Single-molecule spectroscopic determination of lac repressor-DNA loop conformation

Abstract

The Escherichia coli lactose repressor protein (LacI) provides a classic model for understanding protein-induced DNA looping. LacI has a C-terminal four-helix bundle tetramerization domain that may act as a flexible hinge. In previous work, several DNA constructs, each containing two lac operators bracketing a sequence-induced bend, were designed to stabilize different possible looping geometries. The resulting hyperstable LacI-DNA loops exist as both a compact "closed" form with a V-shaped repressor and also a more "open" form with an extended hinge. The "9C14" construct was of particular interest because footprinting, electrophoretic mobility shift, and ring closure experiments suggested that it forms both geometries. Previous fluorescence resonance energy transfer (FRET) measurements gave an efficiency of energy transfer (ET) of 70%, confirming the existence of a closed form. These measurements could not determine whether open form or intermediate geometries are populated or the timescale of interconversion. We have now applied single-molecule FRET to Cy3, Cy5 double-labeled LacI-DNA loops diffusing freely in solution. By using multiple excitation wavelengths and by carefully examining the behavior of the zero-ET peak during titration with LacI, we show that the LacI-9C14 loop exists exclusively in a single closed form exhibiting essentially 100% ET.

FIGURE 1

Design of the fluorophore-labeled 9C14 DNA construct. (A) Model for the looping construct illustrating the orientation of dyad axes of the DNA operators (blue cylinders) relative to the sequence-induced bend. (B) The cocrystal structure of a LacI tetramer-DNA “sandwich” complex, Protein Data Bank entry 1LBG (8). The red duplex DNA is from the cocrystal structure, and the black DNA represents modeled ideal B-DNA extensions. The major groove attachment sites are shown for the donor (green ball) and acceptor (orange ball) fluorophores. The four-helix bundle tetramerization domain (magenta) is connected to the rest of the protein by linkers that are proposed to be flexible. (C) Schematic of the 9C14 sequence showing dimensions in basepairs. The 56-nt PCR primers used to synthesize the full construct are indicated with their internal fluorophores. The total length is 212 bp. (D) Proposed DNA geometries. The closed form loop adopted by 9C14 (red) results from decreasing the radius of curvature of the DNA between the lac operators. The open form conformation is preferred by other constructs with different linkers and was previously proposed to be accessible to 9C14 via DNA twist changes.

FIGURE 2

SM-FRET data from diffusing 9C14 molecules. (A) A typical burst sequence acquired with the 9C14 DNA construct (1 nM) excited at 514 nm. The acceptor counts are shown in red, and donor counts are in green. The LacI concentration is 2.5 nM, and a mix of bursts exhibiting both high and low ET values are observed. The bin size for the transient is 1 ms. Arrows denote bursts that exceed the intensity threshold for analysis. The bursts located near 0.5 s appear principally in the acceptor channel and originate from LacI/9C14 complexes forming a closed loop with an ET efficiency of nearly 1.0. Bursts that appear principally in the donor channel such as the one near 0.15 s belong to free or mislabeled 9C14 molecules. (B) Typical FRET histograms acquired with 514 nm laser excitation for 9C14 both as free DNA (blue) and fully occupied with LacI (gray). Negative efficiency values in the zero-ET peak are due to bursts in which the acceptor channel intensity was less than the average background. The width of the zero-ET peak is further broadened by the subtraction of donor counts from the acceptor channel, which is necessary to correct for nonideal optical filtering, as described in Materials and Methods. The histogram for fully bound 9C14 has a significant zero-ET peak, which is due to a combination of donor-only labeled 9C14 and 9C14 with photobleached acceptors.

FIGURE 3

SM-FRET results from titration experiments acquired with 543 nm excitation. (A) Histograms acquired with three different LacI concentrations illustrating the decrease in the zero-ET peak and the simultaneous increase in the proportion of molecules appearing near 90% ET efficiency. Without LacI (dashed line), only the zero-ET peak is observed. Upon addition of LacI (solid line) a population near 90% ET efficiency appears. This population is assigned to the closed form 9C14-LacI loop conformation. In the fully bound histogram (dotted line) 60% of the recorded single-molecule bursts show high ET efficiency. The remaining bursts are assigned to the zero-ET peak. (B) Graph of the fraction of bursts with ET values above 50% versus LacI concentration. The trend illustrates the direct relationship between the number of bursts with efficient ET and the LacI concentration. Note that at 5.0 nM LacI, a decrease in the fraction with high ET efficiency is observed, which may be due to quenching of the acceptor by nonspecific binding of LacI to 9C14.

FIGURE 4

Effect of excitation power on the appearance of molecules with high ET efficiency. This plot demonstrates the strong laser excitation power dependence on the observed SM-FRET histogram. The fraction of molecules in the fully bound sample that exhibit ET efficiency above 50% is plotted as a function of laser excitation power under ambient conditions (○) and with the addition of an oxygen scavenging system (•). Under ambient conditions, as the power is increased there is a steep decrease in the fraction of molecules exhibiting efficient ET. With the addition of the glucose oxidase- and catalase-based oxygen scavenger, the power dependence is significantly diminished, leveling off above 50 μW.

FIGURE 5

Corrected excitation spectra of the donor (dotted line) and acceptor (solid lines). The excitation spectra are plotted in units of M⁻¹ cm⁻¹ rather than emission intensity to highlight the disparity in the absorptivity of the two dyes at the excitation wavelengths used. The inset provides an expanded view of the acceptor excitation spectrum in the region near the two laser wavelengths to illustrate the large increase in absorbance (an order of magnitude) with 543 nm excitation relative to 514 nm excitation.

FIGURE 6

Analysis of SM-FRET histograms. SM-FRET histograms from free 9C14 (dotted line) and fully bound 9C14 (solid line) were acquired with (A) 514 nm and (B) 543 nm excitation. A difference histogram was constructed for the zero-ET peak at each wavelength by subtracting the histogram of the free 9C14 sample (dotted line) from that of the fully bound sample (solid line). The difference histograms (negative-going bars) for each wavelength were fit to Gaussian distributions (negative-going solid lines). With 514 nm excitation, the difference histogram is symmetrically positioned around zero. In contrast, the difference histogram obtained with 543 nm excitation is shifted toward positive values due to selective depletion of the zero-ET peak on its right side. The parameters obtained from the Gaussian fit in (B) along with parameters obtained from fitting data acquired with donor-only labeled DNA are used to decompose the zero-ET peak at all LacI concentrations (Table 1). (C) SM-FRET histogram of free 9C14 DNA (bars) acquired with 543 nm excitation. The fit to the zero-ET peak (black line) is a sum of Gaussian distributions representing donor-only labeled 9C14 (shaded line) and double-labeled 9C14 (dotted line). (D) Histogram from fully bound 9C14 acquired with 543 nm excitation. The zero-ET peak is fit with a single Gaussian distribution, and the high efficiency peak is fit with a β-distribution function. The Gaussian has the same center and width as those required to obtain a good fit to a FRET histogram acquired with a donor-only labeled sample. The β-distribution function has been used previously to describe SM-FRET histograms (11).