Rous Sarcoma Virus Synaptic Complex Capable of Concerted Integration Is Kinetically Trapped by Human Immunodeficiency Virus Integrase Strand Transfer Inhibitors*

Background: Structure of the three-domain Rous sarcoma virus integrase with viral DNA is lacking. Results: Soluble (>1.5 mg/ml) and stable synaptic complex using Rous sarcoma virus integrase, DNA, and HIV-1 strand transfer inhibitors was produced. Conclusion: Two integrase dimers assemble onto viral DNA to produce a synaptic complex. Significance: This is the first report producing a high concentration of soluble and kinetically stabilized RSV synaptic complex. We determined conditions to produce milligram quantities of the soluble Rous sarcoma virus (RSV) synaptic complex that is kinetically trapped by HIV strand transfer inhibitors (STIs). Concerted integration catalyzed by RSV integrase (IN) is effectively inhibited by HIV STIs. Optimized assembly of the RSV synaptic complex required IN, a gain-of-function 3′-OH-recessed U3 oligonucleotide, and an STI under specific conditions to maintain solubility of the trapped synaptic complex at 4 °C. A C-terminal truncated IN (1–269 residues) produced a homogeneous population of trapped synaptic complex that eluted at ∼151,000 Da upon Superdex 200 size-exclusion chromatography (SEC). Approximately 90% of input IN and DNA are incorporated into the trapped synaptic complex using either the C-terminally truncated IN or wild type IN (1–286 residues). No STI is present in the SEC running buffer suggesting the STI-trapped synaptic complex is kinetically stabilized. The yield of the trapped synaptic complex correlates with the dissociative half-life of the STI observed with HIV IN-DNA complexes. Dolutegravir, MK-2048, and MK-0536 are equally effective, whereas raltegravir is ∼70% as effective. Without an STI present in the assembly mixture, no trapped synaptic complex was observed. Fluorescence and mass spectroscopy analyses demonstrated that the STI remains associated with the trapped complex. SEC-multiangle light scattering analyses demonstrated that wild type IN and the C-terminal IN truncation are dimers that acted as precursors to the tetramer. The purified STI-trapped synaptic complex contained a tetramer as shown by cross-linking studies. Structural studies of this three-domain RSV IN in complex with viral DNA may be feasible.

The retrovirus integrase (IN) 3 is responsible for insertion of the viral DNA genome into host chromosomes. Rous sarcoma virus (RSV) IN possesses the canonical three-domain structure consisting of an N-terminal domain (amino acids 1-44), the catalytic core domain (amino acids 50 -214), and the C-terminal domain (amino acids 222-286) (1)(2)(3). Human immunodeficiency virus type 1 (HIV) IN (amino acids 1-288) has similar size analogous domains and linkers between its domains (4,5). The prototype foamy virus (PFV) intasome has been crystallized and its x-ray structure determined (6). The PFV IN (amino acids 1-392) also has these three similar size domains but has an extra N-terminal extension domain (amino acids 1-44) that binds 4 -5 bp in the distal region of the viral DNA in the crystal (6). In addition, the linkers between the N-terminal and catalytic core domains and in particular between the catalytic core and C-terminal domains are significantly longer in the distantly related four-domain PFV IN (6,7) than the corresponding linkers in RSV or HIV IN. Structure-based sequence alignments of full-length RSV, HIV, and PFV INs provide an overview of these important structural features (8).
Physical identification of an assembled IN-DNA complex capable of concerted integration has been successful. The HIV synaptic complex (SC) contains the viral DNA and IN (Fig. 1) and is the first complex in the integration pathway, detectable by native agarose gel electrophoresis (9,10). In a time-dependent manner, HIV SC associates with the target DNA promoting the concerted integration reaction, producing the strand transfer complex (9,10). In both of these complexes, a tetramer of HIV IN is associated with two LTR DNA ends (9,11). Fluorescence resonance energy transfer (11,12) and atomic force microscopy experiments (13) demonstrated the physical inter-actions that occur between tetrameric HIV IN and the LTR ends. Interestingly, HIV IN monomers were shown to efficiently utilize oligonucleotide (ODN) substrates for concerted integration (14). Similarly, PFV IN monomers are responsible for assembly of the PFV intasome (6,15,16), a term encompassing both the SC and strand transfer complex.
Strand transfer inhibitors (STIs) directed against HIV IN are effective inhibitors in combination with other anti-retroviral drugs to prevent HIV replication in humans (17). Three IN STIs, raltegravir (RAL), elvitegravir (EVG), and dolutegravir (DTG), have been approved by the Food and Drug Administration for HIV treatment. The STIs also inhibit strand transfer mediated by PFV IN and inhibit its replication (18). Localization of these STIs within the active site of the surrogate PFV intasome (6,19) has provided significant structural insights into their mechanisms for inhibiting strand transfer and the development of drug resistance. STIs also have varying capabilities to kinetically "trap" the HIV SC (10,20,21), apparently due to slow dissociation properties of the STI (22)(23)(24)(25)(26). The HIV IN STIs appear to trap SC by making contacts with both IN and DNA, as shown for the PFV intasome (6,19).
Besides their ability to bind within the active site of PFV IN (6) and to inhibit Spumavirus prototype foamy virus replication (18), HIV IN STIs also have the ability to differentially inhibit the replication of Alpharetrovirus, Betaretrovirus, and Gammaretrovirus (27). These studies suggest that RSV IN would also be susceptible to inhibition by STIs. We demonstrate here that STIs at low nanomolar concentrations inhibit the concerted integration catalyzed by RSV IN as effectively as observed previously with HIV IN using similar large size DNA substrates (ϳ1 kb) (10,21
Concerted Integration Assays-Strand transfer assays were performed using a linear 3.6-kb DNA donor substrate that possessed a single U3 long terminal repeat (LTR) DNA end and labeled with 32 P at the 5Ј LTR end (29). The substrate was produced by NdeI digestion of a circular plasmid generating a 2-bp 3Ј-OH recessed U3 end. The U3 end sequence was modified on the cleaved strand at nucleotide position 6 (T to A) producing a gain-of-function (GU3) substitution that possesses severalfold higher catalytic activity than the WT U3 sequence (29). Briefly, RSV IN (20 nM) and donor DNA (0.5 nM) were preassembled at 14°C for 15 min in 20 mM HEPES, pH 7.5, 10 mM MgCl 2 , 300 mM NaCl, 5 mM DTT, and 8% polyethylene glycol 6000. Upon addition of supercoiled target DNA (2.7 kb)(1.5 nM), strand transfer was for 30 min at 37°C. Reactions were stopped with EDTA to a final concentration of 25 mM, and samples were deproteinized with 0.5% SDS, 1 mg/ml proteinase K for 1 h at 37°C. Strand transfer products were separated on a 1.3% agarose gel, dried, and analyzed by a Typhoon Trio Laser Scanner (GE Healthcare).
Solution conditions to efficiently promote concerted integration by HIV IN using ODN substrates (14) were further optimized for RSV IN. The assay with 3Ј-OH recessed ODN substrate (18/20 bp) containing RSV GU3 LTR sequences typically contained 2 M LTR substrate and 4 M IN in 20 mM HEPES, pH 7.5, 125 mM NaCl, 10 mM MgCl 2 (or MgSO 4 ), 5 mM DTT, and 10% (v/v) dimethyl sulfoxide (DMSO). After initial preincubation of the IN/DNA mixture at 14°C for 15 min, the supercoiled target DNA (10 nM) was added, and strand transfer was carried out at 37°C for 45 min. The reactions were deproteinized as described above. Strand transfer products were separated on a 1.8% agarose gel, stained with SYBR Gold (Invitrogen), and analyzed by a Typhoon Trio Laser Scanner (GE Healthcare).
Assembly of STI-trapped RSV SC-We assembled STItrapped SC in solution from RSV IN, 3Ј-OH recessed GU3 ODN and STIs. Optimized assembly reactions contained GU3 20R (15 M) and IN(1-269) (45 M) in 20 mM HEPES, pH 7.5, 100 mM NaCl, 100 mM (NH 4 ) 2 SO 4 , 1 M nondetergent sulfobetaine (NDSB-201), 5 mM DTT, 10% (v/v) glycerol, 10% (v/v) DMSO, and 125 M STI. The assembly process continued for 3 to 24 h at 4°C in the dark because some STIs are light-sensitive. Variations from these standard assembly conditions were described in the text or figure legends as indicated. The trapped SC produced in the presence of an STI was analyzed by Superdex 200 size-exclusion chromatography as described below. Fractions containing the trapped SC eluting at ϳ150 kDa were collected and analyzed as described. The STIs RAL, MK-2048, and MK-0536 were kindly provided by Merck, and DTG was supplied by GlaxoSmithKline. EVG and additional DTG were also purchased from Selleck Chemicals.
Mass Spectrometry and Fluorescence Analyses-STI-trapped RSV SC was purified by SEC. Fractions containing the trapped SC were collected and analyzed for the presence of STIs. The running buffer for SEC was devoid of glycerol as it may interfere with mass spectroscopy analysis. Samples were extracted with 3 volumes of HPLC-grade acetonitrile. The samples were vortexed for 5 min followed by centrifugation at 2000 ϫ g for 5 min. The supernatant was dried under heated nitrogen and reconstituted in 100 l with HPLC-grade acetonitrile for LC/MS/MS analysis. Standards containing MK-2048 were analyzed (100 or 500 ng/ml in acetonitrile). MK-2048 in the samples was identified via an LC/MS/MS system that consisted of an LC-20AD pump (Shimadzu, Kyoto, Japan), an HTC PAL autosampler (Leap technologies, Carrboro, NC), and a Sciex API-4000 mass spectrometer in ESI mode (AB Sciex, Foster City, CA). The multiple reaction monitoring transition for MK-2048 was m/z, 462.06 Ͼ 303.9. The mobile phases consisted of 0.1% formic acid in HPLC grade water and 100% HPLC grade acetonitrile with an Armor C18 reverse phase column (2.1 ϫ 30 mm, 5 m, Analytical Sales and Services, Pompton Plains, NJ) at a flow rate of 0.35 ml/min. The starting phase was 10% acetonitrile for 0.9 min, then increased to 90% acetonitrile over 0.4 min, and maintained for 0.2 min before returning to 10% acetonitrile over 0.4 min, and then held for 1.6 min at 10% acetonitrile.
The presence of RAL and MK-0536 within the trapped RSV SC was determined by fluorescence spectroscopy. Samples were prepared by acetonitrile extraction as described earlier.
The excitation and emission maxima for RAL and MK-0536 were empirically determined using a Fluoromax-3 spectrofluorimeter (Jobin Yvon, Inc., Edison, NJ). RAL exhibited excitation and emission maxima at 310 and 410 nm, respectively. These maxima for RAL matched a previous report (30). The excitation and emission maxima for MK-0536 were 350 and 455 nm, respectively.
Protein-Protein Cross-linking-The oligomeric form of RSV IN(1-269) in the DTG-trapped SC was determined. Fractions spanning the DTG-trapped SC peak were cross-linked with EGS at 22°C for 10 min. An equivalent amount of IN(1-269) was purified by SEC and cross-linked with EGS in parallel. The SEC running buffer contained 0.4 M NaCl and lacked (NH 4 ) 2 SO 4 . The cross-linking reactions were quenched with 100 mM Tris, pH 7.6, for 10 min on ice and analyzed by SDS-PAGE. The gel was stained with Krypton (Thermo Scientific).
SEC-MALS Analyses-The absolute molecular weights of RSV IN(1-269, 1-274, and 1-286) as well as STI-trapped SC were determined by SEC on Superdex 200 (10/300) connected in-line with multiangle light scattering (MALS) mini-DAWN TREOS equipped with a 658-nm laser (Wyatt Technology Corp., Santa Barbara, CA) coupled with refractive index measurement by Optilab T-rEX (Wyatt Technology). The molar mass of IN and trapped SC was calculated from the observed light scattering intensity and differential refractive index using Zimm scattering model in ASTRA software (Wyatt Technology). To determine the absolute molar mass of trapped SC, a dn/dc of 0.185 and 0.170 was used for protein and DNA components, respectively.

Inhibition of RSV IN Concerted Integration by STIs-Clinical
HIV IN STIs bind in the active site of IN within the PFV intasome (6,19). RSV replication was inhibited by RAL but not EVG, and this resistance to EVG mapped to residue Ser-150 of RSV IN (27). Consistently, WT RSV IN(1-286)-concerted integration activity using a 3Ј-OH recessed GU3 substrate (3.6 kb) was inhibited by RAL and MK-2048 but not by EVG in vitro ( Fig. 1). EVG at high concentrations (Ͼ500 nM) was necessary for noticeable inhibition of either concerted or circular half-site (CHS) integration (Fig. 1, lanes 9 -11), although RAL was effective at 25 nM (Fig. 1, lanes 5-8). A second generation STI, MK-2048, was also effective at low nanomolar concentrations (Fig. 1, lanes 12-15). We further demonstrated that IN(1-286)catalyzed concerted integration was inhibited by several different STIs, including DTG at low nanomolar concentrations (Table 1). An equivalent level of inhibition for RSV-concerted integration activity was observed with the different STIs in comparison with inhibition of HIV IN (10,21). Very similar inhibitory data using RAL, EVG, and MK-2048 were also obtained using RSV IN(1-269) (data not shown). These results show that STIs bind effectively in the active site of RSV IN to inhibit concerted as well as CHS integration.  (Fig. 3I). The IN preparations were free of contaminating DNA endonuclease activities as determined by using supercoiled DNA as substrate ( Fig. 2A, lane 5-7). Additionally, no strand transfer activity was observed with the substrate containing nonspecific DNA sequences (21R-NSP) ( Fig. 2A, lanes  14 -16).
Formation of Trapped RSV SC Is Related to Dissociative Halflife of HIV IN STI-We had previously shown that HIV SC can be kinetically trapped by STIs differentially, as analyzed by native agarose gel electrophoresis (21). For example, MK-2048 was more effective than RAL for producing trapped HIV SC. MK-2048, MK-0536, and DTG possess severalfold longer dissociative half-lives from HIV IN-DNA complexes than those observed with either RAL or EVG (22)(23)(24)(25)(26).
We examined whether HIV IN STIs could kinetically stabilize the RSV SC in solution. We investigated a variety of parameters necessary to permit isolation of this complex including the following: 1) concentrations and types of salt to allow IN-DNA binding; 2) reagents that allow high solubility of these formed nucleoprotein complexes; 3) temperature; 4) length of IN with C-terminal truncations; 5) molar ratios of IN, GU3 substrates, and STIs; and 6) determining that added reagents necessary for proper assembly conditions did not modify the inhibitory properties of STI within the active site of RSV SC.
The optimized solution condition (see "Experimental Procedures") to produce STI-trapped RSV SC for 24 h at 4°C included three components at molar ratios of RSV IN(1-269) (45 M monomer), 20R DNA (15 M), and STI (125 M). After assembly of the IN-DNA-STI complex, the samples were analyzed by Superdex 200 SEC operating with a running buffer lacking STI (Fig. 3). The above mentioned STIs (MK-2048, MK-0536, and DTG) possessing longer dissociative half-lives observed with HIV IN-DNA complexes were also the most effective in producing RSV-trapped SC (Fig. 3, A-C, respectively) and inhibition of RSV IN concerted integration activity ( Table 1). The trapped RSV SC elutes from the Superdex 200 column very near the 158-kDa marker (Fig. 3, A-D). The calculated mass of the trapped RSV SC with the tetrameric IN(1-269) and GU3 20R is ϳ146 kDa. Although RAL effectively inhibited concerted integration, it was not as effective in producing trapped SC (Fig. 3D); EVG essentially lacked the ability to produce this IN-20R-STI complex (Fig. 3E) as it does not effectively inhibit strand transfer ( Table 1). The presence of an STI was strictly required for producing the trapped SC with IN(1-269) and 20R (Fig. 3F). Formation of the STI-trapped SC with GU3 18R was similar to that observed with 20R, but starting at 22R the efficiency was decreased, and precipitation started to occur more readily (data not shown). Finally, a 21R-NSP substrate containing nonspecific DNA sequences (Fig. 3G) or a 29R-NSP (data not shown) was not capable of producing the complex using RSV IN(1-269) and MK-2048. As mentioned earlier, no strand transfer activity was observed with the substrate containing nonspecific DNA sequences ( Fig. 2A,  lanes 14 -16). In summary, the ability of an STI to kinetically

Retrovirus Integrase-DNA-strand Transfer Inhibitor Complex
stabilize the RSV SC was associated with STIs that possess longer dissociative half-lives. The presence of glycerol in the assembly buffer significantly increased the solubility of the assembled complex, although DMSO only slightly aided solubility when glycerol was present. We determined that (NH 4 ) 2 SO 4 significantly influenced the solubility of RSV IN itself and also in the assembly mixture to produce trapped RSV SC. Increasing the temperature above 4°C caused a slow but steady increase in the production of IN-DNA complexes that precipitated out-ofsolution. Finally, as described later, RSV IN(1-269) was chosen because of the apparent homogeneous physical nature of this trapped SC upon SEC analysis (Fig. 3).
A time course experiment was performed to determine the rate of assembly of the STI-trapped SC at 4°C. Three identical assembly samples were produced with RSV IN(1-269), and each sample was analyzed by SEC after 3, 6, and 24 h of incubation (Fig. 4). All three samples essentially produced the same quantity of trapped SC as well as incorporation of IN and 20R into this complex. The results suggested that formation of SC at 4°C and its capture by MK-2048 to produce a stable complex is rather rapid at this temperature. The isolation of the complex in the SEC running buffer lacking STIs also suggests that a kinetically trapped SC structure was produced. Trapped RSV SC Lacks Strand Transfer Activity and Contains STI-Capture of an STI in the active site of IN would inhibit the strand transfer reaction (Table 1). We produced the trapped SC with RSV IN(1-269) (45 M) in the presence of MK-2048 and subjected the sample to SEC. As expected, the fractions containing the purified trapped SC lacked strand transfer activity (Fig. 6, top panel; bottom panel, lanes 2-8). A series of controls were performed. An independent assembly reaction using RSV IN(1-269) (Fig. 6, lane 11) and the original IN/DNA assembly mixture without STI possessed normal strand transfer activity (Fig. 6, lane 12). The original IN/DNA mixture with MK-2048 prior to SEC analysis lacked activity (Fig. 6, lane 13). Finally, an identical IN/DNA assembly mixture lacking MK-2048 as described above was also subjected in SEC (Fig. 3F). Combined aliquots of several different fractions containing either free IN or free DNA possessed strand transfer activity (Fig. 6, lanes 9 and 10) demonstrating that the assembly conditions and SEC did not destroy the ability of IN to promote strand transfer activity.

Equilibrium and Stability of Kinetically Trapped RSV SC-
We determined that the STIs remained associated with the trapped RSV SC. Three sets of RSV IN(1-269)   or MK-2048 were incubated for 24 h, and the individual trapped SCs were purified by SEC. The trapped SC fractions with each inhibitor were pooled and subjected to extraction by acetonitrile to isolate each STI. Fluorescence excitation and emission spectra analysis for RAL and MK-0536 extracted from their respective trapped SCs are shown (Fig. 7, A and B, respectively). Extraction of this same complex yielded MK-2048 as detected by mass spectrometry (Fig. 7D). In summary, these qualitative analyses demonstrated that each independently isolated trapped SC contained the inhibitor.  -(1-286) was 58,500 Ϯ 305 Da (Fig. 8A), 62,900 Ϯ 805 Da, and 65,450 Ϯ 70 Da, respectively. As a control, we also determined that the assembly buffer to produce the trapped SC did not disrupt the dimeric structure of RSV IN     ( Fig. 8B) and 20R (data not shown) in the presence of MK-2048 under standard assembly conditions with assembly times varying from 3 to 24 h. The mass of trapped SC using the 18R was 151,000 Ϯ 2,000 Da (Fig. 8B) and with 20R it was 154,800 Ϯ 12,000 Da. (6,9,11). We determined that RSV IN(1-269) exists predominantly as a tetramer in DTG-trapped SC (Fig. 9, lanes 7-9). At 2 mM EGS, the major cross-linked IN species was a tetramer in all three separate SEC fractions of trapped SC. At 1 mM EGS, a trimer is also evident in trapped SC (Fig. 9, lanes 4 -6). With IN in the absence of DNA, a crosslinked dimer and tetramer were observed (Fig. 9, lanes 2 and 3).  HIV IN STIs are interfacial inhibitors (34,35). Interfacial inhibitors target molecular machines consisting of two or more independent components and, in the case of STIs, viral DNA and IN within the context of the assembled SC (Fig. 3I). Soaking the crystal containing the PFV intasome with STIs allowed high resolution analysis of these contacts (6). The interactions of RAL, EVG, and DTG within the catalytic site of the PFV intasome (6,19,36,37) define the interfacial inhibitor interactions with IN residues and the terminal 3Ј-adenylate on the catalytic strand that is displaced by STIs. Modeling of these inhibitors into the HIV IN active site (5,25,38) suggests some differences with inhibitor interactions within IN and the displaced 3Ј-adenylate, but they all possess the interfacial features observed with the PFV intasome. This same inhibitory mechanism appears likely to occur within the active site of the STI-trapped SC. RSV SC is formed in the assembly solution (Fig. 6, lane 12) but is not stable upon SEC (Fig. 3F) unless an STI is present in the assembly mixture (Fig. 3, A-D). Previous fluorescence energy transfer studies using RSV IN(1-286) and fluorophorelabeled GU3 DNA ODN substrates clearly established that assembly of SC occurs in solution under noncatalytic conditions (31). Production of this soluble RSV SC in the presence of STIs highly supports the ability of STIs to trap the molecular machine in a specific inactive form (Fig. 6) previously observed with other macromolecular machines, like inhibitors of DNA topoisomerase-DNA complexes (35).

RSV IN Tetramer Is Associated with Trapped SC-A hallmark of HIV and PFV SC and strand transfer complex is the presence of an IN tetramer
EVG inhibits RSV IN concerted integration only at very high concentrations (Table 1) and fails to produce a stabilized RSV SC (Fig. 3E) (24,26) appear to possess key properties in producing a highly stabilized RSV SC, although RAL is ϳ70% as effective as the slow-dissociating STIs (Fig. 3). This property of slow dissociation for STIs would likely map to interactions of these STIs with RSV IN residues and the displaced 3Ј-adenylate on the catalytic strand similar with PFV IN and modeled with HIV IN. Atomic resolution structure of the kinetically trapped RSV SC will provide further insights into these interactions and a structural view of this canonical three-domain IN-DNA complex.
IN monomers are the precursor to the tetramer in the PFV intasome (15). The HIV IN monomer also has the capacity to form a tetramer for concerted integration (14, 39) suggesting its   (Fig. 8). All three proteins exist as dimers even if IN(1-269) was pretreated with the assembly buffer used to produce trapped SC, prior to SEC-MALS analysis. A dimer of IN is always observed upon time-dependent dissociation of the trapped RSV SC at 4°C (Fig. 5). Together, these data suggest that a dimer is the precursor to the tetramer observed in the STI-stabilized SC (Figs. 8A and 9).
Previous results with RSV IN(1-270) demonstrated that IN with the C-terminal truncation possesses similar properties to those associated with IN(1-286), including robust concerted integration activity using a larger sized (Ͼ1 kb) GU3 substrate (2). We further explored whether IN(1-269) and IN(1-274) displayed different properties than those associated with IN(1-286). Using a large size single end-labeled blunt-ended substrate (29), the 3Ј-OH processing activity of all three proteins were very similar ( Table 2). As mentioned before, all three INs displayed similar efficiencies for concerted integration using GU3 20R ( Fig. 2A, lanes 8 -10). However, the two C-terminal truncated INs had decreasing efficiency for concerted integration with the blunt-ended substrate (20B) in comparison with full-length IN (Fig. 2A, lanes 11-13). Hence, the truncated INs are partially defective in coupling 3Ј-OH processing and concerted integration activities. These data suggest that the tail region of RSV IN, similar to HIV IN (42,43), enhances the functions of IN for concerted integration. Of note, RSV IN(1-274) and IN(1-286) both form two trapped SC species in contrast to IN(1-269) (Fig. 10), suggesting the tail region may further affect assembly and possibly concerted integration in the presence of ODN substrates ( Fig. 2A). Further studies are necessary to understand how the tail region affects the assembly process and concerted integration. In

3-OH processing activities of RSV IN
The 3Ј-processing activities of WT RSV IN(1-286) and two C-terminal deletion constructs were determined as described under "Experimental Procedures." The standard deviation was determined from at least three independent experiments.