Diversity in lac Operon Regulation among Diverse Escherichia coli Isolates Depends on the Broader Genetic Background but Is Not Explained by Genetic Relatedness

Abstract

Transcription of bacterial genes is controlled by the coordinated action of cis- and trans-acting regulators. The activity and mode of action of these regulators can reflect different requirements for gene products in different environments. A well-studied example is the regulatory function that integrates the environmental availability of glucose and lactose to control the Escherichia colilac operon. Most studies of lac operon regulation have focused on a few closely related strains. To determine the range of natural variation in lac regulatory function, we introduced a reporter construct into 23 diverse E. coli strains and measured expression with combinations of inducer concentrations. We found a wide range of regulatory functions. Several functions were similar to the one observed in a reference lab strain, whereas others depended weakly on the presence of cAMP. Some characteristics of the regulatory function were explained by the genetic relatedness of strains, indicating that differences varied on relatively short time scales. The regulatory characteristics explained by genetic relatedness were among those that best predicted the initial growth of strains following transition to a lactose environment, suggesting a role for selection. Finally, we transferred the lac operon, with the lacI regulatory gene, from five natural isolate strains into a reference lab strain. The regulatory function of these hybrid strains revealed the effect of local and global regulatory elements in controlling expression. Together, this work demonstrates that regulatory functions can be varied within a species and that there is variation within a species to best match a function to particular environments.IMPORTANCE The lac operon of Escherichia coli is a classic model for studying gene regulation. This study has uncovered features such as the environmental input logic controlling gene expression, as well as gene expression bistability and hysteresis. Most lac operon studies have focused on a few lab strains, and it is not known how generally those findings apply to the diversity of E. coli strains. We examined the environmental dependence of lac gene regulation in 20 natural isolates of E. coli and found a wide range of regulatory responses. By transferring lac genes from natural isolate strains into a common reference strain, we found that regulation depends on both the lac genes themselves and on the broader genetic background, indicating potential for still-greater regulatory diversity following horizontal gene transfer. Our results reveal that there is substantial natural variation in the regulation of the lac operon and indicate that this variation can be ecologically meaningful.

Keywords: lac operon regulation.

FIG 1

Phylogeny based on the core genome shared between 96 diverse natural isolates of E. coli. Strains whose lac regulatory function was determined and whose lacI-ZYA region was transferred to the reference strain, REL606, are indicated by the red symbols in columns labeled “expression” and “transfer,” respectively. The former group of strains represents a random sample of the complete phylogeny (Fig. S1). The lac regulatory function was also measured for three strains for which we do not have genome sequence and, therefore, are not included here, B156, B1167, and TA263 (Table S1). The lac operon of TA263 was also transferred to REL606. Phylogeny construction is described in Materials and Methods.

FIG 2

Empirical and modeled gene regulatory profiles. Expression of a lac reporter was determined during growth in glucose supplemented with combinations of the inducers cAMP (millimolar) and IPTG (micromolar). Expression was measured from a chromosomally integrated reporter at mid-log phase and is reported in arbitrary fluorescence units (AFU). Solid symbols indicate expression predicted at each measured inducer combination using a simple regulatory model fitted to the observed data (see the supplemental material) (14). Dashed lines connect model estimates and empirically determined expression values. The three profiles shown here are for a lab strain (REL606) and two natural isolate strains (M646 and E1002) and have profiles that differ in the sufficiency of IPTG to induce lac expression to a high level. Additional profiles are shown in Fig. S2.

FIG 3

lac regulatory logic of E. coli strains. Models describing the lac expression phenotype were fitted for each of the tested E. coli strains (Fig. S1). Model parameters were used to determine the ratio of log expression at low IPTG-low cAMP, high IPTG-low cAMP, and low IPTG-high cAMP combinations to the high IPTG-high cAMP combination, giving parameters, π1, π2, and π3, and respectively (14). Combinations of these parameters describe a particular regulatory logic input function. For example, low values of π1, π2, and π3 indicate high lac expression only when both IPTG and cAMP levels are present, reflecting an AND-type logic function. Black symbols indicate parameter estimates of natural isolate strains. Green and red points indicate estimates of the lab strains REL606 and MG1655, respectively. The gray point indicates an E. albertii strain, B156, that does not encode several components of the canonical lactose utilization system, including a LacI repressor, and therefore expresses the reporter at high levels regardless of IPTG (see also Fig. S2).

FIG 4

Relationship between regulatory parameters and lag time following transition from a glucose- to lactose-supplemented growth environment. (A) Gompertz fits to growth data of natural isolate strains and hybrid strains containing the lacI-ZYA region from a natural isolate strain replacing the same region in the REL606 genetic background. Growth is in lactose following a transition from a day of growth in glucose. (B) Partial least-squares (PLS) regression indicating contribution of regulatory model parameters to the largest four components explaining the variation in lag time. (C and D) Changes in expression landscapes dependent on changing the two parameters, a and η, that explain the most lag time variation. Parameters are changed between the extremes of their estimated ranges and preserving their negative correlation. Other parameters are as for REL606, except that α is increased in panel D so that maximum expression level is comparable. Panels C and D correspond to landscapes associated with short and long lag times, respectively.

FIG 5

Association between polymorphism in the lacI-ZYA region and variation in estimated regulatory parameters. For each parameter, mutual information was estimated between estimates and genetic variation at each site in the genetic region (see Materials and Methods for details). All polymorphic sites are plotted. The dashed line indicates a significance cutoff at a P value of 0.05. Gray symbols indicate parameter-polymorphism associations below this cutoff, and colored symbols indicate associations above this cutoff. Only parameters with at least one significant association are colored in the legend. These significant associations primarily affect basal lac expression (γ) and the dissociation constant of IPTG from LacI (Km_IPTG). The left panel presents the entire region considered. The right panel provides higher resolution around the key regulatory area between lacI-Z indicated by the box in the left panel. Transcription factor binding sites and the promoter region are indicated in the right panel (binding site information from Regulon DB). RNAP, RNA polymerase; txn, transcription.

FIG 6

Effect of genetic background on lac expression. (A) Schematic of expression comparisons. The lac expression profile was obtained from a common reference strain (REL606), different donor natural isolate strains, and a hybrid constructed by swapping the donor strain’s lacI-ZYA genes into REL606 (details are in Materials and Methods). (B) Example expression profiles of one comparison set. In this case, the hybrid strain has an expression profile more similar to that of the recipient background strain (REL606) than of the donor (B921). (C) Dendrograms clustering for each of five donors the set of three strains based on Euclidean distances among modeled expression landscapes. The height of dendrograms is scaled to the distance between strains. Expression profiles for each strain are presented in Fig. S5.