Jul 18, 1977 - were centrifuged at 200 for 5 min at 3000 rpm in a Sorvall GS3 rotor. This step pelleted over 98% of the large (DNA-containing) cells, but left ...
Proc. Nati. Acad. Sci. USA Vol. 74, No. 10, pp. 4228-4232, October 1977 Biochemistry
Nonspecific DNA binding of genome-regulating proteins as a biological control mechanism: Measurement of DNA-bound Escherichia coli lac repressor in vivo* (minicells/repressor-DNA interactions/effective ion activity in E. coli/reprressor-inducer-operator-DNA binding constants) --" .
2
.
2
.
.
YING KAo-HUANG, ARNOLD REVZINt, ANDREW P. BUTLERt, PAMELA O'CONNER§, DANIEL W. NOBLE, AND PETER H. VON HIPPEL Institute of Molecular Biology and Department of Chemistry, University of Oregon, Eugene, Oregon 97403
Communiated by V. Boekelheide, July 18,1977
scription of the operon by DNA-dependent RNA polymerase (5-7). The affinity of repressor for operator in vitro, and the level of repression of the operon in vivo, are allosterically modulated by the binding to repressor of small molecule inducers structurally related to allolactose, the "natural" inducer which occurs as an early intermediate in the metabolic degradation of lactose (5, 8). Nonoperator DNA also binds lac repressor, albeit much more weakly than does operator (9-12), and this nonspecific affinity is not changed by the binding of inducer to the repressor (9, 11, 12). Furthermore, RNA polymerase also binds to DNA sites other than promoters, and its affinity for nonpromoter sequences may be modified by interaction with a factor (1, 4, 13). Based on these observations and direct in vitro measurement of most of the relevant binding constants, a model has been developed to account quantitatively for the observed levels of repression and derepression of the lactose operon; this model includes, as a central feature, nonspecific binding of regulatory proteins to the E. coli chromosome (1, 11). The validity of such models is not crucially dependent on knowledge of the absolute binding constants of (e.g.) the lac repressor (R) and repressor-inducer complex (RI) to operator (0) and nonoperator DNA (D) sites; it is the relative binding constants that matter, and these have been established by a variety of in vitro measurements (see ref. 1). However, these binding constants are very dependent on salt concentration; for example, the association constant for lac repressor with nonoperator DNA is decreased over 1000-fold by increasing the Na+ concentration from 0.12 to 0.24 M (12, 14). Thus, because the effective ionic activity of the intracellular environment of E. coli is not known, the possibility exists that it might be high enough to totally dissociate lac repressor (and perhaps other regulatory proteins as well) from the nonspecific DNA sites. Then, of course, all proteins not specifically bound to target sites would be free in the cytoplasm, and coupled equilibria involving nonspecific sites would be inoperative. To determine the fraction of lac repressor molecules that is bound to the bacterial chromosome in vivo, we have measured, and report here, the distribution of repressor between parent cells and the DNA-free "minicells" formed during cell division by the minicell mutant of E. coli isolated by Adler et al. (15). Abbreviations: R, repressor; I, inducer; 0, operator; D, DNA; IPTG,
ABSTBACr Binding of genome regulatory proteins to nonspecific DNA sites may play an important role in controlling the thermodynamics and kinetics of the interactions of these proteins with their specific target DNA sequences. An estimate of the fraction of Escherchia coli lac repressor molecules bound in vivo to the operator region and to nonoperator sites on the E coli chromosome is derived by measurement of the distribution of repressor between a minicell-producing E. coli strain (P678-54) and the DNA-free minicells derived therefrom. Assuming the minicell cytoplasm to be representative of that of the parent E. coli cells, we find that less than 10% of the repressor tetramers of the average cell are free in solution; the remainder are presumed to be bound to the bacterial chromosome. The minimum in vivo value of the association constant for repressor to bulk nonoperator DNA (Kim) calculated from these results is about 103 Mel, and analysis of the sources of error in the minicell experiment suggests that the actual in vivo value of KRD could be substantially greater. The value of KRD, coupled with in vitro data on the ionic strength dependence of this parameter, can be used to estimate that the effective intracellular cation activity of E. coli is no greater than about 0.24 M (and probably no less than 0.17 M) in terms of sodium ion equivalents. Te minicell distribution experiments also confirm that the association constant for the binding of inducer-repressor complex to bulk nonoperator DNA (KRM) is C KRD in vivo. These results are used to calculate minimum in vivo values of KRO and KRpO (association constants for repressor and for inducer-repressor complex binding to operator) of about 1012 M-l and about 109 M- , respectively. The results fit a quantitative model for operon regulation in which nonspecific DNArepressor complexes play a key role in determining basal and constitutive levels of gene expression [von Hippel, P. H., Revzin, Ai, Gross, C. A. & Wang, A. C. (1974) Proc. NatL Acad. Sci. USA 71,4808-4812]. Most well-characterized genome-regulating proteins, in addition to binding tightly to their specific target DNA.sequences (e.g., operators, promoters), show an appreciable "nonspecific" affinity for other DNA regions as well. Though this binding is much weaker than that to the target sites, it has been suggested that the great preponderance of nonspecific binding sites may complex, in vivo, most of the regulatory protein molecules not bound at functional sites. This binding will then control the concentration of regulatory protein free in the cell, and as a consequence participate in the regulation of physiological function (1-4). The lactose operon of Escherichia coli is currently the best molecularly characterized genome regulatory system. The lac repressor controls the level of expression of the lac genes by binding tightly to the operator region and preventing tran-
isopropyl-ft-D-thiogalactoside. * The preceding paper in this series is ref. 1.
t Present address: Department of Biochemistry, Michigan State Uni-
versity, East Lansing, MI 48823. * Present address: Division of Biology, Oak Ridge National Laboratory, Oak Ridge, TN 37830. § Present address: Department of Molecular, Cellular and Developmental Biology, University of Colorado, Boulder, CO 80302.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U. S. C. §1734 solely to indicate this fact.
4228
BiochemWq: ICao-Huang et al.
Proc. Natl. Acad. Scf. USA 74 (1977)
35%" pellet was resuspended in 0.7 ml of TMS buffer and divided into two equal portions. These were assayed for repressor by measuring bound radioactive inducer according to the procedure of Bourgeois (19): [14C]IPTG was added to one half of each extract, while [14C]IPTG plus a 50-fold excess of unlabeled IPTG was added to the other half to permit correction for inducer that may have been mechanically trapped in the precipitate rather than actually bound to repressor. Each sample was made 35% saturated in ammonium sulfate and centrifuged, and the pellet was washed three times with 50% saturated ammonium sulfate, resuspended in TMS buffer, and assayed for radioactivity in a Packard Tri-Carb liquid scintillation counter. Net counts (and thus repressor concentrations) were determined by subtracting the cpm in the control sample containing unlabeled IPTG from the cpm in the corresponding sample containing only radioactive IPTG. P-Galactosidase Assay. #l-Galactosidase activity was monitored by following the hydrolysis of o-nitrophenyl-f3-D-galactoside as described by Platt et al. (20). Aliquots of cell suspension were added to Z buffer to a final sample volume of 1 ml. The cell walls were disrupted either by adding a drop of toluene, vortex mixing for 10 sec, and allowing the toluene to evaporate while shaking at 370 in a shaker bath, or by lysing the cells with lysozyme. Samples were incubated-at 280 and 0.2 ml of the nitrophenyl galactoside (4 mg/ml) was added. The reaction was timed, and after development of sufficient yellow color, 0.5 ml of a 1 M Na2CO3 solution was added to stop the reaction. Optical densities were read at 420 nm and corrected for light scattering by the cell suspension, using 1.75 times the OD recorded at 550 nm (20).
Our data imply that virtually all repressor molecules are bound to DNA, and that nonspecific binding of regulatory proteins must therefore indeed be considered in formulating quantitative descriptions of genome control processes. Furthermore, these results can be used to estimate minimum in dvo values of KRO and KRIO (association constants for the binding of repressor and repressor-inducer complexes to operator). MATERIALS AND METHODS Chemicals and Buffers. All chemicals were reagent or enzyme grade. Unlabeled isopropyl-fl-D-thiogalactoside (IPTG) was purchased from Aldrich, [14C]IPTG from Schwarz/Mann, and penicillin G from Eli Lilly. Lysing buffer contained 0.2 M Tris, 0.2 M KCI, 0.1 mM EDTA, 6 mM 2-mercaptoethanol, and 5% (vol/vol) glycerol, titrated to pH 7.4 with HCl. TMS buffer contained 0.01 M Tris, 0.01 M magnesium acetate, 0.1 mM EDTA, 6 mM 2-mercaptoethanol, 0.2 M KCI, and 5% (vol/vol) glycerol, titrated to pH 7.5 with HCl. Z buffer (used in assaying (i-galactosidase) contained 0.06 M Na2HPO4, 0.04 M NaH2PO4, 0.01 M KCI, 1 mM MgSO4, and 0.05 M 2-mercaptoethanol, at pH 7.0. Bacterial Strains and Media. The minicell-producing E. coli strain K-12 P678-54 (F- lac + ) was the generous gift of Dr. Howard Adler. For comparative purposes, strain CSH46 [Fara A(lac-pro) thi(XcI857St68h80dkscIZ)] (also known as M96) was used to produce repressor in large quantities. Both strains were grown in rich broth containing yeast extract (22 g/liter), NaCl (5 g/liter), and tryptone (36 g/liter); proline (100 mg/liter) and beef bouillon (1 cube/liter) were added for growth of CSH46. Isolation of Minicells. Strain P678-54 was grown in a 9-liter fermentor at 370 to early stationary phase (OD6so = 2). The cells were centrifuged at 200 for 5 min at 3000 rpm in a Sorvall GS3 rotor. This step pelleted over 98% of the large (DNA-containing) cells, but left 10-20% of the minicells in the supernatant (16). The minicells were then pelleted by centrifuging for 10 min at 7000 rpm in the GS3 rotor (at 200), and resuspended in 300 ml of fresh broth to which pencillin had been added to 50 units/ml. The cells were incubated at 370 for 2 hr in this medium; it has been shown that this concentration of penicillin has no observable morphological or physiological effect on the. minicells (16, 17), while the parent (DNA-containing) cells become grossly elongated (17). The large filamentous DNAcontaining cells were then removed by centrifugation in a Sorvall GSA rotor at 3500 rpm for 5 min, the supernatant was sonicated five times for 10 sec each in an MSE sonicator to destroy any residual parent cells (17), and the minicells were pelleted as described above. Counts of minicells were made microscopically using a Petroff-Hauser counting chamber; residual viable parent cells were determined by direct counting and by plating samples of purified minicell preparations. Extraction and Assay of Repressor. Penicillin-treated minicells or parent cells (1.5-2 g) were resuspended in four volumes of lysing buffer. Small amounts of lysozyme and DNase I were added and the cells were stirred in the cold for 1 hr, after which the solution was made 0.01 M in magnesium acetate to activate the added DNase. After incubation for an additional hour the cell debris was pelleted by centrifuging for 25 min at 16,000 rpm (Sorvall SS34 rotor). An aliquot of supernatant was removed for Lowry protein assay (18), and the remaining suspension was brought to 115 mg/ml (23% saturated) ammonium sulfate, stirred 30 min at 40, and centrifuged (16,000 rpm, 15 min, SS34 rotor). The supernatant was then brought to 205 mg/ml (35% saturated) ammonium sulfate, stirred for 30 min at 40, and centrifuged as above. This "23 to
4229
RESULTS The experiments reported here permit measurement of the distribution of lac repressor in the E. colf cell; determination of the repressor content of the DNA-containing parent (P678-54) cells provides a measure of total repressor in the cell, while the repressor content of the DNA-free minicells measures cytoplasmic (free) repressor only. Control experiments, involving the addition of known amounts of purified repressor to the cell extract and comparative experiments with other E. coli lac+ strains [following the procedures of Jobe et al. (21)] were used to test and calibrate the method and to establish that the cells of E. coli strain P678-54 do indeed contain wild-type levels of repressor (about 10 to 20 tetramers per cell). To compare the repressor content of 'P678-54 parent and minicells it is necessary to express repressor content per unit cytoplasm, because comparisons based on cell weight or volume would be biased by the relatively greater proportion of cell wall and membrane materials present in the minicells. To this end intracellular repressor contents have been calculated per mg of soluble protein in each sample, as determined by the Lowry procedure (18).¶ Table 1 shows that the minicell preparations exhibit repressor levels that are 10% or less of the levels characteristic of the parent cells. Because only one (or two to three, depending on growth conditions and stage of chromosomal replication) repressor molecule(s) should be bound to the operator site(s), we infer that almost all the rest of the repressor molecules must be I
The normalization of intracellular repressor contents of parent and minicells in units of mg of soluble protein is further validated by the fact that non-genome-bound proteins such as fl-galactosidase in induced cells (refs. 15 and 22 and our unpublished experiments) are found to be approximately equally distributed between parent and minicells when concentrations are expressed on this basis.
4230
Proc. Nati. Acad. Scf. USA 74 (1977)
Biochemistry: Kao-Huang et al.
Table 1. lac repressor content of minicells and parent cells of E. coli K-12 P678-54 (F- lac+)
Sample,t cpm
Controlt cpm
Net counts,§ cpm/mg
Repressor distribution ratio, minicells to parent cellsl
