S1 nuclease recognizes DNA conformational junctions between left-handed helical (dT-dG n. dC-dA)n and contiguous right-handed sequences.

The ability of negative supercoiling to induce a left-handed helix in the recombinant plasmid pRW777, which contains a tract of 64 base pairs of almost perfect (dT-dG) . (dC-dA) from the mouse kappa immunoglobin gene, was studied. S1 nuclease recognizes and cleaves within the junction region which must exist adjacent to the (dT-dG)n . (dC-dA)n tract when in a left-handed state. The cleavage pattern indicates conformational flexibility and structural differences between the two existing junctions. The 64-base pair alternating copolymer undergoes the supercoil-induced formation of a left-handed state over the superhelical density range of -0.04 to -0.06, indicating that (dT-dG)n . (dC-dA) sequences form a left-handed helix less readily than (dC-dG)n . (dC-dG)n sequences of equivalent length. However, these supercoil densities are within the range found in vivo. Supercoil relaxation and antibody binding studies confirmed that the (dT-dG)n . (dC-dA)n tract in supercoiled pRW777 was in a left-handed helix.

' The abbreviation used is: bp, base pair. sional strain of negative supercoiling was sufficient to came the formation of left-handed helices in (dC-dG), . (dC-dG), tracts under physiological salt concentrations and at negative supercoil densities (0.03-0.07) which are within the range found for the majority of supercoiled DNAs (14). We discovered (4) that the junctions are specifically recognized and cleaved by the single strand specific S1 nuclease and thus have developed an assay for the presence of (dC-dG) junctions when the CG region is left-handed (6).
This assay (4) was used herein to investigate the ability of negative supercoiling to induce a left-handed helix in the (dT-dG),.(dC-dA), regions of pRW777 and to determine if S1 nuclease also cleaves at the junctions which were generated. This plasmid (11) contains an insertion into pBR322 of a 105bp fragment derived from the 3' side of the mouse K immunoglobin gene containing a region of 64 hp of (dT-dG)s2.(dCd A ) 3 p . The fourth dG . dC pair is substituted by a dA . dT pair, thus preserving the alternating purine-pyrimidine sequence. Our results demonstrate that S1 nuclease does recognize and cleave within junction regions between left-handed (dT-dG),. (dC-PA), sequences and nonalternating B-DNA sequences. Furthermore, the junctions have different extents of conformational flexibility as measured by nuclease sensitivity.

MATERIALS AND METHODS
DNA and Enzymes-Plasmid DNA was isolated as described (11). S1 nuclease preparation and characterization was as described (20). Restriction enzymes were from Bethesda Research Laboratories. Topoisomeric samples of pRW777 (11) were generated and characterized essentially as previously reported (21). The mung bean nuclease was isolated and characterized as reported (22).
Nuclease Reactions-S1 nuclease reactions were carried out as follows: 1.5 pg of DNA were incubated at 37 "C for 50 min in the presence of 0.08 units of s1 nuclease in 50 pl of 40 mM Na acetate, 50 mM NaCI, 1 mM ZnSOa, pH 4.6. Reactions were terminated by adding 2 pl of 0.5 M EDTA, pH 7.7, followed by dialysis against 15 mM Tris. C1, 6 mM MgC12, 2.5 mM dithiothreitol, 6 mM NaCI, pH 7.7. After dialysis, 1 unit of either Hind11 or HinfI was added and the samples were incubated at 37 "C for 4 h. Samples were electrophoresed in 2% agarose gels (80 mM Tris acetate, 40 mM Na acetate, 2.5 mM EDTA, pH 8.3, for 5 h at 175 volts. Microdensitometric tracing of a photographic negative of the electrophoretic separations has been described (4,10).
Supercoil Relaxation Studies-Two-dimensional gel electrophoresis was performed essentially as described (23). Electrophoresis in the first dimension was in a 1.5% agarose gel in 80 mM Tris acetate, 40 mM Na acetate, 2.5 mM EDTA, pH 8.3, a t 125 V for 55 h. The strip containing the ladder of separated topoisomers was then cut from the gel, soaked for 1 h in the same buffer plus 1.25 FM chloroquine (Sigma), then embedded in a 1.5% agarose gel also containing 1.25 p M chloroquine. The second dimension electrophoresis then was performed at right angles to the first dimension in the chloroquine containing buffer at 100 V for 48 h. The gel was stained with ethidium bromide and the DNA visualized under uv light.

RESULTS AND DISCUSSION
SI Nuclease Cleavuge of Junctions-S1 nuclease was used to probe for conformational junctions which would exist if the (dT-dG)32-(dC-dA)az region of pRW777 were stabilized in a left-handed state by the torsional strain of negative supercoiling. Fig. l a shows the gel electrophoretic separation of the DNA fragments generated when either pRW777 or the control plasmid pBR322 was treated first with S1 nuclease followed by cleavage with HindII. Both plasmids were at an initial superhelical density of approximately -0.06 under S1 nuclease reaction conditions. For pBR322, two S1 nuclease specific fragments were found at about 870 and 2400 bp. These fragments represent S1 nuclease cleavage at the major inverted repeat of pBR322 (24,25) which lies within the 3255bp HindII fragment. Four other S1 nuclease specific fragments ranging from 696 to 513 bp were found for pRW777. These two sets of doublet bands were as expected if the (dT-dG)s2-(dC-dA)3z region was in a supercoil-induced left-handed state. This type of doublet pattern was shown (4) to be characteristic for left-handed (dC-dG),. (dC-dG). sequences abutting righthanded regions. Indeed, the four bands which possess one HindII terminus and one S1 nuclease-cleaved terminus have the expected lengths (k 4%) if S1 cleavage occurred at the ends of the (dT-dG)a2.(dC-dA)32 sequences (Fig. lb). Other  FIG. 1. pRW777 treated with S1 nuclease followed by cleavage with restriction endonuclease HindII. a, pRW777 (sec-7nd) and pRR322 (third) initially possessed mean superhelical den-5ities of -0.058 and -0.060, respectively, before S1 and HindII creatments. The samples were electrophoresed in a 2% agarose gel. The lirst and fourth contain size markers (from top to bottom): 1133,390,880,610,472,413, and 366 bp. The lengths of the HindII fragments of pRW777 (3255 and 1216 bp) are indicated in parentheses. The lengths (k 4%) found for the SI specific fragments are also given; sizes were determined by using the markers of known lengths (29). b, the top line represents the 1216-bp HindII fragment of pRW777 which contains the alternating (dC-dA). (dT-dG) region. The number of bp from the HindII sites to the first residue in the alternating dinucleotide sequence are given. The arrow represents the T. A substitution for C.G at the fourth repeat from the left. The value in parentheses indicates the distance to the (dT.dA) interruption. The lower two lines represent the mapping of the S1 nuclease cleavage sites as determined from a. mapping studies with HinfI and with AvaII agree with this interpretation.
Effect of Supercoil Density-Supercoiling causes a sharp transition from a right to a left-handed helix in (dC-dG) segments of recombinant plasmids (4,6,(8)(9)(10). Topoisomeric samples of pRW777 were generated and characterized as described (21) to evaluate this behavior with (dT-dG). (dC-dA). Each sample was treated with S1 nuclease followed by HinfI digestion and the products were separated (Fig. 2a). The mean negative superhelical density of the starting pRW777 increases from left to right (lanes [1][2][3][4][5][6][7][8][9]; lane 10 is the pBR322 control. Fig. 2a shows that the S1 nuclease specific sets of doublet bands appear in increasing intensity as the negative superhelical density increases. Fig. 26 shows a plot of the per cent specific cleavage uersus supercoiling. The demonstrated transition represents a supercoil induced shift in the equilibrium between the right and left-handed helical states for the (dT-dGIS2. (dC-dA)s2 sequences. Two other assays for left-handed DNA (described below) confirm this interpretation. Formation of left-handed DNA causes a loss (relaxation) of negative supercoils (about two supercoils lost/turn of helix converted from B to Z) which allows a a. 1 2 3 4 5 6 7 8 9 1 0 FIG. 2. Sensitivity of topoisomeric samples of pRW777 to pRW777 followed by digestion with Hinff. The lengths (in bp) of the S1 nuclease. a, S1 nuclease treatment of topoisomeric samples of Hinff fragments are indicated. The S1 nuclease specific bands are indicated by brackets. The mean negative superhelical density (under S1 nuclease reaction conditions) of each sample is as follows: 0. plasmid, pBR322 (superhelical density of -0.06). treated in an identical manner. b, a plot of specific cleavage by S1 nuclease uersus the mean negative superhelical density (-5) of the pRW777 samples. A photographic negative of a was traced using a Joyce-Loebl microdensitometer (under conditions of linearity). Specific cleavage represents the ratio of the area of the top S1 nuclease specific doublet to the area of the 396-bp band divided by the molar ratio (representing 100% cleavage) of these bands based on their calculated lengths).
by guest on March 25, 2020 http://www.jbc.org/ Downloaded from decrease in the free energy of the supercoil state. Since the decrease in free energy is quadratically related to increasing negative supercoiling (26, 27), the transition found (Fig. 26) is as expected (4,lO). A plateau of S1 nuclease cleavage around 25% is consistent with plateau levels (20-33%) for the supercoil-induced left-handed state for (dC-dG), blocks (4,10).
Thus, these results demonstrate that the torsional strain of negative supercoiling causes the formation and stabilization of a left-handed helical state in regions of (dT-dG). (dC-dA). A titratable negative superhelical density of about 0.06 (under S1 nuclease reaction conditions) is required to completely stabilize the (dT-dG),, . (dC-dA)32 block of pRW777 in a lefthanded state (in good agreement with supercoil relaxation studies described below). The precision (k 4%) of our mapping of the S1 nuclease cleavage sites does not reveal if the dA.dT interruption near one end is within the left-handed portion of this region.
Conformational Flexibility of Junctions-S1 nuclease (Figs. 1 and 2) showed a marked preference for cleavage within one junction (the right hand junction of Fig. 16) as compared to the other (left hand) junction.
Moreover, this preference became more pronounced as the negative supercoiling increased (Fig. 2a). A similar sequence and supercoil dependent junctional flexibility has been found for (dC-dG),,. (dC-dG), junctions (4,10). Thus, conformational flexibility within the junction regions appears to be independent of the type of sequence undergoing the right-handed to left-handed transition and must be related to the neighboring base pair sequence (10). This distinctive hierarchy of S1 nuclease susceptibility indicates sequence-dependent conformational efforts within the junctions. An understanding of this behavior must await further detailed structural studies since evaluation of the sequences at the junctions does not reveal an explanation. However, it should be emphasized that even x-ray crystallographic analysis to atomic resolution on one junction will be only partially informative since each junction seems to respond differently to supercoiling as measured by S1 nuclease sensitivity.
Other Single Strand Specific Nucleases as Probes-The ability of other single strand specific nucleases to interact with junction regions was compared. We have shown that another single strand specific nuclease, the T7 gene three protein (24), will not cleave within junctions flanking (dC-dG),.(dC-dG), sequences when in a left-handed state (4). Similarly, this nuclease did not cleave the junctions existing within pRW777, even at very high negative superhelical densities (data not shown). In contrast, mung bean nuclease will recognize and cleave within (dC-dG), I (dC-dG), junctions but not within (dT-dG),. The BAL31 nuclease has recently been shown (29) to specifically recognize and cleave the salt-induced junctions between right and left handed-helices. The advantages of this nuclease are its pH optimum near neutrality and its insensitivity to very high ionic conditions. Unfortunately, the BAL31 enzyme is not a useful probe for the types of junctions in pRW777 since (dT-dG),,.(dC-dA), is only partially lefthanded in saturated NaCl solution (11) and the presence of an exonucleolytic activity (29) precludes its use as a high J. Klysik and R. D. Wells, unpublished results. resolution (to within a few base pair) tool.
In summary, S1 nuclease is the most sensitive and generally useful probe for these junctions, and thus for left-handed DNA, as compared to these other nucleases.
Since three single strand specific nucleases, isolated from widely different sources, specifically cleave conformational junctions (albiet with different recognition features), we conclude that the structural aberrations possess elements of nonhelical structure as found in random coil polynucleotides.
Supercoil Relaxation of pR W777-Two-dimensional gel electrophoresis studies (23) were performed on a family of topoisomers of pRW777 in order to demonstrate directly the presence of a left-handed segment. This assay is similar conceptually to the one-dimensional electrophoretic assay for the supercoil-induced B to Z transition used previously by this laboratory (4-7, 10, 28, 30) but has the advantage of providing higher resolution, especially for topoisomers with greater supercoil densities. Fig. 3 shows the mobilities in the first dimension of topoisomers of pRW777 and pRW451, a control plasmid, as determined by this technique. For pRW451, a smooth curve of increasing mobility with decreasing linking number is observed. For pRW777, however, a sharp break in this pattern is observed between the topoisomer with 20 negative supercoils (topoisomer -20) and that with 21 negative supercoils (topoisomer -21). Instead of having a greater mobility than topoisomer -20, topoisomer -21 migrates at the same rate as topoisomer -19 in the first dimension. Similarly, topoisomer -22 co-migrates with topoisomer -17 and topoisomers -23, -24, -25, -26, and -27 comigrate with topoisomer -16 in the first dimension. The presence of chloroquine in the second dimension reduces the degree of supercoiling of the topoisomers. Thus, under the conditions of electrophoresis in the second dimension, the (dT-dG)3z.(dC-dA)3z tract in, for example, topoisomer -27 reverts from a left-handed to a right-handed helical form and the most relaxed topoisomer. Topoisomer mobilities from different experiments were normalized using the distance between the most relaxed topoisomer and the topoisomer with 10 negative supercoils (topoisomer -10) as a normalization factor. Mobilities were determined from three separate experiments for each plasmid. T = number of supercoils, thus the values to the right of zero represent negative supercoils (i.e. right-handed supercoils). 0, pRW777; X, pRW451, a control plasmid (5) containing the 174-bp HhaI fragment from pBR322 cloned into the filled in BamHI site of pBR322. Other details are described under "Materials and Methods." Left-handed (dT-dG)n.(dC-dA), its mobility in this dimension is normal (slightly faster than topoisomer -26). It is apparent that the superhelical density of topoisomer -21 is sufficient to initiate the right to lefthanded transition in the (dA-dC)3z. (dT-dG)3z tract and that the superhelical density of topoisomer -27 is required before the maximum relaxation of 11 supercoils is observed.
A relaxation of 11 supercoils agrees well with the 11.6 superhelical turns that would be relaxed if the entire (dT-dG)3P' (dC-dA)32 tract underwent a transition from a righthanded B-helix to a left-handed Z-helix. However, junction regions must exist between the right and left-handed helices, and it is quite possible that this (dT-dG)32. (dC-a),, region adopts some other type of left-handed helical conformation. Also, the conclusion of a relaxation of 11 supercoils makes the assumption that a topoisomer with -16 superhelical turns will have the same electrophoretic mobility in the first dimension as a topoisomer with -16 superhelical turns and a region of left-handed helicity, i.e. that the left-handed region does not substantially affect the electrophoretic mobility. This assumption appears to be reasonable, especially since the lefthanded region comprises < 1.5% of the plasmid. The superhelical density of the topoisomers over which the relaxation of supercoils is observed (-0.044 to 0.060) is in excellent agreement with the superhelical density range over which the (dT-dG)s,.(dC-dA)32 region undergoes the right to lefthanded transition as detected by the S1 nuclease cleavage of junction regions (Fig. 2b). Any difference between the two could be due to the different ionic conditions and/or pH at which the two sets of experiments were performed. For topoisomers -21 through -26, it can be seen that less than the maximum observed number of 11 superhelical turns are relaxed. Three possible interpretations of this are, first, for these topoisomers some, but not all, of the (dT-dG)3z. (dCa),, region is in a left-handed conformation. Second, for these topoisomers the (dT-dG),,-(dC-dA)32 region adopts intermediate conformations. Third, these topoisomers represent time averaged equilibria between the right and left-handed state of the (dT-dGfs2. (dC-dA)32 region, where the transition between the two states is fast compared to the time of electrophoresis. The fact that S1 cleavage is always observed in the same position on the DNA throughout the range of superhelical density at which the right to left-handed transition occurs (Fig. 2a) favors the third interpretation, or possibly the second interpretation if the intermediate conformations extended throughout the (dT-dG)32. (dC-dA)32 region.
The use of this supercoil relaxation assay to monitor B to Z structural changes, in addition to mapping the junctions with a single strand specific nuclease, is important since other types of supercoil-induced structural changes (i.e. cruciforms, nonpaired loops due to slippage, etc.) have been reported (24,25,(31)(32)(33). Also, this relaxation assay reveals that cruciform formation is unlikely since the extent of relaxation would be approximately half of that found; moreover, there is no apparent reason why this sequence should form a cruciform. Furthermore, the antibody binding studies (described below) confirm our conclusions on the formation of left-handed helices.
Energetics of Transition-As stated, a negative superhelical density of 0.06 was required to complete the right to lefthanded helical transition for the (dT-dG)jz. (dC-dA)32 block of pRW777. From previous data (lo), we can estimate that this level of supercoiling would completely stabilize a (dC-dG),,(dC-dG), tract in a left-handed state when n 3 8 (assuming a similar free energy of junction formation for both copolymers). Thus, the (dC-dG) copolymer much more readily adopts the supercoil-induced left-handed state than the (dT-dG) -(dC-dA) polymer.
Along with the free energy of superhelix formation (26, 27) and our previous estimates of about 5 kcal/mol for the junction free energy (10,30), these data can be used to estimate the free energy difference between the right and left-handed state for the 64-bp copolymer of pRW777. At the transition midpoint (-0.0491, the free energy difference will be zero between this sequence in a right-handed state with the plasmid at this density and the copolymer tract in a left-handed state with the plasmid at a density corresponding to a loss of 11.6 turns ( 5 ) . Thus, after subtracting the AG of junction formation, the AGRL is simply equal to the negative of the free energy difference between the two supercoiled states of the plasmid. This calculation gives an estimate of AGRL = 0.70 kcal/mol of bp for the (dT-dG)3z. (dC-dA)32 tract. Consistent with the above comparison, this value is significantly higher than the 0.44 kcal/mol of bp estimate for (dG-dC),. (dG-dC), containing plasmids where n = 10, 28, and 32 (30). 3 Antibody Binding Studies-The capacity of pRW777 to bind antibodies, both poly-and monoclonal, raised versus brominated (dG-dC),.(dG-dC), was tested by the gel retardation assay of Pohl et al. (34). When pRW777 was supercoiled at a density of approximately -0.075, substantial binding was observed to polyclonal antibodies (data not shown). However, when the plasmid was relaxed, no binding was found. Identical results were observed for pRW751 which has been demonstrated to contain regions of left-handed helices by a number of techniques (6). When pRW777 (at supercoil densities of either 0 or -0.075) was tested for its capacity to bind monoclonal antibodies, no gel retardation was found. Other studies (35)4 have revealed that different monoclonal antibodies interact with various parts of helices and probably recognize differences between types of left-handed helices. Hence, this suggests that the TG.CA tract in pRW777 is in a left-handed structure which differs from the canonical Zstructure found for short oligomers of (dC-dG) (reviewed in Refs. [1][2][3]. While this manuscript was in preparation, two papers appeared which extend our previous work (11,(15)(16)(17) on the left-handed properties of (T-G) . (C-A) sequences. Polyclonal antibodies were shown (18) to bind to negatively supercoiled, but not relaxed, pAN064 (which is similar to pRW777 (11)) and supercoil relaxation studies were reported (19) on pDHfl4, a pBR322 containing a polylinker derivative containing (dT-dG)30. (dC-dA),O polymer. The experiments described herein with a different plasmid are in excellent agreement conceptually with those data (18, 19) but utilize the approaches of both papers (as well as the use of monoclonal antibodies) and furthermore evaluate the properties of the junctions between the left and right-handed segments.
Biological Role-The biological role of the 64-bp (dT-dG) . (dC-dA) tract which occurs approximately 300 bp on the downstream side of the mouse K2 V-region (36) is uncertain. Similar sequences which appear to be an evolutionarily conserved repeat family in Xenopus, pigeon, slime mold, yeast, mouse, and human have been found between the human 6 and p globin genes (37). This copolymer tract was found also in the spacer of a sea urchin histone gene repeat (32), at boundaries between human fetal globin genes which neighbor short tracts (n z 3) of (dG-dC),.(dG-dC), (381, and in telo- and dG residues and this sequence is highly repeated in the human genome (40). The role of these simple repeating sequences as recognition signals for recombination events leading to gene conversion and unequal crossover has been postulated (32, 41). Since this unusual but ubiquitous sequence has the capacity to adopt a left-handed structure, its biological properties may be due to its conformational pliability.