key: cord-0256791-1z074bs4 authors: Lyonnais, Sébastien; Sadiq, S. Kashif; Lorca-Oró, Cristina; Dufau, Laure; Nieto-Marquez, Sara; Escriba, Tuixent; Gabrielli, Natalia; Tan, Xiao; Ouizougun-Oubari, Mohamed; Okoronkwo, Josephine; Reboud-Ravaux, Michèle; Gatell, José Maria; Marquet, Roland; Paillart, Jean-Christophe; Meyerhans, Andreas; Tisné, Carine; Gorelick, Robert J.; Mirambeau, Gilles title: The HIV-1 ribonucleoprotein dynamically regulates its condensate behavior and drives acceleration of protease activity through membraneless granular phase separation date: 2021-10-26 journal: bioRxiv DOI: 10.1101/528638 sha: 8bd970c9525370b33e328d45c1ab1e7d74b92a0c doc_id: 256791 cord_uid: 1z074bs4 A growing number of studies indicate that mRNAs and long ncRNAs can affect protein populations by assembling dynamic ribonucleoprotein (RNP) granules. These phase separated molecular ‘sponges’, stabilized by quinary (transient and weak) interactions, control proteins involved in numerous biological functions. Retroviruses such as HIV-1 form by self-assembly when their genomic RNA (gRNA) traps Gag and GagPol polyprotein precursors. Infectivity requires extracellular budding of the particle followed by maturation, an ordered processing of ~2400 Gag and ~120 GagPol by viral protease (PR). This leads to a condensed gRNA-NCp7 nucleocapsid and a CAp24-self-assembled capsid surrounding the RNP. The choreography by which all of these components dynamically interact during virus maturation is one of the missing milestones to fully depict the HIV life cycle. Here, we describe how HIV-1 has evolved a dynamic RNP granule with successive weak-strong-moderate quinary NC-gRNA networks during the sequential processing of the GagNC domain. We also reveal two palindromic RNA-binding triads on NC, KxxFxxQ and QxxFxxK, that provide quinary NC-gRNA interactions. Consequently, the nucleocapsid complex appears properly aggregated for capsid reassembly and reverse transcription, mandatory processes for viral infectivity. We show that PR is sequestered within this RNP and drives its maturation/condensation within minutes, this process being most effective at the end of budding. We anticipate such findings will stimulate further investigations of quinary interactions and emergent mechanisms in crowded environments throughout the wide and growing array of RNP granules. Biomolecular condensates (BCs) are membraneless, intracellular assemblies formed by 2 the phenomenon of liquid-liquid phase separation (LLPS) [1] [2] [3] [4] [5] . Several types of such 3 assemblies have been observed inside eukaryotes with a variety of suggested functions. 4 These range from adaptive cellular response to physiological stresses via formation of 5 stress granules [6] [7] [8] [9] , to meeting the demands of intracellular transport or signalling, 6 amongst many other functions [3] . They have also importantly been linked to disease 7 [10, 11] . Fundamentally, due to their capacity to concentrate biomolecules, a suggested 8 principal function of BCs has been that they regulate enzyme biochemistry [12] [13] [14] [15] [16] . 9 Many condensates sequester mRNAs and associated RNA-binding proteins into what 10 are termed RNA granules [17] [18] [19] [20] [21] [22] [23] [24] . The material properties of such granules can vary 11 depending on composition and biological functionality [25] -from dynamic architectures 12 with liquid-like phases to non-dynamic gel-like phases [26] . Phase transitions between 13 liquid-to gel-like phases due to condensate aging have also been observed [27] . 14 The concept of quinary interactions [28, 29] -the emergent sum of many weak inter- 15 actions that may occur in a crowded biomolecular environment -has been suggested to 16 promote the assembly of highly stable, but dynamic and transient multi-macromolecular 17 complexes without any requirement for membrane compartmentalisation [30] [31] [32] [33] [34] . Com- 18 patible with this concept, multivalent molecules that enable assembly of dense networks 19 of weak interactions are emerging as major molecular drivers that underpin the forma-20 tion of BCs [35] [36] [37] [38] . In particular, cooperation between long polymers, such as RNAs, 21 together with folded proteins and intriniscally disordered proteins (IDPs) may be an 22 essential feature of many condensates [3, 39, 40] . Furthermore, constituent binding affin- 23 ity, valency, liquid network connectivity and critical post-translational modifications all 24 play a role in regulating BCs [41] [42] [43] [44] [45] [46] [47] [48] . 25 Recently, constituents of RNA-containing viruses, such as HIV-1 and SARS-CoV-2, 26 have been shown to phase separate into biomolecular condensates inside cells [49] , using 27 their repertoire of IDPs [50] in conjunction with the RNA-binding capacity of their 28 nucleocapsid proteins to interact with genomic RNA elements [51] [52] [53] [54] [55] [56] . 29 Even though an HIV-1 particle is derived from the self-assembled Pr55Gag shell 30 and is ultimately enveloped by a lipid membrane, the concept of quinary interactions 31 is clearly applicable in describing its dynamic assembly at the mesoscopic scale -since 32 it forms a confined RNP gel phase in a highly crowded space, within a limited time 33 frame and in a cooperative manner. Pr55Gag is composed from N-to C-termini of ma-34 trix (MAp17), capsid (CAp24), spacer peptide SP1, nucleocapsid (NC), spacer peptide 35 SP2 and p6 protein. Key players here consist of NC protein intermediates with their 36 variable nucleic acid (NA) binding properties that are dependent upon their process- 37 ing state [57] [58] [59] [60] . Tethered within the virion by approximately 2400 GagNC domains, 38 the two single strands of 9.2 kb-long gRNA specifically scaffold Pr55Gag self-assembly. 39 Subsequently, the HIV-1 RNP complex engages a granular condensation during the se-40 quential proteolysis of the Pr55Gag RNA-binding domain into the mature nucleocapsid 41 protein (NCp7) by the viral protease (PR) [59, 61, 62] . PR is derived by autoprocessing 42 of a smaller number of GagPol within the Pr55Gag assembly that additionally contain 43 reverse transcriptase (RT) and integrase (IN) . Approximately 60 PR homodimers are 44 potentially available to catalyze maturation, which principally requires 12000 cleavage 45 events. Cleavage of GagNC by PR generates first NCp15 (NCp7-SP2-p6) bound to, and 46 forming with the gRNA an RNP intermediate that physically detaches from the remain- 47 ing outer MA-CA-SP1 shell. The second cleavage between SP2 and p6 releases NCp9 48 (NCp7-SP2) (Figure 1a ). Single-stranded nucleic acids (ssNA) stimulate both cleavage 49 events in vitro [58, 63, 64] . The third cleavage produces the mature 55 amino acids (aa)- 50 long NCp7 and SP2. Within the virus, NCp15 seems to condense gRNA less well than 51 NCp9 and NCp7 [65] . Yet NCp9 does not appear as functional as NCp7 [66] . NCp15 52 and NCp9 are short-lived species not detected during typical virus production [60] . Why 53 such intermediates are maintained along the HIV-1 maturation process remains unclear. 54 HIV-1 PR is an aspartyl-protease, enzymatically active only as a homodimer. Re-55 combinant PR is stabilized in vitro by high ionic strength (>1 M NaCl) and catalysis 56 is strongly activated under acidic conditions (pH 5.0 or even lower). Lower salt (0.1M 57 NaCl) and increasing the pH to 6.0 limits the acidic catalysis and shifts the equilib-58 rium towards the monomer [67] . At quasi-neutral pH, in low salts and an excess of PR, 59 the in vitro cleavage of Gag follows the sequential mechanism described above leading 60 to NCp7 and the condensed RNP [68] . RNA or ssDNA promote NCp15 cleavage in 61 vitro [58, 69] , while recent reports have shown that direct RNA-PR contacts enhance 62 the enzyme activity [64] . Consequently, PR appears to engage in an intricate partner-63 ship with NC and gRNA during viral maturation that remains incompletely understood. 64 HIV-1 NCp7 contains a small globular domain formed with two zinc fingers (ZFs) that 65 generate a hydrophobic pocket with two aromatic residues (Phe16 and Trp37). This 66 platform stacks with unpaired nucleotides, preferentially guanosines exposed in RNA or 67 ssDNA secondary structures, while basic residues stabilize the complex through electro-68 static interactions with the NA backbone. With particular stem-loops in gRNA or DNA, 69 this results in the formation of specific complexes [70, 71] . NCp7 is also a highly mobile 70 and flexible polycationic condensing agent; like polyamines, transient protein:NA elec-71 trostatic contacts neutralize phosphate backbone repulsions lowering the overall energy 72 of the RNP complex [59, 72, 73] . 73 The binding properties of the various maturation states of the nucleocapsid protein 74 to nucleic acids vary [74] [75] [76] . In vitro, these properties induce a massive co-aggregation 75 of recombinant NCp7 and NCp9 with NA templates [57, 60, 77] . This quinary interaction 76 capability guides the matchmaking/NA chaperone activity by facilitating intra-and in-77 termolecular RNA-RNA interactions required for functional gRNA folding [78] . Such 78 crowding effects rely on basic residues particularly concentrated in the two small flexi-79 ble domains, the (1-14) N-terminal domain and the (29) (30) (31) (32) (33) (34) (35) linker between the ZFs [72] . 80 NCp15 shows slightly different NA binding and chaperone properties but is essentially 81 characterized by a reduced ability to aggregate NA [57, 60, 79] , properties recently cor-82 related with a direct fold-back contact between the p6 and ZF domains [60] . NCp9 83 shows an enhanced NA affinity due to a slower dissociation rate, as well as dramatically 84 enhanced NA aggregating activities [57, 60, 73] . Alanine substitution of acidic residues 85 in p6 converts NCp15 to a NA-aggregating protein similar to NCp9, while addition of a 86 p6 peptide lowers the RNA chaperone activity of NCp7 in vitro [60] . This suggests that 87 SP2 contains an additional NA-interaction domain, which may be masked or modulated 88 with another NCp7 domain by intra-or intermolecular protein contacts between p6 and 89 the NC domain. 90 HIV-1 maturation is mandatory for viral dissemination following sequential processes 91 of protein and RNA self-assembly, coordinated in space and time by the enzymatic 92 activity of viral PR [61, 62, 80] . The slow in vitro kinetics of Gag proteolysis supports 93 a general scheme for PR to be auto-processed during the completion of budding thus 94 driving viral maturation within free, released particles in a computed time-scale close to 95 30 min [81] . This model is, however, inconsistent with many observations from electron 96 microscopy that shows i) a huge majority of free but freshly released particles in a mature 97 form containing condensed RNP [82] , ii) both capsid and budding defects in presence 98 of PR inhibitors [83] , and iii) budding and maturation defects for critical NC mutants, 99 whereas Western blots from cell extracts detects PR-processed Gag products [82] . Such 100 findings suggest a much closer overlap between budding and maturation than generally 101 supposed. Importantly, suppressing both PR cleavage sites in NCp15 abolishes viral 102 infectivity [65, 84] and results in an abnormal virion core morphology [65] . In contrast, 103 suppression of the NCp7-SP2 cleavage site shows little effect on virus morphology and 104 infectivity in single-cycle assays, but reverts to WT (e.g. containing NCp7) after several 105 rounds of infection [84] . A "roadblock" mechanism affecting RT activity on a NA 106 template has been shown to be imparted by NCp9 as well as by NCp15, which could limit 107 large-scale viral replication, highlighting NCp7 as the optimized cofactor for accurate 108 RNP folding and viral fitness [66] . 109 The present study highlights how HIV-1 gRNA becomes condensed by NC proteins 110 through the action of the RNP-sequestered PR enzyme. Reconstituted systems that 111 model non sequence-specific binding on a large scale allowed us i) to detail the quinary 112 effects and their variations engaged in this dynamic process as well as ii) to focus on 113 PR action in such a quinary interaction context. 114 Proteins, Nucleic Acids and Reagents 116 Proteins. The HIV-1 NC proteins and proviral plasmids were based on the pNL4-3 117 sequence (GenBank accession number AF324493). Recombinant wild-type and mu-118 tants of NCp7, NCp9 and NCp15, respectively 55, 71 and 123 amino acids in length, 119 were expressed and purified as described [60, [85] [86] [87] . The CA-SP1-NC-SP2-p6 pro-120 tein expression construct was generated by PCR amplifying pNL4-3 using Gag∆MA 121 sense primer 5'-GAT CTG GGT ACC GAG AAC CTC TAC TTC CAG ATG ATA 122 GTG CAG AAC, NL43 OCH antisense primer 5'-GCT TGA ATT CTT ATT GTG 123 ACG AGG GGT CGC TGC and cloning the resulting product into the homologous 124 KpnI and EcoRI sites of pET32a (Novagen, Madison, WI). Expression construct for 125 NCp15(1) (expressing NCp15 that can be cleaved only to NCp9) and NCp15(2) (ex-126 pressing uncleavable NCp15) were generated starting with the NC-SP2-and NCp15-127 containing proviral plasmids of Coren et al. [84] , respectively. The two constructs were 128 generated by PCR amplifying the appropriate plasmids with NL4-3 NC sense primer 129 5'-CGT GGG ATC CTT AGA GAA CCT CTA CTT CCA GAT ACA GAA AGG 130 CAA TTT TAG, NL4-3 NCp15 antisense primer 5'-GTA CGT GTC GAC TCT CTA 131 ATT ATT GTG ACG AGG GGT CGC T and cloning into the homologous BamHI and 132 SalI sites of pET32a. Site-directed mutagenesis of the wild-type NL4-3 NCp7 construct 133 to generate the K3A/F6A/Q9A mutant was performed using the Agilent QuickChange 134 Site-Directed Mutagenesis kit, with verification by NA sequence analysis, for the gen-135 eration of the recombinant expression plasmid, as described [88] . The K3A mutation 136 results from changes to nucleotides 1927 through 1929 from AAA to GCC, F6A results 137 from nucleotides 1936 and 1937 being changed from TT to GC, and Q9A results from 138 nucleotides 1945 and 1946 being changed from CA to GC. Proteins were expressed and 139 purified as described [60, [85] [86] [87] . Proteins were stored lyophilized and then suspended 140 at a concentration of 1 mg/mL in a buffer containing 20 mM HEPES pH7.5, 50 mM 141 sodium acetate, 3 mM DTT, 20% (v/v) ethylene glycol, 200 µM ZnCl2 and stored at 142 -20 • C. The concentrations were determined by measuring the UV absorbance at 280 nm 143 using the following extinction coefficients: NCp7 and NCp7 mutants: 5690 M −1 cm −1 ; 144 NCp9: 11,380 M −1 cm −1 ; NCp15: 12,660 M −1 cm −1 . HIV-1 PR was expressed in 145 E. coli Rosetta(DE3)pLysS strain (Novagen) as inclusion bodies using the expression 146 vector pET-9 and purified as described [89, 90] . The PR domain used here bears the 147 Q7K/L33I/L63I and C67A/C95A protective mutations to respectively minimize auto-148 proteolysis [67] and prevent cysteine-thiol oxidation [91] . PR was suspended, adjusted 149 to 10-20 µM concentration and stored at -80 • C in 50 mM sodium acetate pH5.5, 100mM 150 NaCl, 1 mM DTT, 0.1 mM EDTA, 10% (v/v) glycerol. Nucleic Acids. The circular 7,249 nt M13 ssDNA (m13mp18) was purchased from 152 Bayou Biolabs, the 3,569 nt MS2 RNA from Roche GmBh. Linear m13mp18 molecules 153 were generated by annealing a complementary oligonucleotide to form a restriction site 154 for BsrB I (NEB) as described [92] . The oligonucleotides poly d(A)13 and TAR-RNA (27 155 nt, 5'-CCAGAUCUGAGCCUGGGAGCUCUCUGG-3'), were purchased from Sigma-156 Aldrich, the short RNA fragments corresponding to individual stem-loop motifs of the 157 Psi region: SL1 (17 nt), SL2 (23 nt), SL3 (14 nt) and SL4 (24 nt) were purchased (Mi-158 crosynth) and purified by HPLC (ÄKTA design-Unicorn) on a PA-100 anion exchange 159 column (Dionex). Plasmids used for in vitro transcription of HIV-1 RNAs used in this 160 study have been described previously [93, 94] . Briefly, the pJCB vector was linearized 161 with AflII, XbaI, BssHII, RsaI, or PvuII, and used as templates for the synthesis of 162 RNAs 1-61, 1-152, 1-278, 1-311 and 1-615, respectively, by in vitro run off transcription 163 using bacteriophage T7 RNA polymerase, followed by purification using size exclusion 164 chromatography as described previously [95] . Likewise, plasmid pmCG67 was linearized 165 with AvaII or SalI to produce RNAs 1-1333 and 1-4001, respectively. RNA 1-102 was 166 obtained from a PCR product corresponding to the HIV-1 MAL sequence. AFM imaging 188 NC:NA complexes were assembled under conditions used for EMSA with 1 ng/µl of 189 M13 ssDNA and in a binding solution containing 10 mM TrisAcetate pH 7.0, 50 mM 190 sodium acetate, 2.5 to 5 mM magnesium diacetate and 0.5 mM TCEP. A freshly cleaved 191 muscovite mica surface was pre-treated for 2 min with a fresh dilution of spermidine 192 (50 µM), extensively rinsed with water and dried under a nitrogen flow [96] . A 5 µL-193 drop of the NP complexes was deposited on the surface and incubated for 3-5 min 194 and dried with nitrogen gas. AFM images were carried out in air with a multimode 195 scanning probe microscope (Bruker) operating with a Nanoscope IIIa or V controller 196 (Bruker) and silicon AC160TS cantilevers (Olympus) using the tapping mode at their 197 resonant frequency. The scan frequency was typically 1.0 Hz per line and the modulation 198 amplitude was a few nanometers. A second order polynomial function was used to 199 remove the background with the AFM software. Proteolysis assays 201 NC cleavage and SDS-PAGE analysis. A proteolysis assay of NCp15 bound to 202 ssDNA using recombinant PR was showed previously [58] . The assay was optimized in 203 this study to ensure a detailed analysis of the reaction using SDS-PAGE electrophoresis 204 (Supplementary Fig. 3a -b). Peptides were quantified by fluorescent staining, which 205 allowed accurate measurements in the 25-500 ng range, in agreement with our NP com-206 plex analysis. The standard proteolysis assay contained NCp proteins (6 µM) incubated 207 with NA for 5 min at 37 • C in 10 µl of a PR buffer (MES 50 mM/Tris variable to adjust 208 pH, NaCl 100 mM, DTT 2 mM, BSA 50 µg/ml). Next, PR was added, unless other-209 wise indicated, at a concentration of 600 nM. Reactions were stopped by addition of 210 a SDS-PAGE loading buffer and heat denaturation (5 min at 95 • C), followed by 1 h 211 incubation at 37 • C in presence of 300 mM Iodoacetamide, which prevented protein oxi-212 dation ( Supplementary Fig. 3a ). Samples were separated on 20% acrylamide gels using 213 Tris-Tricine SDS-PAGE in a Hoeffer MiniVE system. After migration at 160 V for 2.5 214 hr, the gels were fixed by 40% ethanol/10% acetic acid for 1 hr and stained overnight in 215 200 mL of Krypton Fluorescent gel stain (Life Technologies) diluted 1/10 in water. Gels 216 were then rinsed with 5% acetic acid and incubated in milliQ water for 30 min before 217 scanning with a Typhoon 8600 imager. Fluorescence counts were quantified using the 218 ImageQuant software (GE Healthcare). Apparent Vmax was measured by dividing the 219 product concentration by the time of incubation with [product]/ [S0] product ratio less 220 than 30%. The PR cleavage assay of Fig. 1e -f was performed by incubating NCp15 (750 221 nM) and M13 ssDNA (1nM) in MES 50mM/Tris pH6.25, 100 mM NaCl, 4 mM MgCl2, 222 2 mM DTT for 15 min at 37 • C in 50 µL. PR (35 nM) was added and the cleavage was 223 carried out at the indicated times. Each reaction was stopped by chilling the tubes on 224 ice while a 5 µl-drop was used to prepare mica for AFM, 15 µl were loaded on the gel 225 for EMSA and the remaining 30 µl were used for SDS-PAGE after treating the samples 226 as previously indicated. All experiments were performed at least in triplicate. FRET assay. The proteolytic activities of PR were determined using the principles 228 of Förster resonance energy transfer (FRET) by cleavage of a fluorogenic peptide sub-229 strate DABCYL-γ-abu-Ser-Gln-Asn-Tyr-Pro-Ile-Val-Gln-EDANS (Bachem, Germany) 230 with DABCYL, 4-(4 ′ -di-methylaminophenylazo)benzoyl; γ-abu, γ-aminobutyric acid; 231 EDANS, 5-[(2-aminoethyl)amino] naphthalene-1-sulfonic acid. Incubation of PR with 232 the probe resulted in specific cleavage at the Tyr-Pro bond and a time-dependent in-233 crease in fluorescence intensity that is linearly related to the extent of substrate hy-234 drolysis. Kinetic experiments were carried out at 30 • C in 150 µL of the PR buffer (50 235 mM MES-Tris combination, 0.1 -1 M NaCl, pH adjusted between 5 and 7), 5.2 µM 236 of the probe and 10-50 nM of PR. The probe was first dissolved in DMSO. The final 237 DMSO concentration was kept at 3% (v/v). Fluorescence intensities were measured in 238 a BMG Fluostar microplate reader. Delay time for the reaction start was calculated as 239 the reaction slope intercept with the x axis. All experiments were performed at least in 240 triplicate. Electron microscopy of HIV-1 particles 242 Maturation mutants. Mutant virions accumulating NCp15 or NCp9 were produced 243 by transfection of mutated pNL4-3 proviral plasmids as described [84] . Plasmids were 244 transfected into HEK 293T cells using Mirus TransIT 293 (Mirus Bio LLC, Madison, 245 WI) according to the manufacturer's instructions. 48 h culture supernatants were clar-246 ified and virions ultracentrifuged and examined by electron microscopy as described 247 previously [97, 98] . At least 180 particles were analysed on the criteria that they were 248 enveloped and a contrast was visible inside. Then the subpopulation of the diffuse cores 249 instead of thin, dark spots was scored with, as discriminating criteria, a diameter equal 250 or larger than 70% of the internal diameter of the particle. Viral particles produced from latently infected cells. Briefly, the latently 252 infected ACH2 cells [99] were grown under standard conditions, were plated onto 10 253 cm cell culture dishes at densities of 4x106 cells and incubated with or without PR in-254 hibitor (10 µM Lopinavir, Sigma). HIV production was activated by adding Vorinostat 255 (10 µM; Sigma). After 2 days, ACH2 cells were fixed with 2.5% glutaraldehyde, washed, 256 dehydrated, embedded in epoxy resin according to standard procedures [100] . Electron 257 microscopy images were obtained with a Tecnai Spirit microscope coupled with a 1376 258 x 1024 pixel CCD camera (FEI, Eindhoven, The Netherlands). We analysed 500 par-259 ticles attached to the membranes after normal production and 120 after production in 260 presence of Lopinavir (respectively 46.1 and 53% of the total number of detectable par-261 ticles). Within each attached population, mainly 91% particles were identifiable, 89.6% 262 containing a dark spot compared to only 1.4% immature for the normal population, 263 while dark spots were not visible after viral production in presence of Lopinavir. Molecular dynamics simulations followed a previously well-established protocol [101] . 266 An initial structure was prepared for the NC-SP2 (RQAN-FLGK) octapeptide ligand in 267 apo-form. Atomic coordinates for the octapeptide were extracted from the 1TSU crystal 268 structure [101, 102] . The standard AMBER forcefield (ff03) [103] was used to describe all 269 parameters. The system was solvated using atomistic TIP3P water and then electrically 270 neutralized with an ionic concentration of 0.15 M, resulting in a fully atomistic, explicit 271 solvent system of approximately 14,000 atoms. Conjugate-gradient minimization was 272 performed. The SHAKE algorithm was employed on all atoms covalently bonded to 273 a hydrogen atom. The long range Coulomb interaction was handled using a GPU 274 implementation of the particle mesh Ewald summation method (PME). A non-bonded 275 cut-off distance of 9Åwas used with a switching distance of 7.5Å. During equilibration 276 the position of all heavy peptide atoms was restrained by a 0.5 kcal/mol/Å 2 spring 277 constant for all heavy protein atoms and the system evolved for 10 ns with a timestep 278 of 4 fs. The temperature was maintained at 300 K using a Langevin thermostat with a 279 low damping constant of 0.1/ps and the pressure maintained at 1 atm for both systems. 280 The system was then equilibrated for 10 ns of unrestrained simulation in the canonical 281 ensemble (NVT) with an integration timestep of 4 fs. The final coordinates were used 282 as input for production simulations. All subsequent simulations were carried out in the 283 NVT ensemble. All production simulations were carried out using ACEMD [104] . An 284 ensemble of 10 x 1 µs production simulations was performed. Coordinate snapshots 285 from production simulations were generated every 10 ps, resulting in a ensemble of 106 286 conformers for analysis. The octapeptide was relabelled as R52-Q53-A54-N55-F56-L57-G58-K59. The con-288 former ensemble was analyzed in a reaction coordinate space consisting of two order 289 parameters: the K59-Q53 C α distance (d KQ ) and K59-F56-Q53 C α angle (θ F QK ). The 290 potentials of mean force (PMF) was calculated by binning the ensemble data into mi-291 crostates corresponding to the given reaction coordinate space and then calculating the 292 mole fraction (ρ) of each microstate using PMF = -kBTln(ρ), where kB is the Boltz-293 mann constant and T the temperature. Corresponding order parameters were calculated 294 for each of the NMR conformers in PDB 1F6U of the NC N-terminus where d KQ was 295 the K3-Q10 C α distance and the θ F QK K3-F6-Q9 C α angle. Conformers were then 296 aligned to the NC N-terminus from PDB 1F6U by the C α atoms of K59-R52 mapped 297 to K3-R10 of 1F6U. The C α RMSD was then calculated as a third order parameter 298 (d n ), its probability density was determined by binning (ρ(d n )) and conformers within 299 the thresholds of 1Å, 1.5Å and 2Å extracted and mapped back to the d KQθ F QK 300 reaction coordinate space. Cleavage of NCp15 to NCp9 and NCp7 underpins Weak-Strong-303 Moderate quinary condensate properties 304 We first focused on the quinary interactions and the architectural behavior of NC:NA 305 complexes by a combination of electrophoretic mobility shift assay in agarose gels 306 (EMSA), atomic force microscopy (AFM) and dynamic light scattering (DLS). Examin-307 ing RNPs with large ssNA templates under increasingly dilute conditions interestingly 308 switched NCp7 binding from NA aggregation (quinary interactions) to intramolecularly-309 folded NP condensates (Figure 1b) . NCp7 binding titrations on a circular M13 ssDNA 310 showed a progressive process of ssDNA migration acceleration in a gel (Figure 1c) , 311 seen by AFM as tightly compact NP structures formed of folded DNA strands coated 312 and bridged with protein ( Figure 1d ). Maximum ssDNA compaction was reached for 313 saturating amounts of one NCp7 over 8-10nt [77] . Additional protein resulted in the fu-314 sion of these NP condensates into very high molecular weight structures that exhibited 315 smearing during electrophoresis. AFM showed a progressive accumulation of protein clusters covering the lattices 317 while the branched and secondary structures of the ssDNA appeared melted or absent, 318 and rather bridged into nucleofilament-like structures ( Supplementary Fig.2 ). Omis-319 sion of magnesium in the buffer (Figure 2g -i) or an excess of NCp7 resulted in fusion 320 of the individual condensates into huge macrostructures with a spheroid shape com-321 parable with previously described NC:NA aggregates [57, 58, 73] . NCp7 mobility was 322 deemed necessary since this fusion was not observed and condensation was delayed at 323 4 • C [79, 86] (Supplementary Fig 1a) . The kinetics of the reaction indicated fast in-324 tramolecular condensation and a slow process of NP condensate fusion (Supplementary 325 Fig. 1b) . Low monovalent salt concentration increased NCp7/ssDNA aggregation and a 326 strong electrostatic competition was observed with Na+ or Mg2+, as expected [73, 105] 327 (Supplementary Fig. 1c) . Mutations of key aromatic residues, Phe16 and Trp37 (Sup-328 plementary Fig. 1d -e) demonstrated that ssDNA condensation not only depend on 329 phosphate backbone neutralization but also on base capture by the ZF domain. The 330 apo-protein SSHS NC mutant [85] promoted DNA aggregation without acceleration of 331 DNA mobility as expected for polycation-induced NA aggregation [106] . Finally, Ala 332 substitution of basic residues in the N-terminal domain and the linker demonstrated 333 these residues to be essential for ssDNA condensation, as expected. Strand circularity, 334 e.g. the ssDNA topological constraint, favored intramolecular ssDNA bridging, whereas 335 intermolecular ssDNA-NC-ssDNA interactions were enhanced with linear M13 ssDNA 336 or MS2 RNA ( Supplementary Fig. 1f) , which demonstrated protein-NA networks involv-337 ing NA-protein-NA and protein-NA-protein interactions, as proposed previously [107] . 338 Binding of NCp9 yielded fast-migrating NP condensates for the lowest protein con-339 centrations (Figure 1c) , indistinguishable by AFM from those formed with NCp7 (Sup-340 plementary Fig. 2f) . However, NCp9-driven NA condensation was seen dramatically 341 associated with a huge fusion process by EMSA (Figure 1c) , DLS (Figure 2i ) and AFM 342 (Figure 1d, Supplementary Fig. 2f-g) . Linearity of the ssDNA template resulted in a 343 huge aggregation, demonstrating the presence of an additional NA binding site in SP2 344 reinforcing NA-NC-NA networks (Figure 2) . In contrast to NCp9 or NCp7, reaching 345 a plateau of one NCp15 per 10-12nt, NCp15 binding to any of the three templates 346 yielded NP complexes of lower gel mobility upon protein addition (Figure 1c) , similar 347 to canonical ssDNA binding proteins [96] . AFM visualization showed passive ssDNA 348 coating instead of bridging compaction within individual complexes for limiting NCp15 349 concentration, which then led to globular structures at saturation (Fig.1d , Supplemen-350 tary Fig. 2h-i) . NCp15 and NCp7 retain equivalent net charges [66] (NCp15 pI 9.93; 351 NCp7 pI 9.59). Therefore, NCp15 binding does not actively compact and aggregate 352 ssNA, confirming previous results [57, 60] . With free NCp15, p6 has been proposed to 353 bind to the NC domain [60] . Like NCp15 from HTLV-1 [108] , HIV-1 NCp15 binding 354 to NA might invoke quinary intermolecular p6-NC contacts instead of quinary NA-NC 355 contacts. These interactions may freeze these globular structures and mask or block 356 the NC residues responsible for NA compaction/aggregation. Followed by EMSA, SDS-357 PAGE and AFM, the dynamics of quinary NC-NA interactions through cleavage of M13 358 ssDNA-bound NCp15 is verified (Figure 1e Transiently unmasked NC binding sites enable modulation of 363 NC:NA molecular interactions 364 A superposition of the N-terminal 3 10 helix from the NMR structures of NCp7-SL2 365 and NCp7-SL3 complexes is shown in Figure 2a -b and reveals two slightly different NA 366 backbone binding motifs for this domain, which could be virtually sandwiched between 367 two RNA stems, providing a bridge to form RNA-NC-RNA networks [107] . Three 368 additional basic residues over the sixteen present in SP2 poorly explain the dramatic 369 enhancement of the NA quinary capabilities of NCp9. Examination of the NCp9 primary 370 sequence reveals that the NC-SP2 cleavage site surprisingly contains 5 of the 8 residues 371 of the NCp7 SL2-binding motif Lys-Gly-x-Phe-x-x-Gln-Arg, but oriented in reverse, 372 from C-to N-terminus (Figure 2a,d) . To determine if conformers of this sequence would be compatible with a NA binding 374 site structurally similar to those of the N-terminal 3 10 -helix, we performed all-atom 375 molecular dynamics (MD) simulations of an NC-SP2 octapeptide cleavage site. Results 376 of the simulations showed a large conformational area corresponding to a predominantly 377 disordered peptide, similar to other disordered peptide regions in HIV-1 [109, 110] . How-378 ever, three conformer populations were found to lie within 2Å to 1Å RMSD with 379 respect to the N-terminal 3 10 -helix (Figure 2c-e) , as expected. The K3A/F6A/Q9A-380 mutation in NCp7 mostly abrogated ssNA aggregation but maintained ssDNA M13 381 condensation, suggesting this triad to be mostly involved in quinary interactions sta-382 bilizing NA:NC networks (Figure 2f-h) . A DLS analysis in low magnesium finally 383 demonstrated a NCp7/NCp9-driven compaction of M13 ssDNA from 100 nm to 70 nm, 384 respective NA ligands. c, The NC-SP2 apo-octapeptide exhibits substantial flexibility and is free-energetically dominated by a turn-like structure with d KQ ∼ 4-6Å structure θ KF Q ∼ 20 • -40 • yielding a PMF of 5-7 kcal/mol. The apo-ensemble conformers sample a region with d KQ ∼10-12Å with θ KF Q ∼140 • -180 • with less frequency (PMF∼1-3 kcal/mol) where fit all NMR conformers of 1F6U, thus compatible with a 3 10 helical structure. d, Illustration of best-fitting structural conformer of the 52-59 segment superimposed head-to-tail with the N-terminal domain of 1F6U. e, Three conformers populations were found to lie within 2Å RMSD (9.1x10 −5 %), 1.5Å (1.3x10 −5 % ) and 1Å (3x10 −7 % ) with respect to the N-terminal 3 10 helix. Mapping the conformers onto the d KQ -θ KF Q order parameters (blue circles) show they occupy the same region of the conformational sub-space.f, K3AF6AQ9A mutations in NCp7 strongly reduced the protein's capability to aggregate circular and linear M13 ssDNA or MS2 RNA (compare with Fig. 1c, NCp7) . g-h In the absence of magnesium, wt-NCp7 (1NCp7/10nt) strongly aggregates M13 ssDNA whereas the K3AF6AQ9A mutant preferentially condenses the ssDNA. i, By DLS in the absence of magnesium, M13 ssDNA (black) is condensed by the K3AF6AQ9A NCp7 mutant (orange), condensed and next aggregated by NCp7 (green) or aggregated by NCp9 (blue). followed by a massive fusion/aggregation of these complexes (Figure 2i ). In contrast, 385 the K3A/F6A/Q9A NCp7 mutant was strongly defective in the fusion/aggregation pro-386 cess. Altogether, these data strongly support a model where the Lys(3/59)-Gly(4/58)-387 x-Phe(6/56)-x-x Gln(9/53)-Arg(10/52) octad would act in both NCp7 and NCp9 as a 388 quinary interaction module, establishing bridges between NC-NA complexes at NA sat-389 uration (Fig.6a) . These positions are highly conserved amongst all the HIV-1 subtypes, 390 except at position 3 where the conservative K and R residues are found equiprobable. Quinary cooperation between NC and RNA drives PR sequestra-392 tion and RNA-length-dependent catalytic acceleration 393 Followed by SDS-PAGE under conditions optimized for peptide quantification (Sup-394 plementary Fig. 3a-b) , in vitro processing of the C-and N-terminal extremities of 395 the NC domain in an environment unfavorable for PR dimers (0.1 M NaCl, pH 6.25) 396 reveals a dramatic acceleration of NCp15, NCp9 and NCp7 production in the pres-397 ence of ssNA templates (Figure 3a, Supplementary Fig. 3c-e) . 100% of ssDNA-or 398 RNA-bound NCp15 were cleaved in two distinct steps producing NCp9 and then NCp7 399 within minutes, confirming a distributive reaction without consecutive cuts upon the 400 same NCp15 copy ( Supplementary Fig. 3c ). Without NA, complete NCp15 cleavage 401 occurred but at a slower rate, only under acidic (pH 5.0) and high salt (1.5M NaCl) 402 conditions ( Supplementary Fig. 3d) , also concomitantly producing a shorter product 403 (NCp7*). Similar effects were observed with NA for NCp9 cleavage ( Supplementary Fig. 404 3e) and the NC-SP2 cleavage appeared 2-3 times slower than that of SP2-p6 either start-405 ing from NCp15 or NCp9, but was completed in minutes, much faster than previously 406 shown [68] . MS2 RNA activation also occurs for the SP1-NC site of a Gag∆MA protein 407 (CA-SP1-NC-SP2-p6), confirming previous results using a Gag∆p6 (MA-CA-SP1-NC-408 SP2) protein [63] (Supplementary Fig. 3f) . 409 We focused on the NC-ssNA NP assemblies and their influence on NCp15 cleavage 410 at pH 6.25 and 0.1 M NaCl. We first compared the influence of large-scale assembly of 411 NCp15 on M13 ssDNA or MS2 RNA versus stoichiometric complexes formed between 412 NC and a TAR RNA stem-loop ( Supplementary Fig. 4a) . The NA concentrations were 413 varied for a fixed concentration of NCp15. A biphasic effect was observed in presence 414 of either long ssNA, PR reaching maximal activity when NCp15 saturated the ssNA 415 lattices (Supplementary Fig. 4b) . A substantial reduction in PR efficacy was observed 416 upon dispersion of NCp15 over the lattice, even though cleavage was maintained at a 417 much higher level than in absence of NA. In contrast, neither a biphasic effect nor a 418 rapid rate was observed in presence of the TAR-RNA. This NA chain-length effect was next followed for NCp15 cleavage, maintaining equal 420 nt concentration for various HIV-1 RNA stem-loops and fragments from 61 to 615 nt 421 (Figure 3b, Supplementary Fig. 4c) . A weak NC substrate, d(A)13 oligonucleotide, was 422 ineffective in stimulating PR activity, whereas the TAR and SL3 RNA led to significant, 423 but incomplete stimulation of the reaction. Above a critical threshold (∼50 nt-length) 424 and for a pH optimum around 6.3 ( Supplementary Fig. 4e-f) , PR activity scaled non-425 linearly with RNA length irrespective of biological origin (Figure 3b) . Without NA, a 426 5-minute incubation between NCp15 and PR in 0.1 M NaCl at pH 6.25 resulted in no 427 cleavage, while the addition of MS2 RNA immediately boosted the reaction (Supplemen-428 tary Fig. 4d ). Diluting PR for fixed NC:PR (10:1) and NA:NC (20nt.:1) ratios revealed 429 a process resistant to dilution for NA larger than 615nt (T1/2 from 0.15 to 1 minute), 430 whereas a strong rate decrease was observed for TAR or cTAR structures (T1/2 extrap-431 olated to 3 h when considering the first quarter of the reaction; (Supplementary Fig. 432 4e) . A regular decrease at pH 5.0 and 1.5 M NaCl was observed in absence of NA (T1/2 433 from 4 to 25 min). These large NA chains greatly stimulated NCp15 cleavage at 0.1 M salt, with a 435 remarkable pH optimum between 6.0 and 6.5 ( Supplementary Fig. 4f) . The NCp7* 436 extra-cleavage, previously described, corresponds to a site at position 49-50 as a result 437 of ZF destabilization at low pH [111] . This product was examined in two NCp15 cleavage 438 mutants ( Supplementary Fig. 4g) and was due to the cumulative effects of both MS2 439 RNA and acidic pH, which clearly "overcut" the NC domain at pH 5.4. A mildly acidic 440 pH appears therefore beneficial in reducing irregular cleavages of NCp7 upon high PR 441 turnover. With such optimized conditions that satisfy both PR efficiency and NC folding 442 while restricting NCp7 cryptic site cleavages, we confirmed the RNA length effect for 443 the NCp9-to-NCp7 reaction ( Supplementary Fig. 4h-i) , leading to a maximal observed 444 rate close to the T1/2-value of NCp15-to-NCp9 reaction rate, under conditions where 445 NCp9 and NA appeared strongly aggregated (see Figure 1 ). 446 Finally, in presence of 1:1 NC ligands (TAR, cTAR, SL3), bound NCp15 appears 447 almost individually distributed in the reaction mix and allow a reaction-diffusion mech-448 anism that accelerate PR turnover but under conditions where native PR is much less 449 stable. Any substance able to increase the local concentration of either the substrate 450 or the enzyme, or both, drives the reaction in the forward direction enhancing enzyme 451 turnover. As such, ssNA length-dependent activation engages a NC crowding effect 452 with NCp15 molecules coating the NA lattice and forming clusters trapping PR inde-453 pendently of the NP complex concentration. These NA-scaffolded clusters allow faster 454 PR turnover, making both SP2-p6 and NC-SP2 cleavages much more efficient. In other 455 words, the NA quinary capabilities of the NC domain induce an RNA-driven sequestra-456 tion effect on PR. To better understand this sequestration phenomenon, we used a FRET-based assay 458 that measures the cleavage rate of a MA-CA octapeptide probe in presence of NA and/or 459 NCp15. The assay firstly confirmed the reduction of PR activity upon pH increase and 460 salt dilution ( Supplementary Fig. 5a-b) . The M13 ssDNA appeared as an effective 461 substitute to high salt and boosted PR activity by a factor of 10 at pH 5.0, three 462 times greater than in presence of 1.5 M NaCl. A similar effect also occurred at pH 6.25, 463 although PR activity was strongly attenuated. These results confirmed that non-specific 464 PR-NA interactions result in enzyme activation [64] . Adding an equimolar amount of 465 NCp15 at pH 5.5 did not affect the reaction. In contrast, addition of NCp15 bound 466 to ssDNA resulted in a total inhibition of the octapeptide cleavage for ∼5 min, before 467 reaching a velocity analogous to that measured in presence of ssDNA alone (Figure 3d , 468 Supplementary Fig. 5c ). PR is thus sequestered into the NP complex and completes 469 NCp15 processing prior to cleaving the MA-CA peptide at a rate comparable with 470 ssDNA alone, the delay time being directly proportional to NCp15 concentration with 471 a fixed NCp15:ssDNA ratio (Figure 3d ). These data were interpreted by devising a two-substrate model of NA-modulated 473 enzyme kinetics (Supplementary Note), which partitions the reaction volume between 474 distinct regions that are either occupied (pervaded) or unoccupied (un-pervaded) by 475 NA (Figure 4a ). Reacting species (S1: MA-CA, S2: NCp15, E: PR) exhibit equilib-476 rium absorption (K S1 , K S2 and K E ) between these regions due to nonsequence-specific 477 NA-binding. In our model, the enzyme escape rate depends on RNP contiguity -the 478 contiguous number (c = [NCp15]n l /[n t ]) of NCp15 molecules bound per NA and not 479 just chain length or NC-NA loading ratio alone. The phase transition is consistent with 480 a minimum critical contiguity threshold (c crit ) required to alter enzyme escape rate. 481 Contiguity is still length dependent: fitting a reduced single-substrate model onto the 482 experimental data (Figure 4b ) yields non-linear dependence of K E on contiguity with 483 exponent ξ ∼0.4. and c crit ∼3. By expanding to a two-substrate competitive assay 484 (Supplementary Note) and incorporating the effects on differential enzyme decay (Fig-485 ure 4d), our model is fitted to, and is compatible with the observed sequestration effect 486 Kinetic model of two-substrate (S1, S2) processing by an enzyme (E) in a RNP. a, Reaction rate is governed by a combination of effective concentration in a volume domain and absorption kinetics for different species. For K E ≫ 1, PR is sequestered into the RNP-pervaded volume whilst for K E ≪ 1 it is forced out. b, A one-substrate model is fit to experimental data to determine non-linear (exponent ξ) K E dependence on the contiguous number c of S2 molecules bound per NA above a critical threshold, ccrit. c, Fitted two-substrate model of sequestration (black) with competitive substrate alone (red) NA-present competitive substrate (brown) data. The early reaction is dominated by high contiguity (c > c crit , K E ≫ 1) inducing enzyme sequestration. This effect dissipates upon processing (c