Photochemical Cross-Linking of DNA with Trioxsalen Nuclei As Probed By

Downloaded from symposium.cshlp.org on May 16, 2009 - Published by Cold Spring Harbor Laboratory Press

Chromatin Structures of Main-band and Satellite DNAs in Drosophila melanogaster Nuclei as Probed by Photochemical Cross-linking of DNA with Trioxsalen

C.-K. J. Shen and J. E. Hearst

Cold Spring Harb Symp Quant Biol 1978 42: 179-189 Access the most recent version at doi:10.1101/SQB.1978.042.01.020

References This article cites 39 articles, 11 of which can be accessed free at: http://symposium.cshlp.org/content/42/179.refs.html Email alerting Receive free email alerts when new articles cite this article - sign up in service the box at the top right corner of the article or click here

To subscribe to Cold Spring Harbor Symposia on Quantitative Biology go to: http://symposium.cshlp.org/subscriptions

Chromatin Structures of Main-band and Satellite DNAs in Drosophila melanogaster Nuclei as Probed by Photochemical Cross-linking of DNA with Trioxsalen

C.-K. J. SHEN AND J. E. HEARST Department of Chemistry, University of California, Berkeley, California 94720

Psoralens, a class of photochemotherapeutic trioxsalen. In particular, these experiments were agents used in the treatment of the skin disease performed to answer the following questions: (1) To psoriasis (Parrish et al. 1974), are important probes what extent can the purified D. melanogaster DNA for genome structures of living organisms because be saturated covalently with trioxsalen molecules? of their ability to covalently cross-link double helices Is there any difference in the saturation values as of DNA (Dall'Acqua et al. 1970; Cole 1970), RNA well as the distribution of cross-links among the (Isaacs et ai. 1977), and DNA-RNA hybrid helices main-band and satellite DNAs? (For a recent review (Shen et al. 1977 ). of D. melanogaster DNA components, see Endow et In the dark, psoralens intercalate into DNA hel- al. [1975].) (2) Since the trioxsalen molecules have ices (Dall'Acqua and Rodighiero 1966; Dall'Acqua been shown to stabilize hairpins or cruciforms et al. 1968). Upon irradiation with long-wavelength formed from inverted repeats (Cech and Pardue ultraviolet (UV) light (360 nm), psoralens react with 1976; Shen and Hearst 1976, 1977a), the question DNA, forming putative covalent cyclobutane bridges of whether such structures exist in the nuclei can between the 3,4- or 4',5'- double bond of the psoralen be addressed. Such hairpins might be unstable under and the 5,6- double bond of pyrimidine bases adja- the conditions of DNA isolation, but, after fixation cent to the intercalary psoralen (Musajo et al. by the cross-linking reaction, detection by electron 1967a,b). If the psoralen molecule intercalates be- microscopy might be feasible. (3) To what extent tween 2 base pairs in which the pyrimidines are are the five D. rnelanogaster DNA components pro- on opposite strands, and if one of the two pyrimi- tected from the photochemical reaction when they dines photoreacts with the 4',5'- double bond of the are in the interphase nuclei? Are the four simple- intercalary psoralen, then the second pyrimidine sequence DNAs protected in different ways than the can react covalently with the 3,4- double bond and complex bulk DNA? Are they different from each form a covalent interstrand cross-link (diadduct) other with respect to the distribution of cross-links? (Dall'Acqua et al. 1971). The answers to these questions provide evidence for Recently, the photochemical cross-linking reac- structural differences between heterochromatin and tion has been used to probe chromatin structure in euchromatin in eukaryotic cells. Drosophila melanogasternuclei (Hanson et al. 1976). By photoreacting the nuclei with the psoralen deriv- EXPERIMENTAL PROCEDURES ative, trioxsalen, and completely denaturing the in situ cross-linked DNA molecules, we have been able Isolation of nuclei and nuclear DNA from D. melano- to visualize the positions of the cross-links directly gaster embryos. D. melanogaster embryos 20-24 under the electron microscope. We found that most hours old were collected and washed with PBS of the DNA regions protected from the photochemi- buffer (0.15 M NaC1, 0.0075 M Na~HPO4, 0.0075 M cal cross-linking have a size of 180-200 base pairs, NaH2PO4, pH 7.0). The eggs were suspended in which is the length of DNA associated with the chro- NaC10 solution (2.5%) for 2 minutes to dechorionate matin subunit, the nucleosome (Hewish and Bur- the eggs. Dechorionated embryos were washed with goyne 1973; Axel et al. 1973; Olins and Olins 1974; PBS buffer and with 95% ethanol. They were then Kornberg 1974; Van Holde et al. 1974; Oudet et al. suspended in 0.3 M sucrose, 0.1 M NaC1, 0.01 M 1975). Subsequently, Wiesehahn et al. (1977) showed NaHSOs, 0.01 M Tris-HC1, pH 7.3, and homogenized that indeed most of these in situ cross-links form in a Potter Elvehjem tissue grinder with a motor- in the internucleosome regions on D. rnelanogaster driven Teflon pestle. The solution was centrifuged chromatin. Thus, the combination of in situ cross- at 1000g for 5 minutes and the resulting pellet was linking and denaturation electron microscopy serves resuspended in 0.3 M sucrose, 0.005 M Tris-HC1, pH as a unique method of preserving a linear record 7.0. It was then homogenized in a Dounce homoge- of the protected regions of DNA in nuclei or cells. nizer with a B pestle, and the nuclei were pelleted We will report here on detailed studies of the pho- at 1000g for 5 minutes. After one more resuspension tochemical reaction of D. melanogaster DNA with and homogenization in 0.3 M sucrose, 0.005 M Tris- 179 Downloaded from symposium.cshlp.org on May 16, 2009 - Published by Cold Spring Harbor Laboratory Press

180 SHEN AND HEARST

HC1, the nuclei were filtered through a nylon cloth Photochemical cross-linking olD. melanogaster DN.4 and pelleted at 1000g. inside nuclei. D. melanogaster nuclei isolated from DNA was isolated from nuclei, with or without 6 g of embryos were washed and resuspended in the cross-linking reaction, as described previously 36 ml of SSC solution made 3 pg/ml in SH-labeled (Shen and Hearst 1977b). trioxsalen (1 mg/ml stock in 100% ethanol, 71,000 The actinomycin D (Act D)-CsC1 density gradient cpm/pg) and irradiated with long-wavelength UV centrifugation of non-cross-linked DNA was per- light at 18~ for 8 hours in the UV apparatus de- formed as described before (Peacock et al. 1974; Shen scribed by Hanson et al. (1976). After each hour of and Hearst 1977b). For DNA isolated from cross- irradiation, 5-ml aliquots were taken out and 3 pg/ linked nuclei, 20 ml of a solution containing 640 ml of fresh trioxsalen was added to the reaction mix- tag DNA and 300 pg Act D was made 1.63 g/cc in ture for continued irradiation. After the photochem- CsC1 and centrifuged in one tube in a 60 Ti rotor ical reaction, samples of cross-linked nuclei were at 40,000 rpm for 60 hours. Pooled fractions from pelleted at 1000g and the DNA was isolated as de- the Act D-CsC1 density gradients were extracted scribed above. with n-butyl alcohol and dialyzed against TE buffer (0.01 M Tris-HCI, 0.001 M EDTA, pH 8.0). Determination of amount of covalently bound trioxsalen. ARer the cross-linking reaction, free triox- Preparative cesium-formate density gradients. The salen was removed from the DNA samples and the Cs-formate density gradient has been used to sepa- amount of covalently bound trioxsalen was deter- rate D. rnelanogaster DNA components (Shen et al. mined from the UV absorbance at 265 nm and 1976) because DNAs of different GC composition are tritium radioactivity as described previously (Wie- resolved in Cs-formate gradients better than in CsC1 sehahn et al. 1977; Isaacs et al. 1977). This method gradients (Hearst and Schmid 1973). In all cases, of calculation probably overestimates the amount 4 ml of 0.01 ~t Tris-HC1, 0.001 M EDTA (pH 8.0) of covalently bound trioxsalen per base pair. We solution containing 10-100 pg DNA was adjusted used 14C-labeled DNA as substrate for the photo- to the desired densities (non-cross-linked DNA, chemical reaction and found that the absorbance ~1.744 g/cc; in situ cross-linked DNA, 1.720 g/cc) of DNA at 265 nm could be suppressed as much by adding the appropriate amount of solid Cs-for- as 10% upon the covalent binding of trioxsalen mate (optical grade; Harshaw Chemical Co.). (C.-K. J. Shen and J. E. Hearst, unpubl.). The DNA solution in Cs-formate was then centrifuged to equilibrium in a Beckman angular 65 tube Electron microscopy. Native spreading was done at 40,000 rpm, 20~ for 60 hours. Fractions were in 40% formamide, 0.1 M Tris-HC1, 0.01 M EDTA pooled and dialyzed against TE buffer. (pH 8.4), according to Davis et al. (1971). To visualize the cross-links, DNA was totally denatured in buff- Determination of buoyant densities in the analytical ers containing formaldehyde and a high concentra- ultracentrifuge. Buoyant densities (p) of DNA in tion of formamide and then spread as described pre- neutral CsCI were determined by first adjusting the viously (Shen and Hearst 1977b). The sizes of the solutions to appropriate densities with CsC1 (Har- loops and double-stranded segments in the dena- shaw Chemical Co.) and then banding them in a tured molecules were measured and analyzed as Beckman Model E analytical ultracentrifuge at described by Wiesehahn et al. (1977). Single- 42,000 rpm, 20~ for 24 hours. Micrococcus lysodeik- stranded fd DNA has been used as the length marker t/cus DNA (p = 1.733 g/ee) was used as the density for both single-stranded and apparently double- marker in most eases. Sometimes the D. melano- stranded regions of denatured molecules. gaster main-band DNA (p = 1.701 g/ee) or satellite II (p = 1.686 g/ec) was used as a marker for cross- linked samples of very low buoyant densities. The RESULTS calculation of buoyant densities has been described before (Hearst and Sehmid 1973). Isolation of Main-band and Satellite DNAs of 1). melanogaster Photochemical cross-linking of purified DNAs of D. melanogaster. DNAs purified from the Act D-CsC1 It was first shown by Peacock et al. (1974) that and/or Cs-formate density gradients were each ad- four simple-sequence satellites could be resolved justed to a concentration of 10 gg/ml in SSC (0.15 from the bulk DNA of D. rnelanogaster by combined M NaC1, 0.015 M Na citrate, pH 8.0) and irradiated use of Act D-CsC1 and netropsin-CsC1 density gra- for 20 minutes, in the presence of 2 pg/ml SH-labeled dients. Ag+-Cs2SO4 and Cs-formate density gradients trioxsalen (sp. act. = 63,000 cpm/pg), with a long- have also been used to fractionate these satellite wavelength UV light with an intensity of 25 mW/ DNAs (Endow et al. 1975; Shen et al. 1976). In this cm 2 (Wiesehahn et al. 1977). Because of the photo- study we combined Act D-CsC1 and Cs-formate den- destruction of free trioxsalen (Isaacs et al. 1977), sity gradients and were able to get reasonably puri- repeated additions of the drug and irradiation were fied satellite and main-band DNAs. Figure la shows necessary (see Results). the equilibrium profile of total D. rnelanogaster Downloaded from symposium.cshlp.org on May 16, 2009 - Published by Cold Spring Harbor Laboratory Press

CROSS-LINKING OF D. MELANOGASTER DNA 181

(b) /%I.701 c) ~0.5~ 1.672

(d) ~ o oJ , (c) "6~c 1.0:~ (e) 7L j k._, ~ o.5 1.688

, i(d) i (f)

Density (gm/cc) 0 5 I0 15 20 ?5 30 Figure 2. Analytical ultracentrifugation of isolated D. mel- Froction Number anogaster DNA components in CsC1 density gradients. (a) Figure 1. Preparative density gradients used to isolate Total nuclear DNA; (b)main-band DNA (pooled fractions main-band and satellite DNAs from D. melanogaster 28-30 from Fig. la); (c) satellite I (pooled fractions 17-20 nuclei. (a) Act D-CsC1 density gradient of total D. melano- from Fig. lb); (d) satellite II (pooled fractions 13-15 from gaster nuclear DNA; (b) Cs-formate density gradient of Fig. lb); (e) satellite III (pooled fractions 11-13 from Fig. pooled fractions 18-21 from a; (c) Cs-formate density gra- ld); (/9 satellite IV (pooled fractions 13-16 from Fig. la). dient of pooled fractions 24-27 from a; (d)Cs-formate density gradient of pooled fractions 18-20 from c. Densities of the solutions before ultracentrifugation: (a) 1.65 g/cc; (b) 1.740 g/cc; (c) 1.747 g/cc; (d) 1.740 g/cc. Photochemical Cross-linking of Isolated D. melanogaster DNA Components The purffied main-band DNA (Fig. 2b) was photo- nuclear DNA in an Act D-CsC1 density gradient. chemically saturated with 3H-labeled trioxsaien by Fractions 13-16, which contained satellite IV, were adding 2 Ilg/ml of %I-labeled trioxsalen after each pooled and dialyzed against TE buffer. Fractions 18- irradiation period of 20 minutes. Figure 3a shows 21 were known to contain satellite I and satellite the number of base pairs per covalently bound triox- II. They were purified from the Act INCsC1 gradient, salen as a function of the number of additions of as described in Experimental Procedures, and put the drug. As can be seen, a saturation level of ap- into a Cs-formate density gradient (Fig. lb). The proximately 5 (base pairs/covalently bound triox- heavy peak (fractions 13-15) and light peak (frac- salen) is reached after six additions. tions 17-20) in Figure lb were pooled separately Samples of cross-linked main-band DNA after and dialyzed against TE buffer. The light side of each addition have been analyzed in the Model E the main peak (fractions 24-27, Fig. la) containing ultracentrifuge. It was found that the buoyant den- satellite III was pooled and put into two successive sity decreases as more trioxsalen molecules bind to Cs-formate gradients (Fig. lc,d). Fractions 11-13 of the main-band DNA, with a value of buoyant density the second Cs-formate gradient (Fig. ld) were pooled change of--0.162 g/cc per trioxsalen/base pair (Fig. and dialyzed against TE buffer. Fractions 28-30 of 3b). This magnitude is slightly lower than that for the Act D-CsC1 gradient (Fig. la) were pooled and satellite-IV DNA of D. melanogaster, for which we purified similarly to yield main-band DNA. found a value of--0.182 g/cc per trioxsalen/base The isolated main-band and satellite DNAs have pair (Shen and Hearst 1977b). been analyzed in the analytical ultracentrifuge (Fig. Because large amounts of purffied satellite DNAs 2). Their buoyant densities and equilibrium profiles were not available, the saturation curves have not in CsC1 are similar to those obtained before (Peacock been established for them except in the case of satel- et al. 1974; Endow et al. 1975; Shen et al. 1976). life IV (Shen and Hearst 1977b). However, we have Downloaded from symposium.cshlp.org on May 16, 2009 - Published by Cold Spring Harbor Laboratory Press

182 SHEN AND HEARST

2~ buoyant densities all decrease with the covalent binding of trioxsalen. Because only one datum is

"u.. 2C available for satellites I, II, or III, precise values of change of buoyant density per trioxsalen/base pair have not been obtained. ID =sl >.

o I0 Electron Microscopic Studies of In Vitro o Cross-linked D. Melanogaster DNAs

5 The five purified D. melanogaster DNA compo- #. nents extensively cross-linked in vitro were spread = for electron microscopy in 40% formamide and 0.1 o 0 0 nr, M Tris-HC1 (pH 8.4). They appeared on the grids Number of Additions of Trioxsolen as linear double-stranded molecules. After denaturation at 37~ in 70% formamide and 1.8% formal-

1,710 dehyde, satellite IV saturated with trioxsalen had (b) A mostly loops of sizes 250 +-50 and 500-+ 50 nucleo- 1.700 tides. The number-average size of the loops was 520 nucleotides. This indicates that the sites cross-linka- >,. 1.69C ble by trioxsalen are distributed nonrandomly and == that a long-range sequence order exists in this satel- ~" Lssc lite (Shen and Hearst 1977b). On the other hand, main-band DNA and satellites I, II, and III all ~ 1.67C showed complete double-strandedness with very few observable small loops after denaturation. This 1.66C ) .... o:o5 ' ' 'o: I0 ...... 0.15 0.20 means that the adjacent cross-links in these exten- Covolently Bound Trioxsolen Molecules per Bose Poir sively treated DNA components are so close to each Figure 3. (a) Covalent binding of trioxsalen molecules to other (less than 300/k) that they cannot be resolved D. rnelanogaster main-band DNA; (b) buoyant density of by the protein monolayer spreading technique. main-band DNA as a function of the amount of covalently bound trioxsalen molecules per base pair. The slope was calculated, using the least-squares method, to be --0.162 Photochemical Cross-linking g/cc per trioxsalen/base pair. of D. melanogaster Nuclei Isolated interphase nuclei were photochemically reacted satellites I, II, and III extensively by irradi- cross-linked with SH-labeled trioxsalen at 18~ Fig- ating each of them by the addition of 2 pg/ml ~H- ure 4 shows the extents of the photochemical reac- labeled trioxsalen every 20 minutes and then deter- tion expressed as base pairs per covalently bound mining the amounts of covalently bound trioxsalen trioxsalen for the seven DNA samples cross-linked after 140 minutes of irradiation. Note that this treat- in situ. It can be seen that DNA inside the nuclei ment was sufficient to allow the main-band DNA was saturated with trioxsalen after 5 hours of irradi- to react with trioxsalen to saturation (Fig. 3a). Table ation with continued addition of trioxsalen and that 1 lists the buoyant densities before and after the the saturation level is approximately 60-65 base extensive photochemical cross-linking reaction, as pairs per covalently bound trioxsalen. In comparison well as the binding (base pairs/covalently bound tri- with the saturation level of purified DNA in vitro, oxsalen) values for the five D. melanogaster DNA 92% of the covalent binding sites of D. melanogaster components. It is clear that all five DNA species DNA are protected from photochemical reaction react covalently with trioxsalen to a level of 3-5 with trioxsalen when the DNA is inside the intact base pairs per covalently bound drug, and their nuclei.

Table 1. Covalent Binding of Trioxsalen to D. melanogaster DNA Components In Vitro and Its Effect on Their Buoyant Densities in CsC1 Density Gradients Buoyant density Buoyant density before binding Base pairs per covalently after binding DNA component (g/cc) bound trioxsalen molecules (g/cc) Main band 1.701 5.0 1.669 Satellite I 1.672 3.4 1.655 Satellite II 1.686 4.4 1.649 Satellite HI 1.688 4.8 1.660 Satellite IV 1.705 3.0 1.646 Downloaded from symposium.cshlp.org on May 16, 2009 - Published by Cold Spring Harbor Laboratory Press

CROSS-LINKING OF D. MELANOGASTER DNA 183

450 (a) 1.700 1.733 40C

g 350 o .9~ 300 ! == -~ 250

c --~ 200 c o 1.686 S g ~50 JD (e) ~. 100 1.705 ; L ~ o o 0 m 50 Density (gm/cc) Figure 5. Analytical ultracentrifugation of in situ cross- linked D. melanogaster DNA components in CsCl density gradients. (a) Main-band DNA; (b) satellite I; (c) satellite DNA Somples If; (d) satellite III; (e)satellite IV. Figure 4. Covalent binding of trioxsaien molecules to D. melanogasterDNA in nuclei. For details of the experiment, see Experimental Procedures and Results. D-CsC1 and Cs-formate density gradients are similar to those of non-cross-linked DNA, and the isolated main-band and satellite DNAs cross-linked in situ Accessibility of Main-band and Satellite DNAs have been analyzed in the analytical ultracentrifuge to the Photochemical Reaction In Situ (Fig. 5). It is concluded that the five in situ cross- To determine the extent to which each DNA com- linked D. rnelanogaster DNA components have been ponent of D. melanogaster is protected from the in reasonably purified and most of them have lower situ photoreaction, we have prepared a large quan- buoyant densities in CsC1 than the native ones. This tity of DNA which had been cross-linked in situ to is most likely caused by the covalent binding oftriox- saturation with trioxsalen (62 base pairs/covalently salen to the DNA. bound trioxsalen). Because only 1.6% of the base From the absorbance at 265 nm and the DNA- pairs have reacted with trioxsalen, it is still possible associated radioactivity, the trioxsalen binding (base to subject this in situ cross-linked DNA to various pairs/covalently bound trioxsalen) has been esti- density gradients in order to purify the five DNA mated to be 62, 31, 39, 93, and 128 for main-band components of cross-linked D. melanogaster nuclei. and satellites I, II, III, and IV, respectively. Table We have used exactly the same procedures as shown 2 lists the buoyant densities and the percentage of in Figure 1 to isolate the in situ cross-linked main- covalent binding sites accessible in situ for these band and satellite DNAs, except that the starting five components. It is clear that satellites I and II densities of the buoyant solutions were lower. The are slightly more accessible to the photoreaction, resulting equilibrium profiles of the preparative Act whereas satellites III and IV are more protected in

Table 2. Saturation Binding of Trioxsalen to D. melanogoster DNA Components In Situ

Buoyant density after in situ In vitro bindin~ cro~-linkin~g Base pairs per covalently sites accessible DNA component (glcc) bound trioxsalen molecule in situ (%) Main band 1.700 62 8.1 Satellite I 1.662 31 11.0 Satellite II 1.679 39 11.3 Satellite III 1.686 93 5.2 Satellite IV 1.705 128 2.4 Downloaded from symposium.cshlp.org on May 16, 2009 - Published by Cold Spring Harbor Laboratory Press

184 SHEN AND HEARST situ than the main-band DNA. The accessibility of 5.0 satellite IV (2.4%) is fourfold lower than that of the main-band DNA (8.1%). 0 - Native and Denaturation Electron Microscopy (b) of In Situ Cross-linked D. melanogaster DNA When the five in situ cross-linked DNA components were spread in 40% formamide and 0.1 M Tris- HC1 (pH 8.4), only linear double-stranded molecules 5.C , ~j~d~ r~ were seen in each of the five samples. Denaturation microscopy has also been used to investigate the distribution of the interstrand cross- (c) links in chromatin. When in situ cross-linked D. melanogaster DNA samples which had been irradi- iz 10.0 ated for 2, 3, and 8 hours (Fig. 4) were denatured by formaldehyde and formamide, loops and apparently double-stranded regions were observed on all of the molecules. Examples of the denatured molecules are shown in Figure 6. The average size of the loops decreases with the irradiation, and the percentage of the apparently double-stranded regions increases. These data have been analyzed and 0 r , , , , ~ -~;'- n are presented in Figures 7 and 8. For unsaturated 200 4.00 600 800 I000 1200 14,00 samples, the distributions of the loop sizes (spacing Loop Size (Nucleotides) between adjacent cross-links) show broad peaks cen- Figure 7. Histograms of the loop sizes of denatured total tered at approximate multiples of 180-200 nucleo- D. melanogaster nuclear DNA samples that were photochemically cress-linked in situ for 2 hr (a), 3 hr (b), and 8 hr (c). Percentage of single-stranded regions in these denatured samples: (a) 93.8%; (b) 84.5%; (c) 72.3%.

tides (Fig. 7a,b). The DNA sample saturated in situ by trioxsalen has only two major peaks in the distribution (Fig. 7c). They occur at approximately 180 + 20 and 360 + 40 nucleotides, respectively. This nonrandom distribution of loop sizes in in situ cross- linked DNA is consistent with our previous results

3.0 ro 2.0 a~ 1.0

0 , n Hn , ,

,, otv % ,.ot 0 0 200 400 600rb-~r 800 I~ I000 ~ 12'00 ~ n IzlO0 Figure 6. Photographs of denatured total D. melanogaster nuclear DNA molecules that were photochemically cross- Double-Stranded Length (Bose Poirs) linked in situ for 2 hr (a), 3 hr (b), and 8 hr (c), respectively, Figure 8. Histograms of the lengths of double-stranded re- with trioxsalen. The bar represents the length of 1000 nu- gions in the denatured total D. rnelanogaster nuclear DNA cleotides. Circular single-stranded fd DNA was used as the samples that were cross-linked in situ for 2 hr (a), 3 hr length marker. (b), and 8 hr (c). Downloaded from symposium.cshlp.org on May 16, 2009 - Published by Cold Spring Harbor Laboratory Press

CROSS-LINKING OF D. MELANOGASTER DNA 185

20.0" (o)

15.0

I0.0 ,o\ Anlnnnn OO50 Cb

0 (c) I0,0

Figure 9. Photographs of denatured D. melanogasterDNA components saturated in situ with trioxsalen. (a) Satellite I; (b) satellite II; (c) satellite IV. The bar represents the length of 1000 nucleotides. The arrows in c point to the positions where the two strands of this particular satellite- 0 160 480 800 1120 IV DNA molecule are linked by trioxsalen molecules. Loop Size (Nucleotides) Figure lO. Histograms of the loop sizes of denatured D. melanogaster DNA components saturated in situ with tri- (Hanson et al. 1976) and suggests that most of the oxsalen. (a) Satellite I; (b) satellite II; (c) satellite III. Per- DNA inside nuclei is protected from the reaction centage of single-stranded regions: (a) 13.2%; (b) 22.9%; (c) 61.9%. of cross-link formation at regular intervals of 180- 200 base pairs. Approximately 28% of the in situ saturated DNA appeared to be double-stranded after denaturation, and there was no distinct pattern in also occur at a frequency of 180-200 base pairs. On the distribution of the sizes of these double-stranded the other hand, satellites I and II have quite distinct regions (Fig. 8a,b,c). loop-size distributions (Fig. 10a,b). A major peak of The five resolved DNA fractions cross-linked in sizes 100-140 base pairs is seen in satellite I (Fig. situ have also been spread under denaturing condi- 10a), whereas satellite II has two broad peaks located tions. The main-band DNA and satellite III have at 100-160 and 220-320 base pairs, respectively similar denaturation patterns to the total DNA (pho- (Fig. 10b). Because of the relatively long distance be- tographs not shown). The molecules of cross-linked tween every two adjacent cross-links, a loop-size and denatured satellites I and II have small loops distribution has not been measured for the in situ separated by apparent double-stranded regions, cross-linked satellite IV. Instead, the mean size of whose sizes range from one-hundred to a few thou- the loops has been calculated for this satellite. A sand base pairs (Fig. 9a,b). Most molecules in satel- value of approximately 10,000 nucleotides was ob- lite IV are long single-stranded DNA or DNA with tained from 30 denatured molecules. Since the num- very large loops (Fig. 9c). ber-average spacing between adjacent cross-links of satellite IV saturated in vitro with trioxsalen is 520 nucleotides (Shen and Hearst 1977b), only 5% of Distributions of lnterstrand Cross-links of the Five In Situ Cross-linked DNA Components the in vitro cross-linkable sites are accessible in situ. This is probably an overestimate because there are Figure 10 shows the loop-size distributions of the quite a few long single-stranded DNA molecules in in situ cross-linked DNAs. Main-band DNA (results the denatured sample of in situ cross-linked satellite not shown) as well as satellite III (Fig 10c) have IV. essentially the same distribution of cross-links as The size distributions of the double-stranded re- the total DNA. This indicates that the trioxsalen- gions have also been examined (Fig. 11). Of these, cross-linkable sites in the chromatin of satellite III main-band DNA and satellite III show random pat- Downloaded from symposium.cshlp.org on May 16, 2009 - Published by Cold Spring Harbor Laboratory Press

186 SHEN AND HEARST

62.b (o) or in vivo accessible sites of DNA to the photoreaction, i.e., the less-protected regions of DNA in the 5.(3 chromatin. Finally, the psoralen presumably sees the most native structures ofchromatin because the photoreaction is performed before the isolation of chromatin which may introduce appreciable altera- 0 ,~l'l , n , tions in the chromatin structure. (b) We have found in the control experiments that both main-band and satellite DNAs of D. rnelanogaster react covalently with trioxsalen to the extent 5.C of 3-5 base pairs per covalently bound trioxsalen. These values are close to the saturation values obtained for proflavin (Peacocke and Skerrett 1956), ethidium bromide (Waring 1965), and toluidine blue (Miura and Ohba 1967) during their primary binding to the DNA helix. It is very likely that most, if not all, of the covalently bound trioxsalen molecules (P photoreact with DNA by first intercalating between the base pairs, as has been proposed for other dye molecules (Lerman 1961) as well as for the psoralens (Dall'Acqua et al. 1968; Cole 1970; Isaacs et al. 1977). The denaturation microscopy showed that the average distance of adjacent cross-links in each of the four DNA components (except satellite IV) saturated in vitro by trioxsalen is less than 300/~. Photochemi- 0 500 I000 1500 :~000 2500 3000 3500 4000 cal cross-linking of satellite IV was studied previously (Shen and Hearst 1977b). It was shown that Double-Stranded Length [Base Pairs) trioxsalen-cross-linkable sites occur in satellite IV Figure 11. Histograms of the lengths of double-stranded at spacings of 250 +- 50 and 500 +- 50 base pairs, and regions in denatured D. melanogaster DNA components saturated in situ with trioxsalen. (a) Main-band DNA; (b) a model of satellite DNA evolution was proposed satellite I; (c) satellite II; (d) satellite III. to explain the regular distribution of cross-links in this satellite DNA. Our in situ cross-linking data of total D. melano- terns which can be approximated by the Poisson gaster DNA are consistent with the data from previ- distribution (Fig. lla,d). Denatured satellites I and ous studies (Hanson et al. 1976; Wiesehahn et al. II croswlinked in situ have much higher percentages 1977) in that 8-10% of the trioxsalen binding sites of double-strandedness: 87% and 77% for satellites on purified DNA are accessible to the photoreaction I and II, respectively. Their size distributions are in situ and the cross-links or clusters of cross-links shown in Figure 11 b and c. It can be seen that occur at spacings of 180-200 base pairs. We have beth satellites display in their distributions broad also found that extensively cross-linked nuclei could peaks that are centered approximately at 250, 500, be digested by micrococcal nuclease to yield specific 750 base pairs, etc. The possible significance of this DNA band patterns on polyacrylamide gels, similar periodicity will be discussed below. to those of non-cross-linked nuclei (Wiesehahn et al. 1977; C.-K. J. Shen and J. E. Hearst, unpubl.). This suggests that the extensive photoreaction did DISCUSSION not alter appreciably the chromatin structure, in The structures of isolated chromatin or chromatin the sense that its accessibility to the micrococcal inside nuclei have been studied by a variety of ap- nuclease has been preserved. However, the satura- proaches, such as nuclease digestion (Clark and Fel- tion-binding value of cross-linked nuclei we have ob- senfeld 1971), methylation in vitro (Bloch and Cedar tained (one covalently bound trioxsalen per 60-65 1976), and noncovalent binding of dye molecules (Mi- base pairs) is about 50% lower than that found by ura and Ohba 1967; Lurquin and Seligy 1972; Brodie Wiesehahn et al. (1977) (one trioxsalen/40 base et al. 1975). The photochemical cross-linking reac- pairs). One or several of the following possibilities tion of DNA by trioxsaien possesses several charac- may be responsible for the difference. First, the de- teristics that make it a useful probe for the chroma- tails of the isolation procedures for the nuclei are tin structure in situ or in vivo. First, this small different. Second, Wiesehahn et al. (1977) repeat- molecule penetrates cells or nuclei very easily. Sec- edly pelleted the nuclei during irradiation, whereas ond, the formation of covalent monoadducts or diad- we constantly agitated the solution of nuclei in ducts (interstrand cross-links) to pyrimidines of a rocking chamber without any pelleting step. DNA generates a permanent record of the in situ Third, a much higher intensity of light was used Downloaded from symposium.cshlp.org on May 16, 2009 - Published by Cold Spring Harbor Laboratory Press

CROSS-LINKING OF D. MELANOGASTER DNA 187 by Wiesehahn et al. (1977) to irradiate the nuclei. ofnucleosomes, our data suggest that trioxsalen can All of these differences could have caused different cross-link the intranucleosomal DNA of satellites extents of conformational alteration and/or migra- I and II easier than the main-band DNA since 87% tion of histories in the chromatin and have thus and 77% of the in situ cross-linked satellites I and modified the accessibility ofchromatin to the photo- II, respectively, appeared double-stranded after de- reaction. naturation. One or both of the following two possibil- Wiesehahn et al. (1977) have demonstrated that ities could explain the difference between satellites in situ the photoreaction occurs preferentially in I and II and the main-band DNA. First, the nucleo- the internucleosomal regions. This would mean that somes of satellites I and II may have different struc- the 180-200-nucleotide-long loops in the in situ tures such that trioxsalen molecules are able to pho- cross-linked and in vitro denatured molecules are toreact with the intranucleosomal DNA much eas- the segment of DNA protected by the histone cores. ier. Second, satellites I and II may have a much The reason that the photoreaction hardly occurs in higher frequency of cross-linkable sites, perhaps be- the intranucleosomal DNA is not clear. We have cause of the large number of alternating AT base shown that trioxsalen can also efficiently cross-link pairs in these satellites relative to main-band DNA. RNA helices (Isaacs et al. 1977) and DNA-RNA hel- Thus, at the low level of photobinding (one cova- ices (Shen et al. 1977) in solution. This suggests that lently bound trioxsalen per 30-40 base pairs for in trioxsalen is not able to differentiate among DNA, situ cross-linked satellites I and II), essentially all RNA, and DNA-RNA helices because of the vibra- covalently bound trioxsalen may exist in the form tional freedom of these helices in solution. Thus, ofdiadducts (interstrand cross-links). Another inter- one possible explanation for the specificity of react- esting feature of the in situ cross-linked satellites ing sites would be that the DNA in the nucleosomes I and II is that they both have heavily cross-linked possesses a rigid secondary structure in which the DNA regions with length multiples of approxi- double bonds of the intercalary trioxsalen and its mately 250 base pairs. These results imply that the neighboring pyrimidines cannot adjust themselves chromatins of satellites I and II are composed of to the right orientations for the cyclobutane-forming subunits containing 250 base pairs of DNA that are cycloaddition to occur. Alternatively, the blocking more easily cross-linked than the nucleosomes con- of the photoreaction could happen at the intercala- taining the main-band DNA. Since the length of tion step. Richards and Pardon (1970) have sug- their intranucleosomal (core) DNA is the same as gested that the amino-terminal tails of the histones that of main-band (140 base pairs) DNA, the chroma- interact strongly with the phosphates in the grooves tins of satellites I and II must have longer internu- of DNA helices. They could prevent the trioxsalen cleosomal spacings (110 base pairs) than that of from getting in between the base pairs. Approxi- main-band (40-60 base pairs). These regions could mately 30% of the in situ cross-linked DNA (27.7% be associated with H1 histones and/or specific non- in total DNA and 33% in main-band DNA) appeared histone proteins that are responsible for the packag- double-stranded and therefore unprotected. The dis- ing of satellite DNA I and II in heterochromatin. tances between the adjacent cross-links in these un- The broadness in the distributions of both the loops protected regions are probably shorter than 300/~, and double-stranded regions reflects the heterogene- the lower limit of loop resolution using the cyto- ity of the inter- and intranucleosomal regions with chrome c monolayer spreading technique. These respect to their accessibility to the cross-link forma- double-stranded regions may represent either the tion. The 120- and 250-base-pair loops may arise DNA devoid of proteins inside the isolated nuclei from a class of nucleosomes with a low frequency or the chromatin portions where chromosomal pro- of intranucleosomal cross-linking. Alternatively, teins do not protect DNA effectively from the they could represent regions where specific chromo- photoreaction. somal proteins, other than histories, protect the The four nuclear satellite DNAs of D. melano- DNA from the photoreaction. gaster have been shown to be highly repetitive and Both the binding values (covalently bound triox- located in the centric heterochromatin of the chro- salen/base pair) and the average distance between mosomes (Rae 1970; Gall et al. 1971; Peacock et al. adjacent cross-links of in situ cross-linked satellite 1974; Goldring et al. 1975). Recently, T. Nelson and IV indicate that, in nuclei, this satellite is inhibited D. Brutlag (pers. comm.) have demonstrated, by mi- with respect to photoreaction severalfold more than crococcal nuclease digestion and Southern's gel-hy- either the main-band DNA or other satellites. Since bridization technique (Southern 1975), that the chro- it is also in the nucleosomal structure, as demon- matins of satellites I, II, and IV are all composed strated by the micrococcal nuclease digestion experi- of nucleosomal subunits. In the limit digests, the ment (T. Nelson and D. Brutlag, pers. comm.), a spe- DNA sizes of their nucleosome cores are the same cific class of nonhistone protein may interact with as that of the bulk DNA (T. Nelson and D. Brutlag, satellite IV in situ and protect most of its internu- pers. comm.), which has a core DNA approximately cleosomal regions from the photoreaction. Our un- 140 base pairs long (G. Wiesehahn and J. E. Hearst, published results have shown that satellite III iso- unpubl.). If these three satellites are also composed lated from cross-linked nuclei could also be cleaved Downloaded from symposium.cshlp.org on May 16, 2009 - Published by Cold Spring Harbor Laboratory Press

188 SHEN AND HEARST

COLE, R. S. 1970. Light induced crosslinking of DNA in by HadII or Hinf restriction enzymes into 350-365- I . base-pair fragments (Manteuil et al. 1975; Shen et the presence of a furoceumarm (psoralen). Biochim. Biophys. Acta 217: 30. al. 1976). Although there are no other experimental DALL'AcQuA, F. and G. RODIGHIERO. 1966. Changes in the results indicating that satellite III is in the nucleoso- melting curve of DNA al%er the photoreaction with mal structure, our in situ data do show that the skin-photosensitizing furocoumarins. Re. Accad. Naz. cross-links formed on satellite III inside the nuclei Lincei 40: 411. are at intervals of 180-200 base pairs. This is evi- DALL'AcQuA, F., S. MARCIANI,and G. RODIGHIERO. 1970. Interstrand crosslinkages occurring in the photoreac- dence suggesting the following: (1) The chromatin tion between psoralen and DNA. FEBS Lett. 9: 121. of satellite III is also composed of subunits contain- DALL'ACQuA,F., M. TERBOJEVICH,and G. RODIGHIERO.1968. ing the histone cores and DNA segments of 180- Light scattering and viscosimetric studies on DNA after 200 base pairs. (2) The internucleosomal DNA of the photoreaction with some furocoumarins. Z. Natur- forsch. 23b: 943. satellite-III chromatin may be as accessible to the DALL'AcquA, F., S. MARCIANI, L. CIAUATrA, and G. RODI- photoreaction as is the bulk DNA chromatin. GmERO. 1971. Formation of inter-strand cross-linkings Satellite DNAs I, II, and IV of D. melanogaster in the photoreactions between furocoumarins and DNA. consist of tandem repeats of simple sequences Z. Naturforsch. 26b: 561. (Peacock et al. 1974; Brutlag and Peacock 1975; DAvm, R. W., M. SIMON, and N. DAVIDSON. 1971. Electron microscope heteroduplex methods for mapping regions Endow et al. 1975). The structure of the nucleosomes of base sequence homology in nucleic acids. Methods containing satellite DNA may be different from that Enzymol. 21: 413. of a random-sequence DNA because of the repetitive- ENDOW, S. A., M. L. POLAN, and J. G. GALL. 1975. Satellite ness of the satellite DNAs. Our experiments do not DNA sequences of Drosophila rnelanogaster. J. Mol. Biol. 96: 665. resolve the details of such a difference, although the GALL,J. G., E. H. COH~.N, and M. L. POLAN. 1971. Repetitive different cross-linking patterns suggest that such dif- DNA sequences in Drosophila. Chromosoma 33: 319. ferences do exist. GOLDmNG, E. S., D. BRtrrLAG, and W. J. PF~COCK. 1975. Arrangement of the highly repeated DNA of Drosophila melanogaster. In The eukaryote chromosome (ed. W. J. Acknowledgments Peacock et al.), p. 47. Australian National University Press, Canberra. We thank Gary Wiesehahn and Drs. Thomas R. HANSON, C. V., C.-K. J. SHZN, and J. E. HF~RST. 1976. Cech and John E. Hyde for many stimttlating discus- Crosslinking of DNA in situ as a probe for chromatin structure. Sc/ence 193: 62. sions and criticisms, Gary Wiesehahn and Dr. Lesly HEARST, J. E. and C. W. SCHMm. 1973. Density gradient Hallick for being extremely helpful in the analysis sedimentation equilibrium. Methods Enzymol. 27: 111. of the data, and Dr. Ignacio Tinoco, Jr. for gener- H~.WISH, D. R. and L. A. BURGOYNE. 1973. The digestion ously letting us use the PDP 8/E computer. We also of chromatin DNA at regular spaced sites by a nuclear thank T. Nelson and Dr. Douglas L. Brutlag for the deoxyribenuclease. Biochem. Biophys. Res. Commu~ 52: 504. communication of their unpublished results. This ISAACS, S., C.-K. J. SHEN, J. E. HEARST, and H. RAPOPORT. work was supported by American Cancer Society 1977. Synthesis and characterization of new psoralen Grant NP-185, National Institutes of Health Grant derivatives with superior photoreactivity with DNA GM 11180, and National Science Foundation Grant and RNA. Biochemistry 16: 1058. KORNB~.RG, R. D. 1974. Chromatin structure: A repeating GB 36799. C.-K. J. S. is a recipient of the Earle C. unit of histones and DNA. Science 184: 868. Anthony Fellowship from the University of LERMAN, L. S. 1961. Structural considerations in the inter- California. action of DNA and acridines. J. Mol. Biol. 3: 18. LURQUIN, P. F. and V. L. SELmY. 1972. Binding ofethidium bromide to avian erythrocyte chromatin. Biochem. Bio- REFERENCES phys. Res. Commun. 46: 1399. MANTEUIL, S., D. H. HAMER, AND C. A. THOMAS, JR. 1975. AXEL, R., W. MELCHIOR, B. SOLLNER-WEBB, and G. FEI~EN- Regular arrangement of restriction sites in Drosophila FELD. 1974. Specificsites of interaction between histones DNA. Cell 5: 413. and DNA in chromatin. Proc. Natl. Acad. Sci. 71: 4101. MIURA, A. ANn Y. OHBA. 1967. Structure of nucleohistone. BLOCH, S. and H. CEDAR. 1976. Methylation of chromatin III. Interaction with toluidine blue. Biochim. Biophys. DNA. Nucleic Acids Res. 3: 1507. Acta 145: 436. BROmE, S., J. GmON, and S. A. LATT. 1975. Estimation of MUSAJO, L., F. BORDIN, and R. BEWLAeQUA. 1967a. Photo- accessibility of DNA in chromatin from fluorescence reactions at 3655/k linking the 3-4 double bond of furo- measurements of electronic excitation energy transfer. coumarins with pyrimidine bases. Photochem. Photo- Nature 253: 470. biol. 6: 927. BRtrrLAG, D. and W. J. PEACOCK. 1975. Sequence of highly MusAJO, L., F. BORDIN, S. CAPORALE, S. MARCIANI, and repeated DNA in Drosophila melanogaster. In The eu- G. RmATrL 1967b. Photoreactions at 3655 ~ between karyote chromosome (ed. W. J. Peacock et al.), p. 35. pyrimidine bases and skin-photosensitizing furocouma- Australian National University Press, Canberra. tins. Photochem. Photobiol. 6: 711. CLARK,R. J. and G. FELSEN~LD. 1971. Structure ofchroma- OLINS, A. L. and D. E. OL~NS. 1974. Spheroid chromatin tin. Nat. New Biol. 229: 101. units (v bodies). Science 183: 330. C~.CH, T. R. and M. L. PARDUE. 1976. Electron microscopy OUD~T, P., M. GROss-BELLARD,and P. CHAMBON. 1975. Elec- of DNA crosslinked with trimethylpsoralen: A test of tron microscopic and biochemical evidence that chro- the secondary structure of eukaryotic inverted repeat matin structure is a repeating unit. Cell 4: 281. sequence. Proc. Natl. Acad. Sci. 73: 2644. PARRISH, J. A., T. B. FITZPATRICK,L. TANENBAUM, and Downloaded from symposium.cshlp.org on May 16, 2009 - Published by Cold Spring Harbor Laboratory Press

CROSS-LINKING OF D. MELANOGASTER DNA 189

M. A. PATHAK. 1974. Photochemotherapy of psoriasis __. 1977b. Detection of long-range sequence order in with oral methoxsalen and longwave ultravioletlight. Drosophila melanogaster satellite DNA IV by a photo- N. Engl. J. Med. 291: 206. chemical crosslinking reaction and denaturation mi- PEACOCK, W. J., D. BRUTLAG. E. GOLDRING, S. APPELS, croscopy. J. Mol. Biol. 112: 495. C. W. HmTON, and D. L. LINIkqLEY. 1974. The organiza- SHEN, C.-K. J., G. WmSZHAHN, and J. E. HEARST. 1976. tion of highly repeated DNA sequences in Drosophila Cleavage patterns of Drosophila melanogaster satellite melanogaster chromosomes. Cold Spring Harbor Syrup. DNA by restriction enzymes. Nucleic Acids Res. 3: 931. Quant. Biol. 38: 405. SHEN, C.-K. J., T. S. Hsimi, J. C. WANG, and J. E. HEARST. PEACOCKE, A. R. and J. N. H. SKzRgs'rr. 1956. The interac- 1977. Photochemical cross-linking of DNA-RNA helices tion of aminoacridines with nucleic acids. Trans. Fara- by psoralen derivatives. J. Mol. Biol. 116: (in press). day Soc. 52: 261. SOUTHERN, E. M. 1975. Detection of specific sequences RAE, P. M. M. 1970. Chromosomal distribution of rapidly among DNA fragments separated by gel electrophore- reannealing DNA in Drosophila melanogaster. Proc. sis. J. Mol. Biol. 98: 503. Natl. Acad. Sci. 67: 1018. VAN HOLDE, K. E., C. G. SAHASRABUDDHE, and B. R. SHAW. RICHARDS, B. M. and J. F. PARDON. 1970. The molecular 1974. A model for particulate structure in chromatin. structure of nucleohistone (DNH). Exp. Cell Res. 62: Nucleic Acids Res. 1: 1579. 184. WARING, S. J. 1965. Complex formation between ethidium SHEN, C.-K. J. and J. E. HEARST. 1976. Psoralen crosslinked bromide and nucleic acids. J. Mol. Biol. 13: 269. secondary structure map of single-stranded virus DNA. WmSEHAHN, G., J. E. HYDE, and J. E. HEARST. 1977. The Proc. Natl. Acad. Sci. 73: 2649. photoaddition oftrimethylpsoralen to Drosophila mela- __. 1977a. Mapping of sequences with twofold symme- nogaster nuclei: A probe for chromatin substructure. try on the simian virus 40 genome -- A photochemical Biochemistry 16: 925. crosslinking approach. Proc. Natl. Acad. Sci. 74: 1363.