Proc. Natl. Acad. Sci. USA Vol. 90, pp. 7809-7813, August 1993 Biochemistry Kinetoplast DNA maxicircles: Networks within networks (trypanosomes/topology) THERESA A. SHAPIRO Division of Clinical Pharmacology, Department of Medicine, The Johns Hopkins University School of Medicine, 301 Hunterian Building, 725 North Wolfe Street, Baltimore, MD 21205-2185 Communicated by Paul Talalay, May 20, 1993 (receivedfor review, April 27, 1993)

ABSTRACT Kinetoplast DNA (kDNA), the mitochondrial trypanosomes, the location of mitochondrial topoisomerase DNA of trypanosomes, is an enormous network of interlocked II with respect to kDNA is unknown. minicircies and maxicircles. We selectively removed miinicir- Early EM studies showed that selective removal of cles from Trypanosoma equiperdum kDNA networks by restric- maxicircles from kDNA networks by restriction enzyme tion enzyme cleavage. Maxicirdes remained in aggregates that digestion left residual catenanes of minicircles (14, 15). were resistant to protease or RNase and contained no residual Treatment with topoisomerase II established unequivocally minicircies, but were resolved into circular monomers by that each kDNA network contained several dozen unit-length topoisomerase II. Maxicirdes thus form independent catenanes circular maxicircles (16). However, the arrangement of within kDNA networks. Heterogeneity in the size, composition, maxicircles within the minicircle scaffold has remained spec- and organization of maxicircle catenanes reflects changes that ulative (10). In nonreplicating networks, maxicircle loops occur during kDNA replication. The rosette-like arrangement protrude from the margins of the matrix. In replicating of maxicircle catenanes is distinctly different from that of networks from trypanosomes (but not from Crithidia), minicircle catenanes. Trypanosome kDNA networks reveal maxicircles are strikingly apparent as a cluster between the unique topological complexity: they are composed of entirely two lobes of the separating daughter networks. dissimilar catenanes that are in turn extensively interlocked T. equiperdum maxicircles are homologous with those of with one another. Trypanosoma brucei (17), but the minicircles are distinctive because they are all of one sequence (18). We identified an Kinetoplastid protozoa are pathogenic to humans, livestock, enzyme that would accomplish the statistically unlikely and commercially important plants. A distinguishing mor- cleavage of minicircle, but not maxicircle, DNA. In this phologic feature of these organisms is their massive mito- paper we describe the maxicircle product produced when Spe chondrial DNA (termed kinetoplast DNA, or kDNA), which I selectively removes all minicircles from T. equiperdum repeatedly proves to be structurally and functionally fasci- kDNA networks. This product is a catenane of interlocked nating. The single mitochondrion of each cell contains one maxicircles. Maxicircles therefore form networks within kDNA network composed of thousands of interlocked min- kDNA networks. icircles and tens of maxicircles (reviewed in refs. 1-3). Depending on the species, minicircles are about 1 kb each and MATERIALS AND METHODS there may be hundreds of different sequence classes in a network; maxicircles are about 20 kb each and of identical Trypanosomes and kDNA Isolation. T. equiperdum trypa- sequence. Maxicircles are analogous to other mitochondrial nosomes (Pasteur Institute strain BoTat 24; African trypa- DNAs in that they code for mitochondrial rRNA and some nosomes closely related to T. brucei) were grown in female mitochondrial proteins. The unique genetic role of minicir- Wistar rats, isolated from blood by means of DEAE- cles has only recently been discovered: they code for guide cellulose, and lysed with SDS. The lysate was treated with RNA molecules that direct the editing of maxicircle tran- RNase and proteinase K, phenol-extracted, and dialyzed scripts to generate fuinctional open reading frames (reviewed (11). Dialysate was layered over sucrose cushions (20% in 10 in refs. 4-6). mM NaCl/100 mM EDTA/10 mM Tris HCl, pH 8.0) and Replication of kDNA entails the production of two daugh- centrifuged (19,000 rpm, 4°C, Beckman SW28 rotor, 1 hr). ter networks identical in size and composition to the parent The pelleted networks were collected, dialyzed, and centri- (reviewed in refs. 7 and 8). Not surprisingly, this process is fuged (12,000 rpm, 4°C, Savant microcentrifuge, 1 hr). highly complex. Covalently closed minicircles are individu- Enzyme Digests. Unless indicated otherwise, digestion was ally decatenated, and they replicate free of the network. for 3 hr at 37°C; buffers provided by the enzyme supplier were Daughter minicircles are segregated from one another and supplemented with 0.5 mM dithiothreitol. DNA was phenol- attached to the periphery ofthe network. Much less is known extracted and ethanol-precipitated between reactions. Net- about maxicircle replication, which involves a rolling-circle works (3 p,g/ml) were decatenated with topoisomerase II mechanism (9). In trypanosomes, the growing network from T4 phage [gift of Ken Kreuzer (19); 2.5 units/ml] for 1 evolves into a dumbbell shape and then divides (10). The hr at 370C in 60 mM KCl/10 mM MgCl2/0.5 mM dithiothrei- numerous topological interconversions necessary to synthe- tol/0.5 mM Na2EDTA, 0.5 mM ATP/40 mM Tris HCl, pH size and remodel kDNA are catalyzed by topoisomerases. A 7.8, with bovine serum albumin (50 ,ug/ml). To analyze for mitochondrial type II DNA topoisomerase was demonstrated residual minicircles, networks (3 p,g/ml) were digested at in Trypanosoma equiperdum by drug inhibition studies (11). 37°C for 3 hr with Spe I (50 units/ml) and for 1.5 hr after Two distinct type II topoisomerases have been isolated from addition ofmore Spe 1 (25 units/ml). An aliquot was taken for Crithidiafasciculata and immunolocalized to the kinetoplast gel analysis and the pH was adjusted to 9.5 with KOH prior (refs. 12 and 13; J. Shlomai, personal communication). In to addition of A exonuclease (33 units/ml). The mixture was incubated at 370C for 1 hr and the reaction was quenched with phenol. 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: kDNA, kinetoplast DNA. 7809 Downloaded by guest on September 23, 2021 7810 Biochemistry: Shapiro Proc. Natl. Acad. Sci. USA 90 (1993) Gel Electrophoresis. Maxicircle DNA was fractionated in A B 0.6% agarose gels (18-21 hr; 3 V/cm; 24°C) in 90 mM Tris/92 1 2 3 4 1 2 3 4 mM boric acid/2.5 mM EDTA, pH 8.3, containing ethidium bromide (1 pg/ml). Size markers for maxicircles were Hin- dIII or Bgl II digests of A DNA. Nicked and covalently closed circular maxicircles were identified by repairing decatenated s networks with ligase, DNA polymerase, and dNTPs; or by f -N nicking decatenated networks with DNase I (data not shown). Markers for minicircles were Hae III or Hpa II digests of 4X174 replicative-form DNA. Electrophoresis of 23.1- * -c minicircle DNA and transfer to GeneScreen (NEN) have 0 ?s - L been described (11). 22.0- Probes and Hybridization. Plasmid pTKH128 (gift of Ken 2.7- Stuart; ref. 20) contains a 7-kb HindlIl T. brucei maxicircle fragment inserted into pBR322. The insert was excised with HindIll, purified by gel electrophoresis, and radiolabeled by the random primer method. Preparation of probes from full-length cloned T. equiperdum minicircle DNA was de- 1.7- scribed previously (11). Membranes were prehybridized at 55°C for 30 min in 500 mM sodium phosphate, pH 7.2/10 mM EDTA/1% bovine serum albumin/7% SDS (50 ,l/cm2) and 1.4 then hybridized for 12-18 hr at 55°C in prehybridization solution (25 p1/cm2) containing 10 pmol of 32P-labeled probe 1.1 - * -N (0.6-6 x 106 cpm/pmol). Filters were washed four times (15 min each, 55°C) in 600 mM NaCl/60 mM sodium citrate, pH :4 7.0/0.5% SDS. 0 9.-- EM. DNA was prepared for microscopy by the formamide technique (21). Samples on parlodion-coated copper grids were stained with uranyl acetate and rotary shadowcast with platinum/palladium. Micrographs were taken on a Zeiss *c transmission model EM 1OA electron microscope. DNA contour lengths were determined with a map measurer (mod- FIG. 1. Analysis of Spe I digests. Purified T. equiperdum kDNA el 11-11, Alvin Co., Windsor, CT). networks were digested with enzymes, and the samples were divided and analyzed for minicircle (A) or maxicircle (B) DNA. (A) DNA was electrophoresed in 1.5% agarose containing ethidium bromide, blot- RESULTS ted, and probed with 32P-labeled minicircle DNA. Networks were Spe I Yields Linear Minicircles and Maxicircle Aggregates. treated with no enzyme (lane 1), BamHI (lane 2), topoisomerase II (lane 3), or Spe I (lane 4). Intact networks are too large to enter the To test whether Spe I selectively cleaves minicircles, purified gel and are variably lost from the slot when the gel is processed for kDNA networks were digested with enzyme and the sample transfer. (B) Same reactions as in A, but DNA was electrophoresed was split to analyze minicircle and maxicircle products. The in 0.6% agarose, blotted, and probed with 32P-labeled maxicircle DNA was separated by gel electrophoresis, blotted, and DNA. Each lane contained 15 ng of kDNA. N, circles containing probed with cloned 32P-labeled minicircle or maxicircle DNA nicks or gaps; L, linearized circles; C, covalently closed circles; S, (Fig. 1 A and B, respectively). As predicted from the se- smear of maxicircle DNA. Scales indicate size of marker fragments quence, the minicircle products of Spe I migrated as a single in kilobases; arrows indicate slots. 1-kb band (Fig. 1A, lane 4) that represented full-length linearized molecules, clearly distinct from nicked and cova- kDNA network preparations, Spe I digests, spreads, EM and lently closed circular monomers generated by topoisomerase film processing), 55 structures (29 plus 26) were evaluated. In II (Fig. 1A, lane 3). However, the maxicircle products ofSpe each experiment, every maxicircle structure within a single I (Fig. 1B, lane 4) were not circular monomers such as those grid space was photographed and analyzed. The mean con- liberated by topoisomerase II (Fig. 1B, lane 3). Instead, tour length of a single maxicircle was determined from 12 maxicircle products remained largely at the origin, indicating circular monomers (8.0 + 0.3 um, a value not statistically that they were intermediate in size between intact networks different between experiments). The monomer content of that were too massive to enter the gel matrix (Fig. 1B, lane each structure was calculated from the total contour length of 1) and linearized maxicircles (Fig. 1B, lane 2). A faint smear the structure divided by the contour length of a monomer. extending from the slot to the level ofnicked monomers (Fig. In addition to diversity in size, heterogeneity was evident 1B, lane 4, and longer exposures) suggested that the Spe I in the composition ofindividual maxicircle structures, which products were heterogeneous. Full-length linearized maxicir- contained both relaxed and supercoiled loops of DNA. In cles, produced by BamHI or Pst I digestion of intact net- smaller aggregates (e.g., trimer, 18-mer), supercoiled loops works, were not cleaved by Spe I, confirming that Spe I has were more common, whereas the largest structures (e.g., no recognition sequence in T. equiperdum maxicircles (data 35-mer, 45-mer) were composed entirely of relaxed DNA. not shown). The overall proportion of supercoiled DNA did not change Visualization of Maxkcircle Aggregates. To obtain further when the DNA was spread for EM in the presence of information about the maxicircle products, purified networks ethidium bromide (100 ug/ml) (data not shown). This sug- were treated with Spe I and prepared for EM. The micro- gests that the relaxed loops seen in Fig. 2 are parts of circles graphs reveal discrete clusters of maxicircles in a field of that are not covalently closed and, further, that the cova- linearized minicircles (Fig. 2). Maxicircle structures occurred lently closed circles are virtually all supercoiled. in a range of sizes, consistent with their behavior in agarose Another qualitative difference between small and large gels (Fig. 1B, lane 4). To determine the number ofmaxicircles structures is in their general organization. The larger struc- in each structure, DNA contour lengths were measured. tures are striking for their highly organized rosette-like ap- From two independent experiments (including different pearance, with loops emanating from dense central cores. In Downloaded by guest on September 23, 2021 Biochemistry: Shapiro Proc. Natl. Acad. Sci. USA 90 (1993) 7811

N .,~~~~~~~~~~~~~~~~~4

y, -~~~~~~qi A .A' i@2/siI,>19

7N~~~~~~~

-. Z s- Y~-- // ~ ~ -'LJ~ ~ g9j,4) ~~ L---, y.

r3.if* / -b:..-

FIG. 2. EM visualization ofmaxicircle aggregates. Purified T. equiperdum kDNA networks were incubated with Spe I and prepared for EM. Micrographs display linearized minicircles (open arrow) and maxicircle aggregates of various sizes. Monomer contents were determined from the total contour length of each structure, divided by the contour length of a monomer. Solid arrows, supercoiled maxicircles; curved arrow in 35-mer, maxicircle attached only to the periphery of the catenane; small arrow in 45-mer, loop of single-stranded DNA. (Bar = 1 in.) most cases, the loops seem to enter and exit at the same point. small structures: in experiments with kDNA networks iso- When several dozen micrographs are arranged by increasing lated several months before preparation for microscopy, or size of maxicircle aggregate, the rosette-like character in- with Spe I digests that were vortexed, >20% of maxicircles creases, and in the largest structures almost all DNA passes were in aggregates of <9. through a core. These cores are not apparent in the smaller The kDNA networks in these experiments derive from aggregates (Fig. 2, compare 18-mer with 35-mer or 45-mer). unsynchronized trypanosomes. This is reflected in the pop- In four large structures containing a total of 136 maxicircles, ulation of intact networks (Fig. 3B), which contains struc- <2% ofthe DNA is in circles attached solely at the periphery tures ranging from single size (nonreplicating; 10-13 ,um in of the catenane (curved arrow in 35-mer, Fig. 2). Occasional circumference) to double size (late replicating; 27 ,um in loops of DNA appear to be single-stranded (small arrow in circumference). Maxicircle aggregates containing -16 mono- 45-mer, Fig. 2). In the five structures suitable for analysis, the mers mimic this size distribution, suggesting that intact maxicircles appear to be joined by a single interlock. kDNA networks contain 16-45 maxicircles, in reasonable Size Distribution of Maxicircle Aggregates. The 55 maxicir- agreement with previous estimates (15, 16). cle structures range from monomers to 45-mers, but 90% of Aggregates Are Interlocked Maxicircles. The isolation pro- total maxicircle mass is in aggregates containing 211 mono- cedure yields networks that appear pure on the basis of EM, mers (Fig. 3A). Smaller structures may be generated by UV absorption spectrum, and electrophoresis in agarose mechanical damage, or may reflect a true occurrence in vivo. gels. However, to determine whether maxicircles are aggre- Although it is impossible to distinguish between these pos- gated by residual protein or RNA, Spe I digests were further sibilities, damage almost certainly contributes some of the processed prior to electrophoresis. Extensive treatment with Downloaded by guest on September 23, 2021 7812 Biochemistry: Shapiro Proc. Natl. Acad. Sci. USA 90 (1993) 40 1 2345

U 30- FIG. 4. Maxicircles are not aggregated £ by protein or RNA. Spe I digests were 20i a treated as indicated to determine whether a - N maxicircles were aggregated by residual pro- tein or RNA. After treatment, the DNA was separated by agarose gel electrophoresis, *.u ... ::.; ....'

...... : blotted, and probed with 32P-labeled maxi- 0 circle DNA. Each lane contained 15 ng of - C kDNA. Lanes: 1, untreated; 2, proteinase K --- -.. fI: I" .---I-: digestion and phenol extraction; 3, RNase A 1-5 6-10 11-15 16-20 21-25 26-30 31-35 36-40 41-45 and T1 digestion; 4, incubation with topo- Number of monomers in catenane m-- L isomerase II; 5, digestion with Pst I, which cleaves maxicircles once. N, C, L, and ar- 40 - row as in Fig. 1. B n by Pst I or topoisomerase II (Fig. 5A, lanes 3 and 4, 30 - U) respectively). These treatments, which dismantle maxicircle catenanes, caused no increase in nicked circular minicircles 0 (Fig. 5B, compare lanes 3 and 4 with lane 2), the expected U) 20 B 1 2 3 4 5 6 7 8 910 11 12 o_cc A 1 2 3 4

10 -

nI I I I n n 1:n * -N * .- Fin U f.1 .l .1 .1 .1 ., in F 1 3 5 7 9 11 13 15 17 19 21 23 25 27 Circumference of intact networks, /cm FIG. 3. Size distribution of maxicircle catenanes and intact ,*. -L networks. kDNA networks were isolated from an asynchronous population of T. equiperdum and processed at once for visualization SpoI + EM before or after with I. - + + by (B) (A) digestion Spe (A) Size X exo + distribution of maxicircle catenanes, based on the number of mono- Pst I - mers in each structure. Data are from 55 structures containing a total Topo II - of 562 monomers. (B) Size distribution of intact networks based on circumference. In this series, all networks with circumference > 20 pm resemble dumbbells. Data are from 54 networks...... *-C proteinase K, followed by phenol extraction and ethanol precipitation, did not release monomers from maxicircle aggregates (Fig. 4, compare lanes 1 and 2). Digestion with -N RNases A and T1 also had no effect (lane 3). Similarly, these treatments did not alter the appearance of the aggregates on --L EM (data not shown). However, topoisomerase II released circular monomers from Spe I digests (lane 4), indicating that the aggregates were catenanes. Another possible explanation for the maxicircle catenanes is that maxicircles were linked together by residual minicir- cles. This is highly unlikely, because (i) maxicircle catenanes were not affected by exhaustive digestion with Spe I and (ii) no interlocked minicircles were detected in electron micro- Spel + + + + graphs of 43 maxicircle catenanes containing a total of 550 Xexo - + + + maxicircles. Nevertheless, to test more rigorously for resid- Pst I - + - ual catenated minicircles, intact networks were digested first Topoll - - + with Spe I to produce maxicircle catenanes and linearized FIG. 5. Maxicircle aggregates contain no detectable minicircle minicircles (Fig. 5, lanes 1). To reduce the strong signal from DNA. Isolated kDNA networks were sequentially digested with linearized minicircles, the Spe I digest was treated with A enzymes, as indicated. Samples were taken after each reaction, exonuclease. The exonuclease had no effect on maxicircle divided, and analyzed for maxicircle (A) and minicircle (B) DNA, as catenanes (Fig. 5A, compare lanes 1 and 2) but converted described for Fig. 1. (A) Purified kDNA networks were digested with linearized minicircles into shorter fragments and revealed a Spe I (lane 1) and then A exonuclease (lane 2). The DNA was small population of nicked minicircle monomers (Fig. SB, phenol-extracted and ethanol-precipitated prior to incubation with Pst I, which cleaves maxicircles but not minicircles (lane 3), or with lane 2). These minicircles were resistant to Spe I, perhaps topoisomerase II (lane 4). Each lane contained 15 ng of kDNA. (B) because the nick was positioned in or near the Spe I recog- Lanes 1-4, as in A; lanes 5-12, Spe I-digested kDNA (0.00005, nition sequence; they were also resistant to A exonuclease, 0.0001, 0.00025, 0.0005, 0.001, 0.01, 0.1, and 1.0 dilutions, respec- which does not act at a nick (22). Finally, after removal ofthe tively, ofthe amount of DNA in lane 1) (15 ng). N, L, C, and arrows exonuclease, maxicircles were released from the catenanes as in Fig. 1. Downloaded by guest on September 23, 2021 Biochemistry: Shapiro Proc. Natl. Acad. Sci. USA 90 (1993) 7813 product if maxicircle catenanes contained interlocked min- of trypanosome network replication (10). Small maxicircle icircles. (In a similar experiment, covalently closed minicir- catenanes, like single-size networks, lack the dense core(s) of cle monomers were not released when Spe I digests were DNA unmistakably present in large structures. (iii) Smaller treated directly with topoisomerase II.) catenanes contain some covalently closed (nonreplicating) Based on serial dilutions of Spe I-digested networks (Fig. maxicircles, whereas the largest structures are composed 5B, lanes 5-12), the nicked circles in lane 2 represent <0.05% entirely of noncovalently closed maxicircles. of all minicircles, or two minicircles per kDNA network During the remarkable process of kDNA replication, two [assuming that each network contains 5000 minicircles (15)]. entirely dissimilar catenanes are synthesized within each Therefore, the signal in lane 2 should have doubled in lane 3 other, at the same time and in the same place, and perhaps by or 4 if maxicircle catenanes each contained just two residual the same replication machinery. Assembly of these inter- minicircles. No increase was detected. Lack of residual locked catenanes is a consequence of topoisomerase action, minicircles in maxicircle catenanes was confirmed by two which must be highly coordinated and tightly regulated. One further experiments. First, Spe I digests were treated with obvious control mechanism is suggested by the two topoi- Bgl II, which cuts minicircles once and maxicircles at three somerases found in C. fasciculata mitochondria. If trypano- sites. Maxicircles were entirely digested, but fuil-length some mitochondria also have two type II topoisomerases, minicircles were not evident. Second, Spe I digests were perhaps one is devoted principally to maxicircle catenanes and incorporated into agarose plugs and the free minicircles were the otherto minicircle catenanes; or one may catenate, and the selectively removed by electrophoresis. The maxicircle ca- other decatenate, circular monomers. In trypanosomes, the tenanes (which remained in the plugs) were then treated with maxicircle catenanes that coalesce late in replication may Pst I orBgl II prior to electrophoresis through an agarose gel. provide a mechanism, obviously distinct from that for min- Maxicircle DNA was fully digested but newly released min- icircles, to ensure equal distribution of maxicircles into the icircle DNA was not detected. daughter networks. Topoisomerase(s) may play a structural role in this process, possibly mediating the migration of DISCUSSION maxicircles to the center of the dividing network. These experiments provide definitive evidence that maxicir- The unique topology of kDNA provokes important ques- cles are not simply threaded, in a random fashion, through a tions on the advantages conferred by, the functional conse- scaffold of minicircles. Instead, they are linked to one quences of, and the mechanisms for transcribing and repli- another within the catenane of minicircles. Vigorous efforts cating these huge interlocked catenanes. to remove protein or RNA did not dismantle maxicircle catenanes (Fig. 4), and multiple different experimental strat- I thank Paul Englund for encouragement and thoughtful critiques, egies, with sensitivity ranging down to 0.2 minicircle per Tamara Doering and Pamela Talalay for carefully reading the manu- maxicircle structure, failed to detect residual minicircles script, Ken Stuart (cloned maxicircle DNA) and Ken Kreuzer (T4 (Fig. 5). T. equiperdum kDNA networks are therefore com- topoisomerase II) for generously supplying essential reagents, and posed of pure (minicircle or maxicircle) catenanes that are Larry Simpson for providing the contiguous T. brucei maxicircle extensively interlocked with one another, clearly an intrigu- sequence. This work was supported by the Public Health Service ing topological arrangement. Whether this arrangement oc- (AI28855), the John D. and Catherine T. MacArthur Foundation, and by a Faculty Development Award from the Pharmaceutical Manu- curs in other kinetoplastids remains difficult to assess be- facturers Association Foundation, Inc. Preliminary studies were cause T. equiperdum is unique in having minicircle DNA of made possible by a special award from the Burroughs Wellcome one sequence class. Foundation. The component catenanes of kDNA networks obviously differ from one another in the size and sequence of their 1. Englund, P. T., Hajduk, S. L. & Marini, J. C. (1982) Annu. Rev. monomers; further, they differ in the number of monomers Biochem. 51, 695-726. per catenane (thousands of mini- versus tens of maxicircles). 2. Stuart, K. (1983) Mol. Biochem. Parasitol. 9, 93-104. We can now appreciate that these catenanes are dissimilar in 3. Simpson, L. (1987) Annu. Rev. Microbiol. 41, 363-382. 4. Benne, R. (1990) Trends Genet. 6, 177-181. their topological organization. Whereas minicircle catenanes 5. Simpson, L. (1990) Science 250, 512-513. (14, 15) resemble a chain-link fence or fishnet, large maxicir- 6. Stuart, K. & Feagin, J. E. (1992) Int. Rev. Cytol. 141, 65-88. cle catenanes are rosettes, with a dense central core and 7. Ray, D. S. (1987) Plasmid 17, 177-190. protruding loops (Fig. 2). In networks from C. fasciculata, 8. Ryan, K. A., Shapiro, T. A., Rauch, C. A. & Englund, P. T. (1988) the "unit structure" of minicircle catenanes is a circle linked Annu. Rev. Microbiol. 42, 339-358. 9. Hajduk, S. L., Klein, V. A. & Englund, P. T. (1984) Cell 36, to three other circles (23). Thousands of these unit structures 483-492. linked together comprise the uniform grid of the minicircle 10. Hoeijmakers, J. H. & Weijers, P. J. (1980) Plasmid 4, 97-116. catenane. However, this unit structure cannot explain the 11. Shapiro, T. A., Klein, V. A. & Englund, P. T. (1989) J. Biol. Chem. core and loop arrangement of large maxicircle catenanes. 264, 4173-4178. One possible model for maxicircles requires that each circle 12. Melendy, T., Sheline, C. & Ray, D. S. (1988) Cell 55, 1083-1088. be linked, at least once, to virtually every other less 13. Shlomai, J., Zadok, A. & Frank, D. (1984)Adv. Exp. Med. Biol. 179, circle; 409-422. likely is a central circle, to which all other circles are linked. 14. Fairlamb, A. H., Weislogel, P. O., Hoeijmakers, J. H. J. & Borst, Several characteristics suggest that the heterogeneity of P. (1978) J. Cell Biol. 76, 293-309. maxicircle catenanes represents different stages of replica- 15. Riou, G. F. & Saucier, J.-M. (1979) J. Cell Biol. 82, 248-263. tion. (i) The size distribution ofmaxicircle catenanes parallels 16. Marini, J. C., Miller, K. G. & Englund, P. T. (1980) J. Biol. Chem. that of intact networks in various stages of replication. 255, 4976-4979. Maxicircle catenanes containing 16-20 monomers are most 17. Simpson, L. (1986) Int. Rev. Cytol. 99, 119-179. 18. Barrois, M., Riou, G. & Galibert, F. (1981) Proc. Natl. Acad. Sci. prevalent (Fig. 3A), analogous to the occurrence of single size USA 78, 3323-3327. networks (10-12 ,um in circumference; Fig. 3B). In both 19. Kreuzer, K. N. & Jongeneel, C. V. (1983) Methods Enzymol. 100, populations the maximum size is about twice the most 144-160. common size, consistent with the largest structures repre- 20. Stuart, K. D. & Gelvin, S. B. (1982) Mol. Cell. Biol. 2, 845-852. senting double-size forms in the final stage of replication. (ii) 21. Davis, R. W., Simon, M. & Davidson, N. (1971) Methods Enzymol. 21, 413-428. The topological arrangement of maxicircles in isolated cat- 22. Weiss, B. (1981) in The Enzymes, ed. Boyer, P. D. (Academic Press, enanes mirrors their arrangement in intact networks. The New York), pp. 203-231. rosette-like appearance ofthe largest catenanes bears striking 23. Rauch, C. A. (1991) Ph.D. dissertation (Johns Hopkins Univ., similarity to the clustered maxicircles seen in the final stages Baltimore). Downloaded by guest on September 23, 2021