Structural organization of avian retrovirus integrase in assembled intasomes mediating full-site integration.

Retrovirus preintegration complexes (PIC) purified from virus-infected cells are competent for efficient concerted integration of the linear viral DNA ends by integrase (IN) into target DNA (full-site integration). In this report, we have shown that the assembled complexes (intasomes) formed in vitro with linear 3.6-kbp DNA donors possessing 3'-OH-recessed attachment (att) site sequences and avian myeloblastosis virus IN (4 nm) were as competent for full-site integration as isolated retrovirus PIC. The att sites on DNA with 3'-OH-recessed ends were protected by IN in assembled intasomes from DNase I digestion up to approximately 20 bp from the terminus. Several DNA donors containing either normal blunt-ended att sites or different end mutations did not allow assembly of complexes that exhibit the approximately 20-bp DNase I footprint at 14 degrees C. At 50 and 100 mm NaCl, the approximately 20-bp DNase I footprints were produced with wild type (wt) U3 and gain-of-function att site donors for full-site integration as previously observed at 320 mm NaCl. Although the wt U5 att site donors were fully competent for full-site integration at 37 degrees C, the approximately 20-bp DNase I footprint was not observed under a variety of assembly conditions including low NaCl concentrations at 14 degrees C. Under suboptimal assembly conditions for intasomes using U3 att DNA, DNase I probing demonstrated an enhanced cleavage site 9 bp from the end of U3 suggesting that a transient structural intasome intermediate was identified. Using a single nucleotide change at position 7 from the end and a series of small size deletions of wt U3 att site sequences, we determined that sequences upstream of the 11th nucleotide position were not required by IN to produce the approximately 20-bp DNase I footprint and full-site integration. The results suggest the structural organization of IN at the att sites in reconstituted intasomes was similar to that observed in PIC.


Retrovirus preintegration complexes (PIC) purified from virus-infected cells are competent for efficient concerted integration of the linear viral DNA ends by integrase (IN) into target DNA (full-site integration). In this report, we have shown that the assembled complexes (intasomes) formed in vitro with linear 3.6-kbp DNA donors possessing 3-OH-recessed attachment (att) site sequences and avian myeloblastosis virus IN (4 nM) were as competent for full-site integration as isolated retrovirus PIC. The att sites on DNA with 3-OH-recessed ends were protected by IN in assembled intasomes from
DNase I digestion up to ϳ20 bp from the terminus. Several DNA donors containing either normal blunt-ended att sites or different end mutations did not allow assembly of complexes that exhibit the ϳ20-bp DNase I footprint at 14°C. At 50 and 100 mM NaCl, the ϳ20-bp DNase I footprints were produced with wild type (wt) U3 and gain-of-function att site donors for full-site integration as previously observed at 320 mM NaCl. Although the wt U5 att site donors were fully competent for full-site integration at 37°C, the ϳ20-bp DNase I footprint was not observed under a variety of assembly conditions including low NaCl concentrations at 14°C. Under suboptimal assembly conditions for intasomes using U3 att DNA, DNase I probing demonstrated an enhanced cleavage site 9 bp from the end of U3 suggesting that a transient structural intasome intermediate was identified. Using a single nucleotide change at position 7 from the end and a series of small size deletions of wt U3 att site sequences, we determined that sequences upstream of the 11th nucleotide position were not required by IN to produce the ϳ20-bp DNase I footprint and full-site integration. The results suggest the structural organization of IN at the att sites in reconstituted intasomes was similar to that observed in PIC.
Integration of the retroviral DNA genome into host chromosomes is an essential step in the life cycle of retroviruses. Significant progress has been made in understanding the organization of cytoplasmic PIC 1 containing the viral DNA genome produced by reverse transcription in retrovirus-infected cells. Within the PIC, the viral IN removes the two terminal nucleotides from the 3Ј-OH ends of the linear blunt-ended viral DNA (1)(2)(3). The two viral DNA ends in the PIC are held together by a protein bridge (4). The PIC are capable of performing the concerted insertion of the viral DNA ends into host DNA in vivo and into exogenously supplied target DNA in vitro, here termed full-site integration (3,(5)(6)(7)(8). The PIC containing the recessed termini are called intasomes (9), which appear analogous to type 1 transpososomes involved in Mu transposition (10 -12).
The retrovirus IN specifically interacts with the viral att sites (ϳ20 bp) located at the DNA termini (1, 6 -8). Characterization of IN interactions with the viral DNA in purified intasomes by MM-PCR footprinting revealed protection and enhancements near the termini (ϳ20 bp from the ends) with an extended region of protection mapping several hundred bp from the ends (6,8). The requirement for a large footprint in relationship to the requirement of having a Ͻ12-bp att site at the termini necessary for efficient integration in vivo is unclear (1,8). DNase I protection analysis of reconstituted AMV and recombinant Rous sarcoma virus IN-DNA intasomes that are capable of efficient full-site integration in vitro showed that the interactions of IN with viral DNA substrates produced a defined outer boundary mapping ϳ20 bp from the DNA ends (13,14). The number and structural configuration of IN subunits that are required with either retrovirus intasomes or reconstituted intasomes performing concerted integration are unknown. Molecular modeling has also been used to describe potential interactions of IN at the viral DNA ends for integration (15)(16)(17)(18)(19)(20). The shared premise of these independent approaches suggest that IN forms multimers at the viral DNA ends for concerted integration.
In this report, we have shown that assembled avian retrovirus intasomes in vitro possess equivalent full-site integration activity compared with that reported for HIV-1, MLV, and Rous sarcoma virus intasomes purified from virus-infected cells. We characterized the Mg 2ϩ and the structural DNA att site end requirements for assembling intasomes using purified AMV IN with linear 3.6-kbp donor DNA substrates for full-site integration (Fig. 1). Although wt U5 att donors possessing 3Ј-OH-recessed termini are competent for full-site integration, IN does not produce the ϳ20-bp footprint even at low NaCl (100 mM) and relatively high IN concentrations (ϳ15 nM) in contrast to ϳ20-bp footprints on U3 att sequences under these same conditions. Under suboptimal assembly conditions for intasomes, DNase I probing demonstrated an enhanced cleavage site 9 bp from the end of the U3 att DNA suggesting that a transient structural intasome intermediate was identified. With a series of small size deletions of wt U3 att site sequences and a single nucleotide change at position 7 from the end, we demonstrated that sequences only up to the 11th nucleotide from the end were required by IN to produce the minimum outer boundary ϳ20-bp footprint observed with full-site inte-gration. In summary, the results suggested that the organization of IN at the att sites in reconstituted intasomes is similar to intasomes purified from virus-infected cells.

EXPERIMENTAL PROCEDURES
Production of DNA Substrates-Linear 4.5-kbp donors containing terminal avian retrovirus (Schmidt-Ruppin strain A) wt U3 and U5 LTRs (each 330 bp in length) were previously described (13). In brief, an NdeI site was placed at the circle junction of the joined U3 and U5 ends. NdeI digestion produced a linear 4.5-kbp donor containing 3Ј-OH-recessed ends. Numbering of the nucleotides originated from the bluntends of the donors. Modifications of the 5th, 6th, and 7th nucleotides on the termini of the catalytic strand are indicated in Table I (21,22). The extension of each LTR blunt-end by one nucleotide in the original viral circle junction DNA clone produced a unique DraI site between the wt U5 and U3 ends.
Another series of large size donors containing wt U5 and mutant U3 att sites were constructed. A single nucleotide change and a series of deletions were introduced into wt U3 terminal LTR sequences. Oligonucleotides containing different mutations in U3 att sequences, immediately adjacent to wt U5 att sequences, were produced for cloning into pGEM-3. The top strand always contained 74 nucleotides and the bottom complementary strand 66 nucleotides to maintain the same spatial arrangements of restriction sites and nucleotide sequences. The arrangements were: BamHI-BglII-U3 sequences-NdeI-U5 sequences-KpnI. The sequence of the top strand for wt U3 and U5 with the above arranged restriction sites is 5Ј-GATCCAGATCTTGTTGCAAGACTAC-AAGAGTATTGCATAAGACTACATATGAAGCCTTCTGCTTCATGCG-GTAC, respectively. The overlapping terminal BamHI and KpnI sites were used for forced ligation of the annealed oligonucleotides into pGEM-3 digested with BamHI and KpnI. The U3 sequences were 36 nucleotides in length including the CA dinucleotide located in the internal NdeI site, whereas the U5 sequences were 22 nucleotides in length. The modifications in wt U3 sequences were: a single nucleotide change C to A at position 7; deletions from 8 to 12, 12 to 15, 15 to 18, 18 to 22, 20 to 26, and 24 to 38. The deleted sequences were always replaced with the next upstream adjacent U3 LTR sequences. All of the plasmid constructs were sequenced to verify the inserted LTR oligonucleotides. NdeI digestion of the constructs produced 2.9-kbp linear donors possessing 3Ј-OH-recessed U3 and U5 ends. ScaI and BglI digestions of the NdeI-digested plasmid were used to isolate unlabeled ϳ1600-bp DNA fragments containing only the U5 and U3 att sites, respectively.
Labeling of DNA Donors-The linear 2.9-and 4.5-kbp donors were 5Ј end-labeled on the noncatalytic strand with [␥-32 P]ATP and polynucleotide kinase. The specific activities were ϳ1,600 -2,000 cpm (Cererkov) per ng of DNA. For assembly and DNase I footprint studies, the 5Ј end-labeled 4.5-kbp donors were digested at either the NheI or XhoI sites to produce 3.6-kbp single-end labeled U3 or U5 donors, respectively (13). The 2.9-kbp donors were digested with either BamHI or KpnI to produce nearly 2.9-kbp single-ended att donors. For measuring the 3Ј-OH processing activity of IN, several of the above 3Ј-OH-recessed 4.5-kbp DNA donors were filled in with [ 32 P]dTTP and unlabeled dATP using Escherichia coli DNA polymerase (Klenow fragment) at 7°C (23).
Purification of AMV IN-IN was purified to near homogeneity as previously described (24,25).
Full-site integration and 3Ј-OH Processing Assays-Assembly of IN for strand transfer with 3.6-kbp single-ended U3 or U5 DNA donors (8 ng) was accomplished in the presence of 320 mM NaCl, 10 mM MgCl 2 , 3 mM dithiothreitol, 8% polyethylene glycol (6,000 daltons), 20 mM HEPES, pH 7.5. The standard reaction volumes were 20 l or higher multiples. The time and temperature for assembly of IN with the donors were routinely 45 min and 14°C, respectively. The reactions were initiated by the addition of 50 ng of supercoiled 2.8-kbp DNA (pGEM-3) as target and immediately incubated at 37°C for either 5 min or as indicated. The reactions were stopped by addition of EDTA, SDS, and proteinase K digestion followed by phenol extraction. Equivalent amounts of labeled products from the reactions were subjected to 1.5% agarose gel electrophoresis. The amounts of donor incorporated into target DNA as full-site and half-site products were determined with a Amersham Biosciences PhosphorImager TM . Half-site integration is defined as the insertion of one donor molecule into the target.
For the 3Ј-OH processing reactions, IN was assembled with single att blunt end-labeled 3.6-kbp donors at 14°C as described above for strand transfer with no target added. The release of the labeled dinucleotides from the blunt-ended LTR donors by IN at either 37 or 14°C for 5 min were measured as acid-soluble counts (Cererkov). The acid-soluble counts were expressed as a percentage of the input donor substrate.
DNase I Footprinting-For assembly at ϳ300 mM NaCl and DNase I footprint analyses, IN and the single end-labeled 3.6-kbp donors (8 ng or 0.17 nM) and the 2.9-kbp U3 deletion mutants donors (10 ng or 0.26 nM) were assembled at 14°C for 45 min (Fig. 1). For some experiments with either the 3.6-kbp donors or the 2.9-kbp donors, different mixtures of labeled and unlabeled donors were mixed together (13,26). An aliquot was removed from the mixture for measuring strand transfer activities just prior to the addition of DNase I to 750 ng/ml for 90 s digestion at 14°C. When measuring strand transfer activity at 50 and 100 mM NaCl concentrations, the DNase I concentrations were reduced to 75 ng/ml for footprint analysis. The DNase I reactions were stopped by the addition of phenol and equivalent amounts of each reaction were subjected to denaturing 10% polyacrylamide gel electrophoresis. The dried gels were analyzed by phosphorimaging and exposure to x-ray films.

RESULTS
Restriction enzyme digestion at two internals sites on linear viral DNA packaged within purified HIV-1 intasomes, followed by size purification of the separated products, demonstrated that the synapsed DNA ends in the intasomes are held together by a protein bridge (4). A similar bimolecular event holding together the two viral DNA ends by purified IN (Fig. 1) mirrors the coordinated integration of two independent viral DNA ends into target DNA. The total amount of viral DNA integrated by HIV-1 and MLV intasomes purified from cells varies between 10 and 50% into target DNA following incubation for 45-90 min at 37°C (4,7,8,27,28). Purified retrovirus intasomes generally produce only concerted integration products equivalent to the full-site integration products described in this report although side reactions (autointegration of viral DNA into itself) can also occur (29,30). 2) equivalent to the amounts observed with purified retrovirus intasomes. With this assay configuration, saturation of the donor substrate (3.6 kbp) containing a single LTR end by IN occurs at ϳ4 nM ( Fig. 2A). Approximately 40% of the donor is incorporated into full-site products with ϳ5% of the donor being incorporated into half-site products at 37°C (Fig. 2B). The results suggest that the reconstitution of AMV IN with viral DNA donors appears equivalent to purified retrovirus intasomes for production of full-site integration products.

Equivalence of Reconstituted Avian Retrovirus Intasomes to Purified Retrovirus Intasomes from Cells for Full-site Integration Activity-Assembled
The divalent metal ion Mg 2ϩ is essential for proper 3Ј-OH processing and strand transfer activities as well as promoting structural changes in IN (1, 31). At optimal AMV IN-DNA  (1) for intasomes requires IN to synapse two viral att DNA ends for full-site integration. The arrow pointing to the right indicates that the assembled complexes were also subjected to DNase I footprint analysis. The addition of a 2.8-kbp circular target substrate (step 2) to the assembled intasomes results in the initiation of strand transfer at 37°C (step 3). The bottom schematics identify the 10-kbp full-site integration product and the half-site product that consists of a nicked circular target with a single donor LTR end inserted. molar ratios, Mg 2ϩ also assists in assembly of intasomes at 14°C. With Mg 2ϩ in the assembly mixtures, a typical IN titration profile for full-site integration was observed (Fig. 2). Without Mg 2ϩ in the assembly mixture, the formation of full-site integration products was diminished ϳ50% compared with that observed when Mg 2ϩ was present (data not shown). The presence of Ca 2ϩ (5 mM) in the assembly mixtures had no or a slightly detrimental effect on the yield of competent intasomes capable of full-site integration after addition of Mg 2ϩ necessary to measure strand transfer activity. The results suggest that Mg 2ϩ assists IN at low concentrations in assembling proper intasomes at 14°C but it is not absolutely essential.
AMV IN Does Not Form a Stable Association with Bluntended att Ends during Assembly at 14°C-The majority of mature cytoplasmic retrovirus intasomes contain 3Ј-OHrecessed LTR ends suggesting the recessed termini are a stabilizing factor for intasomes prior to nuclear import and integration (2,5,7). HIV-1 intasomes that have unprocessed blunt-ends on the viral DNA (because of att site mutations) lack an MM-PCR footprint (9). Recombinant HIV-1 IN forms stable complexes with oligonucleotides possessing 3Ј-OHrecessed LTR ends for half-site strand transfer but not with the same substrate containing blunt-end LTR ends necessary for 3Ј-OH processing (32). Inhibitor studies also suggest that HIV-1 IN forms a unique "strand transfer" conformation upon binding to 3Ј-OH-recessed LTR ends in the presence of target DNA (33,34). We investigated if AMV IN forms a stable structure with blunt-ended LTR donor ends under assembly conditions at 14°C as measured by DNase I footprint protection.
To perform the coupled 3Ј-OH processing and protection studies, we filled-in the 3Ј-OH-recessed ends of the 4.5-kbp DNA containing gain-of-function ("G") sequences at both the U3 and U5 ends (13). After restriction digestion of the 4.5-kbp DNA, the 3.6-kbp donor DNA containing only the labeled bluntended G U5 end sequences was isolated (Table I). A series of three experiments were performed simultaneously with assembled IN-DNA complexes with the blunt-ended substrate. Assembly of DNA at various concentrations of IN was for 45 min at 14°C (Fig. 3A). Aliquots from the assembly mixtures were taken at various times to measure activities at different temperatures. First, at 14°C, minimal 3Ј-OH processing occurs in the assembly mixture even at 21 nM IN, but significant amounts of labeled dinucleotides are released by IN after 5 min incubation at 37°C (Fig. 3A). Second, minimal strand transfer activity also occurred with the same blunt-ended DNA substrate upon incubation at 37°C for only 5 min (Fig. 3B, lanes 6 -10). Prolonged incubations at 37°C (Ͼ30 min) with the blunt-ended donor primarily resulted in half-site products with a minor population of full-site integration products (23). Third, no DNase I footprint protection by IN was observed on the blunt-ended DNA at 14°C suggesting the lack of a stable association by IN under these assay conditions (Fig. 3C). Similarly, negative DNase I footprint protection by IN was obtained using a 3.6-kbp DNA containing either wt U3 or wt U5 blunt-ends that were extended by one nucleotide (Table I). In contrast, IN at the same concentrations and assay conditions using a 3Ј-OH-recessed donor was fully capable of significant full-site integration activity (Fig. 3B, lanes 1 Ϫ Ϫ Non-specific DNA pGEM3 EcoRI site Ϫ Ϫ a wt represents wild-type, whereas G and L represent gain and lossof-function, respectively. The bolded nucleotides represent the 5th, 6th, and 7th nucleotide from the blunt-ended termini. Only the 3Ј-OH catalytic strand is shown. Stand transfer was for 5 min at 37°C. Some of the data were previously published (13) and are summarized in this table.
previously shown that AMV IN was not able to produce a ϳ20-bp DNase I footprint on wt U5 DNA possessing 3Ј-OHrecessed ends at 0.32 M NaCl in the assembly mixture at 14°C even though the strand transfer activities appeared normal at 37°C (13). The possibility existed that IN was not stably associated with U5 at 0.32 M NaCl but could be at lower NaCl concentrations. A variety of new conditions were tested to determine whether IN was able to produce the ϳ20-bp footprint on U5 at 14°C. The NaCl concentrations were varied from 50 to 320 mM using different concentrations of IN (5-42 nM). The ϳ20-bp footprints were not observed at the termini (data not shown). We also tested whether the presence of target facilitated the formation of stable IN-DNA complexes at 14°C (33,34). We observed no DNase I footprints if, target DNA was added after 30 min of assembly time with U5 att DNA followed by 15 min of further incubation at 14°C, prior to DNase I digestion.
In contrast, we were able to demonstrate that IN at low concentrations produced the normally observed ϳ20-bp DNase I footprint using G U3 att DNA containing 3Ј-OH-recessed ends at 50 and 100 mM NaCl (Fig. 4A). When IN was increased to 30 nM under these low salt conditions, the entire length of the DNA was essentially protected from DNase I digestion suggesting nonspecific multimerization of IN subunits on the DNA (data not shown). Aliquots of the same above samples (Fig. 4A) were analyzed for strand transfer activities at 37°C for 5 min (Fig. 4B). Relatively higher quantities of all half-site products at 50 mM (Fig. 4B, lanes 2-4) or 100 mM NaCl (Fig. 4B, lanes  5-7) were observed than with IN at 0.32 M NaCl (Fig. 4B, lanes  8 -10). The quantities of full-site products produced with 4 nM IN at 50, 100, and 320 mM NaCl were 22, 32, and 15% of the total donor incorporated (Fig. 4B, lanes 2, 5, and 8, respectively). The ratio of full-site to half-site integration products produced with the U3 att donor were effected by both IN and NaCl concentrations. In summary, the lack of the ϳ20-bp footprint on the U5 att donor by IN was not apparently because of higher salt concentrations.
DNase I Cleaves at Sensitive att Site Nine Nucleotides from the End Terminus in Suboptimal Assembled Intasomes-Physical interactions of IN with att site sequences suggest that the first ϳ12 nucleotides from the end are functionally important for 3Ј-OH processing and strand transfer activities both in vivo and in vitro (31,35). Detailed MM-PCR footprinting of HIV-1 intasomes revealed the major area of enhancements at nucleotides 9 and 10 from the end on U3 and, nucleotides 11 and 13 nucleotides from the end on U5 (7).
We investigated whether probing of suboptimal assembled avian intasomes by DNase I would provide molecular insights into the contacts between IN and the viral att site near the terminus. Under normal assembly conditions at 14°C and 45 min, complete DNase I protection by IN mapping up to ϳ20 bp of att terminal sequences is observed with wt U3, G U3, or G U5 donors containing 3Ј-OH-recessed ends (Fig. 4) ( Table I). The same footprints were observed on the 32 P 5Ј-labeled wt U3 att donor in the presence of 2-fold excess of unlabeled wt U5 donor; ϳ80% of the observed full-site products are the result of U3and U5-coupled reactions (13). To produce suboptimal assembly and probing conditions, we decreased the assembly time to 10 min and increased the DNase I probing time to 120 s at 14°C instead of the usual 90 s. Six independent assembly tubes were used with the concentration of IN being varied from 5 to 50 nM (Fig. 5, lanes 2-7). A modified DNase I protection pattern was observed with the presence of a newly created but variable appearing band that migrates near the 10th nucleotide T (Fig.  5, lanes 2-7, see arrow) in contrast to the absence of this band in DNase I-treated G U3 DNA with no IN presence (Fig. 5, lane  1). The production of full-site and half-site integration products was essentially normal after 10 min at 37°C incubation (23 and 8%, respectively, at 5 nM IN) (data not shown). The production of this cleavage site was inconsistent between different experiments but appears independent of different IN preparations (data not shown). The cleavage of the bond between the 9th and 10th nucleotide by DNase I in the apparent transient intasomes (Fig. 5, lanes 2-7) and protection of nucleotides up to ϳ20 bp in fully assembled and stable intasomes (Fig. 4) (13,14) suggests that two turns of the DNA helix are captured by as yet, an unknown number of IN subunits. In summary, the transient intasome complex is probably distorting the DNA structure sufficiently thus allowing DNase I to access the intasome 9 nucleotides from the terminus on the noncatalytic strand.
Mapping Internal att Site Sequences Required by IN for Full-site Integration and ϳ20-bp DNase I Footprint-A small size DNA deletion (nucleotides 5-11) of viral U3 DNA terminal sequences in MLV intasomes demonstrated that the enhanced MM-PCR footprint at ϳ20 bp from the end of the viral genome and virus replication required only the first 11 nucleotides; deletion of nucleotides 12-33 had no apparent effects on either the MM-PCR footprint or viral replication (6). We wanted to determine which internal att sequences were required by IN in reconstituted intasomes to produce a DNase I footprint whose outer boundary maps ϳ20 bp from the end. A series of internal deletions in wt U3 and a single nucleotide change at position 7 with a C to A transition were produced. The independent constructs contained deletions from nucleotides 8 -12, 12-15, 15-18, 18 -22, 20 -26, and 24 -36. The deleted sequences were replaced with the adjacent upstream U3 sequences. The sequences of wt U3, att mutants 7 C/A, deletion 8 -12, and deletion 12-15, are shown in Fig. 6. The design of the 8 -12 deletion mutant only allowed the 9, 10, and 11 positions to be changed prior to nucleotide 13. The design of the 12-15 deletion mutant allowed the 12th position to remain wt with changes occurring at the 13th position and beyond. Previous results suggested that the nonsymmetrical nucleotides between avian U3 and U5 att sites at positions 8 and 12 did not appear to significantly influence either full-site or half-site strand transfer activities of IN (21). Each att DNA substrate was assembled with IN at several concentrations and the samples were subjected to DNase I probing and strand transfer.
The full-site integration activities of IN at 8 nM were significantly decreased only in the presence of att mutant 7 C/A and partially (ϳ50%) with the deletion 8 -12 mutants in comparison to either wt U3 or the other att deletion mutants, all of which possessed near wt activities (Fig. 6). Similar quantitative data were also obtained at 16 nM IN. The results suggest that the 7th C nucleotide is critical for full-site activity, whereas nucleotides 9 -11 (Fig. 6) partially influence activities. Nucleotides upstream of position 11 appear to have little or no effect on full-site (Fig. 6) or half-site strand transfer activities (data not shown).
We wanted to determine whether there was a direct relationship between full-site integration activity and the ϳ20-bp DNase I footprint with certain att site sequences. The concentrations of IN were varied from 4 to 12 nM. After assembly, strand transfer analyses were performed on wt U3 and the 7 C/A, deletion 8 -12, and deletion 12-15 mutants (Fig. 7A) with the results being similar to that previously observed (Fig. 6). DNase I footprint protection of att site sequences by IN with the outer boundary mapping at ϳ20 bp from the end (Fig. 7B) paralleled the observed full-site integration activities. IN protected wt U3 att sequences up to ϳ20 bp (Fig. 7B, lanes 4 -7) but no protection was observed with the 7 C/A mutant (Fig. 7B,  lanes 8 -11). Partial or more complete protection of att se- quences in the deletion 8 -12 mutants at the higher IN concentrations was observed (Fig. 7B, lanes 12-15). The deletion 12-15 mutant (Fig. 7B, lanes 16 -19) possessed the same footprint and strand transfer activities as wt U3.
We also probed the other U3 deletion mutants by DNase I. The full-site integration activities of deletion 15-18, deletion 18 -22, and deletion 20 -26 mutants were shown in Fig. 6. At 8 and 16 nM IN, all three deletion mutant att sequences were effectively protected by IN from DNase I digestion up to ϳ20 bp (Fig. 7C). Last, the deletion 24 -36 mutant (Fig. 6) also possessed the ϳ20-bp footprint (data not shown).
We investigated if labeled wt U3 and the 7 C/A and deletion 8 -12 mutants, in the presence of a 2-fold molar excess of either unlabeled wt U3 or wt U5 att donor (1,600 bp) ends, would display full-site integration activities and a ϳ20-bp DNase I footprint. As a control for these properties, all three labeled donors were assayed at 8 nM IN, whereas the mixtures containing unlabeled donors were assayed at 16 nM. The quantities of labeled products obtained at 8 and 16 nM IN under these different conditions were nearly equivalent. With labeled wt U3 and deletion 8 -12 mutant donors, ϳ80% of the labeled products were because of coupled reactions with unlabeled donors as determined by migration of the labeled products on agarose gels (13,26). The full-site integration activities and the ϳ20-bp footprint of these two labeled donors were essentially identical as previously observed in Figs. 6 and 7 (data not shown). The labeled 7 C/A donor mutant mixed with either unlabeled donors was still inactive.
In summary, the results suggested that in addition to nucleotides 5 and 6 from the end terminus (Table I), nucleotide 7 significantly affects the ability of IN to assemble onto the att sites in a stable fashion at 14°C and subsequently full-site integration. Sequences at position 9 -11 also partially decrease the ability of IN to promote full-site integration and the DNase I footprint. Sequences after the 11th nucleotide position in U3 appears to have no apparent effect on the observed ϳ20-bp footprint and full-site integration that is similar to the observed MM-PCR footprint results with the MLV U3 deletion mutants (6). DISCUSSION Assembled intasomes using AMV IN and large size DNA substrates with terminal att sequences were equivalent to retrovirus intasomes purified from virus-infected cells for full-site integration activity. The assembly process was enhanced by the addition of Mg 2ϩ but it was not essential. An ϳ20-bp DNase I protection pattern of wt U3 and G att sites was observed at low IN concentrations at 50, 100, and 320 mM NaCl but was not evident on wt att U5 under similar assay conditions. Several different blunt-ended att site substrates also did not allow the stable assembly of IN resulting in a DNase I protective footprint at 14°C in contrast to footprints observed with 3Ј-OHrecessed ends. Under suboptimal assembly conditions, a specific DNase I enhanced cleavage occurred between nucleotides 9 and 10 suggesting the presence of a transient intasome structure in the assembly process. The ϳ20-bp DNase I footprint and efficient full-site integration activity requires a C at posi- tion 7 and partially nucleotides 9 -11 from the end but, these properties appear to be independent of sequences upstream of the 11th nucleotide. In summary, the assembled intasomes have many of the organizational and enzymatic properties associated with purified retrovirus intasomes.  2 and 3, respectively. Only the markers for wt U3 are presented and marked. The ϳ20-bp footprint boundary for wt U3 is marked by a rectangle on the left and on the right, for the del 12-15 mutant. The del 12-15 mutant experiment was performed on a separate gel and the data were placed at the appropriate location on the shown gel. C, the samples in Fig. 6  resulting in the ϳ20-bp DNase I footprint (Table I) (13). A stable association by IN on blunt-ended G U5 DNA is not apparent at the footprint level at 14°C but IN with this substrate had normal 3Ј-OH processing activity at 37°C (Fig. 3). Several possibilities are plausible for these results: 1) the association and dissociation rates of IN with wt U5 in comparison to wt U3 or G att sites are significantly different at these two temperatures, and 2) the number of IN subunits on U5 are less than on U3 (13). In either case, IN apparently requires nucleotides 5 and 6 as well as nucleotide 7C (Fig. 7) on U3 to form a stable synaptic complex at 14°C resulting in the footprint. The results suggest these terminal sequences including position 7C may serve as "anchoring" nucleotides for IN to initiate binding and to form a stable complex producing the footprint (Figs. 6 and 7). The stable complex includes the 5Ј-2-bp overhang required for strand transfer activities (Fig. 4) (32). The involvement of these terminal nucleotides for 3Ј-OH processing and strand transfer activities have been previously established in several retrovirus systems both in vitro and in vivo (1,7,9,15,22,31,(35)(36)(37)(38).
Are other LTR sequences besides the seven terminal nucleotides required for integration in vivo? Modifications of the U5 end disrupts other critical replication functions including binding of tRNA Trp to U5 RNA for reverse transcription thereby making the analysis of the U5 LTR exclusively for integration impossible (7,39,40). Other in vivo and in vitro studies have indicated IN contacts nucleotides 9 -12 from the end terminus of U3 (6, 7, 36, 40 -42). Random mutagenesis of LTR sequences at internal positions implicated nucleotides 17-20 bp being involved in full-site integration by recombinant HIV-1 IN in vitro (43), and site-directed mutagenesis indicates that nucleotide 41 significantly decreased half-site integration for human T-cell leukemia virus IN (44). Modification of nucleotide 20 in MLV att DNA using recombinant MLV IN did not affect fullsite integration (45). In this report, changing of nucleotides 9 -11 (GAA) to CGT of U3 appear to decrease (ϳ50%) the ability of AMV IN to mediate full-site integration and to form a stable complex as shown in the ϳ20-bp DNase I footprint (Figs. 6 and 7). Modification of these same three nucleotides to ACG also partially affected full-site integration activity (41). Interestingly, the identification of an enhanced DNase I cleavage site between nucleotides 9 and 10 ( Fig. 5) suggests that IN forms a transient intermediate prior to forming a stable intasome. The presence of the ϳ20-bp footprint and efficient full-site integration are not affected by alteration of sequences from the 12th nucleotide and beyond (Figs. 6 and 7). In summary, the stable intasome appears to encompass two turns of the DNA helix by IN with as yet an unknown number of subunits.
The apparent sequence-specific recognition of the first ϳ7 att nucleotides from the end by AMV IN are apparently different from the remaining att nucleotides that encompass the entire ϳ20-bp DNase I footprint on U3 (Figs. 6 and 7). IN may employ both direct readout as well as indirect readout mechanisms (26,38). The catalytic core of IN may employ a direct readout mechanism for binding and catalysis involving the first set of nucleotides by forming specific hydrogen bonding with IN residues (15)(16)(17)(18), whereas the latter set of nucleotides may only require a particular geometrical structure (like a minor or major groove). The lack of detailed structural information of IN-att DNA complexes at the atomic level only permits speculation of which binding mechanisms are involved in assembled intasomes capable of full-site integration.
MM-PCR footprinting of HIV-1 and MLV intasomes isolated from virus-infected cells reveals IN interactions within the first ϳ20 bp but the complete protective footprints extend up to ϳ200 bp from the end terminus (7,8,28). The BAF protein is strongly associated with MLV and HIV-1 intasomes. It may be responsible for the extended footprint and stimulates intermolecular integration (6, 46 -48). The ability of BAF but probably not other cellular factors (49) to possibly interact with internal LTR sequences to synapse the LTR DNA in intasomes is intriguing. The possibility exists that IN is synergistic with other cellular factors like BAF in producing the synapsed viral DNA and the extended footprints observed in MLV and HIV-1 intasomes because, the number of IN dimers in virions have been estimated to be ϳ25-50 (1). Higher AMV IN concentrations (Ͼ20 nM) also produce extended DNase I footprints upstream of the ϳ20-bp footprint (13) but whether this multimerization of IN is specific or nonspecific, contributing to the extended HIV-1 and MLV intasomes MM-PCR footprints on the LTR ends, are unknown. Further studies are required with both integration systems to understand the structural organization of intasomes.