DNA requirements for assembly and stability of HIV-1 intasomes

Abstract

Integration of viral DNA into the host genome is an essential step in retroviral replication that is mediated by a stable nucleoprotein complex comprising a tetramer of integrase bridging the two ends of the viral DNA in a stable synaptic complex (SSC) or intasome. Assembly of HIV-1 intasomes requires several hundred base pairs of nonspecific internal DNA in addition to the terminal viral DNA sequence that is protected in footprinting experiments. We find that only one of the viral DNA ends in the intasome requires long-nonspecific internal DNA for intasome assembly. Although intasomes are unstable in solution when the nonspecific internal DNA is cut off after assembly, they are stable in agarose gels. These complexes are indistinguishable from SSCs with nonspecific internal DNA in Förster resonance energy transfer (FRET) experiments suggesting the interactions with the viral DNA and integrase tetramer are the same regardless of the presence of nonspecific internal DNA. We discuss models of how the internal DNA contributes to intasome assembly and stability. FRET is exquisitely sensitive to the distance between the fluorophores and given certain assumptions can be translated to distance measurements. We anticipated that a set of such distance constraints would provide a map of the DNA path within the intasome. In reality, the constraints we could impose from the FRET data were quite weak allowing a wide envelope for the possible path. We discuss the difficulties of converting the FRET signal to absolute distance within nucleoprotein complexes.

Introduction

Retroviruses such as HIV-1 integrate a DNA copy of their genome into cellular DNA as an obligatory step in the viral replication cycle (reviewed in Ref.1). Retroviral infection introduces the nucleoprotein core of the virus particle into the target cell. The core includes two copies of the viral RNA, reverse transcriptase and integrase together with other viral proteins and cellular proteins acquired from the cytoplasm of the infected cell. Reverse transcription of the viral RNA by reverse transcriptase generates a linear double stranded DNA copy of the viral genome. The viral DNA remains associated with viral and cellular proteins as a large nucleoprotein complex termed the preintegration complex (PIC).² The PIC is transported from the cytoplasm to the nucleus where integrase within the PIC splices the viral DNA into the host genome. Once integrated, the viral DNA is stably replicated along with cellular DNA through multiple cycles of cell division.

Integrase is stably associated with viral DNA within the PIC. The low abundance of PICs in extracts of infected cells precludes direct biophysical analysis of these complexes. However, functional studies demonstrate that integrase remains associated even when challenged with high ionic strength.^{3, 4} This stable association of integrase with viral DNA ends can be reproduced in vitro with purified integrase and synthetic viral DNA ends.⁵ These stable synaptic complexes (SSCs) integrate their viral DNA ends in vitro with all the hallmarks of integration in the cell. Within the SSC two viral DNA ends are bridged by a tetramer of integrase,^{5, 6} forming a nucleoprotein complex called the intasome.

The crystal structure of prototype foamy virus (PFV) integrase in complex with viral DNA⁷ is currently the only available structure for a retroviral intasome. PFV efficiently forms intasomes with oligonucleotide DNA substrates that mimic the ends of the viral DNA and these intasomes are fully competent for integration in vitro. In contrast, HIV-1 integrase requires much longer DNA for assembly of intasomes and this is one of the major factors that has impeded high-resolution structural studies. The role of this nonspecific DNA internal to the viral DNA is unknown, but less than 20 base pairs of the terminal viral DNA are protected in footprinting studies of HIV-1 intasomes.⁵ We have applied Förster resonance energy transfer (FRET) to probe the role of the nonspecific internal DNA in HIV-1 intasome assembly and attempted to map the path of the viral DNA within the intasome. We find that only one of the two viral DNA ends requires long-nonspecific DNA for HIV-1 intasome assembly, and the similar FRET efficiencies between dye pairs near the DNA ends suggest that the core structure of the intasome is not influenced by the nonspecific DNA. FRET is extremely sensitive to the separation of the donor and acceptor fluorophores and, given certain assumptions, the FRET signals of fluorophore pairs located at different positions in the viral DNA can be translated into distance constraints. We discuss technical challenges that prevent the conversion of FRET efficiencies into precise distance measurements in our studies of HIV-1 intasomes. The issues are generally applicable to extracting distance measurements from FRET studies of nucleoprotein complexes, but are frequently ignored resulting in grossly underestimated errors in calculated distances.

Results

SSC assembly, image acquisition, and FRET measurement

Efficient SSC assembly with HIV-1 integrase requires viral DNA substrates greater than several hundred base pairs in length.⁵ However, the requirement for specific viral DNA sequence does not extend beyond the terminal 20 base pairs. This enables fluorescent DNA substrates to be conveniently assembled by incorporating fluorophores into oliogonucleotides corresponding to the terminal viral DNA sequence and ligating them to a restriction fragment of “nonspecific” DNA. The location of the fluorophores in the viral DNA sequence of the substrates we used is shown in Table I. In all cases, Cy3 or Cy5 was incorporated on the nontransferred strand at the positions shown. The fluorescently labeled oligonucleotides were annealed with their complementary strand terminating with a 3′ dideoxyadenosine and ligated to nonviral DNA sequence to make the 1-kb preprocessed DNA substrate with Cy3 or Cy5 at different positions from the viral DNA terminus. The labeling efficiency of the DNAs was consistently close to 90% (Supporting Information, Table SI).

Table I.

Location of Cy3 and Cy5 Fluorophores in the Viral DNA Substrate Pairs

The SSC was assembled from pairwise mixtures of viral DNA substrate containing Cy3 and Cy5 at the same distance from the terminus as shown in Table I. A meaningful FRET signal between the donor (Cy3) and acceptor (Cy5) cannot be measured in solution because only a fraction of the substrate is converted to the SSC, and the SSC aggregates together with unreacted substrate bound by integrase under assembly conditions as evidenced by pelleting of the DNA even with low-speed centrifugation in a benchtop microcentrifuge (data not shown). Treatment with high salt and EDTA greatly reduces the aggregation problem but the remaining small fraction of aggregated material and the low efficiency of SSC assembly still prevent accurate direct FRET measurements in solution. This problem was overcome by resolving the SSC in agarose gels after treatment with high salt and EDTA and directly measuring fluorescence of the SSC bands using a Typhoon 8600 laser gel scanner (GE Healthcare), essentially as described for studies of the multiprotein synaptic complex of the λ att site.⁸ Data was collected by scanning the gel at the donor, acceptor, and FRET channels (see Materials and Methods Section). Representative raw data is shown in Figure 1. FRET can be measured as either the decrease in donor fluorescence (donor-quench-based FRET) or increase in acceptor emission (acceptor-emission-based FRET). Both methods gave similar results and error ranges.

Figure 1.

Representative raw data of FRET signal of SSCs. Complexes were assembled with Cy3 DNA, Cy5 DNA, or a mix of Cy3 and Cy5 DNA and electrophoresed in an agarose gel; in all cases, the fluorophore was located at position 10 (Table I). The images were acquired by scanning the gel at the Cy3 channel (panel A), Cy5 channel (panel B), and the FRET channel (panel C) using a Typhoon 8600 laser scanner. In the Cy3 channel (panel A), DNA containing only the Cy5 fluorophore is not detected. Similarly, DNA containing only the Cy3 fluorophore is not detected in the Cy5 channel (panel B). An appreciable signal was only detected in the FRET channel (panel C) with complexes assembled with a mix of Cy3 and Cy5 DNA, although a weak spillover signal was present with complexes assembled with Cy3 or Cy5 DNA alone. The band migrating more slowly than the SSC is a complex containing two SSCs.

DNA length requirement for HIV-1 SSC assembly

Figure 2(A) shows agarose gel electrophoresis of the SSC assembled with 1 kb DNA, 350 bp DNA, or a mixture of different ratios of the long and short DNAs. The long DNA efficiently forms SSCs, which migrate as a ladder due to self-association.⁶ In contrast, SSC assembly with the short DNA alone is extremely inefficient [Fig. 2(A), lane 1]. When the complexes are assembled from a mixture of long and short DNA new bands corresponding to complexes containing one long (1 kb) and one short (350 bp) DNA are observed. Strikingly, these mixed complexes are formed with similar efficiency as complexes with the long DNA alone. Thus, only one of the two DNAs in the SSC must be long for efficient assembly. Further decreasing the length of the “short” DNA gradually decreased the efficiency of SSC formation of the mixed DNA complexes; reducing the length to 150 bp decreased the efficiency by about twofold, whereas 90 bp DNA formed SSCs with drastically reduced efficiency (Supporting Information, Fig. S1).

Figure 2.

FRET efficiencies of the SSC assembled with long (1 kb) DNA alone and one long plus one short (350 bp) DNA. (A) SSC was assembled with the indicated ratios of long and short DNA labeled with ³²P and electrophoresed in an agarose gel. The SSC migrates as a ladder due to self-association. SSC containing one long plus one short DNA assembles with a similar efficiency as complexes with two long DNAs. (B) Lane 1, SSC with long DNA labeled at position 10 with Cy3; lane 2, SSC with short DNA labeled with Cy5; lane 3, SSC assembled with a mixture of long DNA labeled with Cy3 and short DNA labeled with Cy5; lane 4, blank; lanes 5–7 are the same as lanes 1–3, respectively, except the samples were treated with 10 mM EDTA and 500 mM NaCl before electrophoresis. The gel was scanned at the Cy3 channel. (C) The same gel scanned at the Cy5 channel. (D) Overlay of Cy3 and Cy5 channels.

FRET efficiency of SSC with long–long, long–short, and short–short viral DNA end pairs

The SSC was assembled from long plus long or long plus short viral DNAs containing a Cy3/Cy5 FRET pair at position 10 from the end of the viral DNA (Table I). Complexes were electrophoresed in an agarose gel without or with treatment with NaCl and EDTA; the salt and EDTA treatment causes the ladder of self-associated SSCs to collapse into a single SSC band facilitating the FRET measurements. The gel was scanned at the Cy3 channel [Fig. 2(B)], the Cy5 channel [Fig. 2(C)], and the FRET channel; an overlay of the Cy3 and Cy5 channel signal is shown in Figure 2(D). The FRET signal for the two complexes was indistinguishable and is shown in Figure 4 (below).

Figure 4.

Comparison of FRET efficiencies of SSCs with 1 kb/1 kb, 1 kb/0.35 kb, and 29 bp/29 bp DNAs measured in-gel by acceptor emission fluorescence. The FRET pairs were located at position 10 (Table I).

The SSC is not detected in solution in the absence of the long-nonspecific internal DNA. To distinguish between the internal DNA length requirement for assembly and stability, we assembled the SSC with a 1-kb DNA with photocleavable bond between the terminal viral DNA sequence and the internal DNA segment [Fig. 3(A)]. This viral DNA substrate contained a nick opposite the photocleavable bond so that exposure to UV light results in double strand cleavage. Severing the nonspecific internal DNA of the assembled SSC in the reaction mixture resulted in dissociation as monitored by agarose gel electrophoresis (data not shown). Therefore, the nonspecific internal DNA is required for both assembly and stability.

Figure 3.

Intasomes with two short DNAs are stable in agarose gels. (A) 1 kb DNA with photocleavable spacer used for SSC assembly. (B) SSCs were electrophoresed in a first dimension in an agarose gel without deproteinization. The gel was irradiated with UV light to cleave the photocleavable spacer and then electrophoresed in the second dimension. The SSC stays intact after cleavage as evidenced by retarded mobility in the second dimension. (C) Same as panel B, except the gel was soaked in SDS before the second dimension to dissociate the SSC. The absence of the 1 kb DNA in the second dimension shows that cleavage was complete. Complexes assembled with 1 kb DNA labeled with Cy3 and Cy5 at position 10 (Table I) containing a photocleavable spacer were electrophoresed in a two-dimensional agarose gel in the same way and scanned at the Cy3 channel (D), Cy5 channel (E), and FRET channel (F). The spots migrating more slowly than the core SSC are complexes containing two or more SSCs.

Although the SSC is not stable in solution after cleavage of the nonspecific internal DNA, we find that it is stable in agarose gels (Fig. 3, panels B and C). After a first dimension of native gel electrophoresis, the gel was irradiated with UV to cleave the nonspecific internal DNA from the SSC and then electrophoresed in a second native dimension. The SSC mobility remained retarded in the second dimension, demonstrating that once formed it is stable in agarose gels after cleaving the nonspecific internal DNA. When the gel is also soaked in SDS before the second dimension, the DNA within the SSC runs at 29 bp demonstrating that cleavage is essentially complete [Fig. 3(C)].

The stability of the cleaved SSC in agarose prompted us to measure the FRET of the cleaved SSC assembled with DNA containing the same fluorophore pair as for the SSC containing two long and one long plus one short DNA (Fig. 3, panels D, E and F). The FRET of this cleaved SSC was indistinguishable from the FRET of the other two SSCs within the limits of the experimental error (Fig. 4).

Distance constraints within the SSC derived from the donor-quenching FRET measurements

We attempted to map the path of the viral DNA within the HIV-1 intasome by measuring the FRET signal with donor and acceptor fluorophores placed at various locations (Table I) on the two viral DNAs within the complex. FRET is exquisitely sensitive to the distance between the fluorophores and, given certain assumptions, can be translated to distance measurements. We hoped that a set of such distance constraints would provide a map of the DNA path within the complex.

The SSC was assembled with viral DNA ends containing the Cy3/Cy5 pairs at the positions shown in Table I and FRET efficiency was measured for each dye pair (Fig. 5). FRET was greatest for dye pairs at 10 base pairs from the ends of the DNA, suggesting that these dyes are closer together in the SSC than the pairs located at more internal positions or at the end of the DNA. However, FRET efficiency (E) depends not only on the distance between the donor and acceptor fluorophores but also on the dipole orientation factor (κ²), which cannot be directly measured. The distance (R) between donor and acceptor can be calculated from the FRET efficiency as

Figure 5.

FRET efficiencies of the SSC assembled with FRET pairs at different positions from the end of the viral DNA measured by donor quenching. The data is derived from SSCs assembled with a pair of 1 kb DNAs.

(1)

where R₀ is the Förster radius for κ² = 2/3. If we make the assumption that the fluorophores freely rotate, κ² = 2/3 and the apparent distance separating them (R) can be calculated from Eq. (1). The Förster radius for freely rotating Cy3 and Cy5 is 60.1 Å.

Although κ² cannot be directly measured, its upper and lower bounds can be estimated from the anisotropy measurements.⁹ We first tested whether fluorescence polarization measurements are compromised by secondary light scattering in agarose gels because SSC anisotropy, like FRET, is most conveniently measured in gels to circumvent the problems of SSC assembly efficiency and aggregation in solution. Preliminary experiments established that anisotropy information is not lost in agarose gels making it feasible to use anisotropy to place bounds on κ². The anisotropy data is shown in Figure 6. Both Cy3 and Cy5 exhibit considerable anisotropy when coupled to the oligonucleotides. The degree of anisotropy depends on the location of the dye in DNA and as expected is greater in the 1-kb DNA substrate and the SSC. This anisotropy necessitates that the dipole orientation factor be taken into account when converting the measured FRET to distance measurements. Anisotropy of donor, acceptor, and FRET signal in the SSC with dye pairs at the four different locations is shown in Figure 7 and the calculated upper and lower limits of κ² and the resulting uncertainty in R is tabulated in Table II. Figure 8 shows the calculated distance separation for each dye pair calculated from the donor quench-based FRET efficiencies. The solid error bar represents the error assuming a value of 2/3 for κ². The error in the distance measurements becomes considerably greater when the uncertainty in κ² is taken into account (dotted error bars). Similar results were obtained from acceptor emission FRET measurements (Supporting Information, Fig. S4).

Figure 6.

Cy3 and Cy5 fluorescence anisotropy. Cy3 (panel A) and Cy5 (panel B) anisotropy was measured in solution for single strand oligonucleotide, double strand oligonucleotide, and 1 kb DNA and in agarose gels for 1 kb DNA and the SSC. Each panel shows four sets of data, one for each position of the dye from the DNA end. For the measurements in agarose gels, the sample was electrophoresed in duplicate. One lane was stained with ethidium bromide to identify the position of the band and the corresponding slice of agarose was cut out from the other lane. The gel slice was inserted into a quartz fluorometer cell (Starna Cells, CA) and the space between the gel slice and the cell was filled with 80% glycerol. The anisotropy was then measured using a spectrofluorometer.

Figure 7.

Fluorescence anisotropy of donor, acceptor, and FRET at each position along the viral DNA within the SSC.

Table II.

Anisotropy Measurements

Position	rd	ra	rFRET	βda	κ2 range	Distance uncertainty (%)
1	0.275	0.292	−0.026	58.6°	0.15–2.50	−22/+25
10	0.297	0.313	0.22	25.5°	0.15–3.20	−22/+30
17	0.314	0.316	0.23	25.1°	0.13–3.27	−23/+30
26	0.298	0.318	−0.007	55.7°	0.12–2.70	−25/+26
Example 1	0.400	0.400	–	–	0.00–4.00	−100/+35
Example 2	0.100	0.100	0.05	35.2°	0.44–1.70	−7/+17

The anisotropy of SSC donor (r_d), acceptor (r_a), and FRET (r_FRET) signal were measured. The angle between the donor and acceptor dipoles (β_da), the upper and lower bounds of κ², and the resulting distance uncertainty were calculated from the anisotropy measurements.⁹ Examples 1 and 2 are hypothetical cases.

Figure 8.

Calculated Cy3–Cy5 separation distance in the SSC for each dye pair assuming a dipole orientation factor of 2/3 (solid errors bars) and taking into account the uncertainty in κ² (dotted error bars).

Discussion

The crystal structure of PFV integrase in complex with viral DNA substrate⁷ is currently the only high-resolution structure of a retroviral intasome. The HIV-1 intasome is likely to be very similar. Indeed, the DNA base separation distances available in the PFV intasome structure corresponding to the dye positions studied here for the HIV-1 intasome are within the error range of the distance estimates obtained. However, biochemical studies suggest there are some differences between PFV and HIV-1 intasomes. For example, PFV integrase efficiently forms intasomes with oligonucleotide viral DNA substrates, whereas HIV-1 integrase requires much longer viral DNA substrates. The current in vitro efficiency of intasome assembly with HIV-1 integrase is also much lower than with PFV integrase. The role of the nonspecific internal DNA is unclear, although the requirement for only one of the DNAs to be long leads us to favor some models over others. We initially speculated that weakly associated integrase protomers might help to stabilize the intasome by bridging the two internal DNA segments, but this model does not easily account for the abrupt change in the efficiency of SSC formation when nonspecific internal DNA extends beyond about 300 bp. The interactions are likely to be transient and atomic force microscopy (AFM) studies, which require high ionic strength during deposition on the mica surface to prevent aggregation, do not reveal interaction of the internal DNA segment with integrase.⁶ Considering that only one of the ends needs to have a long-nonspecific DNA segment, it is tempting to speculate here that there is a mechanistic similarity with the Tn10 transpososome. In this case, one DNA segment internal to the transposon ends folds back on the complex and increases the stability of the transpososome, and unfolding is a prerequisite for normal intermolecular transposition.^{10, 11} This could potentially sequester the target DNA binding site and contribute to blocking autointegration before the PIC encounters chromosomal DNA. However, the observation that the shorter partner end still needs to be much longer than the ∼20-bp terminal sequence indicates an additional stabilizing role for the end-proximal DNA segments.

FRET efficiency measurements provide a powerful tool to obtain distance information on the separation of donor and acceptor fluorophores in nucleoprotein complexes. However, the technique needs to be applied with caution when used for quantitative distance measurements rather than simply monitoring conversions between two different conformational states of macromolecular complexes. The similar FRET efficiencies of intasomes with and without nonspecific internal DNA are consistent with a common core structure irrespective of the internal DNA configuration. However, precise distance constraints within the intasome core cannot be derived from the data. Even in the absence of protein binding, fluorescence of the donor and acceptor is highly anisotropic, which reveals fluorophore rotation constraints thus introducing an inherent uncertainly in conversion of FRET efficiency to distance measurements due to uncertainty in κ². Therefore, the error in previously reported¹² distance constraints between viral DNA ends in SSCs and strand transfer complexes (STCs) (3–6Å) is grossly underestimated; Bera et al.¹² measured similar anisotropy values for the Cy3 and Cy5 labeled DNA substrates as reported here, but did not include the resulting error in their distance calculations. In principle, the problem may be overcome using different dyes or a flexible tether that interacts less with DNA, although a long tether adds an additional variable. We found that Atto dyes coupled to DNA via a tether rotate freely and the κ² = 2/3 approximation is valid (Supporting Information, Fig. S5). However, when these DNAs are incorporated into SSCs with integrase, free rotation of the Atto dye is constrained (Supporting Information, Fig. S5) and the error range of the calculated distance constraints remained large. The technical challenges that confound the interpretation of FRET data in our studies of HIV-1 intasomes are generally applicable to extracting distance information from FRET studies of nucleoprotein complexes. Unfortunately, these factors are frequently ignored resulting in grossly underestimated errors in calculated distances.

Materials and Methods

DNA substrates

Unlabeled and Cy3 and Cy5 labeled LTR-U5 oligonucleotides were purchased from Integrated DNA Technology (Iowa). The DNA sequences are listed in Table I. Fluorescent label was incorporated at internal positions in oligonucleotides by replacing the base at that position with a single Cyanine dye phosphoramidite during synthesis. The labeling efficiency was determined by measuring the ratio of absorbance at 260/549 nm for Cy3 and 260/646 nm for Cy5 containing oligonucleotides. The DNA substrates used for assembly of complexes with integrase were prepared as described.⁵ Briefly, the processed viral DNA end sequence with a 3′ ddA was made by transferring ddATP to the oligo 5′ pAATTCTTTTAGTCAGTGTGGAAAATCTCTAGC using terminal deoxynucleotidyl transferase and annealing with the Cyanine dye labeled oligo. The annealed duplex was ligated with a 970-bp DNA fragment excised from modified pCR2.1 plasmid DNA by cleavage with EcoR I and Pst I. The ∼1-kb DNA was finally purified by agarose gel electrophoresis.

Protein expression and purification

His-tagged HIV-1 integrase W235F was purified as described for wild type integrase.¹³ Briefly, the integrase was expressed in E. coli BL21(DE3) and the cells were lysed in buffer containing 0.1 M NaCl. The lysate was centrifuged and integrase was extracted from the pellet in buffer containing 2 M NaCl. Then, the protein was purified by nickel-affinity chromatography and the His-tag was removed by thrombin digestion. The aggregated portion of the integrase preparation was removed by gel filtration on a Superdex-75 column (GE Healthcare).

Complex assembly

Typical reaction mixtures (25 μL final volume) were assembled by incubating 400 nM integrase on ice in 20 mM HEPES pH 7.5, 12% DMSO, 5 mM DTT, 10% PEG-6000, 10 mM MgCl₂, 20 μM ZnCl₂, and 200 mM NaCl (final), followed by addition of 50 nM viral DNA substrate. These components were preincubated on ice for 30 min and then the reaction was initiated by transfer to 37°C and incubation was continued for 2 h. The reactions were stopped by addition of 10 mM EDTA and incubated at 30°C for an additional 30 min. Then, the mixture was centrifuged at 10,000g for 40 min, the supernatant was removed, and the pellet was resuspended in 10 μL suspension buffer (20 mM Hepes, pH 7.5, 500 mM NaCl). Then, the reaction mixture was analyzed by gel eletrophoresis.

Electrophoresis of complexes and image acquisition

Sample (10 μL) prepared as described above was loaded onto a 0.8% agarose gel in TBE buffer containing 1 M urea and electrophoresed at 5.0 V/cm for 2 h. The gel was scanned by a Typhoon 8600 (GE Healthcare). Cy3 donor fluorescence was measured by exciting with 532 nm green laser and detecting with a 580 nm_BP 30 emission filter. Cy5 fluorescence was measured by exciting with 633 nm red laser and detecting with a 670 nm_BP 30 emission filter. FRET was measured by exciting at the Cy3 excitation wavelength and detecting at the Cy5 emission wavelength.

To excise the complexes from agarose gels, an aliquot of the same mixture was loaded side by side, stained with ethidium bromide, and the corresponding band position in the gel was excised.

Apparent FRET and anisotropy measurements

Calculation of FRET efficiency from donor quenching and acceptor emission, and the uncertainty in converting the FRET data to distance constraints are described in detail in Supplementary Information Text and Supplementary Information Figures.

Glossary

Abbreviations

EDTA

ethylenediaminetetraacetic acid

SDS

sodium dodecyl sulfate

HEPES

hydroxyethyl piperazineethanesulfonic acid

DMSO

dimethyl sulfoxide

DTT

dithiothreitol

TBE

tris-borate electrophoresis buffer

Supplementary material

Additional Supporting Information may be found in the online version of this article.