne laser; Wyatt Technologies,) in a 1-µl quartz cuvette (JC-006, Wyatt Technologies) at 20°C as previously described [9]. By assimilating the protein in solution to spheres, the diffusion coefficients (D) were correlated to the hydrodynamic radius (R h ) of the molecules in solution by the Stokes-Einstein equation: In this equation, k represents the Boltzmann constant, while T represents the absolute temperature, and µ is the viscosity of the solvent. Before sample acquisition, the buffer was filtered through 0.02 µm filters (Millex ®) and the offset of the solvent was measured for subsequent sample data treatment.

Plasmid construction for spliced env, ss-and bp-Purines mutant RNA production
The wild type (WT) plasmid (RCR001; Figure 2A) was used for the in vitro transcription of the MPMV unspliced full-length Psi RNA, as previously described [24]. Plasmids for the in vitro transcription of the MPMV spliced env RNA (FN42; Figure 2A) and the ss-and bp-Purines mutants were created using spliced overlap extension (SOE) PCR, as previously described ( Figure 3A; [24,27,61,77] table 3), on the MPMV full-length molecular clone KAL01 as the template [53].
A second round of amplification was carried out using the products of PCRs A & B with primers OTR 1004 and OTR 1379. The resulting product was cleaved with HindIII and XmaI and ligated into the similarly-digested pUC-based vector, pIC19R [24,27,60,61,63]. The sequence of FN42 was then confirmed via sequencing (Macrogen, South Korea) and subsequently used for in vitro transcription. Mutations in the ss-and bp-Purines present in the MPMV packaging determinants were also introduced employing the same strategy, but using SJ2 as the template along with primers listed in Supplemental table 3.

In vitro transcription and purification of unlabeled and [α-32 P]-labeled RNA.
All plasmids containing the WT (RCR001) and mutant MPMV packaging sequences under the influence of the T7 promoter were linearized by cleaving the plasmid DNAs with SmaI and used for in vitro transcription (MEGAscript™ T7 Transcription Kit, Thermo Fisher Scientific). Briefly, the linearized plasmids were incubated at 37ºC for 4 hours in the presence of NTPs, 10X reaction buffer and T7 RNA polymerase, followed by a 15-minute incubation at the same temperature with 1µl of TURBO DNase (2U/ µl). Quality of the in vitro transcribed RNA was determined by testing 2µl of the product on 8%, denaturing (8M urea) polyacrylamide gels following which the remainder of the product was ethanol precipitated overnight at -20 ºC. The precipitated RNAs were then purified by gel filtration chromatography using TSK Gel G2000SW columns (Tosoh Bioscience) in 0.2 M sodium acetate (pH 6.5) and 1% (v/v) methanol, as previously described [9,[78][79][80][81][82]. Finally, RNAs collected from appropriate fractions were pooled, ethanol precipitated, and examined for purity and integrity using 8M urea denaturing 8% polyacrylamide gels.
In vitro transcription in the presence of [α-32 P]-ATP was performed to prepare internallylabeled RNAs, as described previously [9,[79][80][81][82]. Following in vitro transcription, the RNA samples were DNase treated, and the labeled RNAs purified by electrophoresis on a 8% polyacrylamide gel under 8M urea denaturing conditions. RNAs from the bands were excised and extracted in 300 μl of buffer containing 500 mM ammonium acetate, 1 mM EDTA and 0.1% SDS overnight at 4ºC, followed by ethanol precipitation, and resuspension in 10 μl Milli-Q water.

Band-shift and competitive band-shift assays
For band-shift assays, the radiolabeled unspliced, full-length WT (RCR001) RNA (50,000 cpm) and yeast tRNA (2 g) were denatured at 90C (2 minutes) followed by chilling on ice (2 minutes). The denatured RNAs were re-folded in 1X dimer buffer (30 mM Tris pH 8.0, 300 mM NaCl, 5 mM MgCl 2 , 5 units of RNase inhibitor (RNasin, Promega), 0.01% Triton-X 100, in a total volume 10 l) at 37C for 30 min. Next, Pr78 Gag (diluted in 30 mM Tris (pH 8.0), 300 mM NaCl, 5 mM MgCl 2 , 10 mM DTT, and 0.02 mg/ml BSA in a final volume of 10 l) in increasing concentrations (0 to 2000 nM) was mixed with the refolded RNA. The mixture was incubated at 37C for 30 minutes to allow for binding, followed by incubation on ice for 30 minutes. Samples were separated on 1% agarose gels using TBM buffer (0.5X Tris-Borate, 0.1 mM MgCl 2 ) at 150 V for 4 hr at 4C. The gels were then fixed in 10% trichloroacetic acid 26 (TCA) for 10 minutes, dried under vacuum, and analyzed using a FLA 5000 (Fuji) scanner.
Bands on the gels were quantified using ImageQuant software. The experimental data were fit with Hill's equation shown below using the GraphPad Prism 5 software.
In the Hill's equation, B max represents the maximum specific binding while h is the Hill's coefficient.
For competitive band-shift assays, 50,000 cpm of radiolabeled, unspliced, full-length WT (RCR001) RNA were denatured and refolded in the presence of either spliced env RNA (FN42) or increasing concentrations of unlabeled competitor RNAs (0 to 400 nM), as described above.
The refolded RNAs were then mixed with Pr78 Gag at a final concentration of 500 nM in a volume of 10 l, and incubated at 37C for 30 min to allow for binding and then on ice for 30 min for stabilization. The reaction mixtures were then electrophoresed on 1% agarose gels using TBM buffer (0.5X Tris-Borate, 0.1 mM MgCl 2 ) at 150 V for 4 hr at 4C, fixed in 10% trichloroacetic acid (TCA) and quantified as described above. The percentage of bound RNA were calculated from triplicate gels in each case and the statistical significance between the WT MPMV RNA (RCR001) and the spliced env RNA (FN42) and competitor RNAs were determined using a paired two tailed t-test.

RNA footprinting and high-throughput selective 2'-hydroxyl acylation analyzed by primer extension (hSHAPE)
Biochemical analysis of the structures of the mutants was performed using the hSHAPE methodology that allows structural investigation at each nucleotide following structuredependent modification of the RNA [9,24,27,61,63,75]. hSHAPE experiments to biochemically validate the predicted secondary structure of the spliced env RNA were performed using the same methodology as explained above, except in the absence of any protein. RNA was subjected to modification with appropriate controls and followed by reverse transcription using two sets of primers labelled with both VIC and NED (OTR_312-322 and OTR_497-518; Supplemental table 3). QuShape analysis was performed as above and mean reactivity data from three independent experiments were obtained for secondary structure analysis (Supplemental table 4).

Conflict of Interest
The authors declare no conflict of interest.         (1) the ssPurines loop (U 191 UAAAAGUGAAAGUAA 206 ), located upstream the mSD and hence found in all viral RNAs, and (2) the single-stranded A 252 AGUGUU 258 loop, which corresponds to the last two purines of the bpPurines and extends into a GU-rich region. The latter is located downstream of the mSD; thus, it is present only in the genomic and not the spliced RNA. Both these regions act redundantly during the gRNA packaging process. We propose that the MPMV Pr78 Gag may first bind to the ssPurines loop to distinguish viral RNAs from cellular mRNAs. This is probably followed by binding to the A 252 AGUGUU 258 loop, multimerizing Gag binding onto the gRNA in a synergistic manner to further enhance the ability to specifically capture the unspliced RNA, containing both single-stranded purine loops, over the spliced RNAs, which 35 lacks the single-stranded A 252 AGUGUU 258 loop. This ensures that the MPMV gRNA is packaged preferentially into the viral particles, excluding cellular and spliced RNAs. Number versus size distribution. The blue, red and green peaks represent three independent Pr78 Gag samples tested in binding buffer.