New insights into protein-DNA interactions obtained by electron microscopy.

The electron microscopic study of DNA-protein complexes can yield valuable information that is often not easily available by other methods. In this article we give a number of examples that were chosen to illustrate the utility of electron microscopy. Along with the strategy used are protocols that allow such experiments to be carried out. The first example employs the following strategy. Points of close proximity between nucleic acid and protein within a bacteriophage or virus are made permanent by crosslinking. Bacteriophage or virus are then partially disrupted so that individual components can be visualized. With bacteriophages, such experiments show which DNA end first enters the host on infection and therefore can in principle indicate which phage genes would be first available for transcription. This type of experiment can also show which DNA end is first to be encapsulated during formation of the bacteriophage. Information on direction of encapsulation and indirectly, direction of replication of the rolling circles that lead to concatermeric DNA to be encapsulated, can also be derived. Such experiments can additionally accurately define the degree of DNA permutation, if present, within a bacteriophage population. Finally, examples are shown for in vitro reactions involving DNA, RecA, RecO, RecF, RecR, and SSB that lead to a further understanding of recombinational repair. Additionally antibody-gold labeling is used to locate various proteins in such complexes.

Daughter origins can be held together by O protein in phage lambda replicative intermediates.

When lambda replicative intermediates are incubated with initiator protein O, complex molecules are observed in which O interacts with both daughter origin segments and with growing points. In the simplest of these molecules it appears that daughter origins and growing points may all be bound together at a single point. When replicative intermediates are sequentially incubated with single-stranded binding protein and O protein, simpler structures are observed. In this case, both daughter origins are bound together by O protein. This result mimics that found when plasmid containing two tandem lambda origin sequences is reacted with O protein. In this case double origin binding produces a DNA loop. Double origin binding, as demonstrated in this investigation, creates the potential for topological domains which will have important effects on the ability of daughter origins to initiate replication.

DNA looping induced by bacteriophage lambda O protein: implications for formation of higher order structures at the lambda origin of replication.

A plasmid has been constructed, pOri2, which contains two lambda replication origin sequences separated by 1068 bp; both lambda sequences having the same orientation. When lambda initiation protein O is reacted with linearized pOri2 and examined by electron microscopy it is found to contain a looped area in which two parts of the plasmid are bound together by the O protein complex. Length measurements show that the O protein binds at the expected positions of the lambda origin sequences and that the looped area represents the DNA segment between the two O protein binding domains. Similar looping occurs in reactions with supercoiled pOri2 or if an amino-terminal fragment of O protein is used. When looped molecules are reacted with psoralen, crosslinked by irradiation with uv light, and then denatured, it is found that the looped area is more thermostable than the rest of the molecule. This indicates that the DNA within the looped segment is torsionally constrained while that outside the loop is free to rotate and suggests that simultaneous binding of O to two origins fixes the linkage number of the intervening DNA. The double origin binding ability of O may be diagnostic of the details of the reaction of O with a single origin sequence. A model is presented that rests on the assumption that O can produce microscopic looping between O protein binding sites within a single ori sequence.

Initiation protein induced helix destabilization at the lambda origin: a prepriming step in DNA replication.

The interaction of the lambda phage initiator protein, O, with the lambda origin sequence, ori, has been investigated. Binding of O, or its amino-terminal fragment, causes a major structural change within a 60 bp AT-rich region just to the right of the O-binding site. ATP or other molecular energy sources are not required. The modification, as assayed by nuclease sensitivity, is reduced when certain ori mutant sequences, which bind O but fail to replicate, are substituted for the wild-type sequence. The modification of DNA structure caused by the interaction of O is absolutely dependent on the presence of superhelical tension at the lambda origin sequence, and has several properties consistent with a strand separation reaction. We propose that this modification is a fundamental prepriming event that is the first stage in initiation of bidirectional replication in lambda after O binding.

Reinitiation at the lambda DNA origin accompanies the host SOS response.

Abnormal reinitiation of replication from lambda origins has previously been found during infection in the presence of caffeine or cis-diamminedichloroplatinum II (cis-Pt) or when lambda infects a P2 lysogen. It was further shown that the reinitiations arising from cis-Pt treatment took place during the SOS response induced by the template damage caused by the drug. It is now shown that SOS induction by uv irradiation of the host also results in reinitiation events and that it is the SOS response itself rather than some other direct effect of the damaged host template that is responsible for the phenomenon. Parental sections of lambda replicative intermediates can supercoil, whereas daughter segments cannot. To explain the control that prevents reinitiation, it is proposed that normally the origin sequence has to be under superhelical tension to be a suitable substrate for the initiating machinery; once a round is in progress, the daughter origin sequences would not be under such tension and would therefore be inactive. It is shown that in an SOS environment the proposed requirement for a superhelical origin sequence is relaxed and consequently the control against reinitiation lost. Under such conditions, primary growing points that have encountered template lesions terminate and a new wave of replication initiates.

Electron microscopic identification of supercoiled regions in complex DNA structures.

When intracellular lambda replicative intermediates (theta structures) are intercalated with psoralen and then irradiated with long wavelength ultraviolet light (u.v.), interstrand crosslinks are produced. After purification and denaturation of these theta structures, a global difference in denaturation can be observed by electron microscopy; parental sections are essentially native whereas daughter segments are highly denatured. This difference can be explained if parental sections are covalently continuous (and therefore able to supercoil) and daughter segments are not. Due to the higher thermal stability of supercoiled DNA, parental DNA will remain native while daughter sections will denature. Because these structures are crosslinked, the thermal treatment does not lead to dissociation of the highly denatured daughter strands. Experiments with simple negatively supercoiled plasmid circles support the above conclusions. When circles are crosslinked with psoralen-u.v. and then denatured, they remain native because of the higher thermal stability of covalently closed structures. If the circles are linearized before heating but after the psoralen-u.v. treatment, the thermal stability effect is eliminated and the molecules become highly denatured. In this case, however, the crosslinking density is found to be higher than in samples linearized before psoralen-u.v. treatment. This, therefore, shows that crosslinking density also reflects the superhelical state of the molecule at the time of psoralen-u.v. treatment. Two different properties can be used to discriminate between supercoiled and covalently discontinuous domains in complex DNA structures. First, supercoiled regions remain native while covalently discontinuous segments denature following a thermal treatment. This effect requires that covalent continuity exists up to and during the heating treatment. Second, because negative superhelicity enhances psoralen intercalation, crosslinking density is higher in these regions. Even if supercoiled domains are destroyed after the psoralen-u.v. treatment, the imprint of superhelicity is retained and can be recognized as a higher than normal crosslinking density.

Reinitiation during lambda DNA replication resulting from either cis-Pt treatment or infection of a P2 lysogenic strain.

Nested areas of replication are observed in phage lambda replicative intermediates and arise from reinitiation from the lambda origin. Reinitiation occurs when the first round of lambda replication takes place in the presence of the drug cis-Pt or when lambda infects a host which has been preincubated with the drug. In the latter case it is shown that the infection proceeds during the expression of SOS functions induced in the host as a result of the drug treatment. When lambda infects a host lysogenic for phage P2, an interference process occurs which prevents formation of lambda phage. The lambda DNA does, however, undergo at least one round of replication but is abnormal in that lambda origins reinitiate to form nested areas of replication similar to those resulting from exposure to the drug cis-Pt.

Physical evidence for the temporal transition of transcription in bacteriophage lambda.

A high proportion of intracellular lambda DNA molecules are found to have D-loops, when isolated under four different conditions: (1) lambda Ots after 7 min at 31 degrees C in the presence of chloramphenicol; (2) lambda Ots after 7 min at 31 degrees C without chloramphenicol; (3) lambda Ots after 30 min at 42 degrees C; and (4) lambda cIIcIII after 50 min at 37 degrees C. The great majority of these D-loops contain RNA and are produced by E. coli RNA polymerase. In the presence of chloramphenicol, D-loops are mostly limited to the immediate early regions of the major leftward and rightward operons. At early times, with no chloramphenicol present, D-loops map primarily within the delayed early regions of the two major operons. At late times, D-loops are found mostly within the major late operon of the bacteriophage DNA. This physical evidence corroborates evidence of the temporal transition in lambda transcription obtained by other means. Chloramphenicol is shown to block the transition from immediate early to delayed early transcription.

Construction and characterization of a chimeric plasmid composed of DNA Pfrom Escherichia coli and Drosophila melanogaster.

A chimeric plasmid has been constructed in vitro from colicin E1 factor (Col E1), nontransmissible R-factor RSF-1010, and Drosophila melanogaster DNAs by the sequential action of Escherichia coli endonuclease RI(Eco RI) and T4 phage DNA ligase. The chimeric plasmid was assembled in two stages--first, a composite plasmid consisting of Col E1 and RSF 1010 was constructed, followed by partial digestion of the composite with Eco RI (in order to open one of the susceptible cleavage sites) and ligation with an Eco RI-digested D. melanogaster DNA preparation. The chimeric plasmid was selected and amplified in vivo by sequential transformation of E. COLI C with the ligated mixture, selection of transformants in medium containing streptomycin plus colicin E1, followed by amplification in the presence of chloramphenicol and purification of the extracted plasmid by dye-buoyant density gradient centrifugation in ethidium bromide-CsCl solution. Treatment of the chimeric plasmid with Eco RI yields three fragments with mobilities corresponding to the linear forms of the constituents--COL E1, mol wt 4.2 times 106, RSF 1010, mol wt 5.5 times 106 and D. melanogaster DNA, mol wt 4.0 times 106. The buoyant densities of the three constituents are respectively 1.706, 1.719, and 1.697 g/cm3, while the buoyant density of the composite factor is 1.712 and that of the chimeric plasmid is 1.705. Serratia marscesens endonuclease R (Sma) which introduces a single cut in Col E1, but not in RSF 1010, converts the chimeric plasmid to a single linear molecule (mol wt 13.7 times 106) and sequential digestion with both Sma and Hin III yields two distinct fragments, mol wt 3.7 and 10.0 times 10.6, respectively; this implies that the two sites are unique and occur at distinctly different positions. Sequential digestion with both Hin III and Eco RI reveals that the Hin III cut is in the D. melanogaster segment; neither Col E1 nor RSF 1010 contain sites susceptible to digestion with Hin III. In the presence of chloramphenicol, the chimeric plasmid continues toreplicate for 9 hr while bacterial chromosomal DNA replicates at a much slower rate. As in the case of the composite plasmid, continued synthesis is the presence of chloramphenicol suggests that the replicator of Col E1 is functional in the chimeric plasmid as well. Examination of the chimeric plasmid by partial denaturation mapping permits identification of its constituents, each of which presents a characteristic profile. The D. melanogaster segment reveals a wealth of detail at the molecular level pertaining to the distribution of AT-rich regions.