Nonhomologous recombination in human cells.

Nonhomologous recombination (NHR) is a major pathway for the repair of chromosomal double-strand breaks in the DNA of somatic cells. In this study, a comparison was made between the nonhomologous end joining of transfected adenovirus DNA fragments in vivo and the ability of purified human proteins to catalyze nonhomologous end joining in vitro. Adenovirus DNA fragments were shown to be efficiently joined in human cells regardless of the structure of the ends. Sequence analysis of these junctions revealed that the two participating ends frequently lost nucleotides from the 3' strands at the site of the joint. To examine the biochemical basis of the end joining, nuclear extracts were prepared from a wide variety of mammalian cell lines and tested for their ability to join test plasmid substrates. Efficient ligation of the linear substrate DNA was observed, the in vitro products being similar to the in vivo products with respect to the loss of 3' nucleotides at the junction. Substantial purification of the end-joining activity was carried out with the human immature T-cell-line HPB-ALL. The protein preparation was found to join all types of linear DNA substrates containing heterologous ends with closely equivalent efficiencies. The in vitro system for end joining does not appear to contain any of the three known DNA ligases, on the basis of a number of criteria, and has been termed the NHR ligase. The enriched activity resides in a high-molecular-weight recombination complex that appears to include and require the human homologous pairing protein HPP-1 as well as the NHR ligase. Characterization of the product molecules of the NHR ligase reaction suggests that they are linear oligomers of the monomer substrate joined nonrandomly head-to-head and/or tail-to-tail. The joined ends of the products were found to be modified by a 3' exonuclease prior to ligation, and no circular DNA molecules were detected. These types of products are similar to those required for the breakage-fusion-bridge cycle, a major NHR pathway for chromosome double-strand break repair.

Polarity in adenovirus recombination.

The distributions of the crossovers necessary to generate ts+ genomes have been examined in a collection of clonally unrelated ts+ recombinants from a set of ts X ts adenovirus crosses. In a cross between two parents that are grossly heterologous between map units 80.2 and 91.5, the distribution of crossovers was significantly skewed toward the left-hand end of the genome, with a declining frequency proceeding rightward. This gradient of recombination was modified by the removal of the right-hand heterology and by the presence of another region of heterology between map units 3.67 and 10.11. In a cross where the ts markers were flanked by both heterologies, no gradient was observed and ts+ recombinants were characterized by a higher rate of supernumerary crossovers. In a cross designed so that one ts marker was internal to two heterologies, crossovers were found disproportionately between the second ts marker and the nearby heterology. In addition, ts+ recombinants formed by crossing over internal to the heterologies again were accompanied by a high frequency of supernumerary crossovers. Finally, ts+ recombinant frequencies in crosses identical except for the presence of either one or two flanking heterologies were markedly lower in the latter case. These data, taken together, suggest that a major pathway of adenovirus recombination initiates at, or near, the molecular termini and is perhaps driven by the displaced single strands produced during DNA replication. Internal initiation, on the other hand, may employ these single strands to form genetic "patches."

End-joining of DNA fragments in adenovirus transfection of human cells.

Overlapping terminal fragments of adenovirus DNA transfected into human cells either recombine to form standard unit-length genomes, or can join end-to-end to produce internally redundant, viable, genomes. The end-joining reaction in human HeLa and A549 cells is almost as efficient as the recombination reaction, and is relatively insensitive to the nature of the ends, as pairs of fragments terminating in several different single strands or in blunt ends can join. In contrast to the results from transfection with SV40, the ends are usually modified, for example by the loss of 3' single strands or the repair of 5' single strands. The ability to recover viable redundant molecules is not confined to any one area of the adenovirus genome, but can occur in the E1 and L2 regions as well as in the E2b region. The redundant genomes contain extra splice signals and may have the capacity to encode fusion proteins.

The creation of adenovirus genomes with viable, stable, internal redundancies centered about the E2b region.

During the course of constructing new adenoviral strains by overlap recombination, we have discovered that internally redundant viable genomes can be created by end-to-end joining of the input DNA molecules. The cellular functions responsible for the end-joining activity frequently ligated the overhanging single strands of the complementary ends to form a novel restriction site at the junction. In 2 of the 17 cases analyzed in detail by restriction digestion, and some sequence determinations, the cellular functions had repaired the ends, presumably prior to end-joining. Four of the isolates had suffered deletions at the junction ranging in size from 13 to 532 bp. The isolate with the largest deletion also had an insertion of 14 bp of unknown origin at the site of the deletion. All of the redundant isolates replicated as efficiently as isogenic unit length strains, and plaque dilution titrations obeyed one-hit kinetics, showing that the redundant genomes were nondefective. Nevertheless unit-length genomes were observed at a low level (some 5 to 10% of the total) in stocks of each isolate before and after plaque purification. They presumably arose by recombination between the redundant sequences either intra- or intermolecularly. Evidence from Southern blot analysis showed that molecules with three copies of the redundant sequences also arose and could be detected both in intracellular and in capsid viral DNA. These species would arise by unequal crossing-over between redundant genomes. The efficient replication of the redundant species demonstrates that the precise spatial relationships between splice donors and acceptors on either strand, in this region of the genome, do not have to be rigidly maintained. These data suggest that it may be possible to place other genetic information between the DNA polymerase and terminal protein precursor genes and have it expressed from the major late promoter in its normal location.

A genetic investigation of the mechanism of adenovirus marker rescue.

We have developed quantitative and segregational methods for investigating the mechanism of genetic exchange in adenovirus marker rescue. Estimates of "marker rescue frequency" (m.r.f.) were used to show that marker rescue increases linearly with increasing dose of fragment up to equimolarity with the full-length genome. The m.r.f. is also affected by the size of the rescuing fragment and the position of the wild-type allele within it, regardless of whether the fragment is terminal or internal. This is compatible with marker rescue being based on homologous exchange between the recombining partners. Examination of individually transfected cells showed that there is very wide variability in the values of the m.r.f.'s. This suggests that marker transfer can occur after replication of the full-length genome has begun, and can occur late into the infectious cycle. Unselected markers on the rescuing fragment were shown to be co-inherited frequently. This suggests that physical linkage is accompanied by genetic linkage. To examine this more closely, a multifactorial marker rescue was performed. The data show unequivocally that markers resident on the same fragment as the selected allele are inherited at high frequency, with a gradient of transfer in which markers closest to the selected marker are transferred most frequently. Markers up to 13 and perhaps as many as 17 kb apart can be inherited together. There are very few examples of the inheritance of distal markers in the absence of proximal ones. These data suggest that large pieces of DNA are transferred in a concerted reaction during marker rescue.

A modified single-strand annealing model best explains the joining of DNA double-strand breaks mammalian cells and cell extracts.

The joining of DNA double-strand breaks in vivo is frequently accompanied by the loss of a few nucleotides at the junction between the interacting partners. In vitro systems mimic this loss and, on detailed analysis, have suggested two models for the mechanism of end-joining. One invokes the use of extensive homologous side-by-side alignment of the partners prior to joining, while the other proposes the use of small regions of homology located at or near the terminus of the interacting molecules. to discriminate between these two models, assays were conducted both in vitro and in vivo with specially designed substrates. In vitro, molecules with limited terminal homology were capable of joining, but analysis of the junctions suggested that the mechanism employed the limited homology available. In vivo, the substrates with no extensive homology end-joined with equal efficiency to those with extensive homology in two different topological arrangements. Taken together, these results suggest that extensive homology is not a prerequisite for efficient end-joining, but that small homologies close to the terminus are used preferentially, as predicted by the modified single-strand annealing model.

The genetic analysis of adenovirus recombination in triparental and superinfection crosses.

Previous genetic and molecular data suggest that adenovirus genomes can undergo several rounds of recombination before being encapsidated (C. S. H. Young and S. J. Silverstein, Virology 101, 503-515). Two predictions of this hypothesis have been tested. The first is that infection with three differentially marked parental viruses should lead to the appearance of recombinants with genetic contributions from all three parents. In a triparental cross, involving two strains with different ts mutations in chimeric Ad5/Ad2+ND1 backgrounds, and a third strain containing both ts mutations in an Ad5 background, it was demonstrated that multiple recombinations, involving distinguishable restriction endonuclease sites and host range markers from all three parents, were common. The second prediction, from previous kinetic data, is that cells are recombinationally proficient from the eclipse period well into the exponential rise period. This has been tested by superinfecting singly infected cultures, both during eclipse and in early exponential phase. Recombinant viruses were produced in these superinfections, demonstrating that the early to late switch in the replicative cycle does not inhibit recombination. From the temporal appearance of recombinants moreover, it seems likely that recombination functions, and the DNA structures necessary to initiate recombination, are present well into the late phase of replication.

Creation and repair of specific DNA double-strand breaks in vivo following infection with adenovirus vectors expressing Saccharomyces cerevisiae HO endonuclease.

To study DNA double-strand break (DSB) repair in mammalian cells, the Saccharomyces cerevisiae HO endonuclease gene, or its recognition site, was cloned into the adenovirus E3 or E1 regions. Analysis of DNA from human A549 cells coinfected with the E3::HO gene and site viruses showed that HO endonuclease was active and that broken viral genomes were detectable 12 h postinfection, increasing with time up to approximately 30% of the available HO site genomes. Leftward fragments of approximately 30 kbp, which contain the packaging signal, but not rightward fragments of approximately 6 kbp, were incorporated into virions, suggesting that broken genomes were not held together tightly after cleavage. There was no evidence for DSB repair in E3::HO virus coinfections. In contrast, such evidence was obtained in E1::HO virus coinfections of nonpermissive cells, suggesting that adenovirus proteins expressed in the permissive E3::HO coinfection can inhibit mammalian DSB repair. To test the inhibitory role of E4 proteins, known to suppress genome concatemer formation late in infection (Weiden and Ginsberg, 1994), A549 cells were coinfected with E3::HO viruses lacking the E4 region. The results strongly suggest that the E4 protein(s) inhibits DSB repair.