Psoralen-crosslinking of soluble and of H1-depleted soluble rat liver chromatin

.A. (‘OX(‘c)XJ

Psoralen-crosslinking Soluble

ET

.-IL

of Soluble and of Hl-depleted Rat Liver Chromatin

A. CONCONI.It. LOSA. TH. KOLLER ANI)

J.

M.

Soao

lnstitut fiir Zellbiologie ETHZ, H6nggerberg CH-8093 Ziirich, Switzerland

We purified soluble rat. liver chromatin and Hl-depleted chromatin and photocrosslinked its DNA with psoralen at pH 7. Digestion of this chromatin with micrococcal nuclease produced a normal nucleosomal repeat. Chromatin was photoreacted in the presence of 0 to 700 mM-NaCl and was fractionated in sucrose gradients containing the same KaCl concentrations. The dissociation of Hl occurred as in the non-crosslinked controls and no preferential dissociation of core histones was observed. The samples between 100 and 500 mv-h’aC1 showed precipitation. In the electron microscope, the fibers appeared indistinguishable from the controls at low ionic strength. In the presence of 40 mM-NaCl. the fibers of the photoreacted chromatin were slightly more compact than t,he controls, and at 500 mM-NaCl. despite the complete dissociation of Hl, there were still apparently intact fibers at this ionic strength. The disruption of the psoralentreated chromatin fibers occurred only in 600 mM-NaCl, as opposed to 500 rnMEaCl in controls. The DNA of all the phot,oreacted samples was spread for electron microscopy under denaturing conditions. They revealed, for all the samples, single-stranded bubbles corresponding to 200 to 400 base-pairs in size. HI -depleted chromatin containing stoichiometric amounts of core histones was photoreacted at pH IO and very low ionic strength. Under these conditions many of the nucleosomes appeared to be unraveled, although to a variable extent. Tn the electron microscope. the purified DNA from these samples showed extensive crosslinking when spread under denaturing conditions. These observations show that histone-DNA interactions different’ from those in intact, nuczleosomrs may be czreated. which allow extensive access of psoralen to the DE!.

1. Introduction Psoralen-crosslinking of DNA in chromatin has been used to probe chromatin structure (Hanson et al., 1976; Cech et al., 1978). The low extent of crosslinking obtained in chromatin DNA has been interpreted as being due to a protection of the DNA within the nucleosome from crosslinking. However, since psoralen interacts

with

DNA

by

intercalation

(Yoakum

& Cole,

1978)

and

since

it is known

that intercalating dyes like ethidium bromide induce nucleosome sliding and a destruction of the higher order structures of chromatin (bard et al., 1979), it is not clear at all whether psoralen-crosslinking of DNA in chromatin does not lead to alterations in chromatin structure. The results described in the main text raised the question of to what extent the psoralen and ultraviolet light treatment promotes protein displacements and structural alterations. Since extensive

PSORALEN-CROSSLIIUKIN(:

OF

(‘HROMATIN

I)NA

921

attempts to produce soluble chromatin from Dictyostelium discoideum failed (R. Losa, unpublished data), we decided to study the effects of psoralencrosslinking on the biochemically well-characterized soluble chromatin of rat liver (Thoma et al., 1979). We show that under our conditions of crosslinking at pH 7. the size of the DNA fragments produced by limited digestion with micrococcal nuclease remains unchanged, that no preferential dissociation of histones occurs and that no distortions of the nucleosomal and rather a stabilization of the higher order structures of chromatin are detected. However, when HI-depeleted chromatin is crosslinked with psoralen at pH 10 and very low ionic strength, conditions that lead to variable unfolding of many of the nucleosomes. then the low extent of orosslinking seen at pH 7 gets lost and is replaced by heavy crosslinking of the chromatin DXA.

2. Methods The preparation of soluble rat liver chromatin and of Hl-depeleted chromatin was as described by Thoma & Keller (1981). Electron microscopy. protein sodium dodecyl sulfate/ polyacrylamide gel electrophoresis, micrococcal nuclease digestion of soluble chromatin and DXA agarose gel electrophoresis were as described (Thoma & Keller. 1977.1981; Thoma et al., 1979,1983; Labhart et al.; 1981). The procedures for psoralen-crosslinking of soluble chromatin and of Hl-deplet,ed chromatin. using a 4.5’,8-trimethylpsoralen stock solution in ethanol (100 pg/ml), were the same as in the main text. The spreading of the DXA under denaturing conditions was as described by Sogo et al. (1979). The standard buffers contained 1 mm-triethanolamine chloride (pH 7), 0.2 m,n-EDT.4 (called “1 mM”) or 1 mMdiethanolamine chloride (pH 10). 0.2 mM-EDTA (called “1 mu” at pH lo), or 5 mMtriethanolamine chloride (pH 7). 0.2 mM-EDTA containing 10 rnq 40 mM, 100 mM, 350 mM, 500 mM, 600 mM. 700 mM-NaCl (referred to according to the Pu’aCl concentration, e.g. “40 mM”). For fixation and visualization of chromatin in the electron microscope, these buffers contained in addition O.lO/,, (w/v) glutaraldehpde (Thoma & Koller. 1981).

3. Results and Discussion (a) Digestion of psoralen-photoreactedchromatin and HI-depleted chromatin with micrococcal nuclease Purified soluble chromatin and Hl-depleted chromatin were dialyzed against “100 mM”. Trimethylpsoralen was added and the samples were irradiated with ultraviolet light (366 nm) for four hours with two subsequent additions of a portion of psoralen stock solution after one and two hours. After dialysis against “100 mM”, CaCl, was added to 1 mM and the samples were digested for increasing times with micrococcal nuclease. Figure Al shows the gel analysis of the DNA fragments produced. The fragments show a typical nucleosomal repeat of the same size as that in the non-crosslinked control (M). Il’ote the low endogenous nuclease activity in these samples (tracks Cl, C2 in Fig. 1). The same result was obtained upon digestion of Hl-depleted chromatin treated with psoralen in “100 mM” (dat,a not shown).

Cl

c2

0-5

2

5

IO

M

Fro. Al. Digestion \vith ~~I~~I.~~wY~~I nu(.leas~ of ~Aublr (+romat in photocrosslinked with trimethylpsoralen. Rolublr~ c,hromatin I)so’.;tlerr-l.r.osslinkrd in ‘I 00 m>r” \\a~ digested at 37°C’ for of increasing times (as indicated in the I’igurr in min) with 3 x IO 3 units of mic~roc~occal nuclease/~l soluble chromatin. The I1N.A was rlrc,troplrorrsrd in a I W,, agar,~ gel. .\I. control chromatin, not. crosslinked with psor&n. The fastest migrating bands correspond to thr monomer with about 140 base-pairs of D?;A. (‘1 and (‘2 arv control xamples without microc~cc~al nuclrasr. (‘1 being stopped with EDTA at the beginning of digestion. (‘2 baling stopped after 10 mitl

(b) Partial dissociatiw~ HI-depletrd rhromatin

qf psoralen-yhotorPacted at increasing

chromatin and concentrations of Xal’l

in sucrose gradients Hl-depleted chromntin prepared by fractionation containing 500 rn>r-SaC’I (Th oma & Keller, 1981) was dialyzed in parallel against or “10 mM” or “40 mM” or “100 rnM” or “3X) mM”. The samples were “1 rnM”, crosslinked with trimethylpsoralen and were then fractionated in IO?;, t’o 40?() (w/v) sucrose gradients conta,ining the corresponding dialysis buffer. The two top fractions, a portion of the peak fractions containing 15 ,~g of chromatin DSA or t)he pellet fraction in the case of precipit,ated samples were precipitated with %($,. (w/v) trichloroacetic acid and analyzed by gel electrophoresis. Figure AZ(a) shows that no protein could be detected in the top fractions (T) and that all the histones were found in the peak or the pellet (P) fractions. In the pellet fraction of the new bands appeared corresponding to and “3%~ rn>r” gradients. “100 mM” molecular weights greater t,han those of the core histones. Two prominent bands migrated slightly faster than histone HI. We assume that these bands originate from crosslinking of core hist,ones (Yoshikawa rt al.. l!Yi9). The amount of material in the pellet fractions. as compared to the material in the peaks. since the increased with increasing c*oncsrntrations of Sa(‘1 in the gradients, chromatin sediments faster with increasing ionic strength. The same experiment was repeated, but using long Hl-containing chromatin gradients in addition in “SO0 mM”. “600 m&r” and and running t’he summe

l’SOItALE?U’-(‘KOSSLINKISU

OF

(‘HROMATIN

USA

923

mM NaCl

(a)

(b) FIG. 42. Fractionation at increasing ionic strength of (a) Hl-depleted chromatin and of (b) HIcontaining soluble chromatin, photocrosslinked with trimrthylpsoralen. (a) Long Hl-deplet,ed chromatin and (h) HI-containing soluble chromatin was dialyzed at 1°C overnight against the standard buffers indicated in the Figure. Psoralen was then added and crosslinking was performed as drscrihrd in Methods. The data of parallel control samples not treated with psoralen are not shown. since up to “SO0 KIM“ they were the same as described earlier (Thoma Br Keller. 1981). The samples were then fractionated on 10 to 10” 0 suerose gradients containing the corresponding buffer. Fractions from the gradients were precipitated and analyzed bv sodium dodecyl sulfate/polyacrylamide gel elwtrophoresis. ‘I‘. the 2 fractions from the top of the gradients (0.5 ml each) containing the dissociated components. P, material in the absorbance peak recorded at ~54 nm. for “1 mM”, “10 my” and “10 rn.\l” From ” 100” to “‘700 m\l”. P represents the material in the pellet. 11, control rhromatin.

“700 rn>l”. Figure AZ(b) shows that in gradients csontaining 500 mM-KaCl and above most of the Hl has been dissociated, since it appears in the top fractions (T) of the gradients. In gradients containing 700 rnM-NaCl, in addition, a partial dissociation of H2A and H2K is seen. Again. as described above for the Hldepleted chromatin, in gradients containing 100 rnM-NaC1 and above, new bands 32

(d),(e),(f) precipitated

FTC:. ‘43. Electron rni~~osco~~~ of solublr chromatin photocrosslinked with psoralen at increasing ionic strength. I’soralen-photoreac:trd samples and their controls of the experiment of Fig. A2 were fixed with O,l’& g lutaraldehyde in the standard buffers “1” up to “700 mM” and were prepared for rlectron microscopy using Al&m blur-coatrd grids (Labhart & Keller, 1981). (a) “1 mM”; (b) “10 “JM”; (c) “40 mM”: (d) “100 mM”: (P) “350 mzvf”; (f) “500 mM”; (g) “600 mix’)‘: (h) “700 mix”. The bar represents 0.5 pm.

appear that migrate slower than the core histones. The same result was obtained with control native chromatin without crosslinking (not shown, but see Thoma & Keller, 1981) except that these new bands, interpreted as representing crosslinked histones, were absent. We conclude from these experiments that photoreaction with psoralen leads to

l’SORALEN-CROS8LINKIN(:

some protein-protein and this crosslinked chromatin.

OF

crosslinking, material

is

but that similar

CHROMATIS

the partial to that

(c) Electron microscopy of psoralen-crosslinked HI-depleted chromatin and spreading

DNA

926

dissociation of the histones of non-crosslinked control

soluble chromatin, of its DNA

Long chromatin fragments and fragments of Hl-depleted chromatin were crosslinked with psoralen as described above. The samples were then dialyzed in parallel against the standard buffers “1 mM” up to “700 mM” and they were then fixed for over I6 hours at these ionic strengths by adding glutaraldehyde (to a final concentration of 0.1%) to the dialysis solution. The samples were then prepared for electron microscopy, in parallel with controls not crosslinked with psoralen. Whereas in controls only the samples in “100 mM” showed some precipitation, in the psoralen-treated samples in “100 mM” and “350 mM” the precipitation was quite strong. This has to be taken into account, because our specimen preparation technique selects strongly for particles in solution. In Figure A3 we show the ionic strength-dependent folding of psoralen-crosslinked, Hl-containing chromatin. Its appearance in “1 mM” and “10 mM” is indistinguishable from the non-crosslinked controls (not shown) and from the data described by Thoma & Koller (1981). The samples in “40 mM” are slightly more condensed than the corresponding controls. It is interesting to note that even in “500 mM” (dat;t not shown), where all the HI is dissociated (Fig. A2(b)), intact

FIG. AP. DNA psoralen-crosslinked “100 m?l” (experiment of Fig.

A2),

in spread

(a) soluble chromatin under denaturing

and conditions.

(b)

in Hl-depleted chromatin in The bar represents 0.2 pm.

!Mi

.\

(‘OS(‘OSI

E7’

.-I /,

fibers were found in the soluble fraction of the sample. The dissociation of higher order structures seen in controls in “500 m,n” was only seen in “BOO mnl“ (Fig. AS(g)) in the psoralen-t,rc~at,rtl chromatin. Psoral~:n-c.rosslinked H I -depIcted chromatin behaved like t,he controls (data not shown: see also Thoma Ki Kollcr. 1981) except that the samples in “100 mM” frequently showed aggregates OI heavily condensed fragments, which were not seen in the other samples. In parallel to these experiments. the f’soralen-c,rosslinkrd samples were not fixed with glutaraldehydr, but their DSA was extracted and spread for electron microscopy under denaturing conditions. In agreement with published data (Hanson et al., 1976: Cech ef al., 1978). this DNA showed single-st’randed bubbles of about 200 to 466 basr-pairs in size (Fig. A4). We conclude from the electron microscopic analysis that crosslinking with tjrimethylpsoralen under our conditions leads only to small alt,erations and rather slight, condensation and stabilization of the higher order st’ructures of soluble chromatin.

FIG. A5. DNA photoreacted in Hl-depleted chromatin at very low ionic strength and (a) pH 10 and (b) after back dialysis to pH 7, spread under denaturing conditions. The arrow points to nicked circular pBR2.17, which had been psoralen-crosslinked separately to a very low extent in order to demonstrate the spreading of single-stranded regions under the conditions used. The bar represents 0.5 pm.

I’SORALEN-(:ROSSLINKING

(d) Crosslinking

OF

C’HROMATIN

of Hl-depleted soluble chromatin very low ionic strength

DNA

at pH

927

70 and

One model for the structure of chromatin during transcription is that of an unraveled nucleosome (e.g. see Prior et al., 1983). We have shown (Labhart et al.. 1981) that at pH 9 and very low ionic strength many of the nucleosomes in polynucleosomal, HI -depleted chromat,in are unraveled. alt’hough to a variable extent, in a fully reversible manner. The same is true at pH 10 (Losa et al., 1984). In order to test the accessibility of psoralen to DNA in such chromatin samples, long, HI-depleted chromatin was dialyzed against “1 mM” at, pH 10 and was photoreactOed with psoralen. In agreement with numerous experiments performed in this laboratory before, sucrose gradient analysis showed that the core histones remained bound to chromatin in stoichiomrtric amounts (not shown), similar to Hl-depleted chromatin at pH 7 (see Fig. A2(a)). When the DNA was extracted from this sample and spread for electron microscopy under denaturing conditions, no characteristic “nucleosomal” singk-stranded bubbles were found, with most, of t’he DNA appearing as duplex strands (Fig. A5(a)). Free DNA psoralen-crosslinked at pH 10 (not shown) had the same appearance as those shown in Figure AFi(a). A portion of Hl-depleted chromatin at pH 10 was dialyzed back to “1 mM“ at pH 7 before photoreaction wit,h psoralen. The extracted DNA spread under denaturing conditions shows the single-stranded bubbles characteristic for chromatin organized in nucleosomes (Fig. A5(b)), confirming the reversibility of the pH-induced structural changes described by Labhart et al. (1981). l’he results of this experiment demonstrate that alternative histone-DNA interac%ions, as compared to those in normal nucleosomes, do exist that allow extensive access of psoralen to histone-complexrd DNA. We thank P. .J. ,?jess, F. Thoma. R. W. Parish, R. Widmer and R. Lucchini for critical reading of the manuscript, and H. Mayer-Rosa for excaellent technical assistance. This work was supported by Schweizerischer Sationalfonds zur Fiirderung der wissmschaftlichen Forscahung. REFEREYCES L 2 , Cech. T. R., Potter, D. & Pardue. M. L. (1978). Cold Spring Harbor Symp. Quant. Biol. 42, 191-198. Erard, M., Das. G. C., de Murcia, G.. Mazen. A., Pouret, cJ. Champagne, 31, & Daune, 11. (1979). Nucl. Acids Res. 6, 3231-3253. Hanson. C. V., Shen. C.-K. J. & Hearst, J. (1976). Science, 193, 62-64. Labhart. P. & Koller, Th. (1981). Eur. J. Cell. Biol. 24, 309-316. Labhart, P., Thoma, F. & Koller, Th. (1981). Eur. J. Cell Biol. 25. 19-27. Losa. R., Thoma, F. 6 Koller. Th. (1984). J. Mol. Blol. 175, 529-551. Prior, (1. P., Cantor. C. R.. Johnson. E. M.. Littau. V. C. & Allfrey. \‘. G. (1983). Cell. 34. 1033S1042. Sogo. .J. M., Rodeno, P.. Koller. Th., Vinuela. E. & Salas. M. (1979). LVUCZ.Acids Res. 7. 107--120.

Thoma. Thoma, Thoma.

F. & Koller. Th. (1977). Cell, 12. 101-107. F. & Koller, Th. (1981). J. Mol. Riol. 149. ‘709-733. F.. Kollrr. Th. $ Klug, A. (1979). J. Cell Biol. 83, 403-42T.

028

A. (‘ON(‘OXI

E7’ -1 L

Thoma. F.. Losa, R. 8: Keller. Th. (1983). J. Mol. Bid. 167, 619-640. Yoakum, G. H. & Cole. R. S. (1978). Biochim. Biophys. dckz, 521. 528S54B. Yoshikawa, K.. Mori. h’., Sakakihara. S.. Mizuno, Xv. & Song. S.-P (1979). Photobiol. 29. 1127-l 133.

lhiittd by A. Klug

I’hotoch~v/