J. Mol. Biol. (1990) 213, 763-775

lac Repressor Forms Stable Loops in Vitro With Supercoiled Wild-type lac D N A Containing all Three Natural lac Operators Elisabeth R. Eismannl- and Benno Miiller-Hill Institut fiir Genetik der Universitdt zu K61n Weyerlal 121, D-5000 K61n 41, F.R.G. (Received 10 October 1989; accepted 13 February 1990) We have analyzed protein-DNA complexes formed between lac repressor and linear or differently supercoiled lac DNA (802 or 816 base-pairs in length), which carry all three natural lac operators (O1, 02 and Oa) in their wild-type sequence context and spacing and compared them with constructs that contain specifically mutated "pseudo-operators" Oz or 03 . We used gel retardation assays to identify the nature of the complexes according to their characteristic electrophoretic mobility and dissociation rate measurements to determine their stability. With linear DNA we found only indirect evidence for loop formation between 01 and 02. In covalently closed DNA minicircles the formation of a loop between 01 and 02 could be demonstrated by the observation that 01-O2 containing DNA with low negative supercoiling (a= -0"013 and less) is constricted by binding of lac repressor, resulting in an increased electrophoretic mobility. At elevated negative supercoiling (a -- -- 0-025, --0"037, -0-05) 01-02 containing DNA complexed with lac repressor migrates significantly slower than the corresponding 01-DNA, indicating loop formation. The dissociation of lac repressor-operator complexes is decreased with increasing negative supercoiling for all tested operator combinations of 01, 02 and Oa. However, in the presence of at least two natural lac operators on the same DNA minicircle the enhancement of stability is particularly large. This indicates that a DNA loop is formed between these two lac operators, 01 and 02 as well as 01 and 03, since negative supercoiling is known specifically to promote the formation of looped structures. Additionally, we observe a dependence of dissociation rate on the spatial alignment of the operators as a result of changing helical periodicity in differently supercoiled DNA and consider this to be further evidence for loop formation between 01 and 02 as well as 01 and 03.

et al., 1987; Valentin-Hansen et al., 1986) and the gln operon (Reitzer & Magasanik, 1986). In the E. coli lac operon the first operator (Ol), positioned downstream from the promoter, is accompanied by two other operator-like sequences. The second/ac operator (O2), 401 bp$ downstream from the O1 within the 5'-coding region of the/acZ gene (Reznikoff et al., 1974), contains two relevant base-pair exchanges compared to 01 and binds lac repressor with one-fifth of the affinity found for 01 in vitro (Winter & yon Hippel, 1981). The third lac operator (Os) is located 92 bp upstream from O1 and overlaps with the binding site of the catabolite activator protein. Exchanges of several base-pairs result in a low affinity of lac repressor for Os, which by different methods is estimated to be in the range

1. I n t r o d u c t i o n

Proteins that regulate gene expression generally bind to specific DNA sequences close to or within the promoter and thereby influence the activity of RNA polymerase. Additionally, action over long distances is a common regulatory strategy in many prokaryotic and eukaryotic systems and is believed to be due to DNA looping between two distant binding sites (for reviews, see Ptashne, 1986; Schleif, 1987; Gralla, 1989). Evidence for regulation at a distance in prokaryotes came from Escherichia coli operons in which multiple protein binding sites are necessary to obtain full repression: e.g. the gal operon (Fritz et al., 1983; Irani et al., 1983), the araBAD operon (Dunn el al., 1984), the nrd operon (Tuggle & Fuchs, 1986), the deo operon (Dandanell

:~Abbreviations used: bp, base-pair(s); IPTG, isopropyl-fl-D-thiogalactoside; X-gal, 5-bromo-4-ehloro-3indolyl-fl-v-galactoside; PAA, polyacrylamide.

t Present address: Forsehungszentrum Jiilich

Postfaeh 1913, D-5170 Jiilieh, F.R.G. 0022-2836/90/120763-13 $03.00/0

763

1990 Academic Press Limited

764

E. R. Eismann and B. Miiller-Hill

of 16-fold to 1000-fold less than for O1 {Winter & yon Hippel, 1981; Fried & Crothers, 1981). Since /ac repressor-fl-galaetosidase chimaeras require only two lac repressor subunits to bind lac o p e r a , r , Kania & Miiller-Hfll {1977) proposed a model, in which tetrameric lac repressor simultaneously binds with two subunits to lac operator and with the other two subunits to another, lac operator-like, sequence. Meanwhile this proposal was strengthened'by the finding that a lac repressor dimer suffices for lac operator recognition (Lehming et al., 1988). Evidence in vitro for a ternary complex between one lac repressor molecule and two distant binding sites comes from several studies: (1) simultaneous binding of two small DNA fragments containing a /ac operator to a single /ac repressor tetramer have been demonstrated with circular dichroism measurements (Culard & Maurizot, 1981) and filter binding assays (O'Gorman et al., 1980); (2) association and dissociation rate constants of lac repressor and lac operator 01 are influenced by the presence of 02 and/or O3 on the same DNA fragment in a co-operative manner (Pfahl et al., 1979; Winter & yon Hippel, 1981; Fried & Crothers, 1981; Whitson et al., 1986; Whitson & Matthews, 1986; Hsieh et al., 1987); (3) the direct transfer of a repressor molecule between two binding sites on different DNA fragments was also explained by the formation of a ternary complex {Fried & Crothers, 1984). After the demonstration that insertion of an additional/ac operator upstream from a control region enhances repression (Besse et al., 1986; Mossing & Record, 1986; Bellomy et al., 1988), the reported in vitro data evidently became relevant for regulation in vivo. By specific destruction of 02 without altering the amino acid sequence of fl-galactosidase, we demonstrated that 02 indeed participates in repression of the lac operon in vivo (Eismann et al., 1987). We observed a fivefold decrease in repression of fl-galactosidase expression compared to that in wild-type. However, the underlying structural basis remained an open question. KrKmer et al. (1987) used non-denaturing polyacrylamide gel electrophoresis, DNase I protection experiments and electron microscopy to prove DNA loops between tetrameric lac repressor and two perfectly symmetric "ideal"/ac operators (Sadler et al., 1983; Simons et al., 1984) on one linear DNA fragment. Similarly as in the ara and 2-system (Dunn et al., 1984; Hochschild & Ptashne, 1986) loop formation depends on the correct phasing of the two lac operators. This spacing requirement apparently is changed in supercoiled DNA. Here extremely stable loops are formed at all operator distances tested (Kr~mer et al., 1988). The topology of DNA plays a critical role in many functions of DNA. Supercoiling results in a torsionally strained molecule, equivalent to an energetically activated state (Gellert, 1981), which probably facilitates DNA looping. The increased affinities of /ac repressor for supercoiled 2plac DNA compared to the relaxed equivalent (Wang et al., 1974, 1982)

was believed to result from a small unwinding due to repressor binding. Since the DNA that was used carries all natural lac operators (O1, 02 and O3) the effect might also be due to loop formation (see also Kim & Kim, 1982). With the use of a filter binding • ssay, decreases in dissociation rate constants of repressor-operator complexes as a function of supercoil density were observed for plasmids containing various portions of the lac operon (Whitson et al., 1987a,b). Plasmids carrying two primary operator sequences showed the lowest rate constants for dissociation from lac repressor. These findings can be explained by the formation of an intramolecular ternary complex. According to this model the looped DNA segment is stabilized by the combination of increased local concentration of binding sites and torsional stresses on the DNA, which favor binding in supercoiled DNA. Supercoiling also facilitates the formation of a DNA loop between 01 and O3 (Borowiec et al., 1987; Sasse-Dwight & Gralla, 1988) leading to a remarkable increase in the binding affinity of the lac repressor for 03. The formation of this complex is accompanied by structural changes of the intervening DNA, which can be probed with methylation protection experiments and indicate a D N A loop. Recently, Flashner & Gralla {1988) observed co-operative interaction of 01 and 02 in vivo. W e used gel retardation assays {Fried & Crothers, 1981; Garner & Revzin, 1981) to detect D N A loops between the natural lac operators 01 and O2, and between 01 and O~ in vitro. For this purpose we constructed D N A fragments and D N A minicircles that either contain all three natural lac operators at their natural spacing, or that lack 02 or 03 or both, without changing the distance between the remaining binding sites. This was achieved by means of oligonucleotide-directed mutagenesis. Covalently closed D N A minicircles with various degrees of negative supercoiling form D N A loops between 01 and 02 and probably also between 01 and O3. This is indicated by their characteristic mobility in non-denaturing polyacrylamide gel electrophoresis and by the increasing stability of protein-DNA complexes with increasing superhelical density compared to complexes with one operator (01) alone.

2. Materials and Methods

(a) Chemicals and enzymes [32p]Deoxyribonucleotides were obtained from Amersham Buchler (Braunsehweig, F.R.G.); isopropyl~-v-thiogalactoside (IPTG) and 5-bromo-4-ehloro3-indolyl-18-v-galactoside (X-gal) from Bachem Fine Chemicals (Torrance, U.S.A.); ATP, deoxyribonucleotides, dideoxyribonucleotides, dithiothreitol,ethidium bromide, Trizma base and Brij 58 from Sigma Chemie (Miinchen, F.R.G.); agarose and urea from Bethesda Research Laboratories Inc. (Eggenstein,F.R.G.); the chemicals used for automated D N A synthesis from Applied Biosystems (Pfungstazlt, F.R.G.); all other chemicals were obtained from Sigma Chemie (Miinchen,

Loops With Wild-type lac D N A

F.R+G.) or Merck (Darmstadt, F.R.G.). Restriction endonucleases and other enzymes were obtained from Boehringer (Mannheim, F.R.G.), New England Biolabs (Schwaibach, F.R.G.), Bethesda Research Laboratories Inc. (Eggenstein, F.R.G.) and Pharmacia (Freiburg, F.R.G.) and were used as recommended by the manufacturers. The chemicals used for DNA sequencing (Maxam & Gilbert, 1977) are as described by Biichel el al. (1980). Oligonucleotides were synthesized on an Applied Biosystems 380 A DNA synthesizer and purified on denaturing polyacrylamide gels prior to cloning and sequence analysis (Maxam & Gilbert, 1977; Chen & Seeburg, 1985). Purified lax represser was a gift from K. Beyreuther. (b) Construction of plasmids Plasmids pEel23, pEO103, p E e l 2 0 and pEOI00 are derivatives of the plasmid pEE4 (Eismann et al., 1987). In the modified polylinker they contain a PvuII-AatII fragment of the wild-type lax operon coding for the laxI N terminus (codon 348-360; overlapping with the 03-region), the lacPO control region (with 01) and the first 212 residues of the laxZ gene (with O2-region). The four variants differ in that they contain or lack 02 and/or 03. The digits 1, 2 and 3 indicate the presence of 01, O2 and 0 3, respectively, the digit 0 indicates the respective mutation (see aTso Fig. lib)). The wild-type control region and upstream sequences were derived from pUR2K (yon Wilcken-Bergmann, 1983) by insertion of the corresponding 178 bp PvuII-EcoRI restriction fragment between a suitable site of the polylinker, generating a NheI restriction site, and the EcoRI site created in the 5' terminus of the laxZ portion. The 619 bp EcoRI-AatII fragment from the 5'-coding region of laxZ of an 0 3 and a O~ plasmid (pEWO123 and pEWO103, respectively; see below) completes the inserts of pEO123]pEO120 and of pEO103/pEO100, respectively. Cutting within the polylinker with NheI and XbaI generates 816 bp (pEO123, pEel03) or 802 bp (pEOl20, pEO100) fragments, which are shorter as a consequence of 03-mutagenesis and subsequent procedures (see below). The oligonucleotide-directed mutagenesis of 02 is described by Eismann et al. (1987). Briefly, a MstII-AalII restriction fragment containing the 02 was subcloned into the polylinker of the vector pEE4. After generation of a 201 nucleotide long single:stranded gap (MstII-HpaII) covering the 02 sequence an oligonucleotide (50 nucleotides long) with the desired 8 point mutations was hybridized to the single-stranded 02-region. This "heteroduplex" DNA was subjected to DNA polymerase/DNA ligase reaction, transformed, amplified and retransformed in a suitable E. coli strain (MPS51; Pfahl el al., 1974) in which the mutated genotype is indicated by a colorless phenotype in the presence of X-gal, while wild-type 02 gives dark-blue colonies due to titration of episomal is lax represser. The 0~ fragment was then reinserted into the /acZ gene, generating pEWOl03, analogous to the O 3 plasmid pEWO123. The 03 mutagenesis essentially followed the same strategy but was performed in the plasmid pEO103. The gap, 180 nueleotides in length (NheI-EcoRI), covered 03, and the otigonucleotide (37 nueleotides) used for mutagenesis contained 4 point mutations. Since 03 is not able to efficiently titrate/ax represser, selection according to the above described principle was not possible. But the newly introduced XhoI and SslI restriction sites facilitated the identification of successfully mutated clones. In order to adjust the size of the NheI-XbaI fragments for circularization to exactly 802 bp, double the distance between 01and 02, the 39 bp

765

fragment between NheI and XhoI, upstream from the 03-region, was exchanged for a synthetic DNA fragment of 14bp, which resulted in the plasmid pE0100. To generate pEO120, the 166 bp NheI-EcoRI fragment of pEel23, containing 03, was replaced by the corresponding fragment of pE0100 containing 0~. Plasmid constructions were done according to standard procedures (Maniatis eta/., 1982). All constructions were verified by restriction mapping and sequence analysis (Maxam & Gilbert, 1977; Chen & Seeburg, 1985). (e) Generation of D N A fragments and D N A minicircles The plasmids were cut with NheI and XbaI and dephosphorylated with alkaline phosphatase. The 816bp (pEO123, pEel03) and the 802 bp (pEel20, pEO100) lax DNA containing restriction fragments were purified on 4% polyacrylamide (PAA) gels (Maniatis et al., 1982) and labeled with [T-32P]ATP using phage T4 polynueleotide kinase. In order to generate DNA minicircles, the labeled fragments (250 ng/ml) were ligated in 50 mM-Tris. HCl (pH 8"0), 6 mM-MgCl2, 20 mM-dithiothreitol, 1 mM-ATP with 1 unit T4 DNA ligase/ml for 12 h at room temperature (Nordheim & Meese, 1988). Ethidium bromide (0'25, 0"5, 0"75 and 1-0 #M) was added to the incubation buffer to obtain predominantly - 1 , - 2 , --3 and - 4 topoisomers, respectively (Wang, 1974). After ligation the DNA was extracted with phenol, precipitated with ethanol and subjected to gel electrophoresis in order to purify the individual topoisomers (Zivanovic et al., 1986; Nordheim & Meese, 1988). Topoisomers - 2 , --3, and - 4 were separated on 4% (w/v) PAA gels (acrylamide :bisacrylamide, 29 : 1) in 45 mM-Tris-borate, 1.5 mM-EDTA (pH 8"3) for 20 h at 12 V/cm. The large size of the DNA minicircles required a compound gel matrix for the separation of topoisomers 0 and - 1. The mixture contained 2-5o//0 PAA (29:1) and 0-5O/o agarose in 90 mm-Tris-borate, 3 mm-EDTA, 5 mM-MgCI2 (pH 8"3) for 5 h at 9 V/cm, and was prepared essentially according to the method of Peacock & Dingman (1968). MgCIz (5 raM) was added in order to differentiate between 0 and --1 topoisomers (Kr~mer et al., 1988). The presence of 0"3/~g ethidium bromide/ml during electrophoresis resulted in the separation of nicked and relaxed DNA minicircles. The gels were pre-run for 2 h to ensure a constant voltage gradient. The topoisomers were identified by their consecutive appearance in iigation reactions containing gradually increasing amounts of ethidium bromide. The topoisomer, which was obtained in the ligation reaction without any DNA binding agent, generally comigrated with the nicked minicircles, indicating that it was near to relaxation. This topoisomer (ALk~-0) was chosen as a reference. (d) Gel retardation assay Binding conditions were essentially as described :by Kr~mer et al. (1987, 1988) with the modification that the incubation time for DNA and protein was extended to 30 min. Binding buffer was 10 mM-Tris. HCI (pH 8"0), 10 mM-KCI, 10 mM-magnesium acetate, 0"1 mM-EDTA, 0"1 mM-dithiothreitol, 50/Jg bovine serum albumin/ml (Riggs et a/., 1970). When all samples had been loaded, eleotrophoresis was performed under the same conditions for a time depending on the matrix used. The analysis of complexes with the approximately 800 bp linear DN~ fragments took 6 h on 4 % PAA gels (29:1). For the separation of complexes with DNA minicircles of the same

766

E. R. Eismann and B. Miiller-Hill

size the mixtures were analyzed on compound 2-5% PAA/0"5% agarose gels for 8h or on 2% PAA/I% agarose gels for 6 h, or on 1-5% agarose gels for 4 h. The buffer was in each case 45 mM-Tris-borate, 1"5 mM-EDTA (pH 8"3). Gels containing PAA were pre-run for 2 h at a voltage gradient of 12 V/cm. After electrophoresis, gels were dried and autoradiographed at -70°C on Kodak X-Omat AR film.

Whe l

Eco R I

Xbo I

pE0123 [ [03[ I0,1

1021

]8,6bp

pEOI03 J l°3J loll

Ixl

I,.~ b,.

pEOnO I [Xl Iod

Io~1

I,o,b,,

Ixl

I,o~,,

~Eo,oo l'lxl I°,1 #-

92 -! . . . . . . . . . .

4o~ -~ ........

i

(a)

le) Measurement of dissociation rates Purified lac DNA fragments or topoisomers were incubated with lac repressor in a molar ratio of approximately l : l (determined by preceding titration experiments) for 30 min at room temperature in binding buffer (Kr~mer et al., 1987). The incubation was continued for the indicated times (min) in the presence of a large molar excess (1000 to 5000-fold) of unlabeled lac operator DNA. After the indicated time, dissociation was stopped by loading the incubation mixture on to a running gel. After electrophoresis, gels were dried and autoradiographed at -70°C on Kodak X-Omat AR film. For quantitative determinations, the autoradiographs were scanned on a Bio-Rad video densitometer 620. The percentage of bound and unbound DNA was calculated for the various time points and dissociation rates were determined as described by Riggs et al. (1970).

3. R e s u l t s

(a) Complexes with linear lac D N A and lac repressor To characterize the interaction between lac repressor and wild-type lac DNA we used nondenaturing, low percentage polyacrylamide gel electrophoresis (Fried & Crothers, 1981; Garner & Revzin, 1981), which allows the separation and identification of protein-free DNA and different prot e i n - D N A complexes due to their characteristic mobilities in the gel. The DNA fragments used contain all n a t u r a l / a c operators O1, 02 and Os in their natural sequence context and distance from each other (pEO123), the O~ mutation (pEO103), the O~ mutation (pEO120) or both mutations (pEO100). T h e y were produced by digestion of the respective plasmids with NheI and XbaI (Fig. 1(a)). The generation of the operator mutations b y oligonucleotide-directed mutagenesis and the construction of the plasmids is described in Materials and Methods. In Figure l(b) the sequences of the three operators and the two respective mutations are given.

Figure 2 shows the titration of the DNA fragments with increasing amounts of lac repressor. A complex with one bound lac repressor t e t r a m e r is defined b y the retarded band, which appears with p E O l 0 0 at all tested lac repressor concentrations (lanes p to f). In the presence of 03 (pEOl03) a second, slower-moving band is detectable at higher lac repressor concentrations (lanes i and j), p r o b a b l y indicating the additional binding of a second lac repressor t e t r a m e r to 03 on the same DNA fragment, since its gradual appearance coincides with the increase in /ac repressor concentration

Ol

S'--~AATTGTGAGC~GATAACAATT---3"

0 2

S'---AAATGTGAGCG

AGTAACAACC---

3'

0 2- S ' - - - q A ^ G G T T A A T G A A T A . G C A C C C - - - 3 "

03

S'---OGCAGTGAGCGCAACGCAATT---3

0 3

S . . . . T c GAT..CGAGCT C A A C G C A A T T - - -

' 3"

(b) Figure 1. (a) Schematic representation of DNA fragments used for gel retardation and for generation of DNA minicircles. ( x ) The region where the corresponding operator sequence was specifically destroyed by oligonucleotide-directed mutagenesis. Distances in bp from center to center of operator symmetry are given below. (b) Sequences of 01, 02, O~, Oa and 0~. The center of symmetry is marked by an asterisk. To facilitate comparison, the O~ and O~ sequences represent the lower strand in 5'-*3' direction. Bold capital letters indicate the homology to the Old sequence. Exchanged bases in the mutated sequences O~ and O~ are marked with a dot. The mutations of the O~ sequences are generated in part by oligonucleotide-directed mutagenesis (bold dots) and in part by fill-in reaction of the newly introduced XhoI restriction site (CTCGAG--*CTCGATCGAG) (thin dots).

( " t a n d e m " - s t r u c t u r e ; see also Discussion and Kr~mer et al., 1987). DNA fragments t h a t carry 01 and 0 2 (pEO123, pEO120) form no complex corresponding to the occupancy of a single operator by one lac repressor tetramer. Instead, at low lac repressor concentrations only a smear can be seen (lanes b and 1). Higher concentrations yield bands representing the " t a n d e m " structure (lanes c, d, e and m n, o). The small differences in mobility of the various t a n d e m bands originated from the independent occupancy of O1 and 02 (pEO123, lanes d and e; pEO120, lanes n and o) or 01 and 0 3 (pEO103, lanes i and j) result from the different localization of operators, as indicated by the symbols on the left side of Figure 2. Similar effects have been seen with DNA fragments carrying two "ideal" lac operators (Kr~mer et al., 1987). In the presence of all three lac operators (pEO123) a t v e r y high lac repressor concentrations, a third band is visible, which probably represents a complex with three independently bound /ac repressor tetramers per DNA molecule. We take the presence of a smear at low repressor concentrations as an indication of the formation of a t e r n a r y complex between one lac repressor t e t r a m e r and the two operators O1 and 0 2 in the incubation mixture. However, this complex is not stable enough to be resolved as a discrete band in the gel. At higher concentrations of lac repressor the t e r n a r y complex is displaced by the t a n d e m structure. In

Loops With Wild-type lac D N A

767

ng

O0 0 O0 0

IOC R

0 0

Figure 2. Titration of lac operator carrying linear DNA fragments with lac repressor on a 4 % PAA gel. The fragments were .prepared and end-labeled as described in Materials and Methods. DNA-fragments (2 fmol, 1 ng) were incubated with the indicated amounts of lac repressor (1 ng= 7 fmol) in l0/~1 of binding buffer. Incubation, electrophoresis and autoradiography were performed as described in Materials and Methods. The symbols on the top indicate the arrangement of the various lac operators (03, 0 l, 02 from left to right). The symbols on the left indicate the proposed structures (see the text). order to analyze the stability of preformed complexes, we determined their dissociation in the presence of a 5000-fold molar excess of unlabeled competitor DNA ("ideal" lac operator containing fragment, 38 bp in length). Dissociation was stopped by loading portions of the incubation mixture onto a running gel. The half-lives of those complexes, which are characterized-by discrete bands, were in the range one to two minutes (data not shown). Preformed complexes between lac repressor and various lac operator fragments could be induced to dissociate nearly completely by the addition of 1 mM-IPTG (data not shown).

(b) Topoisomeric minicircles of D N A for gel retardation assays DNA in E. coli is known to be supereoiled and the important role of supercoiling in DNA loop formation has been demonstrated repeatedly (Hahn et al., 1986; Whitson et al., 1987a,b; Borowiec et al., 1987; Kr~mer et al., 1988; Sasse-Dwight & Gralla, 1988). Hence, we decided to analyze the interaction between the three natural lac operators and lac repressor on DNA minicircles with increasing superhelical density by the technique of topoisomer gel retardation (Nordheim & Meese, 1988; Kr~mer et al., 1988). To approach the situation in vivo as closely as possible, the wild-type sequences and distances between 01 and 02 as welt as between O 1 and Oa were maintained. In order to avoid any interference of the expected loop formation between 01 and 02, the size of the DNA minicircles was designed to be at least 802 bp, twice the distance between O1 and 02. The ligation of the NheI-XbaI DNA fragment of pEO123 and pE0103 resulted in 816 bp DNA minicirctes, the pEO120 and pEO100 derivatives contain 802 bp as a consequence of 03

mutagenesis and subsequent alterations (as described in Materials and Methods) without changing the distance between 01 and 02. Topoisomeric DNA minicircles, that differ only in their linking number iLk) can be generated by the addition of increasing amounts of the intercalating agent ethidium bromide during the ligation reaction (Wang, 1974). In the case of the approximately 800 bp DNA minicircles, a linking number difference (ALk) of - 4 with respect to the relaxed topoisomer corresponds to the estimated superhelical density of chromosomal E. coli DNA in vivo of a - ~ - 0 " 0 5 (_0"01); that is, the number of superhelical turns per 10 bp (Sinden et al., 1980). Topoisomers with different linking numbers can usually be separated in the same low percentage polyacrylamide gels that are used for gel retardation assays (Shore & Baldwin, 1983; Horowitz & Wang, 1984; Zivanovic et al., 1986), since the difference in linkage (ALk) is composed of changes in helical twist (ATw) and in spacial writhe (AWr) (Fuller, 1971; Crick, 1976), which influence the electrophoretic mobility. However, the commonly used 4 % polyacrylamide gels turned out to be unsuitable for preparing quantitative amounts of 800 bp topoisomers 0 and - 1, because the bands do not separate sufficiently from each other even after 24 hours at 12V/cm. According to Peacock & Dingman (1968), the addition of agarose is recommended ff PAA concentrations lower than 4~/o are desired, and with such compound aerylamide]agarose gels, the separation was achieved (see Materials and Methods). Similar difficulties were encountered with the characterization of the protein-DNA complexes. By the use of certain types of compound gel, containing 2"5~o PAA and 0"5 ~/o agarose (Fig. 3; Fig. 4(a') and (b')) or 2 % PAA and 1~/o agarose (Fig. 4(a) and (b)), or with 1.5~/o agarose gels (Fig. 4(c), (d) and (e)), protein-bound DNA and free DNA could be separated.

768

E. R. Eismanr~ and B. Miiller-Hill (c)

(o)

o

120 1 ,- If) ~ J o

ng/OCR

--

nic

]I.,-

,.- m O410 , " U~ ~

nic

]i..

:..

io b c

.

.

#ef

¢ ~ i

:,

.

:..:

/ k /rn

.

nopq~

s t I

[b)

(d)

Figure 3. Titration of/ac operator carrying DNA minicircles with increasing superhelieal density with lac represser on 2"5%PAA/0"5°/o agarose compound gels. The DNA minicircles were prepared as described in Materials and Methods. Purified topoisomers (2 fmot, 1 ng) with the indicated linking number difference (AL/c)were incubated with the indicated amounts of lac represser in 10 pl of binding buffer. Incubation, electrophoresis and autoradiography were performed as described in Materials and Methods. As indicated by the symbols on the top of each panel the DNA minieireles contain (a) all 3 natural lac operators (123), (b) Ol and O3 (103), (c) O l and 02 (120), (d) only 01 (100). Positions of free nicked DNA (nic) and of complexes with increased mobility (~) are indicated (see the text).

The different sizes of the minicircles of pEO123 and pEO103 (816bp) compared to pEO120 and pEO100 (802 bp) have a remarkable influence on the mobility of 0 and - 1 topoisomers relative to the relaxed nicked DNA {Fig. 3(a) to (d), lanes a and e). The difference of half an integral number of helical turns (14 bp) obviously results in ligation products that differ in their hydrodynamic volume under these conditions. The influence of the matrix and the ionic conditions on the relative electrophoretic mobility (Zinanovic eta/., 1986) is reflected by the observation that, in compound gels with other concentrations, these topoisomers show an inverse electrophoretic behavior (Fig. 4(a) and (b)). In 2 % PAA/I~/o agarose gels, the 816bp topoisomer, isolated as that closest to the relaxed state (0), clearly separates from the nicked DNA (Fig. 4(a)), whereas the - 1 topoisomer now comigrates with nicked DNA (Fig. 4(b)). For the preparative separation of these topoisomers, compound gels with 5 mM-MgCI2 were used that sufficiently differentiate between both linking numbers (see Materials and Methods).

(c) Complexes with lac D N A minicircles and lac represser DNA minicircles with 01 alone are slightly retarded upon binding of lac represser, irrespective of their superhelical density (Fig. 3(d); 100). This also holds true for the nicked DNA minicircles, which unavoidably contaminate each topoisomer preparation. The co-migration of the 0 topoisomer with the nicked DNA indicates that it is near to relaxation (lanes a to d), while - 1 topoisomer is obviously far from relaxation under these conditions (lanes e to h). The presence of 01 and 02 on the same minicirele results in a different retardation pattern (Fig. 3(c); 120). The topoisomers - 2 , - 3 and - 4 are retarded to a larger extent than those of 01 minicireles (lanes i to t), whereas the 0 topeisomer show an increased eleetrophoretie mobility upon binding of lac represser (lanes a to d). These electrophoretic properties are characteristic for looped structures in covalently closed DNA minicircles (KrEmer et al., 1988; see Discussion). DNA minicircles with 01 and 03 show a behavior similar

Loops With Wild-type lac DNA

769

min fOC R

~Lk

0

.

.

.

.

I

..

,

,

-2

-3

-4

Figure 4. Dissociation of lac repressor from DNA minicircles with increasing superhelieal density. Purified topoisomers (2 fmol, 1 ng) with the indicated linking number difference (ALk) were incubated in the absence ( - ) or in the presence ( + ) of an equimolar amount of active la~ repressor (calculated from titration experiments as in Fig. 3) for 30 rain at room temperature. The incubation was continued for the indicated times (rain) in the presence of a molar excess (1000 to 5000-fold) of unlabeled lac operator DNA ((a) and (b): 38 bp ideal la~ operator fragment; (c), (d) and (e): 3000 bp DNA fragment with 2 ideal lac operators). The electrophoresis was performed in 2% PAA/I% agarose (a), (b), in 2"5O/o PAA/0'5 % agarose (a'), (b') or in 1-5% agarose (c), (d), (e). The buffer was in each case 45 mM-Tris-borate, 1-5 mM-EDTA (pH 8"3), to allow separation of all components of the mixture as fast as possible. For further details, see Materials and Methods. The symbols on the top of each panel indicate the arrangement of the various operators.

to t h a t of 01 minicircles in t h a t t h e y are retarded to a low extent (Fig. 3(b); 103). However, these results do not exclude the existence of a DNA loop between 01 and 03 (see Discussion). DNA minicircles with all three natural lac operators (Fig. 3(a); 123) also form loops between 01 and 02 according to the characteristic criteria outlined above. The mobility of DNA with low superhelical

density (0, - 1 topoisomer and nicked DNA as well) is accelerated by binding of lac repressor (lanes a to d, e to h); DNA with moderate and high superhelical density ( - 2 , - 3 , - 4 topoisomer) is retarded to a degree, significantly higher t h a n t h a t for 01-03 minicircles (lanes i to l, m to p, g to t). This strongly suggests t h a t a.loop is formed between 01 and 02 at all tested lac repressor concentrations. These loops

E. R. Eismann and B. Miiller-Hill

770

seem to be more stable than those formed with less supereoiled DNA as indicated by the sharper bands. However, it should be emphasized that even the circularization itself suffices to facilitate DNA loop formati0n and allows the ternary complex to move as a detectable band in the gel, which is not observed with linear DNA (see Fig. 2). This might be explained by the spacial proximity of 01 and 02 as a eonse~luence of circularization. (d) Stability of complexes with lac repressor and lac D N A minicircles In order to analyze the stability of the complexes, we determined the dissociation rate of the pre-

(o)

formed complexes in the presence of a large molar excess of competitor DNA containing "ideal" lac operator sequences (Riggs et al., 1970). We used gel retardation as described above to differentiate between protein-free and protein-bound DNA (Fig. 4). For each individual topoisomer the gel matrix was chosen that yields fast and good separation of bound and unbound DNA to avoid dissociation and protein degradation during electrophoresis. For 0 and - 1 topoisomers, compound gels of various concentrations were used (Fig. 4(a), (a'), (b) and (b')), while the --2, - 3 and - 4 topoisomers were analyzed on 1"5% agarose gels (Fig. 4(e), (d) and (e)). Different competitor DNAs were necessary depending on the mobility of the DNA minicireles in

b)

123

I00.

I00"

50

50

103

[~-

•4 ~ . ~ ...........

~.,

0

\"

---.,...~

~

\\

•

~)

"x

......

~

~" 0

\ \ m "'"

\\.~j

5 ,;

.c:

g,

(c) e2o

(

~m

I( I00

}

io

III~]

d)

I00

(

=

J

~

I00-~

°°[

SO "°'"0"""

2

t~ \~

"~.

I0L

ri \ i

I .

-2

~

'..-3 \\

..... i

;1'5:30

' 60

120

180

s ,5 Time (rain)

Figure 5. Rates of dissociation of preformed complexes of/ac repressor with DNA minicireles. From autoradiographs (as in Fig. 4) the percentage of bound DNA was determined densitometrically (Bio-Rad video densitometer model 620) for the various DNA minicircles: (a) 01, 02 and 03 (123), (b) 01 and 03 (103), (c) 01 and 02 (120), (d) 01 (100) and each individual topoisomer: ALk=0 ( - - O - - ) ; A L k = - I (. . . . . IN . . . . ); A L k = - 2 ( S ); A b k = - 3 (...../%.....) ALk-- - 4 ( - - - A - - - ) . The values for 180 min ((a) and (b)) were inferred from a separate experiment.

Loops With Wild-type lac D N A 4. D i s c u s s i o n

Table 1

Half-lives in minutes of complexes of lac repressor with D N A minicircles calculated from logarithmic plots in Figure 5 ALk

123

103

120

100

010203

0103

0102

01

0 -1 -2

45 (50) 30 (70) 160

10 (40) 50 (40) lO (40)

-3 --4

220 270

75 (50) 95 (70)

771

65 (40) 35 (30) 65 (85) 340 530

--2. This dependence of dissociation rate on the spatial alignment of the operators also illustrates that the nature of the complexes we observe is indeed a looped structure, at least concerning O1 and O2 and probably also O1 and 0 3. In contrast, the stability of complexes with O1 minicircles gradually increases with increasing supercoil density, because there exist no spatial restrictions for the loops, lac repressor forms between 01 and any non-specific DNA in the vicinity. (d) Conclusions The participation of 02 in repression in vivo (Eismann et al., 1987) and its significant role for the formation of stable lac repressor-lac DNA complexes in vitro (this work) confirm the early proposal that tetrameric lac repressor interacts cooperatively with two binding sites by looping out the intervening DNA to regulate transcription of the wild-type lac operon. Our in vitro data suggest that, in vitro, a loop can also be formed between O1 and 03. A DNA loop may strengthen repression in several ways. It increases the local concentration of lac repressor in the vicinity of O1 and thereby enhances the degree of occupation at O1, which overlaps with the promoter sequence, and inhibits the start of transcription. Furthermore, the co-operative interaction may also promote the occupation of 0 2 itself, which would interfere with transcriptional elongation. This mechanism is also proposed by Flashner & Gralla (1989) on the basis of expression studies in vivo. A DNA loop between the two closely adjacent binding sites O1 and O3 would drastically change the DNA conformation (see also Borowiec et al., 1987), which would certainly influence the affinity of RNA polymerase to the promoter sequence (Spassky et al., 1988). Finally, co-operativity between 01 and O a leads to an increased occupation of 03, which would interfere with activation of transcription due to its overlap with the catabolite activating protein binding site. We analyze and discuss these different ways of interactions in more detail in a study about the contribution of each

particular lac operator to repression in vivo (Oehler et al., 1990). The importance of negative supercoiling in this complex regulatory system would be to facilitate loop formation by providing the energy for the required conformational alterations in the double helix. We thank K. Beyreuther for the gift of lae repressor, H. Kr~/mer and B. yon Wileken-Bergmann for discussions, and B. Jack for critically reading the manuscript. This research was supported by grants from Deutsche Forschungsgemeinschaft through SFB 74 and 243.

References Bellomy, G. R., Mossing, M. C. & Record, M. T., Jr (1988). Biochemistry, 27, 3900-3906. Besse, M., yon Wilcken-Bergmann, B. & Mfiller-Hill, B. (1986). EMBO J. 5, 1377-1381. Borowiec, J., Zhang, L., Sasse-Dwight, S. & Gralla, J. D. (1987). J. Mol. Biol. 196, 101-111. Biiehel, D. E., Gronenborn, B. & Miiller-Hill, B. (1980). Nature (London), 283, 541-545. Chen, E. J. & Seeburg, P. H. (1985). DNA, 4, 165-170. Crick, F. H. C. (1976). Proc. Nat. Acad. Sci., U.S.A. 73, 2639-2643. Culard, F. & Maurizot, J. C. (1981). Nucl. Acids Res. 9, 5175-5184. Dandanell, G., Valentin-Hansen, P., Love Larsen, J. E. & Hammer, K. (1987). Nature (London), 325, 823-826. Dunn, T. M., Hahn, S., Ogden, S. & Schleif, R. F. (1984). Proc. Nat. Acad. Sci., U.S.A. 81, 5017-5020. Eismann, E., yon Wilcken-Bergmann, B. & Miiller-Hill, B. (1987). J. Mol. Biol. 195, 949-952. Flashner, Y. & Gralla, J. D. (1988). Proc. Nat. Acad. Sci., U.S.A. 85, 8968-8972. Fried, M. & Crothers, D. M. (1981). Nucl. Acids Res. 9, 6505-6525. Fried, M. & Crothers, D. M. (1984). J. Mol. Biol. 172, 263-282. Fritz, H.-J., Bickn~se, H., Gleumes, B., Heibach, C., Rosahl, S. & Ehring, R. (1983). E M B O J . 2, 2129-2135. Fuller, F. B. (1971). Proc. Nat. Acad. Sci., U.S.A. 58, 815-819. Garner, M. M. & Revzin, A. (1981). Nucl. Acids Res. 9, 3047-3060. Gellert, M. {1981). Annu. Rev. Biochem. 50, 879-910. Goulet, I., Zivanovic, Y. & Prunell, A. (1987). Nucl. Acids Res. 15, 2803-2821. Gratla, J. D. (1989) Cell, 57, 193-195. Hahn, S., Hendrickson, W. & Schleif, R. F. (1986). J. Mol. Biol. 188, 355-367. Hoehschild, A. & Ptashne, M. (1986). Cell, 44, 681-687. Horowitz, D. S. & Wang, I. C. (1984). J. Mol. Biol. 173, 75-91. Hsieh, W.-T., Whitson, P. A., Matthews, K. S. & Wells, R. D. (1987). J. Biol. Chem. 262, 14583-14591. Irani, M. H., Orosz, L. & Adhya, S. (1983). Cell, 32, 783-788. Kania, J. & Miiller-Hill, B. (1977). Eur. J. Biochem. 79, 381-386. Kim, R. & Kim, S.-H. (1982). Cold Spring Harbor Syrup. Quant. Biol. 47, 451-454. Kr~mer, H., Niemiiller, M., Amouyal, M., Revet, B., yon Wileken-Bergmann, B. & Miiller-Hill, B. (1987). EMBO J. 6, 1481-1491.

Loops With Wild-type lac D N A Kr~mer, H., Amouyal, M., Nordheim, A. & Miiller-Hill, B. (1988). E M B O J . 7, 547-556. Le Bret, M. (1984). Biopolymers, 23, 1835-1867. Lehming, N., Sartorius, J., Niem511er, M., Genenger, G., yon Wilcken-Bergmann, B. & Miiller-Hill, B. (1987). EMBO J. 6, 3145-3153. Lehming, N., Sartorius, J., Oehler, S., yon Wilcken-Bergmann, B. & Miiller-Hill, B. (1988). Proc. Nat. Acad. Sci., U.S.A. 85, 7947-7951. Majumdar, A. & Adhya, S. (1984). Proc. Nat. Acad. Sci., U.S.A. 81, 6100-6104. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982). Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Maxam, A. & Gilbert, W. (1977). Proc. Nat. Acad. Sci., U.S.A. 74, 560-564. Mossing, M. C. & Record, T. M., Jr (1986). Science, 233, 889-892. Nordheim, A. & Meese, K. (1988). Nuci. Acids Res. 16, 21-37. Oehler, S., Eismann, E. R., Kritmer, H. & Miiller-Hill, B. (1990). EMBO J. 9, 973-979. O'Gorman, R. B., Dunaway, M. & Matthews, K. S. (1980). J. Biol. Chem. 255, 10100-10106. Peacock, A. C. & Dingman, C. W. (1968). Biochemistry, 7, 668-674. Pfahl, M., Stockter, C. & Gronenborn, B. {1974). Genetics, 76, 669-679. Pfahl, M., Guide, V. & Bourgeois, S. (1979). J. Mol. Biol. 127, 339-344. Ptashne, M. (1986). Nature (London), 322, 697-701. Reitzer, L. J. & Magasanik, B. (1986). Cell, 45, 785-792. Reznikoff, W. S., Winter, R. B. & Hurley, C. K. (1974). Proc. Nat. Acad. Sci., U.S.A. 71, 2314-=2318. Riggs, A. D., Suzuki, H. & Bourgeois, S. (1970). J. Mol. Biol. 48, 67-83. Sadler, J. R., Sasrnor, H. & Betz, J. L. (1983). Proc. Nat. Acad. Sci., U.S.A. 80, 6785-6789.

775

Sasse-Dwight, S. & GraUa, J. (1988). J. Mol. Biol. 202, 107-119. Schleif, R. (1987). Nature (London), 327, 369-370. Shore, D. & Baldwin, R. L. (1983). J. Mol. Biol. 170, 983-1007. Shore, D. Langowski, J. & Baldwin, R. L. (1981). Proc. Nat. Acad. 8ci., U.S.A. 78, 4833-4837. Simons, A., Tils, D., yon Wilcken-Bergmann, B. & MiillerHill, B. (1984). Proc. Nat. Acad. Sci., U.S.A. 81, 1624-1628. Sinden, R. R., Carlson, J. O. & Pettijohn, D. E. {1980). Cell, 21,773-783. Spassky, A., Rimsky, S., Bue, H. & Busby, S. (1988). E M B O J . 7, 1871-1879. Tuggle, C. K. & Fuchs, J. A. (1986). E M B O J . 5, 1077-1085. Valentin-Hansen, P., Albrechtsen, B. & Love-Larsen, J. E. 0986). EMBO J. 5, 2015-2021. yon Wilcken-Bergmann, B. (1983). Ph.D. dissertation, Universiti~t zu KSln. Wang, J. C. (1974). J. Mol. Biol. 89, 783-801. Wang, J. C. {1985). Annu. Rev. Biochem. 54, 665-697. Wang, J. C., Barkley, M. D. & Bourgeois, S. (1974). Nature (London), 251, 247-249. Wang, J. C., Peck, L. J. & Becherer, K. (1982). Cold Spring Harbor Syrup. Quant. Biol. 47, 85-91. Weintraub, H. (1985). Cell, 42, 705-711. Whitson, P. A. & Matthews, K. S. (1986). Biochemistry, 25, 3845-3852. Whitson, P. A., Olson, J. S. & Matthews, K. S. (1986). Biochemistry, 25, 3852-3858. Whitson, P. A., Hsieh, W.-T., Wells, R. D. & Matthews, K. S. (1987a). J. Biol. Chem. 262, 4943-4946. Whitson, P. A., Hsieh, W.-T., Wells, R. D. & Matthews, K. S. (1987b). J. Biol. Chem. 262, 14592-14599. Winter, R. B. & yon Hippel, P. H. (1981). Biochemistry, 20, 6948-6960. Zivanovie, Y., Goulet, I. & Prunell, A. (1986). J. Mol. Biol. 192, 645-660.

Edited by R. Schleif