Volume 4 Number 11 November 1977

Nucleic Acids Research

Urea-induced structural changes in chromatin obtained by sedimentation

Rodney E. Harrington * The University of Tennessee-Oak Ridge Graduate School of Biomedical Sciences and the Biology Division, Oak Ridge National Laboratory, Oak Ridge, TN 37830, USA Received 8 August 1977

ABSTRACT

Sedimentation coefficients have been determined for fractionated preparations of whole and stripped (depleted of very lysine-rich histones and non-histone proteins) chicken erythrocyte chromatin fragments in 0-10 M urea. Significant differences in urea effects are observed between these preparations; differences which can be interpreted structurally by use of Kirkwood's dynamical theory of the translational frictional coefficient. This type of analysis implies that urea-induced chain-swelling in stripped chromatin is due largely to the urea effect upon the constituent v-bodies, whereas the much larger swelling observed in whole chromatin appears to involve also the effect of urea upon the region between adjacent v-bodies. I NTRODUCTION It is now generally accepted that'the structural level immediately above DNA in eukaryotic chromatin is a repetitive "string of beads" array of nucleohistone subunits called v-bodies or nucleosomes (1, 2). Changes in physical properties of chromatin with urea concentrations up to several molar have been observed in hydrodynamic (3-5), spectroscopic (6-1 0), low-angle X-ray scattering (11), thermal denaturation (8, 9, 12), electron microscopic (11, 13, 14) and nuclease digestion (1 5, 16) studies. Despite this wealth of information, the structural basis of urea-induced destabilization of chromatin remains imperfectly understood. We have measured sedimentation coefficients of chicken erythrocyte chromatin over a urea range of 0-10 M. These preparations consisted of fractionated fragments of weight-average size approximately v13 as determined by sedimentation analysis of the constituent DNA, assuming 210 nucleotide pairs per subunit (17-19). Both whole and stripped chromatin preparations were investigated, the latter depleted of the very lysinerich histones HI and H5, and probably of most of the non-histone protein complement as

well. C) Information Retrieval Limited 1 Falconberg Court London Wl V 5FG England

3821

Nucleic Acids Research EXPERIMENTAL

Chicken erythrocyte nuclei were digested for 5 min with micrococcal nuclease by the procedure described earlier by Olins et al. (20). The degraded nuclei were lysed in 0.2 mM EDTA and centrifuged for 15 min at 12,000 X g. The supernatant was fractionated by ultracentrifugation on a sucrose gradient as described elsewhere (21). The stripped chromatin was prepared by further dialyzing the preparation overnight against 0.7 M KCI, 0.2 mM EDTA, and 0.1 mM phenylmethylsulfonyl fluoride (PMSF) immediately before the sucrose gradient fractionation steps. The size assay for the chromatin was based upon molecular weight of the constituent DNA as determined by sedimentation velocity (22, 23). RESULTS AND DISCUSSION Experimental sedimentation coefficients for whole and stripped chromatin preparations are given versus urea concentration in Figure 1. It is clear that the ureadependences of the sedimentation coefficients are significantly different between these two systems. Sedimentation data were corrected to s20 w on the basis of solvent viscosity and density only. Reliable buoyancy data are not yet available for the chromatin-urea systems; hence, the true urea-dependent partial specific volumes for VI monomers (24) were used for both preparations. However, we do not believe that the large observed differences in urea effects between the whole and stripped chromatin can be attributed to differential buoyancy changes since the largest differences occur at the lowest urea concentrations. If the differences between whole and stripped chromatin in Figure 1 are due to frictional rather than buoyancy factors, the theory of the translational frictional coefficient can be used to infer relative structural changes in this system. For this purpose, we model chromatin as a linear assembly of n v-bodies, each of hydrodynamic diameter b, connected by frictionless DNA linkages or spacers. Kirkwood's theoretical treatment for such a model (25) then gives, for the sedimentation coefficient,

(-pv) F

-c

3822

37TJfflbnNA

[1 +

b

fn

i2

-1 1 (R.. KIMCI

(1

Nucleic Acids Research

25-o

10

o

0~~~~~~~~~~~~~~~~~~

0~~~~~~~~~~~~~~~~ O

1

2

4 5 6 UREA (molarity)

3

7

8

9

0o

Figure 1. Sedimentation coefficient data s20 w for whole (0) and stripped (0) chicken erythrocyte chromatin. Solid lines are least-squares fits to cubic polynomials used for evaluating estimated spacer dimensions x (see text). in which p is solution density, v is partial specific volume of the assembly, Mc is its molecular weight, T1 is solvent viscosity, NA is Avogadro's number, and (R;i) is the mean reciprocal center-to-center distance between the ith and ith beads. If Mc = nMv + (n - 1 )M x where M and M A are the molecular weights of v-bodies and spacer regions respectively, and if it is assumed that the partial specific volumes and hydrodynamic diameters of isolated v-bodies are comparable to their values in chromatin, equation 1 can be combined with the sedimentation coefficient for isolated v-bodies, s,to obtain

-c

-v

[I 2n_-f(n,b)n L

-

I

n

Mv

(2a)

inLmV J

For rigid linear and helical conformers, f(n,b) = f(n)/xb where, for each conformer, f(n) is a number whose value depends upon n only, and the spacing between beads, xb, is given as a multiple of hydrodynamic bead diameter b. Hence, with these restrictions, the ratio sc/sv becomes an explicit, although insensitive, function of multimer structure as well as chain length. 3823

Nucleic Acids Research Calculations similar to these have been described by Andrews and Jeffrey (26) for certain general oligomeric structures, and by Shaw and Schmitz (27) to deduce chromatin structure from the molecular weight dependence of the sedimentation coefficient. In this sort of treatment, all structural information comes from f(b). This, in turn, derives entirely from Kirkwood's approximate, linearized treatment of hydrodynamic shielding among the constituent beads of the model assembly. Thus, the method is highly approximate on theoretical grounds, and is further limited by its relatively low sensitivity to model structure and the ambiguity due to the fact that there is, in general, no unique correspondence between sc/sv values and specific geometrical structures. With these caveats in mind, we feel nevertheless that the method can provide useful relative structural information, particularly when structural perturbations are likely to be quite dramatic, as appears to be the case in the work described here. To interpret the data of Figure 1 in structural terms, we must first establish a reasonable conformational model. For present purposes, this can be done sufficiently well with sedimentation-molecular-weight dependence data. We have used the data of Noll and Kornberg (28) along with n = 13, M, = 2.16 X 105 (21) and MA = 4.2 X 104 and 9.9 X 10o for stripped and whole chromatin respectively. These latter values are obtained as follows. Mk for stripped chromatin is taken as the molecular weight of '60 nucleotide pairs of DNA only (28); for whole chromatin, M) is computed as a dimer molecular weight of 5.3 X 105 (27) minus 2Mv. In Table I, the n-dependence of f(n) is given in the form f(n) = Pna (see table) for several geometrical models and for the experimental data of Noll and Kornberg, the latter obtained from equation 2a in the form = 2nx[ f[(n)

(1 +

(v)]

(2b)

A comparison of the theoretical and experimental results shows that, as expected, f(n) is fairly insensitive to the model as between a rigid linear array and rigid closest-packed helices of three to six v-bodies per turn. By a small margin, the experimental data of Noll and Kornberg (28) agree most closely with a helix of five v-bodies per turn. Optimum agreement is obtained with this model and spacer lengths of 2.1 b and 3.7 b for whole and stripped chromatin respectively, but this preferential margin is considerably smaller than experimental error. On the other hand, the results definitely seem 3824

Nucleic Acids Research to rule out a random-flight flexible array as computed by Filson and Bloomfield (29). Some insight into the structural significance of the urea-dependence data of Figure 1 can now be obtained from equation 2a. For our present purposes, it is sufficient to base our calculations upon the rigid five v-bodies-per-turn helical model described above, although we emphasize that the model is somewhat arbitrary and the precise details should not be taken literally. The calculations will be much more sensitive to relative changes in chain length and bead spacing due to urea, however, and we think it instructive to compare whole and stripped chromatin in this fashion. Since we have defined xb as the center-to-center v-body spacing in the model, (x- 1 )b is the effective mean spacer length. The ratio x/xo, where xo is the value of x in the absence of urea, is shown in Figure 2. This quantity decreases almost linearly between 0 and 6 M urea, and is virtually constant at higher urea for stripped chromatin. It increases non-linearly for whole chromatin over the entire urea range, however. TABLE I. The dependence of the geometrical factor f(n) upon chromatin-fragment v-body content n, expressed as an exponential function f(n) = Pn'. The experimental data of Noll andTKornberg (28) are compared with calcuTaled results based upon various geometrical models. P

a

r2*

Experimental: Whole chromatin

0.269

2.003

0.999

Experimental: Stripped chromatin

0.169

2.030

0.997

heoretical: Rigid, linear model

0.573

1.914

0.995

heoretical: Random-flight coil (26)

0.849

1.771

0.999

0.830

1.908

0.997

m = 4

0.801

1.936

0.999

m = 5

0.630

2.026

0.998

m = 6

0.622

2.003

0.998

heoretical: "Contiguous" closest-packed helices (x =1) with m v-bodies per turn m = 3

Theoretical: Extended helices with

m v-bodies

per tum

*

m

=

5

x

=

2.13

0.269

2.003

0.999

m

=

5

x

=

3.73

0.169

2.026

0.998

r2 is the log-log correlation iWtercept.

2O (n o coefficient, r~~~e. e.g., l1og f(n) log n'

evaluated at the

3825

Nucleic Acids Research If the v1 monomers behave hydrodynamically as spheres, their sedimentation coefficients s20 w are proportional to l/b. Thus, the correspondence of for stripped chromatin in Figure 2 implies (s 20 w)/(20,w)O for v1 monomers with that center-to-center v-body spacing in this system is virtually independent of urea over the range studied here; with increasing urea, x decreases in about the same proportion as b increases, hence the urea effect is mainly in the v-bodies. This would seem to indicate that the spacer regions shrink very nearly in proportion to the v-body swelling with urea in the stripped system. Whole chromatin, on the other hand, displays quite different behavior, and the sedimentation data in this case imply a considerable overall lengthening of the chromatin chain with increasing urea concentration. The differences between whole and stripped chromatin observed in this study are evidently associated with the lysine-rich HI and H5 histones and possibly also with non-histone proteins. Our results are not sufficiently definitive to determine whether the urea effects in whole chromatin derive from chain expansion and conformational changes or from an increasing contribution of the spacer regions to the overall frictional coefficient. Since the differences are so dramatic, however, we believe that our results support the notion that these proteins are predominantly associated with the spacer regions rather than with the v-bodies in the chromatin chain.

xl/j

2

0

0

1

2

3

4

5

6

7

8

9

10

UREA (molarity)

Figure 2. Relative spacer dimensions x/xo for whole (0) and stripped (0) chicken erythrocyte chromatin compared to relative sedimentation coefficients (A) (s20,w)/V (s20,w)o for pure chicken erythrocyte v, monomers.

3826

Nucleic Acids Research ACKNOWLEDGMENTS I am grateful to Drs. D. E. Olins and A. L. Olins for many stimulating discussions, and to the Biology Division, Oak Ridge National Laboratory, for research facilities. I am also indebted to E. B. Wright and P. N. Bryan for techical assistance and for preparing chromatin samples. Financial support was through NIH Research Grant GM 19334 (to DEO). Portions of this work were presented to the Cold Spring Harbor Symposium on Quantitative Biology in June, 1977.

During this work the author was on leave (1976-77) from the Department of Chemistry, University of Nevada, Reno, Nevada 89507. Please address correspondence and reprint requests to him there. Oak Ridge National Laboratory is operated by Union Carbide Corporation for the Energy Research and Development Administration. REFERENCES

Elgin, S. C. R. and Weintraub, H. (1975) Annu. Rev. Biochem. 44, 725-774 Felsenfeld, G. (1975) Nature (Lond.) 257, 177-178 Bartley, J. A. and Chalkley, R. (1968) Biochim. Biophys. Acta 160, 224-228 Bartley, J. A. and Chalkley, R. (1973) Biochemistry 12, 468-474 Christiansen, G. and Griffith, J. (1977) Nucleic Acids Res. 6, 1837-1851 Shih, T. Y. and Lake, R. S. (1972) Biochemistry 11, 4811-4817 Rill, R. and Van Holde, K. E. (1973) J. Biol. Chem. 248, 1080-1083 Chang, C. and Li, H. J. (1974) Nucleic Acids Res. 1, 945-958 Wilhelm, F. X., De Murcia, G. M., Champagne, M. H. and Duane, M. P. (1974) Eur. J. Biochem. 45, 431-443 10 Whitlock, Jr., J. P. and Simpson, R. T. (1976) Nucleic Acids Res. 3, 22551 2 3 4 5 6 7 8 9

2266

Carlson, R. D., Olins, A. L. and Olins, D. E. (1975) Biochemistry 14, 31223125 12 Ansevin, A. T., Hnilica, L. S., Spelsberg, T. C. and Kehm, S. L. (1971) Biochemistry 10, 4793-4803 13 Georgiev, G. P., Il'in, Y. V., Tikhonenko, A. S. and Stel'maschchuk, V. Y. (1970) Mol. Biol. 4, 196-204 14 Woodcock, C. L. F. and Frado, L.-L. Y. (1976) J. Cell Biol. 70, 267a 15 Jackson, V. and Chalkley, R. (1975) Biochem. Biophys. Res. Commun. 67, 1391-1400 16 Yaneva, M. and Dessev, G. (1976) Nucleic Acids Res. 3, 1761-1767 17 Senior, M. B., Olins, A. L. and Olins, D. E. (1975) Science 187, 173-175 18 Morris, N. R. (1976) Cell 9, 627-632 19 Olins, D. E., Bryan, P. N., Olins, A. L., Harrington, R. E. and Hill, W. E. (1977) Biophys. J. 17, 114a 20 Olins, A. L., Carlson, R. D., Wright, E. B. and Olins, D. E. (1976) Nucleic Acids Res. 3, 3271-3291 11

3827

Nucleic Acids Research 21 22

23 24 25 26 27

28 29

3828

Zama, M., Bryan, P. N., Harrington, R. E., Olins, A. L. and Olins, D. E. (1977) Cold Spring Harbor Symp. Quant. Biol., in press Bloomfield, V., Crothers, D. M. and Tinoco, Jr., I. (1974) Physical Chemistry of Nucleic Acids, Harper and Row, New York Crothers, D. M. and Zimm, B. H. (1965) J. Mol. Biol. 12, 525-536 Olins, D. E., Bryan, P. N., Olins, A. L., Harrington, R. E. and Hill, W. E. (1977) NucleicAcids Res. 6, 1911-1931 Kirkwood, J. G. (1954) J. Polymer Sci. 12, 1-14 Andrews, P. R. and Jeffrey, P. D. (1976) Biophys. Chem. 4, 93-102 Shaw, B. R. and Schmitz, K. S. (1976) Biochem. Biophys. Res. Commun. 73, 224-232 Noll, M. and Kornberg, R. D. (1977) J. Mol. Biol. 109, 393-404 Filson, D. P. and Bloomfield, V. A. (1968) Biochim. Biophys. Acta 155, 169-182