Conformational Cycling within the Closed State of Grp94, an Hsp90-Family Chaperone

Abstract

The Hsp90 family of chaperones requires ATP-driven cycling to perform their function. The presence of two bound ATP molecules is known to favor a closed conformation of the Hsp90 dimer. However, the structural and mechanistic consequences of subsequent ATP hydrolysis are poorly understood. Using single-molecule FRET, we discover novel dynamic behavior in the closed state of Grp94, the Hsp90 family member resident in the endoplasmic reticulum. Under ATP turnover conditions, Grp94 populates two distinct closed states, a relatively static ATP/ATP closed state that adopts one conformation, and a dynamic ATP/ADP closed state that can adopt two conformations. We constructed a Grp94 heterodimer with one arm that is catalytically dead, to extend the lifetime of the ATP/ADP state by preventing hydrolysis of the second ATP. This construct shows prolonged periods of cycling between two closed conformations. Our results enable a quantitative description of how ATP hydrolysis influences Grp94, where sequential ATP hydrolysis steps allow Grp94 to transition between closed states with different dynamic and structural properties. This stepwise transitioning of Grp94's dynamic properties may provide a mechanism to propagate structural changes to a bound client protein.

Keywords: Grp94; chaperone; conformational change; single molecule FRET.

Figure 1.. Bulk FRET measurements of Grp94 with various nucleotides.

(A) The N91/N91 FRET pair shows anti-correlated changes in bulk fluorescence spectra between apo and AMPPNP conditions. The activity of the labeled protein at 40°C (0.22±0.01 ATP/min/dimer) is similar to the wild-type rate (0.19±0.02 ATP/min/dimer). Inset shows expected distances between the donor and acceptor in the open and closed states (PDB IDs: 2O1V, 5ULS). (B) The N91/N91 FRET pair shows an intermediate population of the Grp94 closed state with ATP at 30°C. Grp94 reopens upon addition of 500 μM AUY922. (C) Temperature-dependent closed state accumulation with ATP. Solid lines are single exponential fits. Error bars on closure rates and activity values are the SEM. Buffer conditions: 25 mM HEPES pH 8.0, 150 mM KCl, 600 μM MgCl₂, 600 μM nucleotides, 0.5 mg/ml BSA, 2 mM BME.

Figure 2.. Single-molecule FRET measurements of Grp94 with various nucleotides.

FRET of individual Grp94 dimers labeled with one donor and one acceptor fluorophore at the N91 position on opposite arms. Samples were preincubated with ATP (left), AMPPNP (middle) and ADP (right). Upper panels: Grp94 adopts discrete conformations with characteristic low and high FRET efficiencies. Solid lines show a double Gaussian fit for ATP and AMPPNP conditions and a single Gaussian fit for ADP conditions. The average and standard deviation of the Gaussian peaks are reported above their distributions. Lower panels: Fluorescence and calculated FRET efficiency traces for example individual molecules under each of the three nucleotide conditions. Donor fluorescence (green lines) and acceptor fluorescence (orange lines) from donor excitation is shown above the calculated FRET efficiencies (black lines). Acceptor fluorescence from direct excitation is shown in red lines. Arrows indicate donor (green) and acceptor (red) photobleaching. Experiments were performed with alternating excitation (donor: 650 μW at 532 nm, acceptor: 150 μW at 633 nm). The histogram data comes from two fields of view for ATP, and one field of view for both ADP and AMPPNP. Buffer conditions 50 mM HEPES, pH 8.0, 150 mM KCl, 0.6 mM MgCl₂, 0.6mM nucleotides, 2 mM BME and 0.5 mg/ml BSA, room temperature.

Figure 3.. Grp94 conformational changes under ATP conditions.

(A) Example trace under ATP conditions. The average and standard deviation of efficiency values are shown for select regions. (B) Average FRET efficiencies are shown for traces that are aligned at their opening time. Experiments were performed with both alternating excitation (blue squares; donor: 650 μW at 532 nm, acceptor: 150 μW at 633 nm) and continuous excitation (black circles; 2.6 mW at 532 nm). Error bars are the SEM. The inset shows a single exponential fit for the transient FRET increase (baseline fit: 0.78±0.01; intercept fit: 0.92±0.01, where the uncertainty is the standard error). (C) Distribution of FRET efficiency values that were recorded zero to 15 seconds prior to opening events. The solid lines show a double Gaussian fit with the average and standard deviation reported for both peaks. Buffer conditions same as Figure 2.

Figure 4.. The C’ conformation of Grp94 is stabilized under mixed nucleotide conditions.

(A) At 75% ADP both the C and C’ FRET states are evident. The solid lines show a triple Gaussian fit. Inset shows the average and standard deviation for the C and C’ efficiency distributions. The histogram data comes from one field of view measured with continuous donor excitation (2.6 mW at 532 nm). (B) Summary of Grp94 FRET states under different nucleotide conditions. The average FRET efficiency for different Grp94 conformations is determined from Gaussian fits (100% ATP is from Figure 3C, 100% ADP is from Figure 2, the mixed nucleotide conditions are from Supplemental Figure 5). Errors bars are the peak standard deviations from the Gaussian fits. Buffer conditions same as Figure 2.

Figure 5.. Grp94 conformational cycling between the C and C’ conformations.

(A) FRET efficiency histogram for a Grp94 heterodimer (wt/E103A) with only one hydrolytically active arm. The solid lines show a triple Gaussian fit. The histogram data comes from one field of view with alternating excitation (donor: 650 μW at 532 nm, acceptor: 150 μW at 633 nm). (B) Average FRET efficiencies for wt/E103A traces that are aligned at their opening time. Experiments were performed with both alternating excitation (blue squares; 2.4 second sampling) and continuous excitation (black circles; 500 ms sampling). Error bars are the SEM. The solid line shows a single exponential fit (baseline fit: 0.81±0.01; intercept fit: 0.96±0.01, where the uncertainty is the standard error). The dashed line shows the fit for wt/wt from Figure 3B. (C) Example trace showing wt/E103A Grp94 cycling between C and C’. The blue circles show FRET efficiency states assigned by the ebFRET program. Rate constants and uncertainties are the ebFRET output from analyzing all 282 traces. Buffer conditions same as Figure 2.

Figure 6.. Kinetic model of Grp94 conformational cycle.

Rate constants in black were determined from analysis of wt/wt homodimers (using data from Figures 2 and 3) and rate constants in red were determined from analysis of wt/E103A heterodimers (using data from Figure 5).