Two-Dimensional Electrophoretic Analysis of Polynucleosomes*

Electrophoretic techniques have been developed to map simultaneously the total distributions of DNA sizes, protein compositions, and nuclease mediated precursor-product relationships present in a heterogeneous population of monoand polynucleosomes. Chromatin, prepared from nuclei after controlled micrococcal nuclcasc dig&ion, is separated by gel electrophoresis at low ionic strength. The DNA or protein components, or both, are displayed next by electrophoresis in a second dimension, using sodium dodecyl sulfate-polyacrylamide slab gels. Monoand polynucleosomes migrate as molecularly dispersed protein:DNA complexes in the first dimension, while protein and DNA species migrate independently in the second dimension. In order to trace precursor-product relationships, electrophoretically separated monoand polynucleosomes are digested in situ prior to performing electrophoresis to display UN4 in the second dimension. This is accomplished by incorporating reversibly inactivated micrococcal nucleate or DNase I within the first dimension gel matrix and running buffer. The enzymes are reactivated after electrophoresis by exposing gels to the appropriate divalent cations. Such a “chromatin fingerprinting” technique can discriminate between various levels of nucleosome organization with regard to differences in core and spacer DNA lengths. This general strategy could be applied to a variety of other problems, including restriction endonuclcase site mapping, -4pplication of the above procedures to a systematic study of bovine thymus chromatin conclusively demonstrates a spectrum of nuclease processing events, many of which were only supposed to occur previously. Most of the bovine genome is shown to consist of 160 base pair cores which are nuclease products of an average repeat of 191 + 8 base pairs. These cores are processed further to 140 base pair and smaller DNA fragments. A minor portion of chromatin consists of a larger repeat of 203 f 9 base pairs which gives rise to two unique classes of mononucleosomes with larger ljN.4 lengths. Other minor classes of repeats exist, with lengths of

and nuclease mediated precursor-product relationships present in a heterogeneous population of monoand polynucleosomes.
Chromatin, prepared from nuclei after controlled micrococcal nuclcasc dig&ion, is separated by gel electrophoresis at low ionic strength. The DNA or protein components, or both, are displayed next by electrophoresis in a second dimension, using sodium dodecyl sulfate-polyacrylamide slab gels. Mono-and polynucleosomes migrate as molecularly dispersed protein:DNA complexes in the first dimension, while protein and DNA species migrate independently in the second dimension.
In order to trace precursor-product relationships, electrophoretically separated mono-and polynucleosomes are digested in situ prior to performing electrophoresis to display UN4 in the second dimension. This is accomplished by incorporating reversibly inactivated micrococcal nucleate or DNase I within the first dimension gel matrix and running buffer. The enzymes are reactivated after electrophoresis by exposing gels to the appropriate divalent cations. Such a "chromatin fingerprinting" technique can discriminate between various levels of nucleosome organization with regard to differences in core and spacer DNA lengths. This general strategy could be applied to a variety of other problems, including restriction endonuclcase site mapping, -4pplication of the above procedures to a systematic study of bovine thymus chromatin conclusively demonstrates a spectrum of nuclease processing events, many of which were only supposed to occur previously. Most of the bovine genome is shown to consist of 160 base pair cores which are nuclease products of an average repeat of 191 + 8 base pairs. These cores are processed further to 140 base pair and smaller DNA fragments. A minor portion of chromatin consists of a larger repeat of 203 f 9 base pairs which gives rise to two unique classes of mononucleosomes with larger ljN.4 lengths. Other minor classes of repeats exist, with lengths of *  I To whom requests for reprints should be addressed.
170 r: 8 and 142 + 3 base pairs, In all, five monomer nucleoprotein particles have been identified, and certain aspects of their precursor-product relationships and protein compositions have been established.
Due to pioneering biochemical and electron microscopic studies, it is now clear that a large proportion of the chromatin of eukaryotic cells consists of a flexible chain of subunits, termed Y bodies, or nucleosomes (3-9). By virtue of the fact that internucleosomal DNA is preferentially hydrolyzed by micrococcal nuclease, nucleosome oligomers and monomers can be released from chromatin after controlled endonuclease treatment. Resulting products have been isolated by sucrose gradient centrifugation or column chromatography and are found to consist of multiples of about 200 base pairs of DNA complexed with chramatin proteins, in particular the histones (6, l&15). The diameter of nucleosome monomers is close to 16). DNA is believed to be wrapped around the outside of a histone cluster (17)(18)(19), yielding a packing ratio of about 7:l (20) and a constraint equivalent to an average of greater than one negative superhelical turn (21,22). At present, there is no evidence that nucleosomes are arranged specifically with regard to DNA base sequence (12,(23)(24)(25)(26). It is likely that each nucleosome consists of 8 histone molecules, 2 each of histones HZA, H23, H3, and H4 (17,27, 28). It has been suggested that the octamer may be organized into two heterotypic tetramers about a dyad axis of symmetry (28a). Histone Hl is believed to lie on internucleosomal DNA and/or along the nucleosomal surface (15,18,29,30).
Several groups have demonstrated that the nucleosome repeat is composed of a nuclease resistant core of about 160 to 130 base pairs, and a nuclease-sensitive spacer (10,11,15,30,31). Recent evidence indicates that variation exists in DNA repeat lengths among different eukaryotes and cell types (32-37). This has been attributed to differences in spacer DNA lengths external to an evolutionally conserved 140 base pair core (32-35, 3'7). However, it is conceivable that more than one type of core may exist within the chromatin of a single cell nucleus. During the course of nuclease digestion several different core DNA lengths are observed simultaneously; no single DNA fragment stoichiometrically accumulates and 11 discrete DNA products are found at the limit of digestion 110-12, 15). It is possible, therefore, that certain of these components may arise from different classes of polynucleosomes.
Indeed , subunits  of PoEynucleosomes   within  transcribed  chromatin  have an altered  conformation  (38, 39), certain  histone  classes contain  subfractions  with different  primary  structures  (40, 411, histones  are known  to  undergo  a variety  of post translational  modifications  (41f, and  non-histone  proteins  may contribute  to nucleosome  variability  ( 12,39,42,43 with the same acrylamide solution after insertion of spacers (0.7 x 0.3 cm1 in order to form sample wells for standards. Electrophoresis was for 12 h at 40 mA. For two-dimensional electrophoresis to display single-stranded DNA, tube gels processed as described below were laid horizontally across preformed 6% acrylamide gel slabs (20 x 17 x 0.3 cm; acrylamide:N,N'-methylenebisacrylamide, 2O:l) made in 10 M urea, 0.1% SDS, 40 rnM Tris, 14.5 rnM sodium borate, 2.5 nnw EDTA, pH 8.3, and polymerized in place using a 4% acrylamide solution made in the above buffer.
The running buffer was as above but without urea; electrophoresis was performed for 3 h at 75 mA. Gels were stained either with Stains-all as described (46) and scanned at 550 nm using a 20-cm Gilford 2520 gel scanner with a 0.05 mm-wide slit, or with 1 pg/ml of ethidium bromide in water (5 &ml for single-stranded DNA) for   appear as skewed spots, while monomer III appears as a crosshatched line. In early digests there is an overlap in the DNA sizes of monomers I, II, and III (Fig. 3A). During the course of digestion there is a gradual reduction in monomer DNA sizes. The cross-hatched line of monomer III turns to eventually become a band. Concomitantly, the DNA size of monomer III changes from a range of about 210 to 160 base pairs to approximately 160 base pairs and monomer II becomes masked (cornpare Fig. 3, B and F; Fig. 4). Nuclease processing of monomer II results in DNA fragments of 160 base pairs from material initially ranging between 185 to 160 base pairs in length (Fig.  3F). Monomer I has a DNA size range of about 175 to 140 base pairs, and becomes trimmed to approximately 140 base pairs in late digests (Figs. 3 and 4). Therefore, the core DNA lengths (15) of monomers I, II, and III are 140, 160, and 160 base pairs, respectively. It is difficult to predict the degree of homogeneity of these monomers with respect to protein composition or particle conformation, or both, from the shapes of their second dimension DNA patterns. DNA size differences for a particle with a homogeneous protein content would be expected to create opposing effects on its electrophoretic mobility; size reduction being accompanied by a reduced net charge.
In contrast to monomer DNA profiles, dimer nucleosomes and higher multimers are separated in the first dimension principally on the basis of DNA size; after electrophoresis in the second dimension a diagonal line exists, as opposed to i FIG. 2. Effect of released protein on DNA mobility in 4% acrylamide, 0.1% SDS gels. Samples: A, purified bovine thymus total histones; B, purified bovine thymus histone Hl; C, chromatin of a 7% nuclear digest; D, purified DNA of Sample C; E, chromatin of a 14% nuclear digest; and F, purified DNA of Sample E. Amido black and Stains-all were used for staining. Electrophoresis was from top to bottom. spots or bands for the DNA components of each oligomer class (Fig. 3). With regard to dimer DNA patterns, a spot appears to move through a less intense diagonal line during the course of digestion; at early periods the spot appears at the top of the diagonal, while at later periods at the bottom. This shows clearly the change in the mass distribution of different DNA lengths for dimer, which is summarized in quantitative terms in Fig. 4.
The slight streaking of oligomers depicted in Fig. 3E was of interest since it occurred only among the longest DNA lengths present within each multimeric class. Furthermore, the appearance of these larger multimers correlated well temporally with the appearance of a faint spot toward the upper left of the cross-hatched line of monomer III DNA (Fig. 3E). These observations suggested that a larger repeat may exist which was more resistant to nuclease and was a precursor to the faint spot neighboring monomer III DNA. Evidence strongly supporting this notion was obtained using different electrophoretie conditions. As shown in Fig. 6, two additional monomers can be resolved in late digests, termed IV and V, with DNA length ranges of 185 to 205 and 190 to 220 base pairs, respectively. These apparently arise from the larger oligomers depicted diagrammatically in Fig. 6B, which yield a repeat length of 203 r 9 base pairs as judged by averaging the The upper left panel shows the separation of chromatin DNA fragments from nuclear digestions of: A: 2%, B, 5%; C, 9%; D, 12%; E, 17%; and F, 21%. Electrophoresis (top to bottom) was performed using a 4% acrylamide, 0.1% SDS slab gel. Hue III restriction endonuclease fragments of PM2 DNA are included as standards. The other panels show two-dimensional separations of the same chromatin samples shown in the upper left panel, where nucleoprotein electrophoresis was the first dimension (lefi to right), and DNA electrophoresis was the second dimension (top to bottom). A one-dimensional standard of the same chromatin sample has been included at the side of each two-dimensional gel as an internal standard. Ethidium bromide was used for staining. differences between successive multiples (32). This value significantly differs from 188 2 3 base pairs, the unit size estimated for the major polynucleosomal class resolved in this experiment (Fig. 6C).
The finding that about 5% of bovine thymus chromatin consists of a different repeat which is more resistant to nuclease prompted us to explore the DNA profiles of yet more extensive digests. As shown in Fig. 7B, a high DNA load of a near limit nuclear digest reveals two additional minor repeats of 170 r 8 and 142 2 3 base pairs, estimated by averaging the differences between successive multiples of trimer through nanomer and monomer through tetramer, respectively (32). These presumably account for less than 1% of the total chromatin DNA. Therefore, the nucleosomal repeats of bovine thymus chromatin range from 203 to 142 base pairs, with the great majority of the nucleosomal DNA centering about a unit size of 191 base pairs.

Quantitation of Poly-and Mononucleosomal
DNA -Under the digestion conditions used, bovine thymus mononucleosomes accumulate with time and show little internal breakdown prior to the depletion of multimers (Fig. 3). Thus, the amount of chromatin organized in nucleosomal structures can be evaluated from the extent of digestion required to quantitatively convert the global repeat of 191 e 8 base pairs to equal masses of 160 and 140 base pair fragments. As shown in Fig.  3F, essentially complete conversion of chromatin to these fragments occurred after 21% digestion, closely agreeing with a theoretical value of 41/191 = 21.5% for the total participation of nuclear DNA in a homogeneous repeat. Although the presence of other DNA repeat lengths complicates this comparison, it appears that only a few per cent, at most, of bovine thymus chromatin can exist in structures other than polynucleosomes. This supposition has been strengthened by quantitative electron microscopy studies4 (56).
The relative proportions of the various mononucleosomes are not constant during the course of digestion. In particular, monomers IV and V appear transiently; they are not visible in early or late digests (Fig. 3) and, under optimum conditions, account for only a few per cent of the total monomer population (Fig. 6). On the other hand, monomer I accumulates during digestion. Quantitation by an analysis of areas of onedimensional gels reveals that monomer I comprises between 28 and 33% of the total monomer population upon proceeding from 2 to 14% digestion (Fig. 1). In agreement with Varshavsky et al. (291, monomer I accounts for a greater percentage in later digests, suggesting that larger monomer(s) may be precursors to this component (Fig. 3F). Estimates of the proportion of monomer II from two-dimensional gels suggest that  (Fig. 3). "Chromatin Fingerprinting" -It seems clear that monomers IV and V exist in chromatin in a tandem arrangement, independent of monomers I, II, and III. The larger monomers appear during digestion only when larger multimers are visible, ruling out a general interspersed organization (Figs. 3E and 6). The question then arises: Do monomers I, II, and III exist in chromatin as independent structures, perhaps tandemly, or do they originate from a structurally and compositionally identical monomer by asymmetric nuclease cleavages or by nuclease processing (or both)? To approach these issues, a chromatin fingerprinting technique has been developed (1). After electrophoretic separation of nucleoprotein complexes, samples are digested further prior to second dimension DNA electrophoresis. This is accomplished by incorporating reversibly inactivated micrococcal nuclease or DNase I within the first dimension gel matrix and running buffer; the enzymes are reactivated after electrophoresis by exposing gels to Ca*+ or Mg*+ ions. The technique thus allows precursor-product relationships between all components to be monitored simultaneously.
The results of three independent time course experiments which employed redigestion with micrococcal nuclease reveal a complexity which is highly reproducible (Fig. 8, A to D, E to H, and I to L). The intact patterns of controls not exposed to Ca*+ ion indicate that during electrophoresis of chromatin the nuclease is inactive and does not alter the details of separation (Fig. 8, A, E, and I). Upon nuclease reactivation, it is clear that each multimer, n, is a precursor to the next smaller multimer, n -1 (Fig. 8, B, H, and K), a point supported earlier by redigestion experiments on pooled oligomers (15,57). It is noteworthy that the 160 base pair cores of monomers II and III are first processed to 140 base pairs and then to two to hotto~n,) was performed usmg a 4% acrylainide, 0.1% SDS slab with ethidium bromide staining. FIG. 7 ( smaller doublets of 124, 120 and 104, 100 base pairs, the same submonomer fragments observed upon redigestion of monomer I (Fig, 8, K andL). We conclude that monomers II and III are precursors to monomer I. Since the above submonomer DNA lengths agree closely with published values for chromatin and unfractionated mononucleosome limit digest bands (10-12, 15, 581, the complexity of submonomer digestion products need not represent structural heterogeneity beyond that contained in monomer I per se. This conclusion is strengthened further by the fact that the three monomers give rise to identical submonomer single-stranded DNA patterns upon DNase I redigestion (Fig. 91, patterns shown earlier to be characteristic of DNase I action on chromatin (59) and unfractionated mononucleosomes (151. Do classes of polynucleosomes exist in the global repeat which are composed only of monomer I? The finding that monomers II and III are precursors to monomer I complicates a solution to this question, particularly if chromatin consisting of monomer I per se is more resistant to nuclease digestion. However, if the major repeat is composed of a distribution of polynucleosomal classes with slightly different repeat lengths, then chromatin fingerprint patterns have great discriminatory potential. This follows from the fact that dinucleosomes and larger multimers are fractionated electrophoretically principally on the basis of DNA length (Figs. 3, 6, and 8). Hence, upon redigestion, different geometric patterns of product DNA fragments would be expected if clustered microheterogeneity exists. Indeed, Fig. 10 shows that different tetramer product DNA patterns can be predicted from basic principles for molecules originally exhibiting length heterogeneity due to exonuclease trimming, spacer length, core organization, or both spacer and core size.
Reinspection of the data of Fig. 8 with reference to the models shown in Fig. 10  During the course of redigestion, nearly all multimers are first H, and K) implies that there is clustered microheterogeneity converted to products showing 160 base pair barriers which in the global repeat due to spacer variability about 160 base then decay to 140 base pair fragments (Fig. 8). Therefore, it is pair cores (Fig. 10, model II). If histone Hl determines the clear that the great majority of polynucleosomes are organized spacer size as suggested by No11 (321, then such microheterowith cores of 160 base pairs. The fact that a diagonal line is geneity might be due to different tandem arrays of the various maintained upon conversion of tetramers to dimers (Fig. 8 III consist of 160 base pair cores, further questions are raised concerning their possible organization in chromatin which remain to be answered.
With regard to monomer I, direct conversion to 140 base pairs, without a pause at 160 base pairs, is seen upon redigestion of small dimers (Fig. 8, B, F, and G). This may be due to end degradation during nuclear digestion which may predispose a direct 140 base pair product upon redigestion (Fig. 10, model I). Alternatively, a minor class of polynucleosomes may be composed of monomer I (Fig. 10, model III). That this species may exist tandemly in a minor portion of chromatin is supported by the 142 + 3 base pair repeat described above (Fig. 71. Other redigestion experiments offer direct support that the larger 203 t 9 base pair repeat is a precursor to monomers IV and V; conversion of large trimers to large dimers and monomers has been observed (not shown). However, due to the low concentrations of monomers IV and V, it has not been possible to map the precursor-product relationships or core structures of these components.
Second Dimension Mapping of Protein Composition-A number of investigators have shown that 140 base pair monomers lack histone Hl, but contain approximately equal proportions of the other histone species (15,28,29,52). However, whether lysine-rich species are present in monomers with longer DNA lengths is unclear; studies on HeLa and erythrocyte mononucleosomes have revealed only traces of Hl and H5 (15,28,52), while larger sized mouse monomers have been reported to be rich in Hl species (29). In the present study, establishing the location of Hl species among bovine nucleosome monomers was particularly attractive in view of the knowledge accrued on precursor-product relationships. To display the chromatin proteins of electrophoretically separated bovine chromatin, a second dimension of electrophoresis using 18% a&amide, 0.1% SDS slabs was employed (54).  In contrast, all histone Hl subfractions are clearly absent from monomer I, and perhaps are lacking from monomer II. One faint non-hi&one protein band which is present between histones Hl and H3 also is absent from monomer I (Fig. DA). Similar separations of briefer or more extensive digests reveal the same qualitative pattern with regard to the histone distribution among mononucleosomes.
Since monomer III was shown to be a precursor to monomer I, it follows that histone Hl becomes released upon digestion of 20 base pairs as suggested by Varshavsky et al. (29).
In order to determine whether monomer II lacked histone Hl, a second dimension of electrophoresis was employed using of Polynucleosomes  Table I summarizes the properties of the five mononucleosomes of bovine thymus chromatin which have been identified in the present study. It is of interest that during the course of nuclease digestion there is partial overlap in the DNA lengths between different monomer classes. In particular, monomers I, II, and III share common lengths; monomers III, IV, and V show similar overlaps. Thus, parameters in addition to DNA size contribute to mononucleonucleosome fractionation by electrophoresis.
These presumably include protein composition, charge, and conformation. The electrophoretic complexity of mononucleosomes cannot be explained solely on the basis of histone Hl content, since both monomers I and II lack these species. Furthermore, Olins et al. (52) have shown that electrophoretic fractionation of erythrocyte mononucleosomes is not entirely due to lysine-rich histone content. Thus, whether the five classes of monomers described above each consist of a histone octamer (17) composed of two copies of each of the four smaller histones remains to be established. chromatin is 191 * 8 base pairs, an average length of about 30 base pairs can be surmised for the spacer which connects 160 base pair cores. Clustered variation in spacer lengths is suggested by the maintenance of diagonal lines in chromatin fingerprints (Fig. 8, B, F, and K; Fig. 10, model ZZ; Ref. 1). Indeed, the DNA length ranges of multimers are sufficiently broad not to preclude the existence of a distribution of repeating units, each differing by several base pairs from the mean value. Variation in the global repeat length may be due to different histone Hl subfractions, if lysine-rich histones are responsible for determining the distance between adjacent cores as suggested earlier (32). Clustering of spacers of similar lengths could be due to specialized organizations of Hl subfractions; chemical cross-linking studies support the existence of homopolymers of lysine rich histone subfractions in chromatin (60).
A paucity of monomer I exists in bovine thymus chromatin; the great majority of polynucleosomes are organized with cores of 160 base pairs. This value, as opposed to 140 base pairs, agrees with the core size predicted for rat liver chromatin by Simpson and Whitlock (31). Although we have established that monomer I accumulates during digestion because it is a processing product of monomers II and III, it is not clear whether monomer III is processed to monomer II or if monomer II exists independently in chromatin.
In addition to the global nucleosomal repeat discussed above, at least three other minor classes with different repeat lengths exist in bovine thymus chromatin.
These findings raise a number of questions with regard to the origins and functions of varied arrangements of nucleosomes. Different repeats may arise from the various cell types of thymus tissue. Recently, nucleosomal repeat lengths have been shown to differ between chicken tissues (61). Conversely, different repeats may exist within single cells and may possibly represent various functional states of the genome. It is conceivable that certain classes may even be organized with subsets of the total nuclear DNA sequences. These possibilities are currently under investigation.
Histone Hl binding domains on spacer DNA have been suggested previously (15,29,31). The results of the present study imply that an additional binding site(s) exists on or near the nucleosomal surface. Hl remains bound to monomer III even after nuclease processing to 160 base pairs. The conversion of this form of monomer III to monomer I is accompanied by digestion of 20 base pairs and release of HI. Less direct findings of Varshavsky et al. (29) support the above results; however, a discrepancy in DNA lengths exists which presumably reflects differences in size calibrations.
Whether single Hl molecules occupy 20 base pairs on one or two terminal sites or become released indirectly are matters for further study. The recent finding of Olins et al. (52) demonstrating equal molar ratios of the five classes of histones in erythrocyte nuclei fw-ther complicates this issue. Acknowledgments -The skillful technical assistance of Messrs. D. Boatwright and S. Albright is gratefully acknowledged. We thank Drs. D. Gray, M. Singer, and B. Hirt for their generous gifts of DNA, and Messrs. T. Webb and C. Moen for photography.
We are also indebted to P and H Packing Co., Inc. of Dallas, Texas for their generous gifts of bovine tissue.

Note Added
in Proof-Recently, we have found that monomers I, II, III, and IV-V are present in the chromatin of embryonic bovine trachea cells (ATCC CCL44). This finding demonstrates that unique mononucleosomes which arise from different parent DNA repeat lengths exist within a single cell type. The DNA sizes of these mononucleosomes are similar to those of bovine thymus chromatin, except that monomer II is trimmed to 140 base pairs. This observation offers further support to the proposal that differences in protein composition or particle conformation, or both, exist between monomers I and II.