- Email: [email protected]
Pancreatic DNAase cleavage sites in nuclei
Cell,
Vol.
IO, 537-547,
Pancreatic
March
1977,
Copyright
DNAase
0 1977
by MIT
Cleavage
Barbara Sollner-Webb and Gary Felsenfeld Laboratory of Molecular Biology National Institute of Arthritis, Metabolism and Digestive Diseases Bethesda, Maryland 20014
Summary The DNA of nuclei is cleaved by a variety of nucleases in such a way that the cuts on a given strand are always separated by an integral multiple of 10 nucleotides. However, the spacing between cutting sites on opposite strands is not known for any nuclease. In this paper, we describe the determination of the spacing, or stagger, between cuts on opposite strands produced by the action of pancreatic DNAase (DNAase I) on nuclei. When nuclei are digested with DNAase I and the resultant DNA is analyzed by gel electrophoresis without prior denaturation, a complex pattern of bands is observed. A method which gives better than 90% recovery of DNA from polyacrylamide gels was used to isolate the individual fractions corresponding to these bands. The structure of the fractions was then determined using singlestrand-specific nucleases to digest singlestranded “tails” and using DNA polymerases to extend recessed 3’-OH termini of partially duplex regions. Our results show that each component consists of a double-stranded region terminating in singlestranded tails at both ends. Although both chains of every duplex are 10. n nucleotides long (n integer), the chains are never completely paired. The experiments with DNA polymerase show an abundance of structures in which the 3’-OH termini of these duplexes are recessed by 8 nucleotides, and by inference, there must be structures with 5’-P termini recessed by 2 or 12 nucleotides. Thus DNAase I acts on nuclei to produce DNA with staggered cuts on opposite strands, separated by (10. n + 8) and (10.11 + 2) base pairs (with 5’-P and 3’-OH termini extending, respectively). Two classes of models of DNA folding in the nucleosome have been proposed by other investigators to account for the presence of DNAase I cleavage sites at 10. n intervals along each DNA chain. One class of models leads to the prediction that cuts should either be unstaggered or separated by 10 nucleotides, while the other class is consistent with staggers of 6 and 4 nucleotides. Neither prediction is verified by our data; however, all these models may be made consistent with the results by assuming that the enzyme’s
Sites in Nuclei
site of recognition on nucleosomal same as its site of cleavage.
DNA is not the
Introduction Nucleases have proved to be valuable probes of chromatin structure. Use of these enzymes has revealed the presence of the approximately 200 base pair nucleosome repeat (Hewish and Burgoyne, 1973; Noll 1974a) and the 140 base pair nucleosome core (Sollner-Webb and Felsenfeld, 1975; Axel, 1975; Shaw et al., 1976). It has also given information about the internal structure of the nucleosome: digestion of nuclei or isolated chromatin with DNAase I (Noll, 1974b; Sollner-Webb, Camerini-Dtero, and Felsenfeld, 1976), staphylococcal nuclease (Axe1 et al., 7974; Sollner-Webb and Felsenfeld, 1975; Camerini-Otero, Sollner-Webb, and Felsenfeld, 1976), DNAase II (Sollner-Webb et al., 1976), or an endogenous endonuclease (Simpson and Whitlock, 1976) all generate characteristic patterns of DNA fragments, which when denatured have single-strand lengths that are multiples of 10 nucleotides. (The DNAase I referred to in this paper is pancreatic DNAase; the DNAase II is spleen acid DNAase). An identical denatured DNA pattern is produced by digestion of DNA complexed with an equimolar mixture of the four histone H2A, H2B, H3, and H4, while a subset of this pattern arises from digestion of DNA complexes with only the arginine-rich histones, H3 and H4 (Camerini-Otero et al., 1976; Sollner-Webb et al., 1976). Such uniform response to nuclease attack must reflect some regularity of DNA and histone arrangement within the nucleosome: the histones protect most of the phosphodiester bonds, leaving accessible to cleavage only sites that are spaced at integral multiples of IO nucleotides. Two classes of models of DNA arrangement within the nucleosome have been proposed to account for the regularity of cleavage by DNAase I (Nell, 1974b; Crick and Klug, 1975; Sobell et al., 1976). Although the models explain the accessibility of each DNA strand at intervals of 10.n nucleotides (n integer), they differ in their predictions concerning the relative position of the cutting sites on the two opposing DNA strands. Variation in this position determines the closest spacing between cuts on opposite strands. When such cuts occur nearby on both strands, overlapping ends are produced in which base pairing may be weak enough to allow that region to denature spontaneously. Partial DNAase I digests may therefore consist of double-stranded fragments with single-stranded tails. The length of the tails is a measure of the separation between staggered cuts on the two strands.
Cdl 536
In this paper, we determine the length of the single-stranded regions in the undenatured DNAase I digestion product and the sizes of the double-stranded regions to which they are attached. This is accomplished both by the use of Sl nuclease to degrade selectively the single-stranded regions and by the use of E. coli DNA polymerase II to extend recessed chains at the 3’-OH end, filling in the single-stranded region. Our experiments reveal structures in which the 3’-OH end is recessed by 8 nucleotides. By inference, structures should exist with the 5’-P end recessed by 2 or 12 nucleotides. These results are not explained by any of the above mentioned models of nucleosome structure in their simplest forms, but all the models can be suitably modified. Results When nuclei are digested by DNAase I and the resultant DNA is denatured and separated on gels, a characteristic fragment pattern is observed (Nell, 197413). Figure IA shows the kinetics of DNA fragment production during DNAase I digestion of nuclei. We have confirmed the sizes of the denatured fragment as multiples of 10 bases in length, with relatively intense 110, 80, 50, and 30 base fragments. (Camerini-Otero et al., 1976). DNAase I cutting sites are located on both DNA strands and can be shown to be separated by 10 or 20 nucleotides over most of the DNA: when care is
Fig&
1. DNAase
I Digestion
taken to retain most of the smaller fragments by avoiding ethanol precipitation as a concentration step (see Experimental Procedures), a majority of the staining material is found in the 10 and 20 nucleotide bands when one third of the starting DNA is acid-soluble (Figure 2). At later stages of digestion, these smaller fragments continue to accumulate. Thus the vast majority of potential sites of DNAase I action occur every 10 or 20 nucleotides along each strand. We noted, however, that the native (undenatured) DNA products of DNAase I digestion of nuclei also form bands in acrylamide gel electrophoresis. The same samples that were run denatured in Figure 1A are run without prior denaturation in Figure 1 B. Bands which migrate as though they are double-strand fragments approximately 10.n base pairs in length are observed, with relatively intense bands at the positions of 120, 110, 90, 80, 60, and 50 base pairs. Each band may be resolved into an apparent doublet. If DNAase I made doublestranded cuts at every site of attack, the patterns of fragments before and after denaturation would be similar. Comparison of Figures IA and 1B shows that they are quite different, and that DNAase I does not operate on chromatin in this manner. The single-strand sizes making up the “native” fragments may easily be seen on two-dimensional gels (Figure 3). In this experiment, DNA from a DNAase I digest of nuclei (called DNAase I digest DNA) is electrophoresed in the native (underna-
Kinetics
Duck erythrocyte nuclei were digested with DNAase I, and the isolated DNA was electrophoresed on 10% denatured before electrophoresis. From left to right are 4, 6, 9, 12, 14, 16, 23, 34, and 53% of the DNA acid-soluble. before electrophoresis. Same samples were run as in (A): at the far right is a staphylococcal nuclease limit digest
acrylamide (B) DNA of chromatin
gels. (A) DNA not denatured as a marked.
Cleavage
Sites
in Nuclear
DNA
539
Figure
3. Two-Dimensional
Native (nondenatured) electrophoresed from trophoresed from top digest is shown at the at the bottom left was the first dimension.
Figure
2. Acrylamide
Gel Electrophoresis
Nuclei were digested with DNAase I, and the DNA was purified by phenol-chloroform-isoamyl alcohol extraction and concentrated by dialysis against polyethylene glycol (four left channels) or ethanol precipitation (four right channels). From left to right are dialysis of 10, 17, 28, and 33% acid-soluble digestion products, and ethanol precipitation of 33, 28, 17, and 10% acid-soluble products. The same gel stained with ethidium bromide reveals nothing smaller than 40 nucleotides except for very faint staining in this size range in the dialysis-concentrated 28 and 33% acidsoluble fractions.
tured) state in the first dimension, and is then denatured and run in the second dimension. The native fragments of approximate length 10.n are seen to be composed of the single strands of size 10.n and smaller multiples of 10 nucleotides. For example, the lower and upper halves of the 50 nucleotide pair doublet are both composed of 40 and 50 nucleotide fragments. It may be shown that the 50 nucleotide pair “native” doublet contains molecules with single-strand regions; similar arguments apply to the larger undenatured fragments. For example, if it were assumed that one band of the doublet contained fully paired duplexes composed of two 50 nucleotide strands, it would follow that the other band of the doublet must contain 50 nucleotide fragments not fully base-paired (Figure 3). It should be noted that fragments 40 or fewer nucleotides in length are largely single-stranded prior to denaturation and migrate slightly faster than the corresponding duplexes (that is, 40 base pairs) in these gels. Under
Gel Electrophoresis
of DNA
DNAase I digest DNA (53% acid-soluble) is right to left. It is then denatured and electo bottom. A marker gel of native DNAase I top. DNA visualized just above the diagonal single-stranded when electrophoresed in
our conditions, no fragments smaller than 30 nucleotides long are observed in duplexes. Single-strand regions in “native” DNAase I digest DNA were detected using single-strand-specific nucleases. When unfractionated DNAase I digest DNA is treated either with Sl nuclease or with staphylococcal nuclease under conditions where only single-strand DNA is hydrolyzed, digestion is observed. With either nuclease, prominent bands on nondenaturing gels undergo a decrease in size consistent with removal of about IO bases, and upon denaturation, this digestion product yields fragments that are approximate multiples of IO nucleotides (Figure 4; Sollner-Webb, 1976). These results confirm that undernatured DNAase I digest DNA has single-stranded ends, but it is not possible to determine the length of these ends precisely by this method. Further experiments were carried out with purified native DNA fractions of the DNAase I digest, isolated from bands cut from preparative acrylamide gels. These DNA fragments were purified, with better than 90% yield, by electrophoresis into hydroxylapatite, and subsequent washing and elution as described in Experimental Procedures. In this way, separate fractions were isolated which contain the lower and upper halves of the approximately 50 base pair doublet (“501” and “LOU”), the lower and upper halves of the approximately 60 base pair doublet (“601” and “LOU”), and the approximately 70 and 80 base pair doublets. The sizes of the single-strand fragments making up these purified fractions were determined by elec-
Cell 540
8070
Figure Digest
4. Single-Strand-Specific DNA
Nuclease
Treatment
of DNAase
Native DNAase I digest DNA (12% acid-soluble) was treated with staphylococcal nuclease under single-strand-specific conditions at 37°C for 30 min with the indicated amounts of nuclease (given as pg staphylococcal nuclease per 100 pg DNA). The DNA was then isolated and electrophoresed on 10% gels in both native and denatured forms. At the left, denatured DNAase I digest DNA was similarly treated and electrophoresed as a control of enzyme activity. “M” is a marker, a reference sample of DNA from chromatin digested to the limit with staphycococcal nuclease under low ionic strength condition; treatment of marker DNA with IO pg of staphylococcal nuclease per 100 pg DNA under single-strandspecific conditions does not alter its electrophoretic pattern.
trophoresis after denaturation (Figure 5) and confirm the conclusions reached above with unfractionated material (Figure 3). These individual “native” fragments were next treated with Sl nuclease to remove single-strand ‘tails.” Acrylamide gel analysis of the DNAase I digest patterns before and after Sl nuclease treatment permits a detailed description of the composition of the individual double-stranded components (Figure 5). For example, both 501 and 50~ of the “50 base pair” doublet in the undenatured gel pattern dissociate on denaturation to a mixture of 50 and 40 nucleotide long single strands in the ratio 4:l (Figure 5). After treatment of 50/ with Sl nuclease, only material about 40 base pairs long remains, and this, when denatured, yields only 40 nucleotide long single strands. The 501 component is thus composed entirely of species that overlap by about 40 base pairs. Treatment of 50~ with Sl nuclease demonstrates that it is composed of about equal amounts of species with 40 and 50 base pair overlaps. Thus neither the upper nor the lower band of the approx-
I
60 50
40
80 70
60
50
40
Denatured
Not Denatured I::= + SI) Figure
5. Sl
Nuclease
Treatment
of Isolated
Bands
Individual bands were isolated from DNAase 1 digest DNA on 10% acrylamide gels. The lower and upper halves of the doublets which migrate as approximately 50 and 60 base pairs (501, 50~1, 601, 60~) and the doublet which migrates as approximately 70 base pairs are electrophoresed in their undenatured (left) and denatured (right) forms, Tracings of parts of the gels are shown by the solid line; migration is from left to right. The native fragments were treated with Sl nuclease, and the resultant DNA was electrophoresed in both the native (left) and denatured (right) form, as shown by the dotted line.
imately 10.n base pair doublets is a homogeneous species. Using similar methods, the lengths of the double-stranded regions in the larger DNAase I digest fragments were also determined. As expected, the double-stranded and single-stranded DNA gel patterns of the Sl-treated material are similar in all cases. a [Conditions for carrying out the Sl nuclease digestion were determined by measuring the extent of reaction as a function of enzyme concentration (Figure 6). It is important to choose conditions such that the fragment pattern is relatively invariant with changing enzyme concentration (see Experimental Procedures). This is necessary to approach complete removal of single-stranded regions without digestion of the double-stranded regions at the ends of the duplex.] Results similar to those in Figures 5 and 6 were obtained when Sl nuclease was replaced by staphylococcal nuclease and the reaction was carried out under conditions (Kacian and Spiegelman,
Cleavage 541
Sites
in Nuclear
DNA
1974) in which the latter enzyme digests single strands selectively (data not shown). Measurements of the denatured products after Sl nuclease digestion suggest that the overlaps in the undenatured DNAase I fragments are not exact multiples of IO base pairs in length, but might be slightly smaller or somewhat heterogeneous (Figure 5). It is not known, however, whether Sl nuclease is complete and absolutely specific in its attack on single-stranded ends of this kind, so that precise determination of the length of these ends is impossible using Sl nuclease alone. We therefore made direct measurements of the size of one class of single-stranded regions (those with a recessed 3’-OH terminus) using a DNA-dependent DNA polymerase: E. coli DNA polymerase II adds nucleotides to a 3’-OH primer to fill in single-stranded regions of a partially duplex DNA template (Wickner et al., 1972a, 1972b). As an initial control of the
bases
completeness of the reaction, Hind III and Eco RI restriction nucleases were used to cut, respectively, SV40 and ColEl DNA, and the products were filled in using DNA polymerase II and radioactive deoxynucleoside triphosphates (Figure 7). The molar ratio of the bases incorporated into Hind Ill cuts is a sensitive assay for incomplete repair of singlestrand regions as well as for 5’ to 3’ exonuclease activity, while G or C incorporation into Eco Rl ends is a sensitive assay for 3’ to 5’ exonuclease activity or other aberrant synthesis. It is seen (Figure 7) that DNA polymerase II incorporates the 4 nucleotides in the expected molar ratios consistent with faithful and complete repair of the singlestranded terminus, and with the absence of significant measurable exonuclease activity that would shorten a protruding 5’ end terminus (the polymerase template) or that would initially shorten a recessed 3’ end before chain elongation occurs. The substrates for the DNA polymerase II incorporation studies were total unfractionated DNAase I digest DNA, as well as the isolated lower and upper halves of the “50” and “60” base pair doublets and the “70” and “80” base pair fragments. After filling in the recessed 3’-OH ends of the native fragments with &“P-labeled nucleoside triphosphates, the DNA was denatured, electrophoresed on acrylamide gels, and visualized by autoradiography (Figure 8). Discrete radioactive fragments are produced which are slightly less than exact multiples of 10 nucleotides in length. We find that the 501 fragment, for example, incorporates additional nucleotides to generate single-strand chains slightly less than 50 nucleotides in length, while the 50~ yields single-strand fragments slightly less than 50 and slightly less than 60 nucleotides in length. The Pol II
Hind
III
Base
5 3
Eco RI
Figure 6. DNA centrations
Fragments
Produced
by Varying
Sl Nuclease
Con-
The lower half of a 50 base pair doublet was treated with increasing amounts of Sl nuclease, and the DNA was denatured and electrophoresed. Flanked by marker total DNAase I digest DNA, from left to right are 0,0.05,0.12, and 0.4 units Sl per pg DNA.
5 3
Ratios
A
G
C
T
3 T-T-C-G-A
1.0
0.86
0.84
0.87
7 C-T-T-A-A
1.0
0.04
0.03
0.84
Figure 7. Polymerase Control Experiments SV40 DNA cleaved with Hind 111 and ColEI DNA cleaved with Eco RI were treated with E. coli DNA polymerase II in three separate reactions in the presence of 3H-dCTP, one 32P-labeled and two unlabeled deoxynucleoside triphosphates. The *H-dCTP incorporation was the same (2 10%) in the three reactions of any one run; incorporation into Hind Ill cut DNA was normalized to the average 3H incorporation. 0.3 units DNA polymerase II were used to fill in 0.5 ng single-strand DNA ends; these conditions are saturating with respect to polymerase. With IO-fold lower polymerase concentration, about 3 fold lower total incorporation was observed, but with the same molar base ratios. Results are normalized relative to adenylate incorporation taken as 1.
Cell 542
80
70
60
50
40
100 90 80 70 Figure Digest
8. DNA Bands
Polymerase
II Reaction
with
Isolated
DNAase
I
The isolated lower and upper halves of the “50” and “60” base pair doublets (501, 5Ou, 601, 60~); the “70” and “80” base pair doublets, and total unfractionated DNAase I digest DNA (T; 23% acid-soluble) were treated with DNA polymerase II (0.5 units per 50 ng DNA) in the presence of 32P-deoxynucleotide triphosphates. The DNA was denatured and electrophoresed on 30 cm 10% acrylamide gels with unlabeled carrier total DNAase I digest DNA. The gel was stained for DNA (left), and the approximate band positions were marked with radioactive ink. An autoradiogram was then made of the gel (right). The center well shows a DNAase I digest of nuclei containing j4C-labeled DNA.
same gel has been stained
for carrier
DNA (Figure
8). The method of calibration is shown in Figure 9. The sizes of the filled in fragments were determined to be between 7.5 and 8 nucleotides larger than the (1O.n) nucleotide long DNAase digest DNA. Thus DNA polymerase II adds about 8 nucleotides to the recessed 3’-OH ends. We conclude that DNAase I makes cuts with a stagger of 8 nucleotides. In interpreting the DNA polymerase II results, it is important to consider the efficiency of the repair process. In control experiments, using as substrates EC0 RI-treated ColEl DNA or Hind illtreated SV40 DNA, total incorporation varied between 40 and 100% of that expected for complete repair of all substrate molecules. In all cases, however, the stoichiometry of incorporated nucleotides corresponded to the known composition of the single-stranded termini. This is evidence, in addition to that shown in Figure 10 (see below), that when DNA polymerase II initiates the repair process on a given substrate molecule, it carries it to completion. The variability in total yield of repaired molecules may reflect the very low template concentration used.
z I m
60 50 40
30 Migration Figure
9. Sizing
of the
Radioactive
Fragments
Densitometer scans of the autoradiogram DNA stain (dotted line) of the same well location of the adjoining radioactive ink semilog plot of migration against size was well using the 10.n base long DNA markers. the radioactive fragments, their sizes were total DNAase I digest as well as for the shown.
(solid line) and total were aligned by the dots (vertical lines). A calibrated for each gel From the migration of determined. Data for a 501 and 601 bands are
When the purified native 501 fragment from DNAase I digests was used as substrate, total incorporation also varied between 40 and 90% of that expected if all 40 nucleotide long chains gained 8 nucleotides in length. Although 501 contains both 40 and 50 nucleotide long single-strand components (Figure 5), the absence of labeled strands larger than 50 nucleotides (Figure 8) indicates that DNA polymerase II was able to add only to the strands 40 nucleotides long. The elongation of the 40 nucleotide long strand from 501 to a length of 48 nucleotides is thus occurring with an efficiency comparable to that of the controls discussed above. It should be noted that virtually no chains shorter than 40 nucleotides are contained in 501 (Figures 5 and 6); there are, for example, not nearly enough small chains so that all radioactive 48 nucleotide fragments could result from addition of 18 nucleotides to 30 nucleotide chains. In contrast, we have not excluded the possibility that elongation by
Cleavage
Sites
in Nuclear
DNA
543
\ Figure
IO. DNA
Polymerase
I and
II Reactions
with
\\
the 501 Band
The lower half of the “50” base pair doublet was treated with polymerase, and the data are presented as in Figure 9. The dotted line is a marker, DNAase I digest DNA (A) The isolated 501 band was treated with E. coli DNA polymerase I at 0.04 units per ng DNA (upper trace) and 0.01 units per ng DNA (lower trace). Incorporation under the latter conditions was 40% of the former. (B) The isolated 501 band was treated with E. coli DNA polymerase II at 0.004 units per ng DNA (upper trace) and 0.001 units per ng DNA (lower trace). Incorporation under the latter condition was 30% of the former.
more than 8 nucleotides occurs in some of the larger native DNAase I digest fragments. Partial contributions to the radioactively labeled bands from species elongated by 10.n + 8 nucleotides (n 3 1) could occur either by filling in of species containing single-stranded regions 10.n + 8 nucleotides in length, or by initiation of the polymerase at internal gaps, followed by strand displacement during polymerization. The behavior of the 501 component provides evidence of chain elongation, but it is indirect, since no labeled chains longer than some of the starting material are produced. One could argue that no net elongation occurs, and that the labeled material is produced by a combined exonuclease and polymerase activity that serves only to shorten and label terminally the 50 nucleotide long strands of 501. This is improbable, since all components of the 501 band have duplex regions of about 40 nucleotides (Figure 5). Thus molecules containing two 50 nucleotide strands would not result in a radioactive 48 nucleotide fragment by this labeling mechanism.
If the amount of incorporation observed above had arisen from duplexes containing one 50 and one 40 nucleotide strand, up to 7 terminal nucleotides would have had to be removed and repolymerized on each of the 50 nucleotide strands (see above). However, the DNA polymerase II control experiments (Figure 7) have ruled out such extensive exonuclease activity within duplex regions. Direct evidence against this mechanism, and in favor of net chain elongation, comes from treatment of 50~ with DNA polymerase II (Figure 8). The 50~ band, like the 501 band, contains only chains that are 40 and 50 nucleotides long. In the case of 50~1, however, structures are present which give rise to radioactively labeled 58 nucleotide long fragments, which are longer than any component strands of the 50~ band. Experiments similar to those using DNA polymerase II (Figure 8) were carried out with E. coli DNA polymerase I at low temperature (5”C), where repair synthesis is favored (Wu, 1970). When unfractionated DNAase I digest DNA and the isolated 501 band are used as substrates, the bulk of the incorporated nucleotides are again contained in fragments 10.n + 8 and 48 nucleotides long, respectively (Figure 10). (Unlike DNA polymerase II, DNA polymerase I acts on the 501 component to produce also a few radioactive chains much larger than any of the starting components, and longer than could result from simple filling in of single-strand tails in “native” 501 DNA. We calculate that
Cell 544
cessed 3’-OH terminus in the complementary strand. What fragment structures can be deduced from this information? DNAase I is known to cut DNA in nuclei at intervais of 10.n nucleotides. Potential cutting sites occur every 10 or 20 nucleotides over most of the DNA (Figure 2). The presence, in the double-stranded DNAase I digest product, of structures with 3’-OH ends recessed by 8 nucleotides thus implies the presence of corresponding structures with 5’-P ends recessed by 2 or 12 nucleotides (Figure 11). While this class of single-strand ends cannot be detected using the polymerase, Sl nuclease should remove them. Using the experimental data and assuming that 5-P ends may be recessed by 2 or 12 nucleotides, it is possible to construct a model of the components, for example, of the 501 band, as illustrated in Figure 12. The 501 band contains both 50 and 40 nucleotide long single strands, in the ratio 4:1 (Figure 5). The amount of DNA polymerase II-induced incorporation is consistent with the presence of most or all of the 40 nucleotide chains in structures recessed by 8 nucleotides at the 3’ end. Digestion with Sl nuclease reveals that all fragments in the 501 fraction have duplex regions about 40 base pairs long, so that the 40 nucleotide chains must all be paired with the 50 nucleotide chains. The remaining 50 nucleotide chains must be paired with each other in conformations inaccessible to DNA polymerase II, since that enzyme acting on the 501 fraction produces no radioactive fragments longer than 50 nucleotides. Thus all 50 nucleotide long strands in 501 must have 5’ recessed ends. As we have pointed out above, such termini should be
recessed by 10.n + 2 nucleotides (Figure 11). In this particular case, the 5’ ends must all be recessed by 12 nucleotides (n = 1), since this is the only consistent way in which a structure containing two 50 nucleotide chains can yield an approximately 40 base pair duplex upon Sl digestion. (It should be noted that the various structures depicted in Figure 12 are not present in equimolar amounts and probably vary in abundance with the extent of digestion.) Using similar arguments, structures may be deduced for the components of the other native DNA gel bands. Those for band 50~ are shown in Figure 12. The 50~ band might reasonably be expected to be contaminated with the slow moving component of 501, which is the duplex of a 40 and 50 nucleotide chain (see Figure 3). However, DNA polymerase II treatment of 501 generates only about a fourth as many labeled 48 nucleotide chains as does treatment of 5Ou, indicating that most of the 40 nucleotide chains present in 501 are unavailable to the polymerase, as shown in Figure 12. What relationship do the structures shown in Figure 12 have to the models of DNA arrangement in nucleosomes discussed in the Introduction? Noll (197413) has suggested that the nucleosomal DNA, lying on a histone core (indicated by the dotted region in Figure 13), is accessible to nuclease only at the phosphodiester bond most exposed to the surroundings. Cutting at these peaks (marked “p” in Figure 13) would create a stagger of 6 and 4 nucleotides oriented in such a way that polymerase would be expected to add 6 nucleotides to 3’ termini. Alternatively, Crick and Klug (1975) have pro50
4-2-t-8-+2*
\F 1 a 5’JllIIIIIIIIIIIIII t
1
3’
40
J15
Proposed
Structures
2
48
2
1
3
t
-
a
42
-
12
38
a
,5ou
2 J
& Figure
11.
DNAase
I Cleavage
I II II
I ’ II
122
4038
501
a12 -1
Sites
DNA is shown (a) with potential cutting sites (arrow) at 10 nucleotide intervals on each strand, staggered so as to create fragments with 3’-OH ends recessed by 8 nucleotides. These ends may be filled in by DNA polymerase, as indicated by (A) (b). Since cuts are often made 10 nucleotides away from another cut, fragments with 5’-P ends recessed by 2 (c) and 12 (d) nucleotides are also inferred.
Figure
12.
Proposed
Structures
for the “50
Base
Pair”
Bands
A summary of experiments with the 501 band is shown on the left. Combining this information with the stagger determined in Figure 11, DNA structures comprising the 501 band are determined as described in the text. On the right are the proposed structures for the 501 and 50~ bands. The zig-zag line represents the nucleotides which may be added by DNA polymerase.
Cleavage 545
Sites
in Nuclear
DNA
minor groove
Y-10 I 1
i;-r--+ /
I f--rpeak
r2~-+----8------2~
I 3’
5’
3’
T &m;&
T $yg.
I’
I
1
I
I
5’ t Figure
5’
13.
Models
P’ for DNAase
A portion of the nucleosome above, and the corresponding
t I Cutting
Action
(See
3’
b Text)
surface is schematically DNA is stretched out below.
shown
posed that the DNA of the nucleosomes is kinked by bending away from the major groove every 10 or 20 base pairs, and that cleavage occurs at these kinks. This would create either ends with no singlestrand tails, or with tails 10.n nucleotides long, which polymerase might fill in. Sobell et al. (1976) recently suggested another kind of kink for nucleosomal DNA, which bends away from the minor groove. The sites of this kink are labeled “minor groove” (m.g.) in Figure 13; the predicted fragment pattern for cleavage at this site is, of course, the same as for the Crick and Klug model. The observed stagger of 8 and 2 nucleotides for the DNAase I cleavage sites is not consistent with the predictions made from any of the models discussed above. All the models can be made consistent with the results, however, if we suppose that the enzyme binds at the accessible site, but cuts one nucleotide away in a specified direction. For example, if binding actually occurs at the peaks (sites marked “p” in Figure 13), but the cuts are made 1 nucleotide to the 5’ side of this site, staggered cuts at 8 and 2 nucleotide intervals (marked by arrows by Figure 13) would be created. On the other hand, if we suppose the enzyme recognition sites are at 10 base pair intervals, as for instance at the points marked “m.g.,” we must postulate that the cuts are made 1 nucleotide to the 3’ side of these sites to account for the observed cleavage pattern. [If the DNA of the nucleosome is lying on, and shielded by, a histone core, the Sobell kink would appear to be more accessible to a nuclease than the Crick-Klug kink (displaced 5 base pairs along the helix from the Sobell kink). Examination of the published kink models shows that the poten-
tial DNAase I cleavage sites are raised to the surface in the Sobell kink, but are virtually buried in the Crick-Klug kink. However, experiments of the kind described here are unable to distinguish between these two types of kinks or any other structure which repeats every 10 base pairs along the DNA backbone. Furthermore, it is not known that DNAase I would favor cutting near a kink over cutting in a linear region of B form DNA.] Our results provide new information about the location of DNAase I cleavage sites within the nucleosome; the cleavage pattern is not predicted by either of the two previously proposed classes of models. As we have shown, either class of models can be used to explain the result with the help of suitable assumptions about the mode of action of pancreatic DNAase. Further understanding of this mode of action would help us to decide whether either of the models is correct, or whether the data must be explained in terms of some other model of nucleosome structure. Experimental
Procedures
Isolation of Nuclei and Chromatin Nuclei were isolated from frozen duck erythrocytes by several successive washings in IO mM Tris-HCI (pH 8), 1 mM MgCI,, 0.25 M sucrose as described previously (Sollner-Webb and Felsenfeld, 1975). The addition of 0.5% Triton X-100 in an early wash had no effect on the digestion results. Chromatin was prepared from salt-washed nuclei as described previously (Camerini-Otero et al., 1976). Enzymic Treatments of Nuclei, Chromatin, and DNA DNAase I (pancreatic DNAase; Worthington, 3750 U/mg) was used to digest nuclei (0.5-3 mg DNA per ml) in 10 mM Tris-HCI (pH 8), 1 mM MgCI,, 0.25 M sucrose at 37°C. Enzyme concentration and time of reaction were adjusted to give the desired amount of acid solubilization, as described previously (Sollner-Webb et al., 1976). Reactions were terminated by addition of 5 mM Na EDTA. DNA was purified by proteinase K (E. Merck) digestion of the reaction mixture, followed by chloroform:isoamyl alcohol:phenol (24:1:24) extraction and ethanol precipitation as previously described (Sollner-Webb and Felsenfeld, 1975). Alternatively, EDTA-terminated reactions were made 0.1% in SDS and extracted with chloroformisoamyl alcohol-phenol, followed by ethanol precipitation as above, with no detectable differences in results. To retain small DNA fragments in higher yield, the ethanol precipitation step of the DNA isolation was replaced by dialysis into 1 mM Tris-HCI (pH 8), 0.1 mM EDTA, concentration against polyethylene glycol 6000 powder (Baker), and another dialysis against the same buffer. Sl nuclease digestions of DNAase I digest DNA (0.04 mg DNA per ml) were carried out in 50 mM Na acetate (pH 4.75), 150 mM NaCI, 0.5 mM ZnSOn at 37°C for 1 hr. Sl nuclease was purified by Drs. Michael Zasloff and Patrick Williamson by the procedure of Vogt (1973) through the sulfopropyl cellulose fractionation; the unit of activity is defined by Vogt. Reactions were performed using about 0.2 units Sl per wg DNAase I digest DNA, but each DNA sample was titrated with the nuclease, using acrylamide gel electrophoresis of the product DNA as an assay for completion of reaction (see Figure 5, where 0.12 units Si per pg DNA were chosen.) Too extensive digestion permits attack on the ends of double-stranded DNA and a blurring of the gel patterns. [To avoid this overdigestion, nuclease to substrate ratios used were about lo4 lower than Shenk et al. (1975) used to digest at sites opposite
Cell 546
nicks.] Reactions were terminated with EDTA, and the DNA was purified as described above, in this and in the digestions described below. Staphylococcal nuclease (Worthington; 30,000 U/mg) was used to digest DNAase I digest DNA (0.2 mg DNA per ml) under single-strand-specific conditions as described by Kacian and Spiegelman (1974), in 0.4 M NaCl, 10 mM Tris-HCI (pH 8), 10 mM MgCI,, 0.1 mM CaCI, at 37°C; each DNA sample was titrated with nuclease as described in Figure 3. Staphylococcal nuclease was used at low ionic strength [I mM Tris-HCI (pH 8), 0.1 mM CaCI,] to digest isolated chromatin to the limit as previously described (Axe1 et al., 1974). The sizes of the native DNA fragments produced were determined previously (Camerini-Otero et al., 1976). Eco RI restriction endonuclease was used to digest ColEl DNA (0.15 mg/ml) in 100 mM Tris-HCI (pH 7.5), 50 mM NaCI, 10 mM MgCI, at 37°C for 15 min. The Eco RI was purified by Dr. James McGhee using a modification of the procedures of Greene et al. (1974) and Thomas and Davis (1975). ColEl DNA was made by Dr. R. D. Camerini-Otero by a modification of the method of Clewell and Helinski (1969). The reaction product was monitored on agarose gels to demonstrate complete reaction, as described below. Hind Ill restriction endonuclease was used to digest SV40 DNA (0.075 mg/ml) in 6.6 mM Tris-HCI (pH 7.5), 50 mM NaCI, 8.7 mM MgCI, at 37°C for 6 hr. The Hind III was purified by Dr. Mark Israel by a modification of the procedure of Smith and Wilcox (1970) through the phosphocellulose column step. Column fractions having only Hind Ill activity, with no detectable contaminating Hind II activity, were selected. The SV40 DNA was made by Dr. Maria Persico-DiLauro by the method of Sebring et al. (1974). Reaction products were monitored on agarose gels to demonstrate complete reaction. E. coli DNA polymerase II was purified by Dr. Sue Wickner by a modification of the procedure of Wickner et al. (1972a), where the unit of activity isdefined. E. coli DNA polymerase II reactions were performed according to Wickner et al. (1972a, 1972b), in 50 mM Tris-HCI (pH 7.4), 5 mM MgCI,, 1 mM dithiothreitol, 30 mM NaCI, 6 PM (approximately 5 +i/nmol), 3ZP-dTTP, 40 PM each of dATP, dGTP, and dCTP, 0.1 mg/ml bovine plasma albumin, with l-10 units polymerase per /*g of DNAase I digest DNA at 37°C for 30 min. Reactions were performed in polymerase excess unless otherwise noted. Incorporations were terminated with 5 fold excess EDTA at 0°C. Carrier DNAase I digest DNA was added, and the DNA was precipitated with 10% TCA at 0°C for 5 min, followed by centrifugation at 12,000 x g for 5 min at 0°C. Immediately the DNA was approximately neutralized with NaOH, made 0.1 M Tris-HCI (pH 8), 0.5 mM EDTA, 0.1% SDS, extracted with phenolchloroform-isoamyl alcohol, and ethanol-precipitated. No detectable depurination (measured by change in single-stranded DNA gel pattern) occurs during this procedure. To measure ratios of incorporated nucleotides, one 3H-labeled (dCTP) as well as one 32P-labeled (dTTP, dATP, or dGTP), nucleoside triphosphates were used at 10 PM concentrations each; the two unlabeled nucleoside triphosphates were at 40 PM each. Three separate reactions, one with each 32P-nucleoside triphosphate in addition to 3H-dCTP, were performed in the ionic conditions described above. E. coli DNA polymerase I reactions were performed according to Wu (1970) in 70 mM KPO, (pH 6.9), 10 mM MgSO,, 15 mM dithiothreitol, 60 mM NaCI, 10 pm (5 &i/nmol)3ZP-dTTP, 20 PM dATP, dGTP, and dCTP, 0.1 mglml bovine plasma albumin at lo50 U polymerase per pg of DNAase I digest DNA at 5°C for 4 hr. This reaction mixture has a conductivity equivalent to 190 mM NaCl. The E. coli DNA polymerase I was purified by Dr. Sue Wickner by a modification of the procedure of Jovin, Englund, and Bertsch (1969), where the unit of activity is defined. Gel Electrophoresis of DNA Fragments Electrophoresis of DNA fragments on 10% gels was carried out as described previously all cases, the acrylamide:methylene-bisacrylamide
polyacrylamide slab (Axe1 et al., 1974). In ratio was 19:l.
Allowing polymerization to proceed for several hours before use further improves the electrophoretic resolution: allowing 12-16 hr for this process is essential for good resolution of single-stranded DNA fragments. Single-stranded DNA electrophoresed under these conditions was resolved as well as it is on formamide gels (Nell, 1974b). The electrophoretic buffer was 3 mM Na EDTA, 90 mM boric acid, and 90 mM Tris (pH 8.0) (Peacock and Dingman, 1967). The sample of DNA was usually applied dissolved in the running buffer diluted 1:9 with water. Single-stranded samples were prepared by briefly exposing the DNA to 0.1 N NaOH and then neutralizing the sample with HCI before application of the gel. Gels were stained with “Stains-all” (Eastman), destained in water, and photographed (Axe1 et al., 1974). Negatives were scanned in a Joyce-Loebl microdensitometer. For the polymerase experiments, where very high resolution is necessary, 30 cm long (1 mm thick) slab gels were used. DNA was visualized by staining in ethidium bromide (30 min. 0.5 pg/ml in H,O) and photographed under an ultraviolet light. DNA bands were marked, and gels were then “fixed” with 20% TCA (30 min) and dried using a BioRad gel drier without the cellophage membrane. Kodak medical X-ray film was used for autoradiography. Two-dimensional electrophoresis was performed using 10% disc gels for the first dimension. Gels were stained for 30 min in ethidium bromide (0.5 fig/ml), and the ionic strength was further reduced by mixing an additional 30 min in 1 mM Tris-HCI (pH 8). After boiling the gels for 5 min, they were laid on a pre-electrophoresed 5 mm thick 10% gel and electrophoresed. Disc gel dimensions were chosen to fit securely into the well of the slab gel. Electrophoresis of restriction enzyme-cut DNA fragments was carried out in 1.6% agarose disc gels, as described above for the acrylamide system, but without aging of the gels. DNA was visualized with ethidium bromide. DNA Extractions from Acrylamide Gels To isolate DNA, singleor double-stranded, from acrylamide gels, a method was developed in which >90% of the DNA is recovered. The DNA gel is stained in ethidium bromide (or only a marker well is stained), and the desired band is cut out. In a disposable pipette, a 1 ml column of washed hydroxylapatite (BioRad) is formed between layers of glass wool. The column is filled with 0.1 M Tris-HCI (pH 8), and the DNA gel slice inserted and crushed against the walls with a glass rod. Glass wool is placed above the gel to hold it in place and electrophoresis is begun. Since liquid flows slowly through this column, either sufficient upper buffer volume or buffer recirculation is necessary. Electrophoresis is carried out at 200 V for 4 hr in 0.1 M Tris-HCI (pH 8); bromphenol blue marker may be used to demonstrate electrical continuity through each column. The acrylamide gel fragments are then removed from the pipette, and the hydroxylapatite column is washed with a few volumes of 10 mM sodium phosphate (pH 6.7) until no further absorbing material elutes. 0.5 M sodium phosphate is used to elute the DNA in a concentrated fraction with an ultraviolet absorption spectrum identical to that of pure DNA. Since this procedure was developed, a similar method for recovering proteins from acrylamide gels has been described (Ziola and Scraba, 1976). Acknowledgments We would like to thank Dr. Sue Wickner for her generous gift of E. coli DNA polymerase II and E. coli DNA polymerase I. We are also grateful to Drs. Michael Zasloff and Patrick Williamson for Sl nuclease, Dr. James McGhee for Eco RI endonuclease, Dr. Mark Israel for Hind III endonuclease, Dr. Dan Camerini-Otero for ColEl DNA, and Dr. Maria Persico-DiLauro for SV40 DNA and j4Cthymidine-labeled cells. Finally, we especially thank Drs. James McGhee and Patrick Williamson for many helpful discussions and suggestions. Received
October
26, 1976
Cleavage 547
Sites
in Nuclear
DNA
References Axel, Ft. (1975). staphylococcal
Cleavage nuclease.
of DNA in nuclei and chromatin Biochemistry 74, 2921-2925.
Vogt, strand them.
with
Axel, Ft., Melchior, W., Sollner-Webb, B., and Felsenfeld, G. (1974). Specific sites of interaction between histones and DNA in chromatin. Proc. Nat. Acad. Sci. USA 77, 4101-4105. Camerini-Otero, R. D., (1976). The organization dence for an arginine-rich
Sollner-Webb, B., and Felsenfeld, of histones and DNA in chromatin: histone kernel. Cell 8, 333-347.
Clewell, D., and Helinski, D. (1969). protein complex in E. coli: purification anopencircularform. Proc.Nat.Acad.
Supercoiled circular and induced conversion Sci. USA62,1159-1166.
Crick,
helix.
F. H., and
Klug,
A. (1975).
Kinky
G. eviDNAto
Nature255,530-533.
Greene, P., Betlach, M., Boyer, H., and Goodman, H. (1974). The EcoRl restriction endonuclease. In Methods in Molecular Biology, 7, DNA Replication and Biosynthesis. Ft. Wickner, ed., (New York: Marcel Dekker), pp. 87-111. Hewish, D., and Burgoyne, L. (1973). Chromatin substructure: the digestion of chromatin DNA at regularly spaced sites by a nuclear deoxyribonuclease. Biochem. Biophys. Res. Commun. 52, 504510. Jovin, T., Englund, P., and Bertsch, L. (1969). sis of DNA: XXVI Physical and chemical studies DNA polymerase. J. Biol. Chem. 244, 2996-3008. Kacian, D. L., and nuclease to monitor them. 58, 534-540. Nell, 251
M. (1974a).
Spiegelman, hybridization
Subunit
structure
Nell, M. (1974b). Internal structure Acids Res. 1, 1573-l 578. Peacock, ribonucleic Biochemistry
S. (1974). reactions
Enzymatic syntheof a homogeneous Use of micrococcal with DNA. Anal. Bio-
of chromatin. of the chromatin
A., and Dingman, C. (1967). acid species by polyacrylamide 6, 1818-1827.
Nature257,249subunit.
Nucl.
Resolution of multiple gel electrophoresis.
Sebring, E., Kelly, T., Thoren, M., and Salzman, ture of replicating simian virus 40 deoxyribonucleic cules. J. Viral. 8, 478-490. Shaw, B., Herman, T., Kovacic, R., Beaudreau, Holde, K. (1976). Analysis of subunit organization erythrocyte chromatin. Proc. Nat. Acad. Sci. USA
N. (1971). acid
Strucmole-
G.,
and Van in chicken 73,505-509.
Shenk, T., Rhodes, C., Rigby, P., and Berg, P. (1975). Biochemical method for mapping mutational alterations in DNA with Sl nuclease. Proc. Nat. Acad. Sci. USA 72, 989-993. Simpson, chromatin base pair Smith, mophilus
R., and Whitlock, J. (1976). Chemical evidence that DNA exists as 160 base pair beads interspersed with 40 bridges. Nucl. Acids Res. 3, 117-127.
H., and Wilcox, K. (1970). influenzae. J. Mol. Biol.
Sobell, H., Tsai, The organization 73, 3068-3072. Sollner-Webb, chromatin. nia. Sollner-Webb, digestion Biochemistry
A restriction 57, 379-391.
C., Gilbert, S., Jain, of DNA in chromatin.
B. (1976). Ph.D. Thesis,
Organization Stanford
B., and Felsenfeld, of nuclei and chromatin 74, 2915-2920.
enzyme
from
He-
S., and Sakore, T. (1976). Proc. Nat. Acad. Sci. USA of histones in nuclei and University, Stanford, Califor-
G. (1975). A comparison of the by staphylococcal nuclease.
Sollner-Webb, B., Camerini-Otero, R. D., and Felsenfeld, G. (1976). Chromatin structure as probed by nucleases and proteases: evidence for the central role of histones H3 and H4. Cell 9, 179-I 93. Thomas, M., and Davis. R. (1975). Studies on bacteriophage lambda DNA with EcoRl restriction J. Mol. Biol. 97, 315-328.
the cleavage endonuclease.
of
V. M. (1973). Purification specific nuclease from 33, 192-200.
Wickner, Ft., Ginsberg, Deoxyribonucleic acid Chem. 247, 489-497. Wickner, R., Ginsberg, cleic acid polymerase 498-504. Wu, R. (1970). 51, 501-521.
Nucleotide
and further Aspergillus
properties oryzae.
Eur.
of singleJ. Bio-
B., Berkower, I., and Hurwitz, J. (1972a). polymerase II of Escherichia co/i, I. J. Biol. B., and Hurwitz, J. (1972b). DeoxyribonuII of Escherichia co/i, II. J. Biol. Chem. 247, sequence
analysis
Ziola, B., and Scraba, D. (1976). Recovery polyacrylamide gels by electrophoresis into Biochem. 72, 366-371.
of DNA.
J. Mol.
of SDS-proteins hydroxylapatite.
Biol. from Anal.
Vol.
IO, 537-547,
Pancreatic
March
1977,
Copyright
DNAase
0 1977
by MIT
Cleavage
Barbara Sollner-Webb and Gary Felsenfeld Laboratory of Molecular Biology National Institute of Arthritis, Metabolism and Digestive Diseases Bethesda, Maryland 20014
Summary The DNA of nuclei is cleaved by a variety of nucleases in such a way that the cuts on a given strand are always separated by an integral multiple of 10 nucleotides. However, the spacing between cutting sites on opposite strands is not known for any nuclease. In this paper, we describe the determination of the spacing, or stagger, between cuts on opposite strands produced by the action of pancreatic DNAase (DNAase I) on nuclei. When nuclei are digested with DNAase I and the resultant DNA is analyzed by gel electrophoresis without prior denaturation, a complex pattern of bands is observed. A method which gives better than 90% recovery of DNA from polyacrylamide gels was used to isolate the individual fractions corresponding to these bands. The structure of the fractions was then determined using singlestrand-specific nucleases to digest singlestranded “tails” and using DNA polymerases to extend recessed 3’-OH termini of partially duplex regions. Our results show that each component consists of a double-stranded region terminating in singlestranded tails at both ends. Although both chains of every duplex are 10. n nucleotides long (n integer), the chains are never completely paired. The experiments with DNA polymerase show an abundance of structures in which the 3’-OH termini of these duplexes are recessed by 8 nucleotides, and by inference, there must be structures with 5’-P termini recessed by 2 or 12 nucleotides. Thus DNAase I acts on nuclei to produce DNA with staggered cuts on opposite strands, separated by (10. n + 8) and (10.11 + 2) base pairs (with 5’-P and 3’-OH termini extending, respectively). Two classes of models of DNA folding in the nucleosome have been proposed by other investigators to account for the presence of DNAase I cleavage sites at 10. n intervals along each DNA chain. One class of models leads to the prediction that cuts should either be unstaggered or separated by 10 nucleotides, while the other class is consistent with staggers of 6 and 4 nucleotides. Neither prediction is verified by our data; however, all these models may be made consistent with the results by assuming that the enzyme’s
Sites in Nuclei
site of recognition on nucleosomal same as its site of cleavage.
DNA is not the
Introduction Nucleases have proved to be valuable probes of chromatin structure. Use of these enzymes has revealed the presence of the approximately 200 base pair nucleosome repeat (Hewish and Burgoyne, 1973; Noll 1974a) and the 140 base pair nucleosome core (Sollner-Webb and Felsenfeld, 1975; Axel, 1975; Shaw et al., 1976). It has also given information about the internal structure of the nucleosome: digestion of nuclei or isolated chromatin with DNAase I (Noll, 1974b; Sollner-Webb, Camerini-Dtero, and Felsenfeld, 1976), staphylococcal nuclease (Axe1 et al., 7974; Sollner-Webb and Felsenfeld, 1975; Camerini-Otero, Sollner-Webb, and Felsenfeld, 1976), DNAase II (Sollner-Webb et al., 1976), or an endogenous endonuclease (Simpson and Whitlock, 1976) all generate characteristic patterns of DNA fragments, which when denatured have single-strand lengths that are multiples of 10 nucleotides. (The DNAase I referred to in this paper is pancreatic DNAase; the DNAase II is spleen acid DNAase). An identical denatured DNA pattern is produced by digestion of DNA complexed with an equimolar mixture of the four histone H2A, H2B, H3, and H4, while a subset of this pattern arises from digestion of DNA complexes with only the arginine-rich histones, H3 and H4 (Camerini-Otero et al., 1976; Sollner-Webb et al., 1976). Such uniform response to nuclease attack must reflect some regularity of DNA and histone arrangement within the nucleosome: the histones protect most of the phosphodiester bonds, leaving accessible to cleavage only sites that are spaced at integral multiples of IO nucleotides. Two classes of models of DNA arrangement within the nucleosome have been proposed to account for the regularity of cleavage by DNAase I (Nell, 1974b; Crick and Klug, 1975; Sobell et al., 1976). Although the models explain the accessibility of each DNA strand at intervals of 10.n nucleotides (n integer), they differ in their predictions concerning the relative position of the cutting sites on the two opposing DNA strands. Variation in this position determines the closest spacing between cuts on opposite strands. When such cuts occur nearby on both strands, overlapping ends are produced in which base pairing may be weak enough to allow that region to denature spontaneously. Partial DNAase I digests may therefore consist of double-stranded fragments with single-stranded tails. The length of the tails is a measure of the separation between staggered cuts on the two strands.
Cdl 536
In this paper, we determine the length of the single-stranded regions in the undenatured DNAase I digestion product and the sizes of the double-stranded regions to which they are attached. This is accomplished both by the use of Sl nuclease to degrade selectively the single-stranded regions and by the use of E. coli DNA polymerase II to extend recessed chains at the 3’-OH end, filling in the single-stranded region. Our experiments reveal structures in which the 3’-OH end is recessed by 8 nucleotides. By inference, structures should exist with the 5’-P end recessed by 2 or 12 nucleotides. These results are not explained by any of the above mentioned models of nucleosome structure in their simplest forms, but all the models can be suitably modified. Results When nuclei are digested by DNAase I and the resultant DNA is denatured and separated on gels, a characteristic fragment pattern is observed (Nell, 197413). Figure IA shows the kinetics of DNA fragment production during DNAase I digestion of nuclei. We have confirmed the sizes of the denatured fragment as multiples of 10 bases in length, with relatively intense 110, 80, 50, and 30 base fragments. (Camerini-Otero et al., 1976). DNAase I cutting sites are located on both DNA strands and can be shown to be separated by 10 or 20 nucleotides over most of the DNA: when care is
Fig&
1. DNAase
I Digestion
taken to retain most of the smaller fragments by avoiding ethanol precipitation as a concentration step (see Experimental Procedures), a majority of the staining material is found in the 10 and 20 nucleotide bands when one third of the starting DNA is acid-soluble (Figure 2). At later stages of digestion, these smaller fragments continue to accumulate. Thus the vast majority of potential sites of DNAase I action occur every 10 or 20 nucleotides along each strand. We noted, however, that the native (undenatured) DNA products of DNAase I digestion of nuclei also form bands in acrylamide gel electrophoresis. The same samples that were run denatured in Figure 1A are run without prior denaturation in Figure 1 B. Bands which migrate as though they are double-strand fragments approximately 10.n base pairs in length are observed, with relatively intense bands at the positions of 120, 110, 90, 80, 60, and 50 base pairs. Each band may be resolved into an apparent doublet. If DNAase I made doublestranded cuts at every site of attack, the patterns of fragments before and after denaturation would be similar. Comparison of Figures IA and 1B shows that they are quite different, and that DNAase I does not operate on chromatin in this manner. The single-strand sizes making up the “native” fragments may easily be seen on two-dimensional gels (Figure 3). In this experiment, DNA from a DNAase I digest of nuclei (called DNAase I digest DNA) is electrophoresed in the native (underna-
Kinetics
Duck erythrocyte nuclei were digested with DNAase I, and the isolated DNA was electrophoresed on 10% denatured before electrophoresis. From left to right are 4, 6, 9, 12, 14, 16, 23, 34, and 53% of the DNA acid-soluble. before electrophoresis. Same samples were run as in (A): at the far right is a staphylococcal nuclease limit digest
acrylamide (B) DNA of chromatin
gels. (A) DNA not denatured as a marked.
Cleavage
Sites
in Nuclear
DNA
539
Figure
3. Two-Dimensional
Native (nondenatured) electrophoresed from trophoresed from top digest is shown at the at the bottom left was the first dimension.
Figure
2. Acrylamide
Gel Electrophoresis
Nuclei were digested with DNAase I, and the DNA was purified by phenol-chloroform-isoamyl alcohol extraction and concentrated by dialysis against polyethylene glycol (four left channels) or ethanol precipitation (four right channels). From left to right are dialysis of 10, 17, 28, and 33% acid-soluble digestion products, and ethanol precipitation of 33, 28, 17, and 10% acid-soluble products. The same gel stained with ethidium bromide reveals nothing smaller than 40 nucleotides except for very faint staining in this size range in the dialysis-concentrated 28 and 33% acidsoluble fractions.
tured) state in the first dimension, and is then denatured and run in the second dimension. The native fragments of approximate length 10.n are seen to be composed of the single strands of size 10.n and smaller multiples of 10 nucleotides. For example, the lower and upper halves of the 50 nucleotide pair doublet are both composed of 40 and 50 nucleotide fragments. It may be shown that the 50 nucleotide pair “native” doublet contains molecules with single-strand regions; similar arguments apply to the larger undenatured fragments. For example, if it were assumed that one band of the doublet contained fully paired duplexes composed of two 50 nucleotide strands, it would follow that the other band of the doublet must contain 50 nucleotide fragments not fully base-paired (Figure 3). It should be noted that fragments 40 or fewer nucleotides in length are largely single-stranded prior to denaturation and migrate slightly faster than the corresponding duplexes (that is, 40 base pairs) in these gels. Under
Gel Electrophoresis
of DNA
DNAase I digest DNA (53% acid-soluble) is right to left. It is then denatured and electo bottom. A marker gel of native DNAase I top. DNA visualized just above the diagonal single-stranded when electrophoresed in
our conditions, no fragments smaller than 30 nucleotides long are observed in duplexes. Single-strand regions in “native” DNAase I digest DNA were detected using single-strand-specific nucleases. When unfractionated DNAase I digest DNA is treated either with Sl nuclease or with staphylococcal nuclease under conditions where only single-strand DNA is hydrolyzed, digestion is observed. With either nuclease, prominent bands on nondenaturing gels undergo a decrease in size consistent with removal of about IO bases, and upon denaturation, this digestion product yields fragments that are approximate multiples of IO nucleotides (Figure 4; Sollner-Webb, 1976). These results confirm that undernatured DNAase I digest DNA has single-stranded ends, but it is not possible to determine the length of these ends precisely by this method. Further experiments were carried out with purified native DNA fractions of the DNAase I digest, isolated from bands cut from preparative acrylamide gels. These DNA fragments were purified, with better than 90% yield, by electrophoresis into hydroxylapatite, and subsequent washing and elution as described in Experimental Procedures. In this way, separate fractions were isolated which contain the lower and upper halves of the approximately 50 base pair doublet (“501” and “LOU”), the lower and upper halves of the approximately 60 base pair doublet (“601” and “LOU”), and the approximately 70 and 80 base pair doublets. The sizes of the single-strand fragments making up these purified fractions were determined by elec-
Cell 540
8070
Figure Digest
4. Single-Strand-Specific DNA
Nuclease
Treatment
of DNAase
Native DNAase I digest DNA (12% acid-soluble) was treated with staphylococcal nuclease under single-strand-specific conditions at 37°C for 30 min with the indicated amounts of nuclease (given as pg staphylococcal nuclease per 100 pg DNA). The DNA was then isolated and electrophoresed on 10% gels in both native and denatured forms. At the left, denatured DNAase I digest DNA was similarly treated and electrophoresed as a control of enzyme activity. “M” is a marker, a reference sample of DNA from chromatin digested to the limit with staphycococcal nuclease under low ionic strength condition; treatment of marker DNA with IO pg of staphylococcal nuclease per 100 pg DNA under single-strandspecific conditions does not alter its electrophoretic pattern.
trophoresis after denaturation (Figure 5) and confirm the conclusions reached above with unfractionated material (Figure 3). These individual “native” fragments were next treated with Sl nuclease to remove single-strand ‘tails.” Acrylamide gel analysis of the DNAase I digest patterns before and after Sl nuclease treatment permits a detailed description of the composition of the individual double-stranded components (Figure 5). For example, both 501 and 50~ of the “50 base pair” doublet in the undenatured gel pattern dissociate on denaturation to a mixture of 50 and 40 nucleotide long single strands in the ratio 4:l (Figure 5). After treatment of 50/ with Sl nuclease, only material about 40 base pairs long remains, and this, when denatured, yields only 40 nucleotide long single strands. The 501 component is thus composed entirely of species that overlap by about 40 base pairs. Treatment of 50~ with Sl nuclease demonstrates that it is composed of about equal amounts of species with 40 and 50 base pair overlaps. Thus neither the upper nor the lower band of the approx-
I
60 50
40
80 70
60
50
40
Denatured
Not Denatured I::= + SI) Figure
5. Sl
Nuclease
Treatment
of Isolated
Bands
Individual bands were isolated from DNAase 1 digest DNA on 10% acrylamide gels. The lower and upper halves of the doublets which migrate as approximately 50 and 60 base pairs (501, 50~1, 601, 60~) and the doublet which migrates as approximately 70 base pairs are electrophoresed in their undenatured (left) and denatured (right) forms, Tracings of parts of the gels are shown by the solid line; migration is from left to right. The native fragments were treated with Sl nuclease, and the resultant DNA was electrophoresed in both the native (left) and denatured (right) form, as shown by the dotted line.
imately 10.n base pair doublets is a homogeneous species. Using similar methods, the lengths of the double-stranded regions in the larger DNAase I digest fragments were also determined. As expected, the double-stranded and single-stranded DNA gel patterns of the Sl-treated material are similar in all cases. a [Conditions for carrying out the Sl nuclease digestion were determined by measuring the extent of reaction as a function of enzyme concentration (Figure 6). It is important to choose conditions such that the fragment pattern is relatively invariant with changing enzyme concentration (see Experimental Procedures). This is necessary to approach complete removal of single-stranded regions without digestion of the double-stranded regions at the ends of the duplex.] Results similar to those in Figures 5 and 6 were obtained when Sl nuclease was replaced by staphylococcal nuclease and the reaction was carried out under conditions (Kacian and Spiegelman,
Cleavage 541
Sites
in Nuclear
DNA
1974) in which the latter enzyme digests single strands selectively (data not shown). Measurements of the denatured products after Sl nuclease digestion suggest that the overlaps in the undenatured DNAase I fragments are not exact multiples of IO base pairs in length, but might be slightly smaller or somewhat heterogeneous (Figure 5). It is not known, however, whether Sl nuclease is complete and absolutely specific in its attack on single-stranded ends of this kind, so that precise determination of the length of these ends is impossible using Sl nuclease alone. We therefore made direct measurements of the size of one class of single-stranded regions (those with a recessed 3’-OH terminus) using a DNA-dependent DNA polymerase: E. coli DNA polymerase II adds nucleotides to a 3’-OH primer to fill in single-stranded regions of a partially duplex DNA template (Wickner et al., 1972a, 1972b). As an initial control of the
bases
completeness of the reaction, Hind III and Eco RI restriction nucleases were used to cut, respectively, SV40 and ColEl DNA, and the products were filled in using DNA polymerase II and radioactive deoxynucleoside triphosphates (Figure 7). The molar ratio of the bases incorporated into Hind Ill cuts is a sensitive assay for incomplete repair of singlestrand regions as well as for 5’ to 3’ exonuclease activity, while G or C incorporation into Eco Rl ends is a sensitive assay for 3’ to 5’ exonuclease activity or other aberrant synthesis. It is seen (Figure 7) that DNA polymerase II incorporates the 4 nucleotides in the expected molar ratios consistent with faithful and complete repair of the singlestranded terminus, and with the absence of significant measurable exonuclease activity that would shorten a protruding 5’ end terminus (the polymerase template) or that would initially shorten a recessed 3’ end before chain elongation occurs. The substrates for the DNA polymerase II incorporation studies were total unfractionated DNAase I digest DNA, as well as the isolated lower and upper halves of the “50” and “60” base pair doublets and the “70” and “80” base pair fragments. After filling in the recessed 3’-OH ends of the native fragments with &“P-labeled nucleoside triphosphates, the DNA was denatured, electrophoresed on acrylamide gels, and visualized by autoradiography (Figure 8). Discrete radioactive fragments are produced which are slightly less than exact multiples of 10 nucleotides in length. We find that the 501 fragment, for example, incorporates additional nucleotides to generate single-strand chains slightly less than 50 nucleotides in length, while the 50~ yields single-strand fragments slightly less than 50 and slightly less than 60 nucleotides in length. The Pol II
Hind
III
Base
5 3
Eco RI
Figure 6. DNA centrations
Fragments
Produced
by Varying
Sl Nuclease
Con-
The lower half of a 50 base pair doublet was treated with increasing amounts of Sl nuclease, and the DNA was denatured and electrophoresed. Flanked by marker total DNAase I digest DNA, from left to right are 0,0.05,0.12, and 0.4 units Sl per pg DNA.
5 3
Ratios
A
G
C
T
3 T-T-C-G-A
1.0
0.86
0.84
0.87
7 C-T-T-A-A
1.0
0.04
0.03
0.84
Figure 7. Polymerase Control Experiments SV40 DNA cleaved with Hind 111 and ColEI DNA cleaved with Eco RI were treated with E. coli DNA polymerase II in three separate reactions in the presence of 3H-dCTP, one 32P-labeled and two unlabeled deoxynucleoside triphosphates. The *H-dCTP incorporation was the same (2 10%) in the three reactions of any one run; incorporation into Hind Ill cut DNA was normalized to the average 3H incorporation. 0.3 units DNA polymerase II were used to fill in 0.5 ng single-strand DNA ends; these conditions are saturating with respect to polymerase. With IO-fold lower polymerase concentration, about 3 fold lower total incorporation was observed, but with the same molar base ratios. Results are normalized relative to adenylate incorporation taken as 1.
Cell 542
80
70
60
50
40
100 90 80 70 Figure Digest
8. DNA Bands
Polymerase
II Reaction
with
Isolated
DNAase
I
The isolated lower and upper halves of the “50” and “60” base pair doublets (501, 5Ou, 601, 60~); the “70” and “80” base pair doublets, and total unfractionated DNAase I digest DNA (T; 23% acid-soluble) were treated with DNA polymerase II (0.5 units per 50 ng DNA) in the presence of 32P-deoxynucleotide triphosphates. The DNA was denatured and electrophoresed on 30 cm 10% acrylamide gels with unlabeled carrier total DNAase I digest DNA. The gel was stained for DNA (left), and the approximate band positions were marked with radioactive ink. An autoradiogram was then made of the gel (right). The center well shows a DNAase I digest of nuclei containing j4C-labeled DNA.
same gel has been stained
for carrier
DNA (Figure
8). The method of calibration is shown in Figure 9. The sizes of the filled in fragments were determined to be between 7.5 and 8 nucleotides larger than the (1O.n) nucleotide long DNAase digest DNA. Thus DNA polymerase II adds about 8 nucleotides to the recessed 3’-OH ends. We conclude that DNAase I makes cuts with a stagger of 8 nucleotides. In interpreting the DNA polymerase II results, it is important to consider the efficiency of the repair process. In control experiments, using as substrates EC0 RI-treated ColEl DNA or Hind illtreated SV40 DNA, total incorporation varied between 40 and 100% of that expected for complete repair of all substrate molecules. In all cases, however, the stoichiometry of incorporated nucleotides corresponded to the known composition of the single-stranded termini. This is evidence, in addition to that shown in Figure 10 (see below), that when DNA polymerase II initiates the repair process on a given substrate molecule, it carries it to completion. The variability in total yield of repaired molecules may reflect the very low template concentration used.
z I m
60 50 40
30 Migration Figure
9. Sizing
of the
Radioactive
Fragments
Densitometer scans of the autoradiogram DNA stain (dotted line) of the same well location of the adjoining radioactive ink semilog plot of migration against size was well using the 10.n base long DNA markers. the radioactive fragments, their sizes were total DNAase I digest as well as for the shown.
(solid line) and total were aligned by the dots (vertical lines). A calibrated for each gel From the migration of determined. Data for a 501 and 601 bands are
When the purified native 501 fragment from DNAase I digests was used as substrate, total incorporation also varied between 40 and 90% of that expected if all 40 nucleotide long chains gained 8 nucleotides in length. Although 501 contains both 40 and 50 nucleotide long single-strand components (Figure 5), the absence of labeled strands larger than 50 nucleotides (Figure 8) indicates that DNA polymerase II was able to add only to the strands 40 nucleotides long. The elongation of the 40 nucleotide long strand from 501 to a length of 48 nucleotides is thus occurring with an efficiency comparable to that of the controls discussed above. It should be noted that virtually no chains shorter than 40 nucleotides are contained in 501 (Figures 5 and 6); there are, for example, not nearly enough small chains so that all radioactive 48 nucleotide fragments could result from addition of 18 nucleotides to 30 nucleotide chains. In contrast, we have not excluded the possibility that elongation by
Cleavage
Sites
in Nuclear
DNA
543
\ Figure
IO. DNA
Polymerase
I and
II Reactions
with
\\
the 501 Band
The lower half of the “50” base pair doublet was treated with polymerase, and the data are presented as in Figure 9. The dotted line is a marker, DNAase I digest DNA (A) The isolated 501 band was treated with E. coli DNA polymerase I at 0.04 units per ng DNA (upper trace) and 0.01 units per ng DNA (lower trace). Incorporation under the latter conditions was 40% of the former. (B) The isolated 501 band was treated with E. coli DNA polymerase II at 0.004 units per ng DNA (upper trace) and 0.001 units per ng DNA (lower trace). Incorporation under the latter condition was 30% of the former.
more than 8 nucleotides occurs in some of the larger native DNAase I digest fragments. Partial contributions to the radioactively labeled bands from species elongated by 10.n + 8 nucleotides (n 3 1) could occur either by filling in of species containing single-stranded regions 10.n + 8 nucleotides in length, or by initiation of the polymerase at internal gaps, followed by strand displacement during polymerization. The behavior of the 501 component provides evidence of chain elongation, but it is indirect, since no labeled chains longer than some of the starting material are produced. One could argue that no net elongation occurs, and that the labeled material is produced by a combined exonuclease and polymerase activity that serves only to shorten and label terminally the 50 nucleotide long strands of 501. This is improbable, since all components of the 501 band have duplex regions of about 40 nucleotides (Figure 5). Thus molecules containing two 50 nucleotide strands would not result in a radioactive 48 nucleotide fragment by this labeling mechanism.
If the amount of incorporation observed above had arisen from duplexes containing one 50 and one 40 nucleotide strand, up to 7 terminal nucleotides would have had to be removed and repolymerized on each of the 50 nucleotide strands (see above). However, the DNA polymerase II control experiments (Figure 7) have ruled out such extensive exonuclease activity within duplex regions. Direct evidence against this mechanism, and in favor of net chain elongation, comes from treatment of 50~ with DNA polymerase II (Figure 8). The 50~ band, like the 501 band, contains only chains that are 40 and 50 nucleotides long. In the case of 50~1, however, structures are present which give rise to radioactively labeled 58 nucleotide long fragments, which are longer than any component strands of the 50~ band. Experiments similar to those using DNA polymerase II (Figure 8) were carried out with E. coli DNA polymerase I at low temperature (5”C), where repair synthesis is favored (Wu, 1970). When unfractionated DNAase I digest DNA and the isolated 501 band are used as substrates, the bulk of the incorporated nucleotides are again contained in fragments 10.n + 8 and 48 nucleotides long, respectively (Figure 10). (Unlike DNA polymerase II, DNA polymerase I acts on the 501 component to produce also a few radioactive chains much larger than any of the starting components, and longer than could result from simple filling in of single-strand tails in “native” 501 DNA. We calculate that
Cell 544
cessed 3’-OH terminus in the complementary strand. What fragment structures can be deduced from this information? DNAase I is known to cut DNA in nuclei at intervais of 10.n nucleotides. Potential cutting sites occur every 10 or 20 nucleotides over most of the DNA (Figure 2). The presence, in the double-stranded DNAase I digest product, of structures with 3’-OH ends recessed by 8 nucleotides thus implies the presence of corresponding structures with 5’-P ends recessed by 2 or 12 nucleotides (Figure 11). While this class of single-strand ends cannot be detected using the polymerase, Sl nuclease should remove them. Using the experimental data and assuming that 5-P ends may be recessed by 2 or 12 nucleotides, it is possible to construct a model of the components, for example, of the 501 band, as illustrated in Figure 12. The 501 band contains both 50 and 40 nucleotide long single strands, in the ratio 4:1 (Figure 5). The amount of DNA polymerase II-induced incorporation is consistent with the presence of most or all of the 40 nucleotide chains in structures recessed by 8 nucleotides at the 3’ end. Digestion with Sl nuclease reveals that all fragments in the 501 fraction have duplex regions about 40 base pairs long, so that the 40 nucleotide chains must all be paired with the 50 nucleotide chains. The remaining 50 nucleotide chains must be paired with each other in conformations inaccessible to DNA polymerase II, since that enzyme acting on the 501 fraction produces no radioactive fragments longer than 50 nucleotides. Thus all 50 nucleotide long strands in 501 must have 5’ recessed ends. As we have pointed out above, such termini should be
recessed by 10.n + 2 nucleotides (Figure 11). In this particular case, the 5’ ends must all be recessed by 12 nucleotides (n = 1), since this is the only consistent way in which a structure containing two 50 nucleotide chains can yield an approximately 40 base pair duplex upon Sl digestion. (It should be noted that the various structures depicted in Figure 12 are not present in equimolar amounts and probably vary in abundance with the extent of digestion.) Using similar arguments, structures may be deduced for the components of the other native DNA gel bands. Those for band 50~ are shown in Figure 12. The 50~ band might reasonably be expected to be contaminated with the slow moving component of 501, which is the duplex of a 40 and 50 nucleotide chain (see Figure 3). However, DNA polymerase II treatment of 501 generates only about a fourth as many labeled 48 nucleotide chains as does treatment of 5Ou, indicating that most of the 40 nucleotide chains present in 501 are unavailable to the polymerase, as shown in Figure 12. What relationship do the structures shown in Figure 12 have to the models of DNA arrangement in nucleosomes discussed in the Introduction? Noll (197413) has suggested that the nucleosomal DNA, lying on a histone core (indicated by the dotted region in Figure 13), is accessible to nuclease only at the phosphodiester bond most exposed to the surroundings. Cutting at these peaks (marked “p” in Figure 13) would create a stagger of 6 and 4 nucleotides oriented in such a way that polymerase would be expected to add 6 nucleotides to 3’ termini. Alternatively, Crick and Klug (1975) have pro50
4-2-t-8-+2*
\F 1 a 5’JllIIIIIIIIIIIIII t
1
3’
40
J15
Proposed
Structures
2
48
2
1
3
t
-
a
42
-
12
38
a
,5ou
2 J
& Figure
11.
DNAase
I Cleavage
I II II
I ’ II
122
4038
501
a12 -1
Sites
DNA is shown (a) with potential cutting sites (arrow) at 10 nucleotide intervals on each strand, staggered so as to create fragments with 3’-OH ends recessed by 8 nucleotides. These ends may be filled in by DNA polymerase, as indicated by (A) (b). Since cuts are often made 10 nucleotides away from another cut, fragments with 5’-P ends recessed by 2 (c) and 12 (d) nucleotides are also inferred.
Figure
12.
Proposed
Structures
for the “50
Base
Pair”
Bands
A summary of experiments with the 501 band is shown on the left. Combining this information with the stagger determined in Figure 11, DNA structures comprising the 501 band are determined as described in the text. On the right are the proposed structures for the 501 and 50~ bands. The zig-zag line represents the nucleotides which may be added by DNA polymerase.
Cleavage 545
Sites
in Nuclear
DNA
minor groove
Y-10 I 1
i;-r--+ /
I f--rpeak
r2~-+----8------2~
I 3’
5’
3’
T &m;&
T $yg.
I’
I
1
I
I
5’ t Figure
5’
13.
Models
P’ for DNAase
A portion of the nucleosome above, and the corresponding
t I Cutting
Action
(See
3’
b Text)
surface is schematically DNA is stretched out below.
shown
posed that the DNA of the nucleosomes is kinked by bending away from the major groove every 10 or 20 base pairs, and that cleavage occurs at these kinks. This would create either ends with no singlestrand tails, or with tails 10.n nucleotides long, which polymerase might fill in. Sobell et al. (1976) recently suggested another kind of kink for nucleosomal DNA, which bends away from the minor groove. The sites of this kink are labeled “minor groove” (m.g.) in Figure 13; the predicted fragment pattern for cleavage at this site is, of course, the same as for the Crick and Klug model. The observed stagger of 8 and 2 nucleotides for the DNAase I cleavage sites is not consistent with the predictions made from any of the models discussed above. All the models can be made consistent with the results, however, if we suppose that the enzyme binds at the accessible site, but cuts one nucleotide away in a specified direction. For example, if binding actually occurs at the peaks (sites marked “p” in Figure 13), but the cuts are made 1 nucleotide to the 5’ side of this site, staggered cuts at 8 and 2 nucleotide intervals (marked by arrows by Figure 13) would be created. On the other hand, if we suppose the enzyme recognition sites are at 10 base pair intervals, as for instance at the points marked “m.g.,” we must postulate that the cuts are made 1 nucleotide to the 3’ side of these sites to account for the observed cleavage pattern. [If the DNA of the nucleosome is lying on, and shielded by, a histone core, the Sobell kink would appear to be more accessible to a nuclease than the Crick-Klug kink (displaced 5 base pairs along the helix from the Sobell kink). Examination of the published kink models shows that the poten-
tial DNAase I cleavage sites are raised to the surface in the Sobell kink, but are virtually buried in the Crick-Klug kink. However, experiments of the kind described here are unable to distinguish between these two types of kinks or any other structure which repeats every 10 base pairs along the DNA backbone. Furthermore, it is not known that DNAase I would favor cutting near a kink over cutting in a linear region of B form DNA.] Our results provide new information about the location of DNAase I cleavage sites within the nucleosome; the cleavage pattern is not predicted by either of the two previously proposed classes of models. As we have shown, either class of models can be used to explain the result with the help of suitable assumptions about the mode of action of pancreatic DNAase. Further understanding of this mode of action would help us to decide whether either of the models is correct, or whether the data must be explained in terms of some other model of nucleosome structure. Experimental
Procedures
Isolation of Nuclei and Chromatin Nuclei were isolated from frozen duck erythrocytes by several successive washings in IO mM Tris-HCI (pH 8), 1 mM MgCI,, 0.25 M sucrose as described previously (Sollner-Webb and Felsenfeld, 1975). The addition of 0.5% Triton X-100 in an early wash had no effect on the digestion results. Chromatin was prepared from salt-washed nuclei as described previously (Camerini-Otero et al., 1976). Enzymic Treatments of Nuclei, Chromatin, and DNA DNAase I (pancreatic DNAase; Worthington, 3750 U/mg) was used to digest nuclei (0.5-3 mg DNA per ml) in 10 mM Tris-HCI (pH 8), 1 mM MgCI,, 0.25 M sucrose at 37°C. Enzyme concentration and time of reaction were adjusted to give the desired amount of acid solubilization, as described previously (Sollner-Webb et al., 1976). Reactions were terminated by addition of 5 mM Na EDTA. DNA was purified by proteinase K (E. Merck) digestion of the reaction mixture, followed by chloroform:isoamyl alcohol:phenol (24:1:24) extraction and ethanol precipitation as previously described (Sollner-Webb and Felsenfeld, 1975). Alternatively, EDTA-terminated reactions were made 0.1% in SDS and extracted with chloroformisoamyl alcohol-phenol, followed by ethanol precipitation as above, with no detectable differences in results. To retain small DNA fragments in higher yield, the ethanol precipitation step of the DNA isolation was replaced by dialysis into 1 mM Tris-HCI (pH 8), 0.1 mM EDTA, concentration against polyethylene glycol 6000 powder (Baker), and another dialysis against the same buffer. Sl nuclease digestions of DNAase I digest DNA (0.04 mg DNA per ml) were carried out in 50 mM Na acetate (pH 4.75), 150 mM NaCI, 0.5 mM ZnSOn at 37°C for 1 hr. Sl nuclease was purified by Drs. Michael Zasloff and Patrick Williamson by the procedure of Vogt (1973) through the sulfopropyl cellulose fractionation; the unit of activity is defined by Vogt. Reactions were performed using about 0.2 units Sl per wg DNAase I digest DNA, but each DNA sample was titrated with the nuclease, using acrylamide gel electrophoresis of the product DNA as an assay for completion of reaction (see Figure 5, where 0.12 units Si per pg DNA were chosen.) Too extensive digestion permits attack on the ends of double-stranded DNA and a blurring of the gel patterns. [To avoid this overdigestion, nuclease to substrate ratios used were about lo4 lower than Shenk et al. (1975) used to digest at sites opposite
Cell 546
nicks.] Reactions were terminated with EDTA, and the DNA was purified as described above, in this and in the digestions described below. Staphylococcal nuclease (Worthington; 30,000 U/mg) was used to digest DNAase I digest DNA (0.2 mg DNA per ml) under single-strand-specific conditions as described by Kacian and Spiegelman (1974), in 0.4 M NaCl, 10 mM Tris-HCI (pH 8), 10 mM MgCI,, 0.1 mM CaCI, at 37°C; each DNA sample was titrated with nuclease as described in Figure 3. Staphylococcal nuclease was used at low ionic strength [I mM Tris-HCI (pH 8), 0.1 mM CaCI,] to digest isolated chromatin to the limit as previously described (Axe1 et al., 1974). The sizes of the native DNA fragments produced were determined previously (Camerini-Otero et al., 1976). Eco RI restriction endonuclease was used to digest ColEl DNA (0.15 mg/ml) in 100 mM Tris-HCI (pH 7.5), 50 mM NaCI, 10 mM MgCI, at 37°C for 15 min. The Eco RI was purified by Dr. James McGhee using a modification of the procedures of Greene et al. (1974) and Thomas and Davis (1975). ColEl DNA was made by Dr. R. D. Camerini-Otero by a modification of the method of Clewell and Helinski (1969). The reaction product was monitored on agarose gels to demonstrate complete reaction, as described below. Hind Ill restriction endonuclease was used to digest SV40 DNA (0.075 mg/ml) in 6.6 mM Tris-HCI (pH 7.5), 50 mM NaCI, 8.7 mM MgCI, at 37°C for 6 hr. The Hind III was purified by Dr. Mark Israel by a modification of the procedure of Smith and Wilcox (1970) through the phosphocellulose column step. Column fractions having only Hind Ill activity, with no detectable contaminating Hind II activity, were selected. The SV40 DNA was made by Dr. Maria Persico-DiLauro by the method of Sebring et al. (1974). Reaction products were monitored on agarose gels to demonstrate complete reaction. E. coli DNA polymerase II was purified by Dr. Sue Wickner by a modification of the procedure of Wickner et al. (1972a), where the unit of activity isdefined. E. coli DNA polymerase II reactions were performed according to Wickner et al. (1972a, 1972b), in 50 mM Tris-HCI (pH 7.4), 5 mM MgCI,, 1 mM dithiothreitol, 30 mM NaCI, 6 PM (approximately 5 +i/nmol), 3ZP-dTTP, 40 PM each of dATP, dGTP, and dCTP, 0.1 mg/ml bovine plasma albumin, with l-10 units polymerase per /*g of DNAase I digest DNA at 37°C for 30 min. Reactions were performed in polymerase excess unless otherwise noted. Incorporations were terminated with 5 fold excess EDTA at 0°C. Carrier DNAase I digest DNA was added, and the DNA was precipitated with 10% TCA at 0°C for 5 min, followed by centrifugation at 12,000 x g for 5 min at 0°C. Immediately the DNA was approximately neutralized with NaOH, made 0.1 M Tris-HCI (pH 8), 0.5 mM EDTA, 0.1% SDS, extracted with phenolchloroform-isoamyl alcohol, and ethanol-precipitated. No detectable depurination (measured by change in single-stranded DNA gel pattern) occurs during this procedure. To measure ratios of incorporated nucleotides, one 3H-labeled (dCTP) as well as one 32P-labeled (dTTP, dATP, or dGTP), nucleoside triphosphates were used at 10 PM concentrations each; the two unlabeled nucleoside triphosphates were at 40 PM each. Three separate reactions, one with each 32P-nucleoside triphosphate in addition to 3H-dCTP, were performed in the ionic conditions described above. E. coli DNA polymerase I reactions were performed according to Wu (1970) in 70 mM KPO, (pH 6.9), 10 mM MgSO,, 15 mM dithiothreitol, 60 mM NaCI, 10 pm (5 &i/nmol)3ZP-dTTP, 20 PM dATP, dGTP, and dCTP, 0.1 mglml bovine plasma albumin at lo50 U polymerase per pg of DNAase I digest DNA at 5°C for 4 hr. This reaction mixture has a conductivity equivalent to 190 mM NaCl. The E. coli DNA polymerase I was purified by Dr. Sue Wickner by a modification of the procedure of Jovin, Englund, and Bertsch (1969), where the unit of activity is defined. Gel Electrophoresis of DNA Fragments Electrophoresis of DNA fragments on 10% gels was carried out as described previously all cases, the acrylamide:methylene-bisacrylamide
polyacrylamide slab (Axe1 et al., 1974). In ratio was 19:l.
Allowing polymerization to proceed for several hours before use further improves the electrophoretic resolution: allowing 12-16 hr for this process is essential for good resolution of single-stranded DNA fragments. Single-stranded DNA electrophoresed under these conditions was resolved as well as it is on formamide gels (Nell, 1974b). The electrophoretic buffer was 3 mM Na EDTA, 90 mM boric acid, and 90 mM Tris (pH 8.0) (Peacock and Dingman, 1967). The sample of DNA was usually applied dissolved in the running buffer diluted 1:9 with water. Single-stranded samples were prepared by briefly exposing the DNA to 0.1 N NaOH and then neutralizing the sample with HCI before application of the gel. Gels were stained with “Stains-all” (Eastman), destained in water, and photographed (Axe1 et al., 1974). Negatives were scanned in a Joyce-Loebl microdensitometer. For the polymerase experiments, where very high resolution is necessary, 30 cm long (1 mm thick) slab gels were used. DNA was visualized by staining in ethidium bromide (30 min. 0.5 pg/ml in H,O) and photographed under an ultraviolet light. DNA bands were marked, and gels were then “fixed” with 20% TCA (30 min) and dried using a BioRad gel drier without the cellophage membrane. Kodak medical X-ray film was used for autoradiography. Two-dimensional electrophoresis was performed using 10% disc gels for the first dimension. Gels were stained for 30 min in ethidium bromide (0.5 fig/ml), and the ionic strength was further reduced by mixing an additional 30 min in 1 mM Tris-HCI (pH 8). After boiling the gels for 5 min, they were laid on a pre-electrophoresed 5 mm thick 10% gel and electrophoresed. Disc gel dimensions were chosen to fit securely into the well of the slab gel. Electrophoresis of restriction enzyme-cut DNA fragments was carried out in 1.6% agarose disc gels, as described above for the acrylamide system, but without aging of the gels. DNA was visualized with ethidium bromide. DNA Extractions from Acrylamide Gels To isolate DNA, singleor double-stranded, from acrylamide gels, a method was developed in which >90% of the DNA is recovered. The DNA gel is stained in ethidium bromide (or only a marker well is stained), and the desired band is cut out. In a disposable pipette, a 1 ml column of washed hydroxylapatite (BioRad) is formed between layers of glass wool. The column is filled with 0.1 M Tris-HCI (pH 8), and the DNA gel slice inserted and crushed against the walls with a glass rod. Glass wool is placed above the gel to hold it in place and electrophoresis is begun. Since liquid flows slowly through this column, either sufficient upper buffer volume or buffer recirculation is necessary. Electrophoresis is carried out at 200 V for 4 hr in 0.1 M Tris-HCI (pH 8); bromphenol blue marker may be used to demonstrate electrical continuity through each column. The acrylamide gel fragments are then removed from the pipette, and the hydroxylapatite column is washed with a few volumes of 10 mM sodium phosphate (pH 6.7) until no further absorbing material elutes. 0.5 M sodium phosphate is used to elute the DNA in a concentrated fraction with an ultraviolet absorption spectrum identical to that of pure DNA. Since this procedure was developed, a similar method for recovering proteins from acrylamide gels has been described (Ziola and Scraba, 1976). Acknowledgments We would like to thank Dr. Sue Wickner for her generous gift of E. coli DNA polymerase II and E. coli DNA polymerase I. We are also grateful to Drs. Michael Zasloff and Patrick Williamson for Sl nuclease, Dr. James McGhee for Eco RI endonuclease, Dr. Mark Israel for Hind III endonuclease, Dr. Dan Camerini-Otero for ColEl DNA, and Dr. Maria Persico-DiLauro for SV40 DNA and j4Cthymidine-labeled cells. Finally, we especially thank Drs. James McGhee and Patrick Williamson for many helpful discussions and suggestions. Received
October
26, 1976
Cleavage 547
Sites
in Nuclear
DNA
References Axel, Ft. (1975). staphylococcal
Cleavage nuclease.
of DNA in nuclei and chromatin Biochemistry 74, 2921-2925.
Vogt, strand them.
with
Axel, Ft., Melchior, W., Sollner-Webb, B., and Felsenfeld, G. (1974). Specific sites of interaction between histones and DNA in chromatin. Proc. Nat. Acad. Sci. USA 77, 4101-4105. Camerini-Otero, R. D., (1976). The organization dence for an arginine-rich
Sollner-Webb, B., and Felsenfeld, of histones and DNA in chromatin: histone kernel. Cell 8, 333-347.
Clewell, D., and Helinski, D. (1969). protein complex in E. coli: purification anopencircularform. Proc.Nat.Acad.
Supercoiled circular and induced conversion Sci. USA62,1159-1166.
Crick,
helix.
F. H., and
Klug,
A. (1975).
Kinky
G. eviDNAto
Nature255,530-533.
Greene, P., Betlach, M., Boyer, H., and Goodman, H. (1974). The EcoRl restriction endonuclease. In Methods in Molecular Biology, 7, DNA Replication and Biosynthesis. Ft. Wickner, ed., (New York: Marcel Dekker), pp. 87-111. Hewish, D., and Burgoyne, L. (1973). Chromatin substructure: the digestion of chromatin DNA at regularly spaced sites by a nuclear deoxyribonuclease. Biochem. Biophys. Res. Commun. 52, 504510. Jovin, T., Englund, P., and Bertsch, L. (1969). sis of DNA: XXVI Physical and chemical studies DNA polymerase. J. Biol. Chem. 244, 2996-3008. Kacian, D. L., and nuclease to monitor them. 58, 534-540. Nell, 251
M. (1974a).
Spiegelman, hybridization
Subunit
structure
Nell, M. (1974b). Internal structure Acids Res. 1, 1573-l 578. Peacock, ribonucleic Biochemistry
S. (1974). reactions
Enzymatic syntheof a homogeneous Use of micrococcal with DNA. Anal. Bio-
of chromatin. of the chromatin
A., and Dingman, C. (1967). acid species by polyacrylamide 6, 1818-1827.
Nature257,249subunit.
Nucl.
Resolution of multiple gel electrophoresis.
Sebring, E., Kelly, T., Thoren, M., and Salzman, ture of replicating simian virus 40 deoxyribonucleic cules. J. Viral. 8, 478-490. Shaw, B., Herman, T., Kovacic, R., Beaudreau, Holde, K. (1976). Analysis of subunit organization erythrocyte chromatin. Proc. Nat. Acad. Sci. USA
N. (1971). acid
Strucmole-
G.,
and Van in chicken 73,505-509.
Shenk, T., Rhodes, C., Rigby, P., and Berg, P. (1975). Biochemical method for mapping mutational alterations in DNA with Sl nuclease. Proc. Nat. Acad. Sci. USA 72, 989-993. Simpson, chromatin base pair Smith, mophilus
R., and Whitlock, J. (1976). Chemical evidence that DNA exists as 160 base pair beads interspersed with 40 bridges. Nucl. Acids Res. 3, 117-127.
H., and Wilcox, K. (1970). influenzae. J. Mol. Biol.
Sobell, H., Tsai, The organization 73, 3068-3072. Sollner-Webb, chromatin. nia. Sollner-Webb, digestion Biochemistry
A restriction 57, 379-391.
C., Gilbert, S., Jain, of DNA in chromatin.
B. (1976). Ph.D. Thesis,
Organization Stanford
B., and Felsenfeld, of nuclei and chromatin 74, 2915-2920.
enzyme
from
He-
S., and Sakore, T. (1976). Proc. Nat. Acad. Sci. USA of histones in nuclei and University, Stanford, Califor-
G. (1975). A comparison of the by staphylococcal nuclease.
Sollner-Webb, B., Camerini-Otero, R. D., and Felsenfeld, G. (1976). Chromatin structure as probed by nucleases and proteases: evidence for the central role of histones H3 and H4. Cell 9, 179-I 93. Thomas, M., and Davis. R. (1975). Studies on bacteriophage lambda DNA with EcoRl restriction J. Mol. Biol. 97, 315-328.
the cleavage endonuclease.
of
V. M. (1973). Purification specific nuclease from 33, 192-200.
Wickner, Ft., Ginsberg, Deoxyribonucleic acid Chem. 247, 489-497. Wickner, R., Ginsberg, cleic acid polymerase 498-504. Wu, R. (1970). 51, 501-521.
Nucleotide
and further Aspergillus
properties oryzae.
Eur.
of singleJ. Bio-
B., Berkower, I., and Hurwitz, J. (1972a). polymerase II of Escherichia co/i, I. J. Biol. B., and Hurwitz, J. (1972b). DeoxyribonuII of Escherichia co/i, II. J. Biol. Chem. 247, sequence
analysis
Ziola, B., and Scraba, D. (1976). Recovery polyacrylamide gels by electrophoresis into Biochem. 72, 366-371.
of DNA.
J. Mol.
of SDS-proteins hydroxylapatite.
Biol. from Anal.