Noncovalent intermolecular crosslinks are produced by bleomycin ...

2 downloads 32 Views 2MB Size Report
Mar 19, 1979 - Noncovalent intermolecular crosslinks are produced by bleomycin reaction with duplex DNA. (covalently closed circular DNA/DNA-reactive ...
Proc. Nati. Acad. Sci. USA Vol. 76, No. 6, pp. 2674-2678, June 1979 Biochemistry

Noncovalent intermolecular crosslinks are produced by bleomycin reaction with duplex DNA (covalently closed circular DNA/DNA-reactive chemicals/anticancer drug/electron microscopy/antibiotics)

R. STEPHEN LLOYD, CHARLES W. HAIDLE, AND DONALD L. ROBBERSON Department of Biology, The University of Texas System Cancer Center, M. D. Anderson Hospital and Tumor Institute, Houston, Texas 77030

Communicated by Norman Davidson, March 19, 1979

ABSTRACT Reaction of covalently closed circular PM2 bacteriophage DNA with-the anticancer drug bleomycin produces nicked circular (form II) and linear duplex (form III) DNA [Lloyd, R. S., Haidle, C. W. & Robberson, D. L. (1978) Biochemistry 17, 1890-18961. As the reaction proceeds, the frequencies of both form II and form III DNA increase and, concomitantly, an increasing fraction of the DNA mass is found to be in crosslinked structures. Approximately 16% of the PM2 DNA mass is found to be crosslinked after 30 min of reaction with bleomycin at 0.5.ig/ml. ITe proportion of each form found in any given crosslinked structure is directly related to the concentration of uncrosslinked (monomeric) forms. Multiple sites of crosslinking occur, and these frequently extend over a region of approximately 500 nucleotide pairs. The intermolecular crosslinked bonds are dissociated by extensive dialysis or by the addition of salt at high concentration (0.8 M NaCQ), as would be expected if the bonds were noncovalent. Becgpse intramolecular covalent crosslinks between complementary strands are not detected, it is suggested that intermolecular crosslinks are formed by noncovalent association of bleomycin molecules bound to each of the forms of DNA. Various chemical agents such as psoralen, mitomycin C, nitrous acid, and trans-diamminedichloroplatinum (II) are known to produce covalent intramolecular crosslinks in DNA and possibly intermolecular crosslinks between DNA and proteins (1-4). Several of these reagents react with DNA to produce major distortions of the DNA helix structure which in turn are expected to elicit a corrective response by cellular DNA repair systems. By way of contrast, biologically significant noncovalent intermolecular crosslinks in DNA could occur intracellularly. Such crosslinks might present barriers to proper transcription or organization of the genome, yet be undetected by cellular repair systems because they likely would produce little if any helix distortion. One such pattern of noncovalent intermolecular crosslinking of DNA structure occurs when the glycopeptide antibiotic bleomycin reacts with bacteriophage PM2 DNA. MATERIALS AND METHODS Bleomycin Treatment of PM2 DNA and Electrophoresis Analyses. Covalently closed circular PM2 DNA was the generous gift of R. R. Hewitt and was prepared by the procedure described by Salditt et al. (5) as modified by Strong and Hewitt (6). The form I DNA at a final concentration of 501Ag/ml was treated with bleomycin (0.5 Ag/ml) in a total volume of 27 ,l of a solution containing 0.27 mM CaCl2, 25 mM 2-mercaptoethanol, and 20 mM Tris at pH 8.0. The reaction mixtures were incubated for increasing times at 370C and the reaction was stopped by addition of either 30 ,Al of 20 mM Na2EDTA/50 mM Tris, pH 7.8, for examination by electron microscopy or 30 Al of 20 mM Na2 EDTA/50 mM Tris, pH 7.8/10% (wt/vol) sucrose/0.025% bromphenol blue for electrophoresis.

The three forms of DNA were separated by electrophoresis in 1.4% agarose tube gels, 10 cm in length. Electrophoresis was conducted for 5.5 hr at 100 V and 2.5 mA per gel at a temperature of approximately 200C. The mass fraction of each DNA form was determined by scanning spectrofluorometry after staining with ethidium bromide (0.5 pg/ml) for 4 hr (7). Extraction of DNA from high molecular weight DNA bands was performed by two methods. DNA was extracted from agarose gel slices by electroelution as described (8). DNA was also extracted from gel slices by a "freeze-squeeze" technique (Douglas L. Vizard, personal communication). Gel slices were frozen between folds of Parafilm at -20°C for 20 min. The slices were removed and thumb pressure was applied to extrude the solvent and DNA from the collapsed gel. These samples were then prepared for electron microscopy (see below). Electron Microscopy. Samples of bleomycin-treated or untreated PM2 DNA were prepared for electron microscopy by the aqueous basic protein/Kleinschmidt technique described by Davis et al. (9). The grids were rotary-shadowed with Pt/Pd (80:20) and examined in a Philips 300 electron microscope. Sedimentation Analyses. Tritium-labeled PM2 DNA with a specific activity of 20,000 cpm/,g was prepared as described (5) and was the generous gift of Horace B. Gray, Jr. The DNA (3.0 Ag) was treated with bleomycin (0.5 ,g/ml) for 15 min exactly as described above and sedimented through a linear sucrose gradient (5-20%) containing 10 mM EDTA, 50 mM Tris (pH 8.0), and bleomycin at 0.5 'ig/ml. Sedimentation was performed for 140 min at 50,000 rpm and 20°C in the Beckman SW 56 Ti rotor. Fractions (80 gAl) were collected from the top of the gradient by using a Buchler Auto Densi-flow. Untreated samples of DNA were run in parallel gradients. Thermal Denaturation of Bleomycin Crosslinke4 DNA. Samples of PM2 DNA treated with bleomycin were dialyzed against 1 liter of 0.1 M NaCl/0.05 M Tris, pH 8.5/0. A1 M Na2 EDTA (STE). After dialysis, the DNA solutions were heated in boiling water for 30 sec and then immediately immersed in ice water at 0°C for 2 min. The total volume of solution was 50 Al, and the DNA concentration was 2.5 ,g/ml. Samples were prepared for electron microscopy as described above. RESULTS Comparison of Eleetrophoretic with Electron Microscopic Analyses of Bleomycin Reaction Products. When bleomycin reacts with covalently closed circular (form I) PM2 DNA, there is a time-dependent conversion to nicked circular (form II) and linear duplex (form III) DNAs (7). These three forms of DNA can be separated by agarose gel electrophoresis (Fig. IA inset) and the mass fractio1 of each form can be determined by scanning spectrofluorometry (Fig. 1). The mass fraction of form I DNA, as a function of reaction time, determined by this method was consistently lower than the mass fraction estimated

The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U. S. C. §1734 solely to indicate this fact.

Abbreviation: STE, 0.1 M NaCI/0.05 M Tris, pH 8.5/0.01 M Na2EDTA.

2674

Biochemistry: Lloyd et al.

Proc. Natl. Acad. Sci. USA 76 (1979)

2675

z--

o

0.4-

EC.

CU~~~~~~~~~~~~~~~~~~~~~~~~~~~

9-

0o

5

1

0

15

20

25

30 0 5 Reaction time, min

10

15

20

25

30

FIG. 1. Comparison of gel electrophoresis (solid symbols) and electron microscopic analyses (open symbols) of bleomycin reaction with PM2 DNA. The mass fraction of form I PM2 DNA as a function of reaction time with bleomycin is shown in A and the corresponding formation of form II (A, A) and form III (*, 3) DNAs is shown in B. Insets in A illustrate electrophoretic separation of PM2 DNA forms I, II, and III not treated with bleomycin and the appearance of form II PM2 DNA internally crosslinked to an extent that would lead to misclassification as a form I DNA molecule. Inset in B illustrates the appearance of form II (top) and form III (bottom) PM2 DNAs. Electron microscopic analyses were performed prior to electrophoresis.

by electron microscopy (Fig. 1A). Between 500 and 800 DNA molecules were scored for each time point. The mass fractions of form II DNA estimated by spectrofluorometry were consistently higher than those determined by electron microscopy (Fig. 1B). The mass fractions of form III DNA were essentially the same by both methods except for the 15-min sample. In estimating the mass fraction of form I DNA by electron microscopy, we applied the criterion that a circular molecule must have three or more crossovers of the duplex strands to be classified as being covalently closed circular. This criterion is based on results of independent experiments in which covalently closed circular PM2 DNA was completely nicked with DNase I and then examined by the same aqueous Kleinschmidt technique for electron microscopy. When this was done and 577 molecules were classified, 88.6% had no crossovers of duplex strands, 8.8% had one, 2.1% had two, and 0.5% had three or more. Because this criterion appears to provide a valid estimate of the fraction of covalently closed circular DNA, the results of these analyses suggest that some form II DNA may have been assigned as form I DNA by electron microscopy. This could occur if some form II DNA molecules contained intramolecular crosslinks. Such contacts arising from crosslinking of duplex strands at different sites on the same circular molecule would be scored as crossovers in the procedure applied here. An example of the appearance of one such molecule is shown in Fig. 1A inset. There is no confusion in classification of linear duplex forms (Fig. 1B) because two termini can be identified by electron microscopy, and one expects good agreement for estimates of the mass fractions of this form of DNA determined by either gel electrophoresis or electron microscopy (Fig. 1B).

expected for a simple crossing-over of duplex strands (with diameters of approximately 20 nm) as occurs in the case of catenation. In addition, the molecules frequently appeared to be associated at more than one site (Figs. 2 and 3). The preparation of closed circular PM2 DNA utilized in these reactions contains 2.7% of the mass as dimer catenanes which probably represent contamination by the phage isolated with intracellular PM2 DNA.

The frequency and mass fraction of DNA that is associated in oligomers increased with increasing time of reaction with bleomycin above the initial concentration of catenanes (Fig. 4A). Between 500 and 800 DNA molecules were scored for each time point. Because individual duplex molecules will not be observed in contact with each other in the Kleinschmidt procedure applied here (9), we conclude that these associations reflect a crosslinking reaction mediated by bleomycin. The frequencies and mass fractions of oligomeric forms consisting of three or more molecules also increased with increasing time of reaction with bleomycin (Fig. 4B); oligomers consisting of three or more molecules were not detectable in the unreacted

\~~~~~LN ~ ~ ~

~

Bleomycin Reaction with PM2 DNA Produces Intermo-

lecular Crosslinks. Examination of the bleomycin reaction mixtures by electron microscopy revealed an increased frequency of apparent catenations or associations between each of the three forms of DNA (Fig. 2). In addition to associations between two molecules, more complex structures were observed involving three or more molecules of PM2 DNA (Fig. 3). The region of physical contact between associated molecules (indicated by arrows in Fig. 2) was often considerably greater than

FIG. 2. Electron micrographs illustrating the appearance of bleomycin-crosslinked dimers of PM2 DNA. Topological forms I, II,

and III can be found in crosslinked (x) structures as indicated. Arrows indicate regions of extensive crosslinking. Bar represents 1lym.

Bioc'hemistry: Lloyd et al.

2676

Proc. Natl. Acad. Sci. USA 76 (1979) 0.8

II II / 2)

A

B

.II

Zc

c 0.6

.o I,

0

t

/

0 0.4

U -Y

0

T

-/-0

a II

III

a:

/ 0 / .

12 .

0

I

J

IT

X~~~~~~~~~~~

0.21

/

.1

- J

.

L'1 /

II x IL--.

FIG. 3. Electron micrographs illustrating the appearance of bleomycin-crosslinked oligomners containing more than two PM2 DNA molecules. Topological forms of DNA involved in individual crosslinked structures are indicated. Bar represents 1 im.

material. The sigmoid form of the curves in Fig. 4 probably reflects a bimolecular reaction mechanism for the crosslinking events. In this regard, it is found that the mass fractions of each form of bleomycin-treated DNA that remained uncrosslinked (monomeric forms) as a function of time (Fig. SB) parallelled the proportions of these different forms observed in the crosslinked structures (Fig. 5A). Thus, the probability that a particular DNA form will be found in any given crosslinked structure is directly related to the proportion of that respective uncrosslinked form relative to the other uncrosslinked forms. I3H]Thymidine-labeled PM2 DNA was reacted with bleomycin for 15 min and the DNA forms were resolved by velocity sedimentation in sucrose gradients containing bleomycin (Fig. 6A). The DNA, initially containing 85% form I DNA, was converted to a high percentage of form II DNA. The bleomycin in the gradient did not promote DNA strand scission because it was inactivated by the high EDTA concentration in the medium. In addition to a residual form I DNA peak, a peak of DNA sedimenting more slowly than form II DNA was partially resolved. Aliquots of DNA from selected fractions across the gradient were prepared for electron microscopy and scored for the presence of crosslinked species. Crosslinked species were found to be enriched in fractions at the trailing edge of the form II DNA (Fig. 6B). These fractions directly correspond to the slowest sedimenting peak. In addition to dimer crosslinked molecules, higher oligomers up to and including pentamers were observed in this portion of the gradient. The crosslinked species were partially separated into

0.6

oa

5

X

--o

E

\\ / Form II

0.4

0

Form

Cu0.2 O

Fomr

10 15 20 25 30 0O 5 Reaction time, min

,

0 16

C

30

families of molecules in which oligomers involving form I molecules sedimented more slowly than those containing only form II and III DNAs. Bands of higher molecular weight DNA presumably representing crosslinked species were also detected by agarose gel electrophoresis with or without inclusion of bleomycin in the gels and electrophoresis buffer (Fig. 6C). The DNA in the four higher molecular weight bands above the form II band in Fig. 6C were extracted either by electroelution or by a "freezesqueeze" technique and immediately examined by electron microscopy. Several attempts to purify crosslinked species by these procedures yielded only monomer species of forms I, II, and III DNAs with approximately 5% of the DNA mass in each band appearing as multiple-length linear molecules. Multiple Sibs of Crosslinking Occur in Bleomycin-Treated PM2 DNAs. After 5 min of bleomycin reaction, approximately 11% of the total PM2 DNA mass was noted by electron microscopy to be crosslinked. Examination of the individual crosslinked structures such as those shown in Figs. 2 and 3, leadE to a minimal estimate of the corresponding number of crosslinks. Physical contacts between or within molecules that extend a distance corresponding to the combined diameters of the strands (40 nm) were scored as single crosslinks. F rrr 1

A

F- orm

rf

a

b 4

IK

3

S.

2

2

FI I

0.

6

V,

el'

"

I~~~

__.o_1

A

20 25

FIG. 5. (A) Relative proportions (no. of molecules of specific form in oligomers/total no. of molecules in oligomers) of forms I (0), 11 (A ), and III (3) of PM2 DNA observed in crosslinked structures as a function of reaction time with bleomycin. (B) Mass fraction of uncrosslinked forms (from top to bottom at zero time, form I, form II, and form III) at corresponding times.

0 0.12 * >

10 15

-

F orm,

-

Form

III

R3

B

0 00

:30.14 .~0.12. 0.10

C

M c

1 I6 20 24 28 32 36 40 44 H8 r2

0.08

Fra i-ior t0.06

~~0.01 ~~~~ 0'

20 20

00 0

56

_10

_15

20

25

30

Reaction time, min

FIG. 4. Mass fraction (0) of PM2 DNA in crosslinked oligomers and the frequency (0) of oligomers as a function of reaction time with bleomycin. (A) Total oligomers. (B) Oligomers containing three or more molecules. Note that same numbers apply to both variables in each panel.

FiCT. 6. Sedimentation and electrophoretic analyses of bleomycin-crosslinked species. (A) Velocity sedimentation of [3H]thymidine-labeled PM2 DNA (3,ug) treated with bleomycin (0.5 pg/ml) for 15 min in a 5-20%/ linear sucrose gradient containing bleomycin. Sedimentation is from left to right. (B) Mass fraction of crosslinked oligomers as determined by electron microscopy from aliquots of the DNA fractions shown in A. (C) Electrophoretic separation of PM2 DNA after reaction with bleomycin (0.5 yg/ml) for 5 (lane a) or 15 (lane b) min. The positions of forms I, II, and III DNAs are indicated relative to four discrete higher molecular weight species indicated by the numbers

1-4.

Proc. Natl. Acad. Sci. USA 76 (1979)

Biochemistry: Lloyd et al.

2677

Examination of 111 molecules that occur in oligomers at 0 min of reaction with bleomycin revealed thak9&8 of thlz1

ecules were associated through topological bsias wo observed for molecules interlocked in catenations; only 2% were in contact over a region extending to approximately 500 nucleotide pairs. At 5 min of reaction, random sampling of 159 molecules in oligomers revealed 42% of the associations to be either as topological bonds or as single crosslinked bonds, 53% as a minimum of two adjacent crosslinked bonds with a region of contact extending continuously over approximately 500 nucleotide pairs (Fig. 2 upper right with regions indicated by arrows), and 5% as a minimum of three adjacent crosslinked bonds with a region of contact extending continuously over approximately 1000-2000 nucleotide pairs (Fig. 2, lower right, indicated by arrows). Assuming that adjacent regions of contact of 500 nucleotide pairs between associated molecules are produced by a minimum of two bleomycin crosslinks, we estimate that for 5 min of reaction there is-a minimum of 1.4 bleomycin crosslinks per molecule for those molecules that occur in oligomeric association (corrected for the initial content of catenanes). For molecules in this sample containing at least two crosslinks, 63% of the associations extend over a region of 500 nucleotide pairs. Measurements of 30 such crosslinked regions demonstrate that these segments comprise 5.0 + 1.1% of the PM2 genome length. This corresponds to 500 ± 110 nucleotide pairs of DNA assuming a length of 10,000 nucleotide pairs for PM2 DNA (10). After 30 min of reaction, random sampling of 210 molecules physically associated in oligomers revealed 28% of the associations as topological bonds or as single crosslinks, 61% as two adjacent crosslinks, and 11% as three adjacent crosslinks. In this case, we estimate a minimum of 1.8 bleomycin crosslinks per molecule, for molecules that occur in oligomeric association. For molecules that contain at least two crosslinks, 80% of the crosslinks extend over a region of 500 nucleotide pairs. Thus, as the reaction proceeds, the average number of crosslinks per molecule in an oligomer increases and, in addition, the frequency of adjacent crosslinks increases. The increased frequency of adjacent crosslinks could reflect a cooperative nature of the crosslinking reaction to favor neighboring and possibly unique sites on the PM2 genome. Bleomycin-Mediated Crosslinks Are Noncovalent. When PM2 DNA was allowed to react with bleomycin for 2.5 min followed by extensive dialysis against STE, no crosslinked forms were detected by electron microscopy, although a low frequency of catenated forms (2.7% of the PM2 DNA mass) persisted. Crosslinks also were eliminated by addition of NaCl to a final concentration of 0.8 M prior to examination by electron microscopy. Thus, it appears that the crosslinks are sensitive to dialysis or high salt concentration and do not possess the stable properties expected for covalent bonds. Electron microscopic examination of reaction mixtures that contained 12 mM EDTA or from which mercaptoethanol was deleted revealed no crosslinked forms of PM2 DNA (less than 2% for 300 molecules examined). Reactions conducted with DNA at 150 ,ug/ml instead of 50,ug/ml contained the same frequencies of crosslinked forms. Electron microscopic analyses of reactions conducted at 4°C revealed that the mass fraction of crosslinked oligomers was approximately one-fifth that observed for the reaction at 37°C. DNA breakage frequencies were also decreased at 40C to the same extent. Additionally, omission of CaCl2 from the reaction mixture and reduction of the Tris concentration from 20 to 5 mM had no effect on the breakage or crosslinking reactions. The CaC12 was included in the reaction to complex any trace amounts of EDTA remaining after DNA purification. Whereas intermolecular associations appear to be noncovaIi

FIG. 7. Electron micrograph illustrating the appearance of single-strands of PM2 DNA collapsed as bushes (indicated by arrows). Bar represents 1 Am.

lent, it is possible that intramolecular covalent crosslinks may be formed by reaction of bleomycin with PM2 DNA. To test this possibility, we purified linear duplex form III PM2 DNA, produced after 15 min of reaction with bleomycin, by agarose gel electrophoresis and electroelution of 36 combined gel slices. The sample was concentrated and then thermally denatured. If intramolecular covalent crosslinks between complementary strands of form III DNA had been formed by reaction with bleomycin, such molecules would be expected to renature rapidly upon quenching of the denatured form III DNA. The DNA strai.Is of noncrosslinked form III DNA will become physically separated and, when examined by the aqueous Kleinschmidt technique for electron microscopy, will appear as collapsed bushes (11). Examination of one sample of thermally denatured form III DNA derived from bleomycin reaction revealed no duplex molecules and approximately 10,000 single-strand bushes (Fig. 7). We conclude that intramolecular covalent crosslinks between complementary strands that are stable under these denaturing conditions occur at a frequency of less than 0.0002 per molecule of form III PM2 DNA. DISCUSSION that bleomycin reaction with double-stranded It is now evident DNA results in a spectrum of structural damages. These alterations in DNA structure range from removal of bases and introduction of alkali-labile sites (12-16) to single-strand and site-specific double-strand scissions of the phosphodiester backbone (7, 8, 13-17). We have reported herein our initial observations of yet another level of DNA structural alteration involving the formation of noncovalent crosslinks. Little, if any, DNA helix distortion may result from the formation of such crosslinks; however, this type of modification in chromosomal DNA might be expected to result in increased folding or packing of DNA chains. Although this probably would not elicit a response by cellular DNA repair systems, transcription and other relevant cellular functions such as segregation of mitotic chromosomes and availability of the DNA for binding nuclear proteins probably would be disrupted by bleomycin-mediated crosslinks. If these crosslinks are subsequently found to be sequence specific, disruption of cellular function would be considered more likely. D'Andrea and Haseltine (18) have recently used a cloned DNA segment from the lactose promoter-operator region of Escherichia coli to demonstrate that high concentrations of bleomycin fragment duplex DNA predominately at the dinucleotide sequences G-C and G-T. In the case of double-strand breaks produced by bleomycin reaction with PM2 DNA, we have previously shown that 11 specific sites on the genome are preferentially cleaved (7, 8). These sites probably correspond to a specific base sequence or set of sequences in the DNA that specifies breakage by bleomycin. To account for the observed specificity of duplex-strand scissions as well as the kinetics of both single- and double-strand scissions under different experimental conditions, it was postulated that a single component in the bleomycin mixture could self-associate to form a dimer molecular species (7, 16). We suggest that a similar noncovalent association of bleomycin molecules can provide the molecular

2678

Biochemistry: Lloyd et al.

FIG. 8. Model for bleomycin-mediated crosslinks. (A) Bleomycin monomers bound to native form I DNA associate to form a noncovalent crosslink. Superhelical turns in native PM2 DNA are not indicated. (B) Alternative scheme of crosslink formation in which bleomycin monomers associate first to form a dimer species which then binds to native form I PM2 DNA. The complex then binds to a second molecule of PM2 DNA. Binding of bleomycin at more than one site on DNA molecules in either A or B provides for the formation of additional crosslinks within the same complex structure or formation of higher oligomers through association with other bleomycin-bound PM2 DNA molecules.

basis for the formation of intermolecular crosslinks observed here. Although each form of DNA is found in crosslinked structures (Fig. 5A), our recent unpublished results demonstrate that only form I DNA can initiate the formation of crosslinks with bleomycin. Thus, the presence of forms II and III DNA in crosslinked structures must have resulted from fragmentation of form I DNA after the crosslinking events. The binding of bleomycin to multiple sites on form I DNA as shown diagrammatically in Fig. 8A (with superhelical turns not drawn in the figure for simplicity) provides a conformation of bleomycin that favors dimer formation of the DNA-bound molecules. Within the framework of this hypothesis, we expect the crosslinks to be noncovalent and therefore sensitive to dilution or dialysis as well as to high salt concentration. Incubation in high-salt medium has been observed to preferentially suppress duplex strand scission and this was postulated to result from dissociation of a dimer species of bleomycin (7). Bleomycin monomers could bind singly to separate DNA molecules and then associate to form a noncovalent crosslink (Fig. 8A). At this time, we must also consider the possibility that dimers of bleomycin could bind to the DNA without producing strand breakage but could provide a type of association that leads to crosslinks (Fig. 8B). In this regard, it is of interest that examination of the mass fraction of form I DNA after bleomycin reaction (Fig. 1A) revealed a discrepancy between analyses by electron microscopy and agarose gel electrophoresis. This discrepancy probably arose from intramolecular crosslinking within form II DNA that was incorrectly scored as form I DNA by electron microscopy; however, the discrepancy was rather small. If bleomycin is bound to DNA in an asymmetric and unique configuration, intramolecular crosslinks arising by association of bound bleomycin molecules on the same DNA will be possible, although less likely, under the same circumstances that permit intermolecular crosslinks. Such intramolecular crosslinks require folding of the DNA molecule to accommodate the bleomycin dimer configuration by a process that is expected to be energetically unfavorable. In examination of the crosslinked structures by electron microscopy we have noted that the crosslink frequently involves more than a simple point of physical contact. Often, these regions of contact correspond to approximately 500 nucleotide pairs of DNA. Because we observe molecules crosslinked at multiple adjacent sites (Fig. 2), we expect that the formation of one intermolecular crosslink would favor the formation of additional adjacent crosslinks, especially if crosslinking occurs at unique sites on the genome.

Proc. Natl. Acad. Sci. USA 76 (1979)

Finally, it should be emphasized that high molecular weight bands of DNA, presumably representing crosslinked species, are detected by agarose gel electrophoresis (Fig. 6C). Attempts to extract crosslinked species from the gel DNA bands failed to reveal crosslinked species; instead, monomer-size forms of DNA were detected. Monomer-size forms of PM2 DNA are unexpected at these high molecular weight positions in the gel and we interpret this result as a physical dissociation of crosslinked species by the extraction procedures utilized. Crosslinked species are detected, however, by velocity sedimentation in sucrose gradients containing bleomycin (Fig. 6A). The crosslinked species are found to sediment more slowly than form II PM2 DNA. This unexpectly large reduction in sedimentation velocity could result from an increase in the frictional coefficient due to molecular expansion or a reduction of partial specific volume of DNA upon binding bleomycin at a substantial number of sites. Partially collapsed DNA structures are frequently detected by electron microscopy of these sucrose gradient fractions as well as the unfractionated reaction mixtures (see Fig. 3 lower left and right). Although there may be other reasons for the unexpected sedimentation behavior of the crosslinked forms, it is apparent that they represent a unique molecular species that can be physically separated from the other DNA forms. We thank Dr. Norman Davidson for his helpful suggestions. We also thank Dr. Douglas L. Vizard for his help with the "freeze-squeeze" method of extracting DNA from agarose gels and Ms. Janie Finch for help in preparation of the manuscript. This investigation was supported, in part, by Grants CA-13246 and CA-16527 awarded by the National Cancer Institute and G-441 from the Robert A. Welch Foundation. R.S.L. is a Predoctoral Fellow with the Robert A. Welch Foundation.

1. Cole, R. S. (1970) Biochim. Biophys. Acta 217,30-39. 2. Iyer, V. N. & Szybalski, W. (1963) Proc. Natl. Acad. Sci. USA 50, 355-362. 3. Geiduschek, E. P. (1961) Proc. Natl. Acad. Sci. USA 47,950955. 4. Ewig, R. A. G. & Kohn, K. W. (1978) Fed. Proc. Fed. Am. Soc. Exp. Biol. 37, 1410. 5. Salditt, M., Braunstein, S. N., Camerini-Otero, R. D. & Franklin, R. M. (1972) Virology 48,259-262. 6. Strong, J. E. & Hewitt, R. R. (1975) Isozymes: Developmental Biology, ed. Markert, C. (Academic, New York), Vol. 3, pp. 473-483. 7. Lloyd, R. S., Haidle, C. W. & Robberson, D. L. (1978) Biochemistry 17, 1890-1896. 8. Lloyd, R. S., Haidle, C. W., Robberson, D. L. & Dodson, M. L., Jr. (1978) Curr. Micro. 1, 45-50. 9. Davis, R. W., Simon, M. & Davidson, N. (1971) Methods Enzymol., eds. Grossman, K. & Moldive, K. (Academic, New York), pp. 413-428. 10. Stuber, D. & Bujard, H. (1977) Mol. Gen. Genet. 154, 299303. 11. Davis, R. W. & Davidson, N. (1968) Proc. Natl. Acad. Sci. USA

60,243-250.

12. Haidle, C. W., Weiss, K. K. & Kuo, M. T. (1972) Mol. Pharmacol.

8,531-537.

13. Muller, W. E. G., Yamazaki, Y., Breter, H. J. & Zahn, R. K. (1972) Eur. J. Biochem. 31, 518-525. 14. Lloyd, R. S., Hewitt, R. R. & Haidle, C. W. (1977) Fed. Proc. Fed. Am. Soc. Exp. Biol. 36,694. 15. Povirk, L. F., Wubker, W., Kohnlein, W. & Hutchinson, F. (1977) Nucleic Acids Res. 4, 3573-3580. 16. Lloyd, R. S., Haidle, C. W. & Hewitt, R. R. (1978) Cancer Res.

38,3191-3196.

17. Haidle, C. W. (1971) Mol. Pharmacol. 7,645-652. 18. D'Andrea, A. D. & Haseltine, W. A. (1978) Proc. Natl. Acad. Sci.

USA 75, 3608-3612.