Proc. Nat. Acad. Sci. USA Vol. 68, No. 11, pp. 2839-2842, November 1971

Isolation of Catenated and Replicating DNA Molecules of Colicin Factor El from Minicells (E. coli/sucrose gradients/ethidium bromide/CsCl/electron microscopy) JOSEPH INSELBURG AND MOTOHIRO FUKE* Department of Microbiology, Dartmouth Medical School, Hanover, New Hampshire 03755; and Department of Biological Chemistry, Harvard Medical School, Boston, Massachusetts 02115 Communicated by Luigi Gorini, September 14, 1971

ABSTRACT Various catenated and replicating mole- at 37°C) of minicells previously incubated in spheroplast cules of colicin El isolated from minicells have been medium was introduced into the DNA extraction procedure identified in regions of neutral sucrose density gradients or cesium chloride-ethidium bromide density gradients. before the addition of Sarkosyl NL30 and subsequent phenol These colicin molecules have been found in those regions extraction (2, 11). Neutral sucrose density gradient sedimen- of the gradient where pulse-labeled DNA accumulates. tation analysis (2) and cesium chloride-ethidium bromide (2,7-diamino-10-ethyl-9-phenylphenanthridium bromide) den- Adler et al. (1) isolated a mutant of , P678-54, sity-gradient analysis of DNA (2, 12) and electron that produced significant numbers of small, DNA-less progeny microscope techniques (7, 13, 14) were also described in ("minicells") under normal growth conditions. Subsequent detail. Buffer solutions used are Tris-saline (TS) 0.05 M work has shown that if or episomal are carried Tris-0.05 M NaCl (pH 8.0); Tris-EDTA-Saline (TES), by that mutant, they can segregate into (2-6) and replicate which is TS + 5 mM EDTA; and saline-sodium citrate in (2, 6, 7) the minicells. In this report minicells containing solution (SSC), which contains 0.15 M NaCl and 0.015 M colicin factor El (Col El) DNA have been used to examine sodium citrate (pH 7.4). DNA replication. Two principal advantages of studying DNA replication in minicells are first, the small size of the Col El RESULTS DNA monomers [4 X 108 daltons (8)1 and second, the absence It can be estimated that a Col El DNA molecule (4 X 106 of chromosomal DNA, which otherwise interferes with the daltons) (7, 8) should complete a round of replication in 5 identification of extra-chromosomal DNA. The study of the sec if it replicates as rapidly as chromosomal DNA [the replication of Col El DNA in minicells has shown that (a) all E. coli of 2.5 X 109 daltons (9) replicates in of the observed DNA in minicells carrying Col El DNA is about 40 min (15)1. If the molecule did replicate at that rate Col El DNA (2, 7); (b) the Col El DNA principally exists as in minicells, then by pulse-labeling of the replicating Col El covalently-closed circular and open circular molecules, DNA for about 5 sec, it should be possible to obtain as much though linear monomers are also found (2, 7); (c) the Col El as 50% of the total radioactively labeled DNA in a replicating DNA can undergo at least two complete rounds of replication form. Minicells carrying replicating Col El DNA were (2); and (d) replicating structures identical to the forked purified and labeled with [3Hjthymidine for different periods circular molecules originally found by Cairns (9), as well as of time. The DNA was then extracted and either examined those similar to the structure of rolling circles as postulated in a neutral sucrose density gradient or in a cesium chloride- by Gilbert and Dressler (10), can be identified (7). These ethidium bromide density gradient. The results of the neutral observations have led to a further examination of Col El sucrose density-gradient analysis (Fig. 1) indicate that a DNA replication, a part of which is the subject of this report. broad peak of DNA, sedimenting more rapidly than the MATERIALS AND METHODS covalently-closed circular form of Col El (24 S) (2), was All experiments were performed with purified minicell labeled during the 5-sec pulse of [3H]thymidine and then preparations (1 viable bacterium per 106-107 minicells) became progressively less distinct as the labeling period was isolated from log-phase cultures of E. coli strain P678-54 extended. Four separate experiments gave identical results. (Col El) (2) grown in Tris-Casamino acid-glucose medium The fast-sedimenting peak accounted for 20-30% of the supplemented with 20 ug/ml of threonine and leucine and DNA that was pulse labeled in 5 sec. As the DNA was purified with 1 jig/ml of vitamin B1 (2). The detailed procedures for in the presence of Pronase, Sarkosyl, and phenol, the rapid minicell purification, [3Hlthymidine labeling of Col El DNA sedimentation property should not be due to the binding of in the above medium supplemented with 250,ug/ml of deoxy- protein to DNA as has been described by Clewell and Helinski adenosine and 0.5 jg/ml of thymidine, and DNA extraction (17). The results of the cesium chloride-ethidium bromide from minicells treated with spheroplast medium [lysozyme- density-gradient analysis, shown in Fig. 2, indicate that EDTA-pancreatic ] are as described (2, 11). during a short pulse-labeling a significant amount of labeled Pronase treatment (1 mg/ml of predigested Pronase for 15 min DNA bands at an intermediate density between the positions of covalently-closed circular molecules (fractions 10-15) * Present address: Laboratory of Molecular Biology, N.I.A.M.D., and of open circular and linear molecules (fractions 27-32) National Institutes of Health, Bethesda, Md. 20014. (12). On prolonged labeling, the DNA in the intermediate- 2839 Downloaded by guest on September 26, 2021 2840 Biochemistry: Inselburg and Fuke Proc. Nat. Acad. Sci. USA 68 (1971) DNA, which maintains its sedimentation characteristics upon recentrifugation in an identical sucrose gradient, was further analyzed. DNA labeled for 1 hr was fractionated in a neutral sucrose-density gradient and divided into two fractions (b and c in Fig. 3a). Both fractions were examined in a cesium O chloride-ethidium bromide density gradient. The result (Fig. 3b and c) shows that much of the fast-sedimenting DNA bands at the intermediate-density position. Hudson and Vinograd (18, 19) have shown that several 0.4.0 Ul) forms of catenated mitochondrial DNA from HeLa cells H can sediment faster than covalently-closed circular molecules ~00- in a sucrose gradient, and can be found in cesium chloride- o _ 5 IS 25 5 15 25 ethidium bromide gradients at a density between that of FIG.°U 12. Nb ;. .ic ' open and covalently-closed circular molecules. Tomizawa H R P. 0 and Ogawa (20) and Levine et al. (21) have found that replicating, forked circular structures of bacteriophage 0 lambda and SV40 , similar to those originally found by Cairns (9), also sediment in neutral sucrose at a position faster than covalently-closed circles. In view of those findings, 4.0- and since a previous report indicated that a major replicating form of Col El DNA in minicells is a forked circular structure it 0.0 (7), was expected that both replicating and catenated 5 15 25 tO 20 30 molecules might be present in the regions of the sucrose FRACTION NUMBER density or cesium chloride-ethidium bromide gradients where FIG. 1. Neutral sucrose density-gradient analysis of Col El labeled DNA accumulated during a short pulse. Col El DNA DNA labeled for various times. Purified P678-54 (Col El) mini- cells were incubated with [3H] thymidine (250/uCi/ml for the 5-sec pulse, l00 Ci/ml for all other experiments) for the time indicated. 1.I 20mi Cold TES buffer containing 0.04 M NaN3 was added to stop the 8.0stO labeling of the minicells, which were washed four times in TES containing 0.02 M NaN3. DNA was incubated in spheroplast 6.0I medium, with successive additions of Pronase and Sarkosyl. Each 4.0- DNA isolation step lasted about 15 min at 370C. DNA was ex- w tracted with TES-saturated phenol at room temperature, dialyzed H 2.0- against 0.1 X SSC overnight at 4VC, and concentrated to 0.2 ml, z using Carbowax 6000 when necessary. The DNA was layered on a i 6.0 5 I A cr 6:0 _ 10 20 !0\ 40 4.6 ml, 5-20% TES-sucrose gradient and centrifuged at 84,000 X w g in a Spinco SW50.1 rotor at 15'C for 4.5 hr. in a Spinco model a. (f) L2 ultracentrifuge. The two principal peaks in the gradients are H marked and are, from right to the 18S and 24S cova- z left, open 3 lently-closed circular forms of Col El DNA. A third broad peak is 0 seen to the left of the 24S peak in the gradients containing 5-sec -J 4.0W 10min and 30-sec pulse-labeled DNA. The gradients of the 5-sec and 1.5-hr labeled DNA have been superimposed (a) to facilitate 0 comparison. E. coli phage P1 [32P]DNA, with a reported sedi- IL mentation coefficient of about 43 S (2, 16), was added to each tube 0 as a sedimentation marker and its position in the gradient is marked by an arrow. (a) 5 sec (0), total cpm 190, total fractions 2.010.0 30; 1.5 hr (0), total cpm 22,640, total fractions 30. (b) 30 sec, total cpm 390, total fractions 35. (c) 1.5 min, total cpm 630, total frac- 10 20 30 40 tions 35. (d) 4 min, total cpm 1408, total fractions 30. FRACTION NUMBER FIG. 2. Cesium chloride-ethidium bromide density-gradient centrifugation of Col El DNA labeled for various times. Col El density position decreases in relative amount, whereas the DNA in minicells was labeled and purified as described in Fig. 1, region containing covalently-closed circular molecules, as dialyzed against 1 X SSC, added to a solution containing CsCl, well as that containing both open circular and linear mole- 4.36 g; ethidium bromide, 1.9 ml of 700 ,ug/ml in sodium phos- cules, increases. In a pulse-chase experiment in which the phate buffer, (pH 7.0) (11), and brought to a refractive index of DNA of minicells was labeled for 1 min, and was then chased 1.3860 as measured in an Abbe 3L refractometer. The solution was in a 50 Ti rotor at a similar in the of the centrifuged Spinco fixed angle 105,000 X g for for 40 min, change banding properties 48 hr at 5°C, and fractions were collected by drop collection from DNA in the cesium chloride-ethidium bromide density the bottom of the tube. Refractive indexes of selected fractions gradient was observed. were always taken. E. coli [32P] DNA (2) was added to each gradi- It can be seen in Fig. 1 that the material labeled for 1.5 hr ent to mark the position of linear DNA. (a) 5 sec, total cpm 656, contains a small, broad band of fast-sedimenting DNA in total fractions 50. (b) 1.5 min, total cpm 2675, total fractions 52. addition to the 24S and 18S DNA. This fast-sedimenting (c) 20 min, total cpm 1985, total fractions 52. Downloaded by guest on September 26, 2021 Proc. NVat. Acad. Sci. USA 68 (1971) Colicin El from Minicells 2841

isolated from minicells was fractionated by either neutral sucrose .density-gradient centrifugation or cesium chloride- ethidium bromide density-gradient centrifugation, and the LUJ pooled fractions of the sucrose gradient that sedimented just z Pe ahead of the position of the covalently-closed circular mol- ecules and the pooled fractions obtained from the inter- LUJ 0. 12.0- mediate-density position of the cesium chloride-ethidium (I) bromide gradient were examined by electron microscopy (Table 1). Examples of typical catenated Col El DNA O8.0 b 1525 -00 4 molecules obtained from both fractions are shown in Fig. 4. -J~~~~~~~~~~~~. DISCUSSION 0 work Col El DNA structures H 0 In previous (7), replicating ill FIG 3.A9.omclrdeehdu broiedniygain minicells were detected by density labeling techniques. The 0A DNA molecules found were principally in the form of repli- cating, Cairns type, double-forked circular structures. Several molecules in the form of open circles with long FRACTION NUMBER tails were found and tentatively identified as replicating FIG. 3. A cesium chloride-ethidium bromide density-gradient "rolling circle" structures. The present work shows that analysis of Col El DNA obtained from a neutral sucrose density replicating molecules of Col El DNA can be efficiently gradient. (a) Col El [3H]DNA labeled for 1 hr was isolated from isolated (see Table 1) from the region of accumulation of minicells, prepared as described in Fig. 1, and fractionated in a pulse-labeled Col El DNA that sediments a little faster than neutral sucrose density gradient. P1 [32P] DNA was used as a sedi- 24 S, the position of covalently-closed circles in a neutral mentation marker and its position in the gradient is indicated. between the Appropriate fractions of the gradient were pooled, dialyzed sucrose density gradient, or that band positions against 0.1 X SSC, and concentrated with Carbowax 6000. of open and covalently-closed circular DNA in a cesium Material from pooled fractions b (9-14) and c (15-21) were chloride-ethidium bromide density gradient. Replicating, analyzed by cesium chloride-ethidium bromide density-gradient twisted circular DNA and replicating catenated molecules not centrifugation. (a) total cpm 18,781, total fractions 30. (b) total previously reported were found together with the two replica- cpm 3,088, total fractions 54. (c) total cpm 11,317, total fractions ting structures that had been described (7). These results 53.

FIG. 4. Electron micrographs of catenated Col El DNA obtained from neutral sucrose density gradients or cesium chloride-ethid- ium bromide density gradients. Downloaded by guest on September 26, 2021 2842 Biochemistry: Inselburg and Fuke Proc. Nat. Acad. Sci. USA 68 (1971)

TABLE 1. Molecular forms of Col El DNA DNA replication process (18, 19, 24, 25). The ability to observed by electron microscopy isolate significant numbers of catenated structures from Number of molecules minicells should facilitate the resolution of the problem of their formation. From cesium chloride- We are grateful to Mrs. Virginia Johns for patient and expert From ethidium assistance and to Dr. Martin Lubin for use of his electron micro- scope and associated facilities. This work is supported by Grant Type sucrose bromide Al 08937-03 VR from the National Institute of Allergy and Twisted circle 332 93 Infectious Diseases. Open circle 46 230 1. Adler, H. I., W. D. Fisher, A. Cohen, and A. A. Hardigree, Linear fragment 29 67 Proc. Nat. Acad. Sci. USA, 57, 321 (1967). Circular dimer 5 0 2. Inselburg, J., J. Bacteriol., 102, 642 (1970). Catenated (all types) 122 82 3. Roozen, K. J., R. J. Fenwick, Jr., S. Levy, and R. Curtiss, open-open 39 20 III, Genetics, 64, S54 (1970). open-twisted 67 54 4. Levy, S. B., and P. Norman, Nature, 227, 606 (1970). twisted-twisted 9 5. Kass, K. R., and M. B. Yarmolinski, Proc. Nat. Acad. Sci. 4 USA, 66, 815 (1970). trimer 7 4 6. Inselburg, J., J. Bacteriol., 105, 620 (1971). Replicating (all types) 51 14 7. Inselburg, J., and M. Fuke, Science, 169, 590 (1970). double-forked 15 11 8. Roth, T. F., and D. Helinski, Proc. Nat. Acad. Sci. USA, circle (Cairns) 58, 650 (1967). circle with tail 9 1 9. Cairns, J., J. Mol. Biol., 6, 208 (1963). twisted circle* 19 2 10. Gilbert, W., and D. Dressler, Cold Spring Harbor Symp. catenated (all types) 8 0 Quant. Biol., 33, 473 (1968). 11. Bazaral, M., and D. R. Helinski, J. Mol. Biol., 36, 185 Totals 585 486 (1968). 12. Radloff, R., W. Bauer, and J. Vinograd, Proc. Nat. Acad. Twisted circle is synonymous with Sci. USA, 57, 1514 (1961). "covalently-closed circular" 13. Kleinschmidt, A. D., D. Lang, D. Jackert, and R. K. molecules. In the case of the twisted-circular replicating molecules Zahn, Biochim. Biophys. Acta, 61, 857 (1962). (*), the covalently-closed circular nature of the molecules is 14. Ogawa, T., J. Tomizawa, and M. Fuke, Proc. Nat. Acad. uncertain. In a forthcoming paper a complete presentation will be Sci. USA, 60, 861 (1968). made of the electron-microscopic evidence, demonstrating the 15. Helmstetter, C., S. Cooper, 0. Pierucci, and E. Revelas, variety of replicating and catenated Col El DNA structures Cold Spring Harbor Symp. Quant. Biol., 33, 809 (1968). reported in this table. 16. Ikeda, H., and J. Tomizawa, J. Mol. Biol., 14, 85 (1965). 17. Clewell, D. B., and D. R. Helinski, Biochemistry, 22, 4428 (1970). 18. Hudson, B., and J. Vinograd, Nature, 216, 647 (1967). bring to four the number of duplicating structures of Col 19. Hudson, B., and J. Vinograd, Nature, 221, 332 (1969). El whose role in the replication process of must be 20. Tomizawa, J., and T. Ogawa, Cold Spring Harbor Symp. evaluated. It is expected that since replicating plasmid and Quant. Biol., 33, 533 (1968). episomal DNA molecules can now be obtained relatively easily 21. Levine, A. J., H. S. Kang, and F. E. Billheimer, J. Mol. Biol., 50, 549 (1970). from minicells, the problems relating to their replication pro- 22. Rhoades, M., and C. A. Thomas, Jr., J. Mol. Biol., 37, 41 cess will be more rapidly resolved. (1968). Catenated DNA molecules have been isolated from several 23. Rion, G., and E. Delain, Proc. Nat. Acad. Sci. USA, 62, sources (18, 19, 22-25). The mechanism by which they are 210 (1969). formed of 24. Gordon, C. N., M. G. Rush, and R. W. Warner, J. Mol. and the significance their presence is uncertain. Biol., 47, 495 (1970). It has been speculated that they are formed as a result of a 25. Kontomichalou, P., M. Mitani, and R. C. Clowes, J. breakage and reunion recombination event or during the Bacteriol., 104, 34 (1970). Downloaded by guest on September 26, 2021