In Vivo Demonstration of FNR Dimers in Response to Lower O₂ Availability

Abstract

Escherichia coli FNR is an O₂-sensing transcription factor. In vitro studies indicate that anaerobic iron-sulfur cluster acquisition promotes FNR dimerization. Here, two-hybrid assays show that iron-sulfur cluster-dependent FNR dimers are formed in vivo in response to lower O₂ availability, consistent with the current model of FNR activation.

The Escherichia coli FNR protein is an O₂-sensing global transcriptional regulator (4, 6, 9, 10, 12, 15, 23). Although present at similar intracellular concentrations under aerobic and anaerobic conditions (26), FNR activity is maximal in the absence of O₂. Isolation of the FNR protein under aerobic conditions yields an inactive, monomeric apoprotein (apo-FNR) (5), whereas under anaerobic conditions, an active, dimeric [4Fe-4S]²⁺ cluster-containing protein (4Fe-FNR) is isolated (17, 18). Upon exposure of 4Fe-FNR to O₂, the [4Fe-4S]²⁺ cluster is converted to a [2Fe-2S]²⁺ cluster forming 2Fe-FNR, which is unable to bind DNA (1, 2, 7, 11, 18, 26). If aerobic conditions persist, 2Fe-FNR is converted to apo-FNR, which can be recycled (3, 27). Thus, acquisition of [4Fe-4S]²⁺ clusters in vitro promotes dimerization of FNR, presumably by initiating conformational changes in the dimer interface (17, 20). However, the response of the oligomeric state of FNR to O₂ availability in vivo has not been determined.

Initial attempts to detect FNR dimers under anaerobic conditions in vivo using two bacterial two-hybrid systems, the Ladant (13) and the Stratagene Bacteriomatch I and II systems, were unsuccessful. Both systems are used to screen gene libraries for interacting partners, and thus the plasmid encoding the “bait” protein is present in lower copy numbers than that encoding the “target” protein. This difference in plasmid copy numbers favors the formation of unproductive hybrids. Therefore, the Ladant system was modified to equalize the copy numbers of “bait” and “target” plasmids. The plasmids pKT25 (low copy number) and pUT18 (high copy number) allow the fusion of proteins to domains T25 and T18 of Bordetella pertussis adenylate cyclase (13). When the fusion proteins interact, T18 and T25 form active adenylate cyclase, which produces cyclic AMP, activating transcription of lacZ, which thus acts as a reporter of the “bait”-“target” interaction. Plasmids pGS2083, a pKT25 derivative encoding a T25-FNR fusion, and pGS2086, encoding a T18-FNR fusion, were available. Ligating a PCR-amplified HindIII-BamHI fragment containing the T25-FNR fusion to the “backbone” of pUT18, followed by disruption of the bla gene by insertion of a Kan^r selectable marker at the ScaI site, created the high-copy-number T25-FNR fusion plasmid pGS2092; this was used with pGS2086. Because these plasmids have the same origin of replication, cultures were grown with ampicillin and kanamycin, and the presence of both plasmids in similar amounts was confirmed by agarose gel electrophoresis of plasmid preparations from cotransformants (not shown).

High-copy-number T25 fusion plasmids that lacked fnr (pGS2095) or that encoded T25-Δ29FNR (pGS2093) or T25-FNR* (pGS2094) were constructed, along with partner pUT18 derivatives encoding T18-Δ29FNR (pGS2087) and T18-FNR* (pGS2088). The Δ29FNR protein lacks the first 29 amino acid residues, including three of the four ligands for the iron-sulfur clusters, and thus the protein is monomeric in vitro and inactive in vivo (8, 24). In contrast, the Asp154→Ala substitution in FNR* alters the dimer interface such that FNR* is dimeric under aerobic conditions in vitro and retains significant aerobic activity in vivo (14, 16, 20).

The reporter strain for the Ladant system is E. coli BTH101, a nonreverting adenylate cyclase mutant (13). To study FNR dimerization in BTH101, the chromosomal copy of fnr was deleted by P1-mediated transduction of the fnr deletion from E. coli JRG1728 (fnr lac) (25), yielding JRG5708. The high-copy-number plasmid pairs described above were cotransformed into strain JRG5708 by electroporation and selected on L agar plates (22) containing kanamycin (40 μg ml⁻¹) and ampicillin (100 μg ml⁻¹). The resulting reporter strains were named JRG5709 (T25-FNR, T18-FNR), JRG5710 (T25-Δ29FNR, T18-Δ29FNR), JRG5711 (T25-FNR*, T18-FNR*), and JRG5712 (vector control). Cultures (aerobic, 5 ml medium in 25-ml bottles with shaking at 250 rpm; anaerobic, 7 ml in filled, sealed bottles) were grown in L broth (22) containing ampicillin (100 μg ml⁻¹) and kanamycin (40 μg ml⁻¹) for 16 h at 37°C to stationary phase (aerobic cultures, optical density at 600 nm, ∼2; anaerobic cultures, optical density at 600 nm, ∼0.4). An ∼9-fold increase in β-galactosidase activities (19) in the cultures grown under anaerobic conditions showed that FNR monomers interacted under anaerobic, but not aerobic, conditions (Fig. 1A). The Δ29FNR subunits were unable to interact under either condition. This inability suggests that FNR must obtain an iron-sulfur cluster to dimerize, in accordance with previous observations (18), although the possibility that this variant has an additional conformational defect that inhibits dimerization cannot be excluded. Cultures expressing the FNR* fusions exhibited significant β-galactosidase activity under aerobic conditions (∼5-fold greater than the readout obtained with native FNR), which increased only ∼1.5-fold under anaerobic conditions. Thus, it seems that FNR* is mostly dimeric under aerobic conditions in vivo as well as in vitro. The vectors did not produce any activity in JRG5708.

FIG. 1.

FNR is dimeric under anaerobic conditions and monomeric under aerobic conditions in vivo. (A) FNR dimerization in vivo was measured in strain JRG5708 (BTH101 fnr), expressing the indicated fusion proteins. Cultures were JRG5709 (T25-FNR, T18-FNR), JRG5710 (T25-Δ29FNR, T18-Δ29FNR), JRG5711 (T25-FNR*, T18-FNR*), and JRG5712 (vector control). (B) FNR fusion proteins are functional in vivo. FNR activity was measured using strain JRG1728 (fnr lac) containing plasmid pFF-41.5 (29), which carries lacZ under the control of an FNR-dependent promoter, and plasmids encoding the FNR fusion proteins. Cultures were JRG5720 (T25-FNR, T18-FNR), JRG5721 (T25-Δ29FNR, T18-Δ29FNR), JRG5722 (T25-FNR*, T18-FNR*), and JRG5723 (vector control). Open bars, aerobic conditions; closed bars, anaerobic conditions. Values shown are means (±standard deviations) of results from triplicate assays. Data are representative of three independent experiments.

It was important to establish that the FNR fusion proteins were functional as O₂-responsive regulators. Therefore, the plasmid pairs were introduced into strain JRG1728 (fnr lac) containing pFF-41.5 (a low-copy-number plasmid carrying an FNR-dependent promoter fused to lacZ) (29). The corresponding strains JRG5720 (T25-FNR, T18-FNR), JRG5721 (T25-Δ29FNR, T18-Δ29FNR), JRG5722 (T25-FNR*, T18-FNR*), and JRG5723 (vector control) were assayed for β-galactosidase activity as described above, except that tetracycline (12.5 μg ml⁻¹) was added to the culture medium to select for pFF-41.5. The activity patterns (Fig. 1B) generally followed those found in the two-hybrid data (Fig. 1A), i.e., native FNR was ∼15-fold more active under anaerobic conditions, Δ29FNR was inactive under both conditions, and FNR* exhibited aerobic activity that increased ∼5-fold under anaerobic conditions. However, the enhancement in FNR* activity under anaerobic conditions was significantly greater than that observed for FNR* dimerization (Fig. 1A). It seems that the FNR* dimer is suboptimally configured for interaction with RNA polymerase, and as suggested previously (21), the presence of iron-sulfur clusters is important for the formation of productive FNR contacts with RNA polymerase.

The model for FNR activation suggests that the acquisition of [4Fe-4S]²⁺ clusters promotes dimerization, thereby enhancing DNA binding to facilitate transcriptional regulation (6, 15). This predicts that the responses of the dimerization and FNR activity reporters to changes in O₂ availability should be the same. Therefore, the rates of O₂ diffusion into different volumes of H₂O in shaking (100 rpm) 250-ml conical flasks at 37°C were determined as described previously (28). Oxygen transfer rates of 72.2 μmol min⁻¹ liter⁻¹ for 50 ml, 30.2 μmol min⁻¹ liter⁻¹ for 100 ml, 11.9 μmol min⁻¹ liter⁻¹ for 150 ml, 6.9 μmol min⁻¹ liter⁻¹ for 200 ml, and 3.0 μmol min⁻¹ liter⁻¹ for 250 ml were obtained. Cultures of strains JRG5709 and JRG5720 were grown in shaken (100 rpm) 250-ml conical flasks containing these volumes of L broth (19). The highly aerated cultures exhibited the lowest levels of FNR dimers and FNR activity (Fig. 2). As the O₂ supply became more restricted, the dimerization and activity readouts increased in parallel, with half-maximal readings at an O₂ transfer rate of ∼22 μmol min⁻¹ liter⁻¹.

FIG. 2.

Correlation between FNR dimerization and FNR activity in response to changes in O₂ availability. Cultures were grown with different O₂ transfer rates as described in the text. ▪, strain JRG5709 (FNR dimers measured by two-hybrid system); ▴, strain JRG5720 (FNR activity measured by β-galactosidase expression from pFF-41.5). Values shown are means (±standard deviations) of results from triplicate assays. Data are representative of three independent experiments.

In conclusion, a modified bacterial two-hybrid system that allowed the detection of FNR dimers in vivo under anaerobic conditions is described. The FNR fusion proteins were able to activate expression from an FNR-dependent promoter, and the increase in FNR-dependent in vivo transcription as the O₂ supply was lowered closely followed the increase in levels of FNR dimers detected in the cells. Thus, the data support the model in which iron-sulfur cluster acquisition by FNR under anaerobic conditions promotes dimerization, which in turn enhances DNA binding to facilitate transcription regulation.