Genetic studies of the lac repressor. X. Analysis of missense mutations in the lacI gene.

Approximately 2000 non-suppressible mutations in the lacI gene of Escherichia coli have been extensively analyzed. The majority consists of missense mutations resulting in amino acid substitutions in the lac repressor. We characterized each mutation with respect to the resulting altered phenotype, and also mapped them against a large set of deletions. The correlation of the genetic and physical map reported previously has been used to localize the part of the protein affected by each mutation with a high degree of precision (within several amino acids). In particular, we examined the distribution of mutational sites along the gene leading to the i^?, i^s, i^r and i^ts phenotypes. Certain regions of the protein, such as the amino-terminal end, are very sensitive to amino acid exchanges with regard to the i^? phenotype, whereas other regions are relatively insensitive to substitutions. Of particular interest is the C-terminal half of the gene-protein map, where many I^s, and I^ts mutational sites cluster in very small regions separated by distinct and nearly regularly spaced intervals. The possible significance of these results with respect to repressor structure and function, and to protein structure in general, is discussed. In the following paper we consider the results reported here together with the data from suppressed nonsense mutations, which are described in the preceding paper.     

DNA-binding site of lac repressor probed by dimethylsulfate methylation of lac operator.

In order to compare the structures of the DNA-binding sites on variants of the lac repressor, we have studied the influence of these variants on the dimethylsulfate methylation of the lac operator. Since a bound protein changes the availability of specific purines in the operator to this chemical attack, comparisons of the methylation patterns will show similarities or differences in the protein DNA contacts. We compared lac repressor, induced lac repressor (repressor bound to the gratuitous inducer isopropyl-β-d-thiogalactoside), mutant repressors having increased operator affinities (X86, I12 and the X86-I12 double mutant) and repressor peptides (long headpiece, residues 1 to 59 and short headpiece. residues 1 to 51). All of these repressors and repressor peptides exhibit the same general pattern of protection and enhancement in the operator; however, the short headpiece pattern differs most from that of the repressor while the induced repressor and the long headpiece show intermediate patterns that are strikingly similar to each other. The mutant repressors do not show an isopropyl-β-d-thiogalactoside effect but otherwise are almost indistinguishable from wild-type repressor. These results demonstrate that all molecules bind to the operator using basically the same protein-DNA contacts; they imply that (1) most and possibly all repressor contacts to operator lie within amino acids 1 to 51, (2) inducer weakens many contacts rather than totally disrupting one or even a few and (3) the tight-binding mutants do not make additional contacts to the DNA.These results are consistent with a model in which the amino-terminal portions of two repressor monomers make the DNA contacts. We show that one can understand the affinity of binding as related to the accuracy of the register of the two amino-terminal portions along the DNA. Furthermore, the action of inducer and the behaviour of the tight binding mutants can be accomodated within a two-state model in which the strongly or weakly binding states correspond to structures in which the amino-terminal regions are rigidly or loosely held with respect to each other.     

Genetic studies of the lac repressor. IX. Generation of altered proteins by the suppression of nonsence mutations.

We have generated more than 300 altered lac repressor proteins carrying known amino acid replacements, by employing nonsense mutations at 90 positions in the lacI gene together with eight different nonsense suppressors. This allows the substitution of lysine, serine, tyrosine, leucine and glutamine at virtually all of the respective positions in the repressor, and tryptophan at ten positions in the repressor. Since most of the nonsense sites have been correlated with specific codons in the lacI messenger RNA, in almost all cases the position of the substituted residue is known. This process results in the creation of a large number of mutant phenotypes. We have analyzed the effects of each substitution in vivo, and in several cases studied partially purified repressors in vitro. The properties of the altered proteins have been compared with the position and nature of each exchanged residue. We discuss the implications of these findings with regard to repressor structure in particular, and to protein structure in general. Further applications of the suppression method are also considered.     

Local mobility of the lac repressor molecule

The rotational mobility of lac repressor from Escherichia coli was investigated by nanosecond fluorescence depolarization spectroscopy. A single rotational correlation time (φ) of the repressor was observed by monitoring the emission anisotropic decay of the intrinsic tryptophan fluorescence. The small value of φ (9·5 ns) suggests that one or both of the two tryptophan residues in the repressor are located in a flexible segment of the protein molecule. This segmental flexibility is enhanced by binding of inducer (isopropyl-β-d-thiogalactoside) to the repressor while it is restrained by binding of anti-inducer (glucose) or small DNA fragments, as indicated by the changes in φ. Further time-dependent emission anisotropy studies with an extrinsic fluorescent probe, N-(iodoacetylaminoethyl)-5-naphthylamine-1-sulfonate, covalently attached to the repressor yielded two rotational correlation times. The shorter φ_S (6·7 ns) also corresponds to a segmental flexibility whereas the longer φ_L (118 ns) represents the rotational motion of the entire repressor molecule. Both the values of φ_S and φ_L vary by addition of inducer or anti-inducer in a manner similar to that observed for the intrinsic tryptophan fluorescence but they are insensitive to addition of DNA fragments. The changes in local mobility of the lac repressor molecule observed in these studies may provide some insight into how inducer (or anti-inducer) destabilizes (or stabilizes) the repressor-operator complex.     

Genetic studies of the lac repressor. III. Additional correlation of mutational sites with specific amino acid residues

The complete results of the analysis of over 5300 independently derived nonsense mutations in the lacI gene are presented. These have been mapped and divided into specific sites. A total of 105 nonsense mutations derived from 90 different codons can be distinguished, of which several are the result of tandem double base changes induced by ultraviolet light. With the aid of results determined in a preceding paper (Miller et al., 1977), the majority of these mutations have been assigned to points in the gene coding for specific residues in the lac repressor. This allows a detailed correlation of the physical and genetic map.Recombination studies have been carried out using mutations at known sites. For crosses involving mutations separated by less than 30 nucleotides (the main object of this study), a significant lack of agreement between distance and recombination frequency has been found.     

Heterogeneity of the two tryptophanyl residues in the lac repressor of Escherichia coli.

The wild-type lac repressor of Escherichia coli is a tetrameric protein which contains two tryptophanyl residues per subunit at positions 190 and 209. Solute perturbation studies of the tryptophan fluorescence of the repressor were performed using a polar but uncharged quencher, acrylamide, to prevent possible bias caused by ionic quenchers. The results indicate that the two tryptophan residues have different accessibilities to the quencher. In addition, contrary to a previous report, the accessibility of these tryptophan residues is not altered by isopropyl-β-d-thiogalactoside (IPTG) binding to the repressor. Similar studies with mutant lac repressor containing only a single tryptophan either at positions 190 or 209 suggest that tryptophan 209 is located in a region which is perturbed by inducer binding. That the two tryptophanyl residues have heterogeneous environments was further confirmed by nanosecond fluorescence spectroscopy which showed the wild-type lac repressor exhibiting two excited-state lifetimes, τ₁ = 5.3 ns and τ₂ = 10 ns. In the presence of 10^?3m IPTG, only a single lifetime of 6 ns was observed for the wild-type repressor suggesting that the inducer perturbs the tryptophan residue with the longer lifetime but not the one with the shorter lifetime. This is in accord with the observation that the mutant repressor containing only tryptophan 190 (the Tyr-209 repressor) has a single lifetime of 4.5 ns which is not altered by IPTG binding. The surprising finding that the mutant repressor which contains only tryptophan 209 (the Tyr-190 repressor) shows two excited-state lifetimes has been interpreted to indicate that the repressor either does not exhibit fourfold symmetry in its subunit arrangement or is present in two different conformational states.     

lac Repressor-operator interaction. IX. The binding of lac repressor to operators containing Oc mutations

Representative members of the six classes of operator constitutive (O^c) point mutations, which have been mapped and well characterized in vivo, were crossed into λφ80 lac phages. The phage DNAs containing the O^c mutations were used to measure the affinity of the lac repressor (R) for each O^c operator by determining the half-lives of the different RO^c complexes in vitro. The results provide evidence that: (a) the higher the constitutive level of β-galactosidase in vivo, as the result of an O^c mutation, the lower the affinity of the lac repressor for that O^c operator, with a maximum difference of two orders of magnitude in affinity of the repressor for the highest O^c tested as compared to the wild type O⁺ operator; (b) the six classes of O^c operators appear to be twofold degenerate, in that two members of each class, which were previously distinguished by mapping, have the same affinity for the lac repressor; (c) an inducer and an anti-inducer have the same effect on the RO^c complexes as on the RO⁺ complexes; (d) the relationship between induction ratios in vivo and the binding constant of the repressor for each O^c mutation in vitro does not follow the mass action equation but rather a more complex dependency, which is discussed.These results suggest a functional symmetry in the lac operator.     

Mapping of I gene mutations which lead to repressors with increased affinity for lac operator

A genetic mapping system is used to locate mutations on the lac repressor gene (I) which lead to repressor proteins with an increased affinity for operator DNA. These tight binding repressors (I^tb) are of particular interest since their analysis should allow some conclusions on the mechanism of interaction between repressor and operator. I^tb mutations were found to map in two regions of the I gene. One is near the amino-terminal end, a region which has been shown to be essential for the DNA binding properties of the repressor. The other region in which I^tb mutations were mapped codes for approximately amino acids 255 to 295 of the repressor, a region which had so far not been considered to be essential for the DNA binding properties of the repressor protein.     

lac repressor-operator interaction. Analysis of the X86 repressor mutant

In vitro measurements show that the X86 repressor, which has an increased affinity for the lac operator as compared to wild-type repressor, also has an increased affinity for non-operator sites on Escherichia coli DNA. The rate constant of association of repressor and operator is decreased by E. coli DNA fivefold more for X86 repressor than for wild-type repressor. Low inducer concentrations increase the rate of association of X86 repressor and operator in the presence of E. coli DNA. In a partial equilibrium situation where part of the X86 repressor is bound to the operator, and part to either non-operator sites on E. coli DNA or to an O^c operator, the formation of complexes between X86 repressor and wild-type operator is favored by low inducer concentrations. Repression of the lac enzymes increases drastically in the X86 mutant in the absence of DNA synthesis in vivo. A new explanation for the in vivo characteristics of the X86 mutant is suggested.     

Historical Contingency Causes Divergence in Adaptive Expression of the lac Operon

Populations of Escherichia coli selected in constant and fluctuating environments containing lactose often adapt by substituting mutations in the lacI repressor that cause constitutive expression of the lac operon. These mutations occur at a high rate and provide a significant benefit. Despite this, eight of 24 populations evolved for 8,000 generations in environments containing lactose contained no detectable repressor mutations. We report here on the basis of this observation. We find that, given relevant mutation rates, repressor mutations are expected to have fixed in all evolved populations if they had maintained the same fitness effect they confer when introduced to the ancestor. In fact, reconstruction experiments demonstrate that repressor mutations have become neutral or deleterious in those populations in which they were not detectable. Populations not fixing repressor mutations nevertheless reached the same fitness as those that did fix them, indicating that they followed an alternative evolutionary path that made redundant the potential benefit of the repressor mutation, but involved unique mutations of equivalent benefit. We identify a mutation occurring in the promoter region of the uspB gene as a candidate for influencing the selective choice between these paths. Our results detail an example of historical contingency leading to divergent evolutionary outcomes.     

Sequence-specific interactions of the tight-binding I12-X86 lac repressor with non-operator DNA.

The tight-binding I12-X86 lac repressor binds to non-operator DNA in a sequence-specific fashion. Using the DNA of the E. coli I gene we have investigated these sequence-specific interactions and compared them to the operator binding of wild-type repressor. The specific, non-operator DNA interactions are sensitive to the inducer IPTG. One strong binding site in the I gene DNA was found to be one of two expected on the basis of their homology with the lac operator. The binding of I12-X86 repressor to this site was visualized using the footprinting technique, and found to be consistent with an operator-like binding configuration. The protection pattern extends into an adjacent sequence suggesting that two repressor tetramers are bound in tandem.     

Inactivation of prophage lambda repressor in vivo.

Jacob &; Monod (1961) postulated that prophage A induction results from the inactivation of the λ repressor by a cellular inducer. Although it has been shown that the phage A repressor is inactivated by the recA gene product in vitro (Roberts et al., 1978), we wanted to determine the action of the “cellular inducer” in vivo. Our results have led to a new model, which defines the relationship between the “cellular inducer” and the recA gene product.In order to quantitate the action of the cellular inducer on the λ repressor, we made use of bacteria with elevated cellular levels of the λ repressor (hyperimmune lysogens). We determined the kinetics of repressor inactivation promoted by three representative inducing treatments: ultraviolet light irradiation, thymine deprivation and temperature shift-up of tif-1 mutants.The kinetics of repressor decay in wild-type monolysogens indicate that repressor inactivation is a relatively slow cellular process that takes a generation time to reach completion. Incomplete inactivation of the repressor without subsequent prophage development may occur in a cell. We call this phenomenon detected at the biochemical level “subinduction”. In hyperimmune lysogens. subinduction is always the case.A high cellular level of A repressor that prevents prophage λ induction does not prevent induction of a heteroimmune prophage such as 434 or 80. Although the cellular inducer does not seem specific for any inducible prophage, it does not inactivate two prophage repressors present in a cell in a random manner. We have called this finding “preferential repressor inactivation”. Preferential repressor inactivation may be accounted for by considering that the intracellular concentration of a repressor determines its susceptibility to the action of the inducer.In bacteria with varying repressor levels, a fixed amount of repressor molecules is inactivated per unit of time irrespective of the initial repressor concentration. The rate of repressor inactivation depends on the catalytic capacity of the cellular inducer that behaves as a saturated enzyme. In wild-type bacteria the cellular inducer seems to be produced in a limited amount, to have a weak catalytic capacity and a relatively short half-life. The amount of the inducer formed after tif-1 expression is increased in STS bacteria overproducing a tif-1-modified RecA protein. This result is an indication that a modified form of the RecA protein causes repressor inactivation in vivo.From the results obtained we propose a model concerning the formation of the cellular inducer. We postulate that the cellular inducer is formed in a two-step reaction. The is model visualises how the RecA protein can be induced to high cellular concentrations, even though the RecAp protease molecules remain at a low concentration. The latter accounts for the limited proteolytic activity found in vivo.     

Arginine 197 of lac repressor contributes significant energy to inducer binding. Confirmation of homology to periplasmic sugar binding proteins.

Based on primary sequence homology between the lactose repressor protein and periplasmic sugar-binding proteins (Müller-Hill, B. (1983) Nature 302, 163-164), a hypothetical sugar-binding site for the lac repressor was proposed using the solved x-ray crystallographic structure of the arabinose-binding protein (ABP) (Sams, C. F., Vyas, N. K., Quiocho, F. A., and Matthews, K. S. (1984) Nature 310, 429-430). By analogy to Arg151 in the ABP sugar site, Arg197 is predicted to play an important role in lac repressor binding to inducer sugars. Hydrogen bonding occurs between Arg151 and the ring oxygen and 4-hydroxyl of the sugar ligand, two backbone carbonyls, and a side chain in ABP, and similar interactions in the lac repressor would be anticipated. To test this hypothesis, Arg197 in the lac repressor protein was altered by oligonucleotide-directed site-specific mutagenesis to substitute Gly, Leu, or Lys. Introduction of these substitutions at position 197 had no effect on operator binding parameters of the isolated mutant proteins, whereas the affinity for inducer was dramatically decreased, consistent with in vivo phenotypic behavior obtained by suppression of nonsense mutations at this site (Kleina, L. G., and Miller, J. H. (1990) J. Mol. Biol. 212, 295-318). Inducer binding affinity was reduced approximately 3 orders of magnitude for Leu, Gly, or Lys substitutions, corresponding to a loss of 50% of the free energy of binding. The pH shift characteristic of wild-type repressor is conserved in these mutants. Circular dichroic spectra demonstrated no significant alterations in secondary structure for these mutants. Thus, the primary effect of substitution for Arg197 is a very significant decrease in the affinity for inducer sugars. Arginine is uniquely able to make the multiple contacts found in the ABP sugar site, and we conclude that this residue plays a similar role in sugar binding for lactose repressor protein. These results provide experimental validation for the proposed homology between ABP and the lac repressor and suggest that homology with ABP may be employed to generate additional insight into the structure and function of this regulatory protein.     

Activity changes in lac repressor with cysteine oxidation.

The effects of prior covalent cysteine modification or nonspecific DNA presence on the reaction of lac repressor protein with N-bromosuccinimide have been investigated. At low excesses, N-bromosuccinimide oxidation causes loss of operator DNA binding activity with simultaneous retention of inducer and nonspecific DNA binding activities. Cysteine and methionine are oxidized under the conditions utilized. Covalent modification of the cysteines of repressor prior to reaction decreased the observed loss of operator DNA binding capacity; the presence of nonspecific DNA partially prevented oxidation of the cysteines by N-bromosuccinimide, and concurrent protection of operator binding ability was observed. Methionine oxidation was observed in the cases where protection of the operator DNA binding capacity of repressor was seen. The region surrounding cysteine 107 was found to be influential in maintaining intact operator DNA binding function in repressor. This observation provides chemical evidence for the contribution of the core region of repressor in determining specificity of the protein in binding the lac operator. The protection from oxidation of cysteine residues in the core region by the presence of nonspecific DNA suggests that this binding influences the core region of the protein.     

Binding of lac repressor to the secondary lac operator in Escherichia coli.

Summary In the lac operon, the existence of a secondary repressor binding site, inside Z gene, had been inferred from in vitro binding studies (Reznikoff et al., 1974; Gilbert et al., 1975).A serie of deletions have been constructed from a lac transducing bacteriophage. Some of those deleted bacteriophages have still the property of derepressing a chromosomal lac operon, even though they do not contain any more the lac operator. This phenomenon is an indication that the secondary repressor binding site is also active in vivo.     

Conformational prediction and circular dichroism studies on the lac repressor

Using the protein predictive model of Chou & Fasman (1974b), the secondary structure of the lac repressor has been elucidated from its amino acid sequence of 347 residues. The conformation is predicted to contain 37% α-helix and 35% β-sheet for the repressor, and 29% helix and 41% β-sheet for the trypsin-resistant core (residues 60 to 327). Circular dichroism studies indicate that native lac repressor contains 40% helix and 42% β-sheet, while the core has 16% helix and 54% β-sheet, in general agreement with the predicted conformation. The sharp reduction in helicity for the trypsinized lac repressor could be due to the loss of two long helical regions, 26–45 and 328–344, predicted at both terminals. There are extensive β-sheets predicted in the 215–324 region, which may be responsible for tetrameric stabilization found in both the lac repressor and the core. Residues 17 to 33 were previously predicted by Adler et al. (1972) to be helical and were proposed to bind in the major groove of DNA. However, the present analysis shows that there are two anti-parallel β-sheet regions: 4–7 and 17–24 at the N-terminal as well as 315–318 and 321–324 at the C-terminal of the lac repressor. These β-sheet pairs may assume the twisted “polypeptide double helix” conformation (Carter & Kraut, 1974) and bind to complementary regions in the major groove of DNA. The OH groups of Tyr at the N-terminal and those of Thr and Ser side chains, in both β-sheets at the N and C-terminal ends, could form hydrogen bonds to specific sites on the lac operator. There are 23 reverse β-turns predicted that may control the tertiary folding of the lac repressor, which is essential for operator binding. The behavior of several lac repressor mutants can be satisfactorily explained in terms of polar to non-polar group replacements as well as conformational changes in light of the present predicted model.     

Direct identification of the amino acid changes in two mutant lac repressors.

Deletions extending into the trp operon at one terminus and the lacI control region at the other terminus have been examined. One of these, B116, ends within the trp leader sequence and eliminates the trp attenuator site, placing the synthesis of lac repressor under trp control. We have isolated and characterized the B116 repressor. The protein sequence of the aminoterminus of B116 shows that an additional 16 residues are added to the amino-terminal end of wild-type repressor. Moreover, a valine residue appears in place of methionine at position 17 (the original amino-terminal residue of the wild-type repressor). A comparison of the messenger RNA sequence of the trp leader region and of the I leader region demonstrates that the translation of the B116 repressor is initiated at an AUG codon within the trp leader sequence. The GUG initiation codon at the start point for translation of wild-type repressor is now read as valine, since it appears at an internal position (residue 17 of the altered repressor). The B116 repressor accumulates at levels as high as 1% of the soluble cell protein in trpR^? strains. The efficiency of the trp leader initiation codon in translation suggests that in wild-type strains this AUG is also active in directing protein synthesis, which would result in a polypeptide consisting of 14 amino acids. We have examined the physical properties of the B116 repressor, which shows a marked tendency to form higher aggregates. Other characteristics of B116 are also described.     

Genetic studies of the lac repressor. XIII. Extensive amino acid replacements generated by the use of natural and synthetic nonsense suppressors

We have altered the amino acid sequence of the lac repressor one residue at a time by utilizing a collection of nonsense suppressors that permit the insertion of 13 different amino acids in response to the amber (UAG) codon, as well as an additional amino acid in response to the UGA codon. We used this collection to suppress nonsense mutations at 141 positions in the lacI gene, which encodes the 360 amino acid long lac repressor, including 53 new nonsense mutations which we constructed by oligonucleotide-directed mutagenesis. This method has generated over 1600 single amino acid substitutions in the lac repressor. We have cataloged the effects of these replacements and have interpreted the results with the objective of gaining a better understanding of lac repressor structure, and protein structure in general. The DNA binding domain of the repressor, involving the amino-terminal 59 amino acids, is extremely sensitive to substitution, with 70% of the replacements resulting in the I- phenotype. However, the remaining 301 amino acid core of the repressor is strikingly tolerant of substitutions, with only 30% of the amino acids introduced causing the I- phenotype. This analysis reveals the location of sites in the protein involved in inducer binding, tighter binding to operator and thermal stability, and permits a virtual genetic image reconstruction of the lac repressor protein.     

Repressor--operator interaction in the lac operon. II. Observations at the tyrosines and tryptophans

This paper shows that ¹⁹F-nuelear magnetic resonance spectroscopy on 3-fluoro-tyrosine and 5-fluorotryptophan-substituted wild-type lactose operon repressors from Escherichia coli can be used to examine the interactions with lac operator DNA.A survey of inducer and salt concentration effects on the repressor-operator complex is presented. The data lead us to a scheme for the interactions between the repressor, operator and inducer, in both binary and ternary complexes, that accommodate the results published by others.The complex between the tetrameric repressor and one 36 base-pair operator DNA fragment results in the simultaneous broadening of the resonances from all four N-terminal DNA binding domains. The actual contacts made by these binding domains are similar but probably not identical.The binding of the inducer molecule to the tetrameric repressor results in an allosteric change that can be monitored by the increased intensity of the resonances from individual tyrosine residues in the N-terminal binding domain. This increased N-terminal tyrosine resonance intensity in the complex is transmitted to repressor subunits that have not yet bound an inducer molecule.