Retroviral intasome architecture shapes the dynamics of target DNA search and integration

TL;DR

This study compares the behavior of intasomes from two retroviruses, showing how their structure affects DNA integration speed and efficiency.

Contribution

The first single-molecule analysis of an orthoretrovirus intasome's interaction with target DNA.

Findings

MMTV intasomes remain active for ~80 minutes, much longer than PFV intasomes which plateau in ~2 minutes.

MMTV intasomes integrate more efficiently with manganese than magnesium and search DNA more slowly than PFV intasomes.

The time between integrating the two viral DNA ends is ~3 times slower in MMTV compared to PFV intasomes.

Abstract

Recombinant retroviral intasomes assembled from purified integrase (IN) and oligonucleotides mimicking viral DNA ends (vDNA) faithfully recapitulate concerted integration in vitro. Structural studies of retroviral intasomes have revealed an array of IN oligomer forms, which appear to share a conserved intasome core coordinating the vDNA ends for strand transfer into target DNA. Here we have explored the biochemical and dynamic properties of the mouse mammary tumor virus (MMTV) octameric intasome. We show that MMTV intasomes continue to accumulate concerted integration products for ~80 min in vitro, whereas prototype foamy virus (PFV) intasomes plateau within ~2 min. MMTV integration activity peaks within the range of physiological ionic strength and is more active in the presence of manganese compared to magnesium. Single-molecule images demonstrate that the target DNA search by MMTV intasomes appears rate-limiting, similar to PFV intasomes. The time between strand transfer of the two MMTV vDNA ends into the target DNA is ~ 3 fold slower than PFV intasomes. This is the first report of the dynamics of an orthoretrovirus intasome interacting with target DNA with single molecule resolution.

Retroviruses insert their genetic material into the DNA of infected cells, a process carried out by a viral enzyme called integrase. Integrase molecules assemble with viral DNA ends to form a complex known as the intasome, which joins the viral DNA to the host DNA. Although the overall steps of integration are shared among retroviruses, the details of how different intasomes find and react with host DNA have remained unclear. Here, we characterized the intasome from the mouse mammary tumor virus (MMTV) for comparison to the prototype foamy virus (PFV) intasome. Using biochemical and single-molecule imaging approaches, we found that the larger MMTV intasome remains catalytically active longer than the smaller PFV complex and moves more slowly while searching for a target site on DNA. We found a common mechanism for how the intasomes interact with target DNA, while the details appear to vary based on the size of the complex.

Figures

MMTV intasomes have different integration efficiencies in the presence of manganese or magnesium.MMTV intasome integration to a SC plasmid was performed with increasing concentrations of (A) MgCl2, (B) MnCl2, (C) MgSO4, and (D) NaCl. Integration products were separated by agarose gel electrophoresis and quantified by the Cy5 fluorescent signal in each lane (fmol). Half-site integration (HSI) products where only one vDNA has been joined to the plasmid have the mobility of a relaxed circle. A single concerted integration product (CI1) has the mobility of a linear plasmid. Additional CI events (CI2) result in a smear between CI1 and unreacted vDNA. Autointegration (AI) of vDNA to vDNA has slower mobility than unreacted vDNA. All reactions include 1 mM CaCl2 that carried over from intasome assembly and purification steps. MMTV intasomes are not active in the absence of magnesium or manganese divalent ions. Increasing concentrations of MgCl2 or MgSO4 lead to increasing total CI (CI = CI1 + CI2). AI products were only apparent at the highest concentration of MgCl2. In contrast, MMTV intasome CI was equivalent at all concentrations of MnCl2 assayed, while AI products increased with increasing concentrations of MnCl2. HSI products were unaffected by the increasing concentrations of divalent ions. MMTV intasomes were also assayed with a titration of NaCl in the presence of 20 mM MgSO4. MMTV CI products peaked in the physiologically relevant range of 100-150 mM NaCl. Fewer CI products were observed at >150 mM NaCl. MMTV integration efficiency appeared to be reduced at 50 mM NaCl, a concentration less than physiological, but was not statistically significant. Error bars indicate the standard deviation of at least 3 independent experiments performed with at least 2 independent MMTV intasome preparations.

Time course and catalytic rate of MMTV intasomes.(A) MMTV intasomes were evaluated for integration to a SC plasmid for up to 3 h. Integration products were separated by agarose gel electrophoresis and quantified by the Cy5 fluorescent signal in each lane (fmol). Based on the accumulation of Cy5 CI products, MMTV intasomes retained activity for at least 80 min. (B) The initial rate of reaction was estimated based on velocity vs substrate concentration. Increasing SC plasmid concentrations were converted to nM sites (0.25, 0.3, 0.35, 0.4, 0.45, 0.5, 1, 3 and 6 nM) considering the number of possible MMTV intasome binding sites in the 3 kb target plasmid. The MMTV intasome minimal recognition site is 6 bp, with a likely higher limit of 12 bp. This range of site size yields a Km•app range from 34.5 to 69.0 nM. The calculated kcat•app for MMTV intasomes is 0.01 min-1. (C) MMTV intasomes were assayed for integration to recombinant nucleosomes with 147 bp of linear DNA wrapped around a histone octamer. CI reaction products displayed faster mobility than unreacted target DNA and were visualized and quantified as in (A). Error bars indicate the standard deviation of at least 3 independent experiments performed with at least 2 independent MMTV intasome preparations.

MMTV intasome aggregation.MMTV intasomes were incubated for 1 h at 37˚ C in integration reaction buffer with indicated concentrations of components. All aggregation reactions include 1 mM CaCl2 carried over from the purification buffer. Aggregates were pelleted by centrifugation and analyzed by denaturing PAGE stained with Coomassie blue. (A) The percentage of precipitated MMTV IN (% Aggregation) is relative to the control lane (lane 1). Increasing concentrations of MgSO4 led to decreased aggregation of MMTV intasomes. Addition of 10 mM CaCl2 (final concentration 11 mM CaCl2) reduced aggregation, similar to the highest concentrations of MgSO4. Addition of acetylated BSA was the most effective at preventing MMTV intasome aggregation. (B) MMTV intasome aggregation in the presence of 20 mM MgSO4 was assayed with a titration of NaCl. NaCl concentrations >100 mM also led to a decrease of MMTV intasome aggregation. The difference in aggregation between 100 mM and 200 or 300 mM NaCl was significant (p < 0.05), but there was no significant difference between aggregation in 150, 200, or 300 mM NaCl. Three independent experiments were performed with at least 2 independent MMTV intasome preparations.

MMTV intasome strand transfer kinetics and search dynamics.(A) The time between the two strand transfers (τST) of MMTV vDNA was determined by single-molecule magnetic tweezers. A linear DNA was attached at multiple points to a surface and a paramagnetic bead. Magnets above the microscope stage introduced negative supercoils to the DNA, effectively reducing the apparent height of the bead. The beads were tracked in the x-, y-, and z-axes. The first strand transfer event introduces a nick in the target DNA releasing the supercoils, and changing the z-axis position (blue triangle). The second strand transfer introduces another nick and releases the bead, seen as changes in x-, y-, and z-axes. A representative integration event is shown. The τST is the time between the first and second strand transfer events. (B) The τST measured for 41 independent integration events was used to construct a histogram. For MMTV intasomes τST is 1.31 ± 0.12 s. (C) Fluorescently labeled MMTV intasomes moving on linear DNA were tracked by TIRF microscopy. A 24 kb linear DNA was attached at both ends to a flow cell surface. 50-100 pM fluorescently labeled MMTV intasomes were added to the flow cell. A representative kymograph of a single MMTV intasome in association with target DNA is shown; the x axis of the trace is time and the y axis is the length of the target DNA. (D) The lifetime (τON) of MMTV intasomes in association with target DNA was measured at multiple concentrations of NaCl. Increasing NaCl concentrations had an inverse relationship with the τON of the intasomes. (E) Diffusion coefficients of MMTV intasomes were measured at multiple NaCl concentrations. The diffusion coefficients were similar, indicating a random walk search of ~2.5 kb at 110 mM NaCl. Error bars indicate the standard deviation of at least 3 independent experiments performed with at least 2 independent MMTV intasome preparations.

MMTV intasomes are prone to aggregation in the presence of target DNA.(A) When 2 nM MMTV intasomes were injected in the flow cell they aggregated on the stretched DNA molecules. The images suggest recruitment and accumulation of additional MMTV intasomes binding the target DNA molecules over time. (B) MMTV intasome aggregation was assayed in the absence or presence of 3 kb linear DNA. Wild type and catalytically inactive IN(D122N) intasomes were incubated for 1 h at 37˚C in the presence of 110 mM NaCl. Aggregates were pelleted by centrifugation and analyzed by denaturing PAGE. Both wild type and mutant intasomes displayed increased aggregation when linear DNA was present, indicating that aggregation does not require catalysis. The graph represents the mean of three independent experiments performed with at least two independent MMTV intasome purifications.

Full text

Introduction

Retroviral RNA genomes are copied to linear double stranded DNA (cDNA) by reverse transcriptase [1]. Integration of this cDNA into host chromatin is an essential step in the retroviral life cycle and is catalyzed by the viral protein integrase (IN) [1–3]. During infection retroviral IN performs two catalytic activities. First, a 3’ processing activity cleaves two terminal nucleotides from the cDNA 3’ ends generating terminal recessed hydroxyl groups [3]. Second, a strand transfer activity covalently joins the recessed 3’ hydroxyls to host DNA [3–6]. Depending on the retrovirus, the two strand transfer events are separated by 4–6 bp of host target DNA. The result is the integrated proviral genome flanked by 4–6 nt single strand gaps of host DNA and 5’ dinucleotide flaps of viral cDNA. Host enzymes appear to repair this gapped integration intermediate resulting in a stable integrated provirus flanked by duplications of host DNA [7].

Retroviral INs have three common domains: an amino terminal domain (NTD), catalytic core domain (CCD), and carboxyl terminal domain (CTD) (reviewed in [3]). Some retroviruses, including the spumavirus prototype foamy virus (PFV), have an additional amino terminal extension domain (NED). The CCD includes a DDE catalytic triad that coordinates divalent metal cations. All domains participate in multimerization to form an integration complex. The CCD and CTD also have non-specific DNA binding activity [8]. The linker region between the IN CCD and CTD varies in length between retroviruses and may correlate with the number of IN protomers necessary for an active complex [9–13].

A multimer of retroviral IN binds the viral cDNA ends to form a functional enzymatic complex. This intasome may be assembled in vitro with recombinant IN and oligomers mimicking the viral cDNA ends (vDNA) [14]. In addition, the vDNA may be labeled with small fluorophores or biotin moieties at the distal end [15]. The first retroviral intasome structure revealed a tetramer of PFV IN with two vDNAs [14,16]. A second intasome structure included target substrate and revealed the target DNA is bent to accommodate strand transfer at 4 bp spacing across the major groove [16]. Two “inner” PFV IN subunits are catalytically active and bind both the vDNA and target DNA. Coordinates for only the CCD of the two “outer” PFV IN protomers have been resolved, which appear important for intasome structure but not catalysis [16–19]. Any function for the unresolved domains of the outer PFV IN protomers are unknown but they have been implicated in binding to nucleosomes in vitro [20].

Structures of orthoretroviral intasomes showed a diversity of IN oligomerization [9,10,12,13,21]. The deltaretrovirus human T cell leukemia virus (HTLV-1) intasome is also a tetramer of IN [22,23]. The alpharetrovirus Rous sarcoma virus (RSV) and the betaretrovirus mouse mammary tumor virus (MMTV) intasomes are octamers of IN [9,10,13,24]. The lentivirus Maedi visna virus (MVV) intasome is a hexadecamer of IN [9,10,13]. Single particle cryogenic electron microscopy of lentiviral intasomes revealed tetrameric, dodecameric, and filamentous assemblies for HIV-1 IN as well as dodecameric and filamentous assemblies for simian immunodeficiency virus from red-capped mangabey (SIVrcm) IN [12,25,26]. All of these intasome structures notably display a conserved intasome core (CIC) that is similar to the central structural elements of the PFV IN tetramer [27–29].

Interrogation of integration processes in vivo is challenging, in part because only a fraction of reverse transcription products become integrated proviruses [30–32]. Real-time single-molecule imaging has been utilized to dissect the search and integration dynamics of PFV intasomes with naked DNA in vitro, providing mechanistic insight into the process while complementing studies of integration in the chromatin context [19,33]. These studies demonstrated a target DNA search for 2.1 s while in continuous contact with the DNA backbone, suggesting that PFV intasomes interrogate ~1.6 kb of duplex DNA per search, most likely by rotation coupled diffusion. The average time between the two vDNA strand transfer events was 0.47 s [19,33,34]. Remarkably, PFV intasomes performed 100–300 search events per strand transfer event strongly suggesting that target site selection was rate-limiting [33]. Interestingly, PFV intasomes specifically bind to DNA strand breaks and gaps, perhaps explaining the large number of search events vs strand transfer events [35]. To date, the dynamic interactions with target DNA have only been described for PFV intasomes [19,33,35,36]. We hypothesized that the CIC mediates a common DNA search and integration mechanism across retroviruses, but that oligomer size and domain organization modulate the kinetics of this process.

While some retroviruses employ a host protein as an integration co-factor, none has been reported for MMTV [37–41]. Retroviral integration host co-factors help determine the genomic integration profile relative to chromatin elements such as transcription units, promoters, or CpG islands [42–44]. Among retrovirus integration profiles MMTV displays only modest preference for genomic elements, which agrees with the absence of an integration co-factor [37,45,46]. Despite this integration profile, the MMTV provirus is able to dysregulate cellular proto-oncogene expression leading to mammary adenocarcinomas in mice. Thus, MMTV infection has been employed as a model for breast cancer pathogenesis [38–40]. Integration host co-factors are known to aid in the stability of recombinant intasomes, as in the case of the LEDGF/p75 host co-factor for lentiviruses MVV and HIV-1 and the FACT complex for alpharetrovirus avian leukosis virus (ALV) [47,48]. However, PFV and MMTV intasomes may be readily assembled from recombinant IN and vDNA only [9,10,12,16,49].

Here we examined the biochemical properties, strand transfer kinetics, and search dynamics of the MMTV intasome. We predicted that the larger octameric architecture of MMTV intasomes would slow the DNA search relative to tetrameric PFV intasomes. Our results show that MMTV intasomes indeed display a longer interval between the two strand transfer events and perform a more prolonged DNA search compared to PFV. These findings suggest that while retroviral intasomes share a common search mechanism mediated by the CIC, oligomer size and domain organization diversify their kinetics. Understanding how orthoretroviral intasomes like MMTV differ in dynamics from spumaviral integration complexes will clarify how integrase architecture contributes to viral genome integration and stability.

Results

To characterize the biochemical and kinetic behavior of MMTV intasomes, we first confirmed their assembly, composition, and catalytic competence under standard conditions.

Purified MMTV intasomes catalyze concerted integration

Recombinant MMTV intasomes were assembled with Cy5 labeled vDNA [9,10,14,16,18]. Size exclusion chromatography (SEC) revealed a high molecular weight peak (~530 kDa) consistent with an asymmetric octamer of MMTV IN and two vDNAs (S1A Fig). Two additional peaks were also observed, consistent with a dimer of MMTV IN (~100 kDa) and free vDNA (~19 kDa). The predicted MMTV intasome molecular weight is 312 kDa. The molecular weights determined by SEC fractionation can be influenced by molecular shape. PFV intasomes purified by SEC also displayed a larger molecular weight peak (~305 kDa) compared to the predicted molecular weight of the tetrameric IN with two vDNAs (226 kDa) [33]. Intasomes purified by SEC were analyzed by single molecule mass photometry, which determines molecular weight independent of the shape of the protein complex. MMTV intasomes were determined to be 379 kDa and MMTV IN was 87 kDa consistent with a dimer (S1E Fig). For comparison PFV intasomes were shown to be 225 kDa by mass photometry and PFV IN was 48 kDa consistent with a monomer (S1F Fig). These data confirm the mass and assembly of intasome complexes with the correct multimerization.

MMTV intasome SEC fractions were analyzed for integration activity by agarose gel electrophoresis. Concerted integration (CI^1^) of the two vDNAs into a supercoiled (SC) plasmid results in a linear product (LN) (S1B Fig). Additional concerted integration (CI) into the linear CI^1^ product may produce an array of shorter linear products (Cl^2^). Intasomes may also integrate a single vDNA to the target plasmid, termed half site integration (HSI), appearing as a product with the mobility of a relaxed circle (RC). Gels were visualized by ethidium bromide staining of DNA and Cy5 fluorescence of vDNA and integration products (S1C and S1D Fig). Analysis of SEC fractions indicated that the octameric MMTV intasome was active, which can be easily observed by the generation of Cy5 fluorescent CI^1^ and Cl^2^ products.

Divalent cations are required for the assembly and catalytic activities of retroviral integration complexes [50–55]. HIV-1 IN is catalytically active in the presence of magnesium or manganese, but is not active in the presence of calcium [50,52,56]. This observation provided the foundation for the purification of intasomes in the presence of calcium, effectively preventing catalysis until the addition of magnesium or manganese [9,50,57]. Because the SEC purification buffer contains 1 mM CaCl₂ to stabilize the intasome complex and prevent premature catalysis, a small amount of CaCl₂ is unavoidably carried into subsequent reactions. We therefore examined MgCl₂ and MnCl₂ effects under these assay conditions, where Ca² ⁺ initially occupies the metal-binding sites and must be displaced for catalysis.

The effects of magnesium and manganese on MMTV intasome integration activity were examined in reactions that included 1 mM CaCl_2_, which was carried over from the SEC purification buffer. Cy5 labeled CI^1^ products were observed beginning at a 2 fold molar excess of MgCl_2_ to CaCl_2_, with CI^1^ products continuing to increase up to the maximum of 15 fold molar excess used in these experiments (Figs 1A and S2A). At 10 mM MgCl_2_ we observed the accumulation of Cl^2^ products. Autointegration (AI) products associated with in vitro integration reactions are the result of one vDNA integrating to a second vDNA. AI products were observed in the presence of 15 mM MgCl_2_. In the presence of 1 mM MnCl_2_ we observed Cl^1^, Cl^2^ and HSI products. (Figs 1B and S2B). The increased integration activity in the presence of MnCl_2_ likely reflects an enhanced ability of manganese ions to displace calcium ions, similar to previous studies with isocitrate dehydrogenase and chelators such as EGTA [58]. However, MnCl_2_ also substantially increased the fraction of AI products, which could potentially complicate biochemical and kinetic analysis. In addition, we have found that manganese is incompatible with the oxygen scavenging system (OSS) that is essential for single-molecule imaging analysis. Equivalent concentrations of MgSO_4_ yielded approximately 30% less CI products compared to MgCl_2_ (Figs 1C and S2C). However, MgSO_4_ resulted in virtually undetectable AI products and was used in subsequent reactions. The biochemical cause of the chloride versus sulfate anion dissimilarity in MMTV integration is unknown. PFV intasomes displayed less concerted integration in the presence of NaOAc and MgOAc compared to NaCl and MgSO_4_, suggesting that co-ions can affect multiple retroviral intasomes [53].

MMTV intasomes have different integration efficiencies in the presence of manganese or magnesium.MMTV intasome integration to a SC plasmid was performed with increasing concentrations of (A) MgCl2, (B) MnCl2, (C) MgSO4, and (D) NaCl. Integration products were separated by agarose gel electrophoresis and quantified by the Cy5 fluorescent signal in each lane (fmol). Half-site integration (HSI) products where only one vDNA has been joined to the plasmid have the mobility of a relaxed circle. A single concerted integration product (CI1) has the mobility of a linear plasmid. Additional CI events (CI2) result in a smear between CI1 and unreacted vDNA. Autointegration (AI) of vDNA to vDNA has slower mobility than unreacted vDNA. All reactions include 1 mM CaCl2 that carried over from intasome assembly and purification steps. MMTV intasomes are not active in the absence of magnesium or manganese divalent ions. Increasing concentrations of MgCl2 or MgSO4 lead to increasing total CI (CI = CI1 + CI2). AI products were only apparent at the highest concentration of MgCl2. In contrast, MMTV intasome CI was equivalent at all concentrations of MnCl2 assayed, while AI products increased with increasing concentrations of MnCl2. HSI products were unaffected by the increasing concentrations of divalent ions. MMTV intasomes were also assayed with a titration of NaCl in the presence of 20 mM MgSO4. MMTV CI products peaked in the physiologically relevant range of 100-150 mM NaCl. Fewer CI products were observed at >150 mM NaCl. MMTV integration efficiency appeared to be reduced at 50 mM NaCl, a concentration less than physiological, but was not statistically significant. Error bars indicate the standard deviation of at least 3 independent experiments performed with at least 2 independent MMTV intasome preparations.

MMTV intasomes function best at physiologically relevant ionic strength

The integration activity of MMTV intasomes was assessed over a range of NaCl concentrations (Figs 1D and S2D). The formation of CI products peaked at 150 mM NaCl and appeared to rapidly decrease to nearly undetectable at 300 mM NaCl. We conclude that the maximal MMTV activity occurs within the physiological range of ionic strength. All subsequent biochemical, kinetic and single-molecule studies of MMTV intasome activity were performed in 110 mM NaCl, in a final reaction buffer composition similar to physiological ionic strength.

MMTV intasome integration displays saturation kinetics

MMTV intasome integration to a SC plasmid target DNA was measured over time. MMTV intasomes generated CI products for at least 80 min with product saturation occurring when only 50–60% of the target SC plasmid remained (Figs 2A and S3A). In contrast, HSI products reached a plateau after 5 min and remained at ~10% of the total integration products. These results appear consistent with the conclusion that HSI mostly results from a fraction of defective intasomes rather than an intermediate that progresses to a CI product.

Time course and catalytic rate of MMTV intasomes.(A) MMTV intasomes were evaluated for integration to a SC plasmid for up to 3 h. Integration products were separated by agarose gel electrophoresis and quantified by the Cy5 fluorescent signal in each lane (fmol). Based on the accumulation of Cy5 CI products, MMTV intasomes retained activity for at least 80 min. (B) The initial rate of reaction was estimated based on velocity vs substrate concentration. Increasing SC plasmid concentrations were converted to nM sites (0.25, 0.3, 0.35, 0.4, 0.45, 0.5, 1, 3 and 6 nM) considering the number of possible MMTV intasome binding sites in the 3 kb target plasmid. The MMTV intasome minimal recognition site is 6 bp, with a likely higher limit of 12 bp. This range of site size yields a Km•app range from 34.5 to 69.0 nM. The calculated kcat•app for MMTV intasomes is 0.01 min-1. (C) MMTV intasomes were assayed for integration to recombinant nucleosomes with 147 bp of linear DNA wrapped around a histone octamer. CI reaction products displayed faster mobility than unreacted target DNA and were visualized and quantified as in (A). Error bars indicate the standard deviation of at least 3 independent experiments performed with at least 2 independent MMTV intasome preparations.

Because the integration reaction kinetics were linear when the CI products were below 20 fmol we performed a Michaelis-Menten substrate concentration analysis with supercoiled plasmid DNA (Fig 2B). Sequencing of integration sites in genomic DNA indicated that MMTV has no apparent DNA sequence preference [46]. Structural studies have suggested that the MMTV intasome may occupy a 6–12 bp footprint [9]. Together these observations suggest an ~ 30 fold excess of target sites at these integration assay conditions. Moreover, retroviral intasomes do not turn over. As might be expected, we found that the MMTV intasome substrate-dependent kinetic data fit extremely well to saturable enzymes with a deceleration component at elevated substrate concentration resulting from the lack of enzyme turnover [59] (Fig 2B). We calculated an apparent kcat•app (0.01 min^-1^) that reflects the bulk enzyme rate of integration catalysis (Fig 2B). Because the occupied site size is not precisely known we could only calculate a range for the Km•app (34.5 - 69.0 nM). These results echo previous studies that have suggested retroviral integration is inefficient [30,32]. Since an infected cell is likely to harbor a single integration complex capable of catalyzing integration, the cellular concentration will be well below nM suggesting that successful target site identification will be limiting [33].

Chromatin, rather than supercoiled plasmid DNA, is the natural target of retroviral integration during infection. MMTV intasome integration to a nucleosome target was also assayed (Figs 2C and S3B). The kinetics of concerted integration to a nucleosome were similar to integration to a supercoiled plasmid. Previous studies of PFV intasomes also revealed the same kinetics of concerted integration with either supercoiled plasmid DNA or nucleosomes as the target [20,34]. This suggests that PFV and MMTV intasomes are able to equally utilize supercoiled plasmid DNA or nucleosomes as an integration substrate.

MMTV intasome aggregation can be reduced by altering solvent components

Purified PFV intasomes appear to readily aggregate in solution resulting in substantial loss of integration activity [34]. This aggregation at least partly explains the lack of PFV integration activity after 5 min at 37˚ C. Single-molecule total internal reflection fluorescence (smTIRF) imaging employs an OSS to reduce photobleaching of fluorophores by reactive oxygen species (ROS) [60–62]. Protocatechuate 3,4-dioxygenase (PCD) uses protocatechuic acid (PCA) as a substrate to effectively scavenge free oxygen, which contributes to ROS-induced photobleaching. Interestingly, PCA prevented aggregation of PFV intasomes and enhanced integration efficiency [34]. Acetylated bovine serum albumin (BSA) also enhanced PFV integration efficiency, presumably by stabilizing the PFV intasome similar to other enzymes in solution [63,64].

We evaluated the effects of PCA and acetylated BSA on MMTV integration and found that neither significantly enhanced MMTV integration when included as a buffer component (S4 Fig). Moreover, elevated concentrations of PCA appeared to reduce MMTV intasome CI activity, though this difference did not reach statistical significance.

We evaluated the aggregation of MMTV intasomes in reactions that included CaCl_2_ (1 mM) carried over from the purification process. This analysis relied on centrifugal precipitation of high molecular weight aggregates. We compared the relative quantity of precipitated aggregates in the absence or presence of additives. A significant reduction in aggregation was observed with the addition of acetylated BSA (47%, p = 0.003; Fig 3A, lane 7 compared to lane 1). This result parallels studies with PFV intasomes and suggests acetylated BSA fundamentally alters MMTV intasome solvation [34,63,64]. MMTV intasome aggregation was reduced by 34% with additional CaCl_2_ (10 mM, p = 0.008; Fig 3A). MgSO_4_ also reduced the aggregation of MMTV intasomes by 29% (20 mM, p = 0.05). Elevated NaCl concentrations similarly reduced MMTV intasome aggregation (Fig 3B). However, MMTV intasome integration activity peaked at physiological monovalent salt concentrations (100–150 mM NaCl, Fig 1D) despite the relatively higher aggregation. It is notable that the prevention of MMTV intasome aggregation at the higher NaCl concentrations was similar to that achieved in the presence of 200 µg/ml acetylated BSA.

MMTV intasome aggregation.MMTV intasomes were incubated for 1 h at 37˚ C in integration reaction buffer with indicated concentrations of components. All aggregation reactions include 1 mM CaCl2 carried over from the purification buffer. Aggregates were pelleted by centrifugation and analyzed by denaturing PAGE stained with Coomassie blue. (A) The percentage of precipitated MMTV IN (% Aggregation) is relative to the control lane (lane 1). Increasing concentrations of MgSO4 led to decreased aggregation of MMTV intasomes. Addition of 10 mM CaCl2 (final concentration 11 mM CaCl2) reduced aggregation, similar to the highest concentrations of MgSO4. Addition of acetylated BSA was the most effective at preventing MMTV intasome aggregation. (B) MMTV intasome aggregation in the presence of 20 mM MgSO4 was assayed with a titration of NaCl. NaCl concentrations >100 mM also led to a decrease of MMTV intasome aggregation. The difference in aggregation between 100 mM and 200 or 300 mM NaCl was significant (p < 0.05), but there was no significant difference between aggregation in 150, 200, or 300 mM NaCl. Three independent experiments were performed with at least 2 independent MMTV intasome preparations.

Octameric MMTV intasomes display slower strand transfer kinetics and DNA diffusion than tetrameric PFV intasomes

Retroviral intasomes covalently join each end of the viral cDNA to the host DNA via two independent strand transfer reactions. Single-molecule magnetic tweezers (smMT) analysis was utilized to determine the real-time strand transfer kinetics of MMTV intasomes. In this system a target DNA is tethered at one end to a passivated flow-cell surface by multiple contact points along the backbones of both DNA strands, while the second end is tethered by multiple contact points also along the backbones of both DNA strands to a paramagnetic bead (PMB) [33]. Counter-clockwise turns of a neodymium magnet at low magnetic field force (0.3 pN) induced negative supercoils which are a preferred substrate for MMTV intasomes. The time between the two intasome-catalyzed strand transfers (τ_ST_) may be distinguished with smMT by tracking the vertical and horizontal motions of individual PMBs (Figs 4A and S5A). The first strand transfer is visualized as the PMB position changes on the z-axis when the MMTV intasome has nicked the DNA releasing the negative supercoils. The second strand transfer is observed when the intasome nicks the second DNA strand leading to a double strand break and the PMB position is altered in all axes. MMTV intasomes displayed a τ_ST•MMTV_ that was nearly 3 fold longer than the τ_ST•PFV_ reported for PFV intasomes (Fig 4B,τ_ST•MMTV_ = 1.31 ± 0.12 s; τ_ST•PFV_ = 0.47 ± 0.05 s; mean ± s.e.) [33]. Moreover, the maximum time observed between MMTV intasome strand transfers was 8 s (N = 41), while the maximum time observed for PFV intasome strand transfers was 2.5 s (N = 38). The relaxation time (τ_RE_) that measures the release of the supercoils from the target DNA following the first strand transfer was extracted from the integration data (S5B Fig, 0.21 ± 0.01 s), which is similar to previous results with the PFV intasome (0.26 ± 0.09 s) as well as nickase-induced supercoil relaxation (0.25 ± 0.05 s) [33]. These results are consistent with the conclusion that the strand transfer kinetics of MMTV intasomes are intrinsically slower than PFV intasomes, but this difference is not a result of MMTV intasome interference with the fundamental kinetics of DNA supercoil relaxation. These studies are unable to distinguish whether the longer MMTV strand transfer time is a result of the larger octameric intasome, the longer distance between strand transfer sites on the target DNA (6 bp for MMTV versus 4 bp for PFV), or some other unknown catalytic factors.

MMTV intasome strand transfer kinetics and search dynamics.(A) The time between the two strand transfers (τST) of MMTV vDNA was determined by single-molecule magnetic tweezers. A linear DNA was attached at multiple points to a surface and a paramagnetic bead. Magnets above the microscope stage introduced negative supercoils to the DNA, effectively reducing the apparent height of the bead. The beads were tracked in the x-, y-, and z-axes. The first strand transfer event introduces a nick in the target DNA releasing the supercoils, and changing the z-axis position (blue triangle). The second strand transfer introduces another nick and releases the bead, seen as changes in x-, y-, and z-axes. A representative integration event is shown. The τST is the time between the first and second strand transfer events. (B) The τST measured for 41 independent integration events was used to construct a histogram. For MMTV intasomes τST is 1.31 ± 0.12 s. (C) Fluorescently labeled MMTV intasomes moving on linear DNA were tracked by TIRF microscopy. A 24 kb linear DNA was attached at both ends to a flow cell surface. 50-100 pM fluorescently labeled MMTV intasomes were added to the flow cell. A representative kymograph of a single MMTV intasome in association with target DNA is shown; the x axis of the trace is time and the y axis is the length of the target DNA. (D) The lifetime (τON) of MMTV intasomes in association with target DNA was measured at multiple concentrations of NaCl. Increasing NaCl concentrations had an inverse relationship with the τON of the intasomes. (E) Diffusion coefficients of MMTV intasomes were measured at multiple NaCl concentrations. The diffusion coefficients were similar, indicating a random walk search of ~2.5 kb at 110 mM NaCl. Error bars indicate the standard deviation of at least 3 independent experiments performed with at least 2 independent MMTV intasome preparations.

Previous studies with PFV intasomes suggested that the rate-limiting step for integration was the DNA search process [33,35]. We used real-time smTIRF microscopy to visualize single MMTV intasomes on a 24 kb λ-based target DNA, as previously described [33]. Numerous single particles were observed moving along the DNA (Figs 4C and S6A and S1–S4 Movies). The lifetime of the MMTV intasome-DNA interaction decreased with increasing salt concentration consistent with increasing ionic shielding of the DNA backbone from protein binding activity (Figs 4D, S6B, and S7A-S7D). Within the physiological ionic window, the lifetime of MMTV intasomes on the target DNA appeared to be 4.6 ± 0.6 s, approximately twice as long as PFV intasomes (2.1 ± 0.1 s) [33].

The diffusion coefficient of MMTV intasome particles on the DNA (D_MMTV_ = 0.035 ± 0.02 μm^2^/s, 110 mM NaCl; Figs 4E, S6B, S6C, and S6D) was approximately 2 fold slower than PFV intasomes (D_PFV_ = 0.082 ± 0.005 μm^2^/s, 110 mM NaCl) [33]. These observations are consistent with an increased Stokes’ drag associated with the larger octameric size of MMTV intasomes compared to the tetrameric PFV intasomes. We determined that the diffusion coefficient was constant over a range of ionic strength, consistent with the conclusion that the MMTV intasome maintains continuous contact with the DNA backbone (Figs 4E and S6B). The diffusion coefficient indicates that the MMTV intasome may interact with approximately 2.5 kb of target DNA during an average 4.60 ± 0.57 s lifetime in 110 mM NaCl, most likely in a rotation-coupled site-search capacity. We did not observe any concerted integration events during 322 searches of smTIRF target DNA. The lack of efficient integration into a linear target DNA is typical for retroviral intasomes [33].

Catalytically inactive intasomes were assembled with the point mutant MMTV IN(D122N) and Cy5 labeled vDNA. These inactive MMTV intasomes displayed a similar lifetime (4.71 ± 0.33 s) and diffusion coefficient (0.033 ± 0.02 μm^2^/s) to wild type MMTV intasomes (S6C, S6D, and S7E Figs). These results are consistent with the conclusion that catalysis of strand transfer is independent of DNA binding and diffusion along the target DNA backbone [33].

MMTV intasomes form filaments on target DNA

At relatively low MMTV intasome concentrations (50–100 pM) single particles may be observed interacting with and diffusing along a target DNA by smTIRF. However, when the MMTV intasome concentration is increased to 2 nM, particles appear to aggregate and ultimately form nucleoprotein filaments along the entire length of the 24 kb duplex target DNA (Fig 5A). Moreover, at the initiation of these aggregates there is no visible motion of the intasomes along the DNA. Importantly, we observed little if any integration events that would result in DNA breakage and segregation of the two halves into visible globular aggregates (Fig 5A). We examined the stability of these aggregates by centrifugal precipitation analysis. Intasomes reconstituted with wild type IN or IN(D122N) appeared significantly more prone to precipitation in the presence of linear DNA (Fig 5B). Together these observations suggest that MMTV intasomes may spontaneously form stable aggregates that progress to nucleoprotein filaments when in proximity on target DNA.

MMTV intasomes are prone to aggregation in the presence of target DNA.(A) When 2 nM MMTV intasomes were injected in the flow cell they aggregated on the stretched DNA molecules. The images suggest recruitment and accumulation of additional MMTV intasomes binding the target DNA molecules over time. (B) MMTV intasome aggregation was assayed in the absence or presence of 3 kb linear DNA. Wild type and catalytically inactive IN(D122N) intasomes were incubated for 1 h at 37˚C in the presence of 110 mM NaCl. Aggregates were pelleted by centrifugation and analyzed by denaturing PAGE. Both wild type and mutant intasomes displayed increased aggregation when linear DNA was present, indicating that aggregation does not require catalysis. The graph represents the mean of three independent experiments performed with at least two independent MMTV intasome purifications.

Discussion

The tetrameric PFV intasome was the first high resolution retroviral intasome structure described and has served as a model for studies of intasome mediated integration [14,16]. Two catalytically active inner PFV IN protomers were shown to position the vDNA ends near a target DNA, while the outer two IN protomers were not completely resolved but appear to play structural roles in stability of the complex [14,16]. Several other retroviral intasome structures have been visualized in recent years revealing a spectrum of IN multimers [9,10,12–14,16,18,22,23]. A common feature of all these intasome multimers is a CIC [27]. The roles of additional IN protomers outside the CIC are largely unknown.

Here we characterized the activity of octameric MMTV intasomes under a variety of biochemical conditions. Importantly, calcium cations are essential for the assembly of stable intasomes because they stabilize the IN-vDNA interaction but are incapable of acting as a cofactor during strand transfer [9,50]. These calcium cations must be displaced by magnesium or manganese that can serve as metal ion cofactors capable of coordinating the S_N_2 nucleophilic integration reaction [65]. We noted a significant increase in integration products in the presence of the manganese cation (Fig 1). Manganese cations have been shown to alter the catalytic activity of the transposase from bacterial transposon Tn10, while HIV-1 IN displays equivalent activity in magnesium or manganese [50,52,56,66]. Similarly, PFV IN displays equivalent strand transfer activity in both divalent cations, but exhibits increased 3’-end processing in the presence of manganese [51]. Since the MMTV intasomes were assembled with oligomers that mimic the 3’-end processed vDNA the increased integration activity with manganese cannot result from enhanced 3’-end processing but rather improved strand transfer. Consistent with this conclusion, manganese also led to increased AI products that can only be formed by strand transfer. Manganese cations display a more relaxed catalytic coordination that may facilitate end-to-end joining of vDNA, non-specific DNA endonuclease activity, and/or non-specific alcoholysis [67–69]. However, we regard it more likely that the enhanced catalytic activity exhibited by the manganese cation reflects its ability to displace calcium from the catalytic active site similar to other enzymes [58]. MMTV intasomes also displayed variable AI activity in the presence of MgCl_2_ or MgSO_4_ suggesting that co-ions also unexpectedly affect catalysis. These observations highlight the unique behavior of different retroviral IN enzymes to variably employ different divalent cations and co-ions.

Detailed dynamic processes of PFV intasome association and integration into a target DNA have been visualized by smTIRF and smMT techniques. These previous studies suggested that the PFV intasome binds and moves along a linear target DNA for 2.1 sec by rotation-coupled diffusion [33]. Together with an observed diffusion coefficient of 0.082 ± 0.005 μm^2^/sec, it was calculated that PFV intasomes interrogate ~1.6 kb of DNA during every binding event [33]. Remarkably, only one of every 100–300 binding events resulted in integration [33]. Nevertheless, once an integration site was identified, the time between sequential strand transfer events was 0.47 s on average [33].

Interestingly, the time between MMTV intasome strand transfer events (1.31 s) was ~ 3 fold slower than PFV intasomes. The slower strand transfer kinetics may reflect an intrinsically slower strand transfer chemistry or the relatively increased distance between the strand transfer joining points (4 bp for PFV and 6 bp for MMTV), among other possibilities [70]. The average lifetime of MMTV intasomes on the DNA (4.6 s) was 2.2-fold longer than PFV intasomes. Taken together these observations suggest that MMTV intasomes are catalytically less efficient than PFV intasomes. This conclusion is further mirrored by bulk integration kinetic analysis (Fig 2) that suggested a very slow kcat•app (0.01 min^-1^) for MMTV integration catalysis and relatively high Km•app (34.5 - 69.0 nM).

Multiple mechanisms have been described for molecular searches of DNA [71–77]. An ionic strength-independent diffusion was observed for both MMTV intasomes and PFV intasomes in agreement with one common search mechanism [33]. As predicted, the diffusion of the octameric MMTV intasome on target DNA (D_MMTV_ = 0.035 ± 0.02 μm^2^/s) was significantly slower than that of the tetrameric PFV complex (D_PFV_ = 0.082 ± 0.005 μm^2^/s). These data are consistent with a model of one-dimensional diffusion in continuous contact with the DNA backbone [71]. Using the relationship D = µk_B_T, where µ is the value for mobility, the reduced diffusion coefficient implies a roughly 2.3-fold lower apparent mobility for MMTV intasomes relative to PFV intasomes. The mass-photometry and SEC measurements (1.4-fold and 1.7-fold differences, respectively) suggest that molecular size and shape likely contribute, although they may not fully explain the observed decrease. Additional, transient DNA contacts, particularly involving CTDs of non-catalytic protomers, could further modulate intasome diffusion characteristics. Indeed, PFV intasomes lacking outer-protomer CTDs show diminished nucleosome binding and are not readily visualized on DNA by smTIRF [20]. Thus, both increased size and enhanced DNA engagement likely underlie the slower diffusion of MMTV intasomes compared with PFV.

The combination of 1D and 3D mechanisms enhance DNA site search efficiency. Rotation-coupled 1D facilitated diffusion has been observed for the Type II restriction endonuclease EcoRV, which is also able to perform 3D dissociation/reassociation searching [78,79]. The single-molecule imaging approaches presented here are unable to determine a role of 3D searching during retroviral intasome interaction with target DNA. Studies of HIV-1 IN predating intasome purification methods suggested a quick commitment to target DNA suggesting that this retroviral IN is incapable of 3D searching [80]. However, similar experiments with recombinant PFV IN or intasomes showed it to be more promiscuous for target DNA, supporting a hypothesis that retroviruses may employ multiple target search mechanisms [51,53].

Our in vitro smTIRF platform provides quantitative single-particle measurements of intasome engagement with a defined DNA substrate, allowing direct quantitation of target-search parameters such as diffusion coefficient and lifetime. The values observed for wild type and catalytically inactive intasomes are similar, consistent with target search being largely decoupled from the chemistry step. While this configuration robustly quantifies 1D search dynamics after target engagement, it is not optimized to measure absolute binding/landing rates. Extending such measurements to the nuclear environment would be of considerable interest, as chromatin architecture, macromolecular crowding, and the composition of the pre-integration complex are likely to influence target engagement in vivo. However, current live-cell single-molecule imaging generally does not provide resolution required to directly resolve 1D diffusion along chromatin with the precision achieved in vitro. Nonetheless, cellular imaging approaches can detect higher-order behaviors of viral replication complexes, including the formation of discrete nuclear foci, offering tractable routes to test in vivo correlates of the intasome behaviors described here [81].

There is no known host co-factor for either PFV or MMTV intasomes, unlike lentiviral intasomes that require LEDGF/p75 or gamma retroviral intasomes which require Brd4 [9,10,12,43,82–86]. During infection, PFV and MMTV integration profiles suggest that there is little to no preference for genomic features, such as transcription units or promoters [18,37,44,87,88]. This information offers little insight as to how PFV or MMTV intasomes find a target site in vivo. Although these retroviruses lack the strong chromatin-associated biases observed for lentiviruses and gammaretroviruses, MMTV integration sites exhibit modest local sequence preferences, including enhanced A-form DNA and enrichment of flexible pyrimidine/purine steps at the integration site [24,46]. Host co-factors, such as LEDGF/p75, that include chromatin and DNA binding domains may dramatically alter the search mechanisms of intasomes. Thus, the influence of a host co-factor search mechanism on intasome search dynamics remains to be explored.

Finally, we have found that increasing the concentration of MMTV intasomes results in the formation of stable aggregates and filaments on the target DNA (Fig 5). We note that HIV-1 pre-integration complexes have been observed to merge and form clusters in the nuclei of primary monocyte derived macrophages [81]. In vitro, HIV-1 intasomes have also been reported to undergo liquid-liquid phase separation (LLPS), which may promote higher-order assemblies that enhance strand transfer activity [89]. Moreover, HIV-1 and SIVrcm intasomes that appear to consist of at least a dodecamer of IN subunits also form filaments in solution in the absence of a target DNA [25,26]. This type of aggregation activity suggests that IN associations outside the CIC may reflect both specific and non-specific interactions. Studies of HIV-1 and MLV pre-integration complexes derived from cells showed that several hundred bp of viral cDNA are protected from DNA footprinting, suggestive of protein binding to this region [90–93]. The catalytic IN protomers protect only ~20 bp at the cDNA termini [5]. It is conceivable that filaments of IN nucleate at the cDNA termini and extend inward protecting the nascent viral genome before integration. This phenomenon would be mediated by the inherent non-specific DNA binding activity of the CCD and CTD as well as inter-protomer binding of IN domains.

It is notable that MMTV intasomes resemble PFV intasomes in DNA search mechanics, strand transfer kinetics, lack of a requisite host co-factor, and ease of recombinant intasome reconstitution. Alternatively, MMTV integration complexes more closely resemble HIV-1 integration complexes in IN multimerization beyond the CIC as well as aggregate/filament formation. These observations appear to suggest that the MMTV intasome occupies an assembly and catalytic median between the intasomes of PFV and HIV-1. It is likely that comparison of PFV, MMTV, and HIV-1 intasome dynamics will help to parse the mechanical processes that define integration for distinct retrovirus species.

Materials and methods

Chemicals

Chemicals were at least 98% pure and were purchased from Sigma-Aldrich, Millipore-Sigma, GoldBio, VWR and ThermoFisher. NHS ester Cy3 and Cy5 dyes were purchased from Lumiprobe. DNA oligonucleotides were synthesized by IDT (Newark, NJ).

MMTV integrase purification and labeling of viral DNA

The MMTV IN expression construct included an N-terminal hexa-histidine tag, thrombin site, and human rhinovirus (HRV) 3C protease site. The protein was induced with 0.25 mM IPTG in E.coli Rosetta BL21(DE3) (Novagen) in the presence of 50 µM ZnCl_2_ at 37˚ C overnight. Bacteria were lysed in 20 mM HEPES (pH 7.5), 1 M NaCl, 5 mM CHAPS and 1 mM PMSF. Cells were sonicated and centrifuged at 120,000 x g for 1 h at 4˚ C. The supernatant was applied to Ni-NTA Superflow resin (Qiagen) and proteins eluted with a gradient of 20–200 mM imidazole (pH 8.0). The histidine tag was removed by cleavage with HRV 3C protease overnight at 4˚ C. MMTV IN was further purified with heparin sepharose (GE Healthcare). Pooled fractions were dialyzed in 20 mM HEPES (pH 7.5), 1 M NaCl, 5 mM CHAPS, 2 mM DTT, 0.5 mM EDTA and 10% glycerol. Catalytically inactive MMTV IN(D122N) was prepared by the same purification method. Oligonucleotides of the transferred strand 5’ CAGGTCGGCCGACTGCGGCA 3’ and non-transferred strand 5’ AATGCCGCAGTCGGCCGACCTG 3’ mimic the MMTV U5 end. T indicates the position of a 4-amino-thymine used for conjugation of a Cy3 or Cy5 NHS ester. Labeled oligonucleotides were purified by HPLC Poroshell 120 EC-C18 reverse phase column (Agilent) and 12% urea PAGE. The purified oligonucleotide was annealed to the non-transferred oligonucleotide yielding a vDNA with recessed 3’ hydroxyls.

MMTV intasome assembly

Intasomes were assembled as previously described [9]. Briefly, a 3:1 ratio of purified MMTV IN:vDNA in 20 mM HEPES (pH 7.5), 600 mM NaCl, 2 mM DTT was dialyzed overnight at 4˚ C against 25 mM Tris-HCl (pH 7.5), 80 mM NaCl, 2 mM DTT, 25 µM ZnCl_2_, 10 mM CaCl_2_. The NaCl concentration was increased to 250 mM and incubated 1 h on ice. Intasomes were purified by size exclusion chromatography with a Superdex 200 Increase (10/300) column (GE Healthcare) equilibrated in 25 mM Tris-HCl (pH 7.5), 200 mM NaCl, 2 mM DTT, 25 µM ZnCl_2_, 10 mM CaCl_2_ and 10% glycerol. Individual fractions were frozen with liquid nitrogen and stored at -80˚ C. Frozen MMTV intasomes retained activity for at least 6 months. PFV integrase and assembled intasomes were purified as previously described [15,53].

Single molecule mass photometry

Measurements were performed on a Refeyn OneMP instrument with calibrations done using MassFerence P1 (range 90–1000 kDa) with manufacturer slides and gaskets (Refeyn). Protein or intasomes purified with a Superdex 200 Increase (10/300) column (GE Healthcare) were measured in imaging buffer (25 mM Tris-HCl (pH 7.5), 200 mM NaCl, 2 mM DTT, 25 µM ZnCl_2_, 10 mM CaCl_2_ and 10% glycerol) with concentrations ranging from 0.5 to 20 nM with most measurements performed at 10 nM protein. 60 second movies were acquired at 999 Hz with an exposure time of 0.95 ms by AcquireMP software (Refeyn) followed by processing with DiscoverMP software (Refeyn). All samples were analyzed at least three times.

Integration assays

Integration reactions were 20 mM HEPES (pH 7.5), 110 mM NaCl, 4 µM ZnCl_2_, 10 mM DTT, 20 mM MgSO_4_, 1.8 nM 3 kb supercoiled pGEMT plasmid (Promega), and 20 nM MMTV intasomes in a final volume of 15 µL. Excluding kinetic assays, all integration reactions were incubated for 1 h at 37˚ C. Reactions were stopped with the addition of 0.5% SDS, 0.5 mg/mL proteinase K, 25 mM EDTA (pH 8.0) and incubated for 1 h at 37˚ C. 18 fmol of a 717 bp linear DNA was added to some reactions as a gel loading control. Integration products were resolved by 1.25% agarose gel electrophoresis, stained with ethidium bromide and scanned for Cy5 and ethidium bromide fluorescence (Sapphire Biomolecular Imager, Azure Biosystems). Images were quantified with GelAnalyzer software (Azure Biosystems). Cy5 fluorescent products include unreacted vDNA, concerted integration (CI^1^) of both vDNA resulting in linear products, and half site integration (HSI) of a single vDNA to the target DNA yielding a relaxed circle. Ethidium bromide images visualize the unreacted supercoiled target DNA, linear and relaxed plasmid DNA, and the loading control. Multiple CI events to a target DNA resulted in a Cy5 smear (Cl^2^ products). The Cl^2^ products were included in the calculation of total CI (Cl = CI^1^ + Cl^2^) as previously described [34]. Nucleosomes were assembled from recombinant human histones and the 601 nucleosome positioning sequence as described [20]. All experiments were performed in triplicate with at least two independent preparations of MMTV intasomes. Statistical significance of integration reactions was determined by paired t-test to generate two-tailed P-values (Microsoft Excel).

Aggregation assays

20 nM MMTV intasomes were incubated in 20 mM HEPES (pH 7.5), 110 mM NaCl, 4 µM ZnCl_2_, 10 mM DTT, 20 mM MgSO_4_, except where noted, for 1 h at 37˚ C in 100 µL total volume. Where indicated, 1.8 nM linear 3 kb DNA was included. Samples were centrifuged at 18,000 x g for 30 min at 4˚ C. Pellets were resuspended in denaturing PAGE buffer and resolved by 12% PAGE. Gels were stained with Coomassie Brilliant Blue, scanned (Sapphire Biomolecular Imager, Azure Biosystems), and quantified (GelAnalyzer software, Azure Biosystems). Band intensities were normalized to the sample without MgSO_4_ or the sample with the lowest concentration of NaCl. Statistical significance of aggregation reactions was determined by paired t-test to generate two-tailed P-values (Microsoft Excel).

Single-molecule magnetic tweezers

Single-molecule magnetic tweezers analysis was performed as previously described [33,34]. Briefly, a 7 kb DNA substrate was generated by ligating a linker with multiple biotins to one end and a linker with multiple digoxygenins to the other end. Glass slides were treated with 3-aminopropyl triethoxysilane and passivated with a biotin-PEG SVA/mPEG SVA mix (Invitrogen). Treated glass slides were assembled with double sided tape and an aluminum backing to generate flow channels in a custom microscope slide. The DNA substrate was tethered to the surface via NeutrAvidin (Invitrogen). Paramagnetic beads (Thermo Fisher Scientific) coated with anti-digoxygenin (Novus Biologicals) were injected into the channel. Ten negative supercoils were induced by an equivalent number of counter-clockwise turns of a neodymium magnet above the flow cell slide at 0.3 pN force. Reactions containing 20 nM MMTV intasomes in 30 mM Bis-tris propane (pH 7.5), 110 mM NaCl, 4 µM ZnCl_2_, 250 µM DTT, 20 mM MgSO_4_, 200 μg/mL acetylated BSA, 0.02% IGEPAL were flowed into the channel. Movies were recorded for 30 min at a 100 ms frame rate. The 3D positions of the paramagnetic beads were evaluated using the tracking software Video Spot Tracker (CISMM at UNC-CH). Resultant coordinates were analyzed using custom MATLAB scripts (MathWorks). The time between the two strand transfer events (τ_ST_) was determined for concerted integration events; the initial change in the z-position corresponded to the first strand transfer and subsequent movement in all axes (x-, y- and z-) indicated the second strand transfer. Histograms generated from n events were fit as a single exponential decay to determine the mean τ_ST_ and standard error (Origin, OriginLabs). Binning of histograms was performed as described previously with a bin minimum of 100 ms [33,94,95].

Single-molecule total internal reflection fluorescence

Single-molecule total internal reflection fluorescence (smTIRF) was performed as previously described [33]. Briefly, phage λ DNA was digested with restriction endonuclease XmaJI to generate 24 kb fragment DNAs. Biotin labeled DNA linkers were ligated to the fragment ends. Flow cells were assembled from quartz slides treated with 1 biotin-PEG-SVA:300 mPEG-SVA and glass cover slips treated with mPEG-SVA. NeutrAvidin was injected into the flow cell followed by addition of biotinylated λ fragments stretched by flow.

Imaging was performed with 50–100 pM or 2 nM MMTV intasomes with Cy5 or Cy3 labeled vDNA in 30 mM Bis-tris propane (pH 7.5), 110 mM NaCl, 20 mM MgSO_4_, 4 µM ZnCl_2_, 100 µM DTT, 200 μg/mL acetylated BSA, 0.02% IGEPAL, 20 nM protocatechuate 3,4-dioxygenase (PCD), and 5 mM protocatechuic acid. PCD was prepared as described [61,62]. Fluorescence was detected with a custom built prism TIRF microscope (Olympus IX-71, water-type 60X objective NA = 1.2, 1.6X extended magnification) and recorded on an electron-multiplying charge-coupled device camera (EMCCD, Princeton Instruments, ProEM 512 excelon). Lifetime and diffusion were recorded for 1200 s at a 250 ms frame rate. Following the real-time recordings, the target DNA was stained with Syto 59 Red Fluorescent Nucleic Acid Stain (Thermo Fisher Scientific) and the images overlaid onto the Cy5 or Cy3 movies. Particles were tracked using DiaTrack software (Sydney, Australia) and the results were analyzed with MATLAB (MathWorks) and Origin Pro (OriginLab) [96].

Supporting information

S1 FigPurification and activity of Cy5 labeled MMTV intasomes.(A) SEC chromatogram of Cy5 labeled octameric MMTV intasome (elution at ~10 mL volume), dimeric MMTV IN (elution at ~13 mL) and Cy5 labeled DNA (elution at ~16 mL). Fraction numbers (top) and elution volume (bottom) are indicated. Fluorophore Cy5 is excited at 650 nm. Protein molecular weight markers (670, 158, 44, and 17 kDa) were profiled in the MMTV intasome buffer and generated a standard curve (red line). The standard curve equation Log10 (MW) = -0.2412 (Velution) + 5.1361 (R2 = 0.9712) was used to approximate the molecular weights of the three observed MMTV species (530, 100, and 19 kDa, black arrows). (B) Integration of a viral donor DNA (vDNA) converts a supercoiled (SC) plasmid target to several products that may be resolved by agarose gel electrophoresis. Half-site integration (HSI) of a single vDNA to the plasmid results in a tagged circle with the mobility of a relaxed circle (RC). A single concerted integration of both vDNAs to the plasmid yields a product (CI^1^) with the mobility of linearized plasmid (LN). Additional concerted integration events to CI^1^ result in a smear of products with mobility between linear plasmid and unreacted vDNA (CI^2^). (C) SEC fractions 34–42, 50–52, and 62 were tested for integration with a 3 kb supercoiled plasmid target. Integration reaction products were separated by agarose gel electrophoresis and imaged by ethidium bromide staining. The ethidium bromide image revealed RC, LN, and SC plasmid as well as a linear DNA loading control (LD). (D) The identities of integration products were confirmed by Cy5 imaging of the agarose gel. RC may be the result of HSI or non-specific endonuclease nicking but the presence of Cy5 fluorescence indicates HSI. LN bands correlate with CI^1^ products. Multiple integration events to a single target DNA generate linear fragments with faster mobility than the linear CI^1^ product. This smear of CI^2^ products was more apparent with the Cy5 image. Intermolecular autointegration (AI) uses vDNA as an integration target. The highest integration activity is seen in fractions corresponding to the peak at 10 mL elution volume. (E) MMTV integrase (gray) and intasomes (red) and (F) PFV integrase (gray) and intasomes (purple) were analyzed by single molecule mass photometry to confirm the size-independent molecular weight of monomers or complexes.(TIF)

S2 FigMMTV intasome integration assays with titrations of divalent and monovalent salts.Ethidium bromide gel images correspond to the Cy5 images of Fig 1. Quantitation of the ethidium bromide bands is expressed as the percentage of total fluorescence in the lane. Addition of increasing concentrations of (A) MgCl_2_, (B) MnCl_2_, or (C) MgSO_4_ enhance MMTV intasome integration activity. (D) MMTV intasomes displayed maximal activity in the presence of 100–150 mM NaCl; higher concentrations of NaCl reduced MMTV intasome integration efficiency.(TIF)

S3 FigMMTV intasome integration over time.Ethidium bromide gel images correspond to the Cy5 images of Fig 2. Quantitation of the ethidium bromide bands is expressed as the percentage of total fluorescence in the lane. MMTV intasome integration activity increases over time to (A) supercoiled plasmid and (B) nucleosome bound DNA. Graphs represent the mean of three independent experiments performed with at least two independent MMTV intasome purifications. Error bars indicate standard deviation.(TIF)

S4 FigTitrations of PCA or acetylated BSA during MMTV intasome integration assays.(A) Increasing concentrations of PCA reduced MMTV intasome integration activity. Ethidium bromide gel image and quantitation of the ethidium bromide bands is expressed as the percentage of total fluorescence in the lane (top). The percentage of linear (LN) products appeared to decrease while the percentage of unreacted supercoiled (SC) plasmid increased, suggestive of a decrease in concerted integration as PCA concentration increased. The same gel was imaged for Cy5 fluorescence and quantified (bottom). Integration products were quantified by the Cy5 fluorescent signal in each lane (fmol). There were no significant differences in CI products throughout the titration of PCA, suggesting that any effects of this small molecule on MMTV intasomes are subtle. (B) Similar to the results with PCA, addition of acetylated BSA showed no apparent effects on MMTV intasome integration efficiency. Graphs represent the mean of three independent experiments performed with at least two independent MMTV intasome purifications. Error bars indicate standard deviation.(TIF)

S5 FigRelaxation time of a supercoiled DNA after the first MMTV strand transfer.(A) A linear DNA was attached at multiple points to a surface and a paramagnetic bead. Magnets above the microscope stage were used to introduce negative supercoils to the DNA, reducing the apparent height of the bead. The first strand transfer nicks the DNA and releases the supercoils, producing a change in the z-axis. The second strand transfer breaks the DNA and the bead leaves the field of view. Three representative integration reactions are shown. (B) The relaxation time (τ_RE_) is the time for the paramagnetic bead to move in the z axis from the supercoiled position to the extended position. The τ_RE_ is observed immediately after the first strand transfer when a nick releases the supercoils. The τ_RE_ observed with MMTV intasomes is 0.21 ± 0.01 s (s.e.), consistent with previous τ_RE_ observations of both PFV intasome and a nicking restriction endonuclease (28).(TIF)

S6 FigMMTV intasome search dynamics with naked DNA.(A) Representative traces of a single MMTV intasome in association with target DNA. The x axis of the traces is time and the y axis is the length of the target DNA. (B) The average lifetimes (τ_ON_) of MMTV intasomes in association with DNA and diffusion coefficients at multiple concentrations of NaCl. The τ_ON_ displays an inverse relationship with NaCl concentration. However, the diffusion coefficients remain constant at multiple NaCl concentrations. (C) Catalytically inactive MMTV IN(D122N) intasomes displayed similar τ_ON_ and diffusion coefficients as wild type intasomes at 110 mM NaCl. (D) Diffusion coefficients of wild type and IN(D122N) intasomes in the presence of 110 mM NaCl.(TIF)

S7 FigHistograms of observed τ_ON_ for MMTV intasomes at multiple NaCl concentrations.The τ_ON_ for wild type MMTV intasomes is shown for (A) 25 mM NaCl, (B) 50 mM NaCl, (C) 110 mM NaCl, and (D) 150 mM NaCl. (E) The τ_ON_ of catalytically inactive IN(D122N) intasomes is shown at 110 mM NaCl. Histogram fits from these plots were used to generate average τ_ON_ values.(TIF)

S1-S4 MoviesMMTV intasomes searching target DNA.A 24 kb fragment of phage lambda DNA was attached to a surface at both ends. Cy5 labeled MMTV intasomes (green) were imaged at a 250 ms frame rate for 1200 s. Following video capture, the DNA was stained with Sytox and imaged to visualize the 24 kb target DNA (magenta). The Sytox image was overlaid to the MMTV intasome video (colocalization is white). Movies are shown at 3X speed.(AVI)