Skip to main content
Biophysical Journal logoLink to Biophysical Journal
. 2015 Jul 7;109(1):66–75. doi: 10.1016/j.bpj.2015.05.025

Conformational Transitions in the Glycine-Bound GluN1 NMDA Receptor LBD via Single-Molecule FRET

David R Cooper 1, Drew M Dolino 3, Henriette Jaurich 1, Bo Shuang 1, Swarna Ramaswamy 3, Caitlin E Nurik 3, Jixin Chen 1, Vasanthi Jayaraman 3,∗∗, Christy F Landes 1,2,
PMCID: PMC4572502  PMID: 26153703

Abstract

The N-methyl-D-aspartate receptor (NMDAR) is a member of the glutamate receptor family of proteins and is responsible for excitatory transmission. Activation of the receptor is thought to be controlled by conformational changes in the ligand binding domain (LBD); however, glutamate receptor LBDs can occupy multiple conformations even in the activated form. This work probes equilibrium transitions among NMDAR LBD conformations by monitoring the distance across the glycine-bound LBD cleft using single-molecule Förster resonance energy transfer (smFRET). Recent improvements in photoprotection solutions allowed us to monitor transitions among the multiple conformations. Also, we applied a recently developed model-free algorithm called “step transition and state identification” to identify the number of states, their smFRET efficiencies, and their interstate kinetics. Reversible interstate conversions, corresponding to transitions among a wide range of cleft widths, were identified in the glycine-bound LBD, on much longer timescales compared to channel opening. These transitions were confirmed to be equilibrium in nature by shifting the distribution reversibly via denaturant. We found that the NMDAR LBD proceeds primarily from one adjacent smFRET state to the next under equilibrium conditions, consistent with a cleft-opening/closing mechanism. Overall, by analyzing the state-to-state transition dynamics and distributions, we achieve insight into specifics of long-lived LBD equilibrium structural dynamics, as well as obtain a more general description of equilibrium folding/unfolding in a conformationally dynamic protein. The relationship between such long-lived LBD dynamics and channel function in the full receptor remains an open and interesting question.

Introduction

The N-methyl-D-aspartate receptor (NMDAR) is a member of the glutamate receptor family of proteins, along with the AMPA receptor and the kainate receptor. These receptors mediate ion transport in the cellular membrane and are considered potential drug targets for treatment of a number of neurological diseases including Alzheimer’s and Parkinson’s (1–6). The NMDAR is unique among the glutamate receptors in that it forms hetero-tetrameric complexes using two glycine-binding GluN1 subunits, and two other subunits, either the glutamate-binding subunit GluN2 or the glycine-binding GluN3 subunit (7–16). Each subunit comprises four main domains—the extracellular N-terminal domain, the ligand-binding domain (LBD), the intracellular C-terminal domain, and the transmembrane domain (14–19). The LBD cleft, comprised of two lobes connected by a hinge, is responsible for inducing large conformational changes that control the opening and closing of the cation channel, allowing for the regulation of ion concentration into the cell (19–29). Agonist binding induces a closing of the LBD cleft (see Fig. 1, inset) (24,25), which in turn opens the ion channel in the full-length protein (24–34). The average distance of the isolated LBD cleft closure has been shown to directly correlate to channel transport efficiency for multiple glutamate receptor types (18,35–37), allowing the isolated LBD to be used as a model for further conformational analysis and prediction of channel transport efficiencies.

Figure 1.

Figure 1

Measured ensemble smFRET distribution for the NMDAR LBD with glycine bound. (Dashed curves) Gaussian fits for each state; (solid magenta curve) total sum of Gaussians to provide a visual estimate for the goodness of fits of the seven-state model. (Inset) Structure of the NMDAR GluN1 isolated LBD bound to glycine, based on the PDB:1PB7 crystal structure. The protein strand was mutated at T193 (red marker) and S115 (blue marker) to cysteine to attach the acceptor and donor fluorophores. A His-tag was added to the protein at the N-terminus (yellow marker) to allow for immobilization. (Green-dashed line) FRET distance, which changes as the protein opens and closes the cleft.

The NMDAR has a lower conformational stability and a higher range of conformations than the other glutamate receptors, including reported structure/function fluctuations with time constants as long as seconds (7,31–34). This increased instability likely plays a key role in the functionality of the NMDAR (24–29). Simulations of the NMDAR LBD suggest that the energy landscape of the protein is broad, even when bound to the primary agonist, indicating that there are many conformations that are weakly stable (19,38). In addition to the conformational changes experienced during normal receptor function, the protein can also unfold due to loss of stability of the folded state (18,39,40). Traditionally, structural determination through crystallographic methods has allowed for a detailed picture of only the most stable conformation to be obtained (15–17,38,41). Such information is static and must be paired with electrophysiological measurements to correlate current flow through the channel with static conformations. Due to the complex nature of protein folding, it is believed that the folding energy landscape contains a number of partially stable conformational states and that these energy minima help explain how proteins can fold into nonnative states (42,43). However, a link between the theory of broad conformational distributions and experimental studies of equilibrium transitions among conformations is lacking.

Single-molecule Förster resonance energy transfer (smFRET) experiments allow us to probe protein folding/unfolding dynamics on millisecond to second timescales (35,44–47). By collecting a statistically meaningful distribution of smFRET time trajectories, the entire ensemble equilibrium smFRET distribution, and therefore a detailed understanding of equilibrium dynamics, can be acquired (20,34–38,46). Here, we apply smFRET to record the dynamics occurring in the LBD cleft (Fig. 1), under saturating glycine conditions. At the high concentrations of glycine used in our experiments, the binding rate constant is ∼8 μs−1 (48), which is much faster than our collection rate of 1 ms−1. Therefore, any unbinding event is associated with a corresponding binding event faster than our time resolution, and we instead observe equilibrium transitions among the many conformations that comprise the time-averaged glycine-bound LBD state.

To our knowledge, our results provide several new insights. First, the resulting smFRET distribution for the NMDAR LBD correlates well to the derived energy landscape seen in simulation studies of the glycine-bound NMDAR LBD (38,49), which, along with another recent smFRET study (37), supports the conclusion that we are indeed monitoring LBD cleft conformational states. More importantly, we introduce a model-free step transition and state identification method (STaSI) to identify additional cleft states in the isolated LBD that are too wide to be observed in the full hetero-tetramer. Using STaSI and recent advances in extending smFRET trajectories (50–53), transition analysis confirms that interstate conversions follow equilibrium statistics. Although it is unclear how these long-lived states relate to channel function in the full receptor, the states identified by smFRET correlate well to the broad distribution of states identified in umbrella sampling models of the LBD cleft in both NMDA and AMPA receptors (38,54). In addition, it is possible with both the AMPAR and NMDAR LBDs to relate average smFRET values for a variety of agonist conditions to agonist binding strength and macroscale ion transport across the channel (35–37).

To further test the hypothesis that the observed transitions represent equilibrium dynamics across the LBD cleft, guanidinium chloride (GdmCl) was used to shift the distribution of identified states toward more open, and upon reversal, more closed forms. Although it is unknown whether such denaturants as GdmCl and urea induce protein denaturation through the normal folding pathway or via alternate energy wells (42), it is nonetheless useful to demonstrate that the broad distribution of states identified for the NMDA LBD can be reversibly shifted using such methods. Due to both experimental and computational limitations, most protein folding dynamics studies focus on sub-millisecond timescales (55,56), thus we hope the studies presented here also provide guidance on how to combine denaturants with smFRET to understand folding dynamics that occur on longer timescales. Overall, the observed transition statistics provide insight into the relationship between conformational dynamics and function in the NMDAR LBD as well as providing mechanistic detail about the more general topic of protein-folding/unfolding dynamics.

Materials and Methods

Protein preparation

A pet22B vector encoding N-terminal His-tagged NMDAR GluN1 isolated ligand-binding domain was provided by Eric Gouaux (Oregon Health and Science University, Portland, OR). Serine 115, located at the end of helix C in domain 1, and Threonine 193, located in the middle of helix G in domain 2 (S507 and T701 in the full-length protein), were mutated to cysteines using standard polymerase chain reaction methods with PfuTurbo DNA polymerase AD (Agilent Technologies, Santa Clara, CA) and primers obtained from Sigma-Aldrich (St. Louis, MO). These plasmids were then transformed into Origami B (DE3) Escherichia coli (Novagen, Darmstadt, Germany) and plated onto LB Agar (Thermo Fisher Scientific, Waltham, MA) supplemented with 50 μg/mL ampicillin (Sigma-Aldrich), 15 μg/mL kanamycin (Thermo Fisher Scientific), and 12.5 μg/mL tetracycline (Calbiochem, San Diego, CA). Isolated colonies were grown at 37°C in liquid Miller’s Luria Broth (Thermo Fisher Scientific), and protein production was induced when the culture reached an OD600 of 0.8.

To induce protein production, IPTG (isopropyl β-D-1-thiogalactopyranoside; Thermo Fisher Scientific) was added to a final concentration of 0.5 mM. For unnatural amino acid (UnAA) experiments, S115 and T193 were mutated to the amber stop codon, TAG, while the original stop codon was mutated from amber to ochre, TAA (37). The machinery to incorporate the UnAA was included by cotransforming the Origami E. coli with the pEVOL plasmid obtained from Peter Schultz (The Scripps Research Institute, San Diego, CA) (57). The pEVOL plasmid contains the UnAA p-acetyl-L-phenylalanine tRNA synthetase developed from the Methanococcus jansaschii TyrRS, as well as the suppressor tRNACUA. Nonspecific read-through of the amber stop codon by natural amino acids, while possible, would result in protein that would not be labeled with a FRET pair and would thus be invisible to our experiments. The pEVOL plasmid was maintained using 50 μg/mL chloramphenicol (Acros Organics, Geel, Belgium). For these experiments 0.02% arabinose (Sigma-Aldrich) and 1 mM p-acetyl-L-phenylalanine (RSP Amino Acids, Shirley, MA) was added in addition to IPTG. The E. coli were then allowed to grow for 24 h at 20°C before pelleting the cultures down and storing at −80°C. The UnAA NMDAR was analyzed by mass spectrometry to ensure proper expression (see Table S1 in the Supporting Material).

On the day before scanning, the pellets were thawed and lysed using a cell disruption vessel (Parr Instruments, Moline, IL). Cell debris was pelleted at 40,000 rpm for 1 h at 4°C, and the supernatant was further filtered through a 0.45-μm nylon membrane (Thermo Fisher Scientific). This filtrate was then purified via fast protein liquid chromatography (ÅKTA, Chicago, IL) using a linear imidazole gradient (Buffer A: 200 mM NaCl, 20 mM Tris, 1 mM glycine, pH 8.0; Buffer B: 200 mM NaCl, 20 mM Tris, 1 mM glycine, 400 mM imidazole, pH 8.0), and a HiTrap Chelating column (GE Lifesciences, Pitsburgh, PA) that had been charged with NiSO4 (Mallinckrodt Pharmaceuticals, Dublin, Ireland). Purified protein was then dialyzed for 2 h in PBS (phosphate-buffered saline) supplemented with 1 mM glycine (Thermo Fisher Scientific).

Protein labeling

On the day of scanning, dialyzed protein was quantified and labeled with Alexa 555 and Alexa 647 maleimide (Invitrogen, Grand Island, NY) to label cysteines, or the hydrazide versions of the dyes for the UnAA experiments. Dye was added to give a 1:1:4 protein/Alexa555/Alexa647. After a 30-min labeling time, protein was separated from unreacted fluorophores using a PD-10 desalting column (GE Healthcare Lifesciences). After this, protein was then further dialyzed in PBS with 1 mM glycine for 30 min. After dialysis, protein was then incubated with SulfoLink resin (Thermo Fisher Scientific) for another 30 min. Finally, 1 μg of biotin-conjugated anti-His epitope antibody (Rockland Immunochemicals, Limerick, PA) was added to 500 μL of the prepared protein.

Anisotropy measurements

Polarization anisotropy measurements were performed on the cysteine-labeled NMDAR LBD using an OD470 single-photon counting spectrometer (Edinburgh Instruments, Livingston, UK). A 657.2-nm excitation, 200-ns pulsed laser beam was emitted by a picosecond pulse diode laser (Edinburgh Instruments) and photons were collected at 675 nm. Four decay curves were recorded comprising horizontal and vertical detection polarizations after excitation in both the horizontal and vertical directions. The anisotropy curve was calculated from the decay curves and shown in Fig. S1 in the Supporting Material.

Microscope setup

All single-molecule fluorescence measurements were performed using a custom-built confocal microscope described previously in Landes et al. (35) and Darugar et al. (46). Briefly, the sample was excited with a continuous-wave 532-nm laser (Compass 315M-100 SL; Coherent, Santa Clara, CA) and focused through an oil immersion 100× 1.3 NA objective lens (Carl Zeiss, Oberkochen, Germany) onto the sample with a power density of ∼50 W/cm2 at the sample. The sample position was controlled with a scanning x-y-z piezo stage (P-517.3CL; Physik Instrumente, Karlsruhe, Germany) and the sample was moved to focus the confocal spot onto single molecules. Emitted light was collected back through the same objective and separated with a 640-nm high-pass dichroic mirror (640 DCXR; Chroma Technology, Bellows Falls, VT), then collected by two avalanche photodiodes (SPCM-AQR-15; PerkinElmer, Waltham, MA) tuned to 570 and 670 nm with band-pass filters (NHPF-532.0; Kaiser Optical Systems, Ann Arbor, MI and ET585; Chroma Technology) for donor and acceptor signal collection, respectively.

Sample chamber preparation

A sample chamber was prepared similar to that described previously in single-molecule measurements with the glutamate receptors (35,36). Plasma-cleaned 22 × 22 mm No. 1 glass coverslips were treated with a Vectabond-acetone solution (1% wt/vol; Vector Laboratories, Burlingame, CA) and stored under vacuum. Using silicone templates, a small section of the Vectabond-treated slides were then covered with a mixture containing 5 kDa biotin-terminated PEG (polyethylene glycol 2.5% w/w in MB water; NOF, Burlingame, CA) and sodium bicarbonate (Sigma-Aldrich) and allowed to dry in the dark for 4–6 h. Excess PEG was washed off and the pegylated area was covered with a custom 13-μL HybriWell chamber (Grace Bio-Labs, Bend, OR) fitted with two press fit silicon ports (Grace Bio-Labs) for entry and outlet flow tubes before taking a control image of the pegylated slide with PBS buffer. A solution of 0.2 mg/mL streptavidin (Invitrogen) in PBS buffer was added to the chamber and incubated for 10 min. Approximately 60 μL of 20 nM protein was added to the chamber in three successive injections and incubated for 20 min before being flushed with excess buffer solution and the sample was placed on the piezo stage for binding conformation.

Flow system preparation

To prolong photobleaching lifetimes, all experiments were performed in the presence of an oxygen-scavenging buffer system, consisting of 33% w/w β-D-(+)-glucose, 1% w/w glucose oxidase, 0.1% v/v catalase, 1 mM methyl viologen, and 1 mM ascorbic acid (all from Sigma-Aldrich) in MB water (MB Water Technologies, Chennai, India) saturated with phosphate buffer (50,52). In addition, 1 mM of glycine was added to the oxygen-scavenging system, depending on the experimental conditions. This solution was continuously pumped through the chamber at a rate of 1 μL/min. In all of the denaturant studies an additional flow system was set up containing 8 M guanidinium chloride (Sigma-Aldrich) and 1 mM glycine. For our measurements, the mean lifetime of the protein during observation was 4.96 s, assuming a first-order exponential decay process for photobleaching (see Fig. S2). Concentration of denaturant in the chamber was controlled by varying the flow rate in combination with the oxygen-scavenging system flow to get the desired concentration from the final mixed flow. For the single-molecule denaturant study the system was allowed to equilibrate for 30 min before data collection.

Single-molecule data collection

To obtain smFRET trajectories for the individual protein molecules, a 10 × 10 μm area of the sample was scanned to spatially locate 20–25 molecules. Once a molecule had been located, the stage was moved to place it under the laser focus, and the fluorescence signals of the donor and the acceptor were collected until the fluorophores were photobleached. The emission intensity trajectories were collected at 1-ms resolution and later binned to 10-ms time steps to improve signal/noise. Example trajectories are shown with the collected photon counts for the acceptor and donor channel (see Fig. 4, a and d) used to calculate the observed and denoised FRET signals (see Fig. 4, b and e).

Figure 4.

Figure 4

Two example single-molecule trajectories with the calculated FRET efficiency and state distribution. (a and d) The acceptor and donor signal for a single identified NMDAR LBD. The signal has been background-corrected and blink-filtered and shows a strong anticorrelation between the donor and acceptor, which is indicative of smFRET. (b and e) The corresponding apparent FRET efficiency for the signal in (a) and (d), respectively. The STaSI state at each time point is overlaid. (c and f) The smFRET histogram for each individual trajectory. The STaSI state histogram has also been overlaid and rescaled so that the maximum peak is the same height as the denoised maximum. (a–c) Example protein that remained stable in a high FRET efficiency state before transitioning into a stable lower FRET efficiency state and thus a more open-cleft conformation. (d–f) State-to-state movement is not one-directional and the NMDAR LBD can recover into the original state after visiting the open-cleft conformation.

Denaturation experiments

For the denaturation experiments, an area of 30 × 30 μm was raster-scanned three times and its location was recorded. Power of the laser was set to ∼25 W/cm2 at the sample and the speed of the scan was set to 1000 pixels per s with 256 pixels per line and 256 lines per image for a final per image time of 65.7 s. After the area image had been collected, the flow was adjusted to a new concentration of denaturant and left for 20 min to equilibrate, after which the area was rescanned under the new denaturant conditions. This process built up a series of images of the same molecules under all conditions of denaturant.

Data processing

All data was analyzed with programs written in-house using the software MATLAB (R2009b; The MathWorks, Natick, MA). The signal was processed via the wavelet denoising technique and the FRET efficiency was calculated from the denoised signal using the Förster equation (58). Wavelet denoising should not change the identification of the states, as verified by simulation (Fig. S3). A postprocessing algorithm was used to characterize the trajectories and ensure that they matched the criteria for a single-molecule, single-FRET-pair signal. Of the 238 trajectories collected, 76 were excluded after meeting the conditions of multistep bleaching or abnormally high background noise based on a normal distribution leaving 162 single-donor and single-acceptor-labeled molecules (59).

The raster-scanned area images were also analyzed with MATLAB. Each area was averaged with its repeats to enhance signal/noise. Molecules were located in the acceptor channel via a cutoff value set at the maximum intensity of the control image and grouped according to a Gaussian fitting to exclude clusters and single pixels of bright noise. The areas were spatially corrected to account for drift in the sample over the duration of the experiment. The FRET signal for each molecule was obtained and tracked throughout the experiment to build a denaturation profile of the protein.

Step transition and state identification analysis

The step-transition-and-state-identification (STaSI) method was developed to help cluster FRET traces into the optimum number of states with minimal user input (51), as opposed to other methods such as hidden Markov modeling or change point analysis, which require either foreknowledge of the number of expected states or the data to be raw-time-tagged rather than binned (60–62). STaSI combines the advantages of the bias-free approach of an information theoretic analysis method with the ability to analyze binned data to allow for unbiased conformational state determination. STaSI first identifies all the transition points using Student’s t-test (see Eq. S1 in the Supporting Material) based on similar analysis used in single-photon counting experiments (62,63). These identified transition points break down the FRET trace into different segments, and the t-test is applied to each segment until no further transition points are identified. After that, a hierarchical clustering is applied on the segments. For each clustering level, the difference between every remaining two-segment pairing is measured by its merit of likelihood (see Eq. S2 in the Supporting Material). The two clustered segments corresponding to the lowest difference are grouped into a single state and treated as a singular cluster in the next iteration. This process is repeated until the entire data-set is in a single state. We then applied the minimum description length (MDL) equation (see Eq. S3 in the Supporting Material) to the data set for each state assignment set and determined the optimum number of states. The MDL equation considers both the goodness of fit and the complexity of the fitting model. Simulation tests show MDL avoids overfitting and underfitting even with high noise level (51). More detailed explanation of STaSI can be found in the Supporting Material.

Results and Discussion

We collected 162 molecules over five samples of the isolated NMDAR LBD under conditions of saturating glycine. The FRET efficiencies of the single molecules were combined to produce the distribution shown in Fig. 1. The highest probability FRET value in the ensemble smFRET distribution is ∼0.95. This value corresponds to a cross-cleft distance of 3.2 nm, correlating well with the 3.4-nm cross-cleft distance calculated from the crystal structure for the glycine-bound LBD (PDB:1PB7; Protein Data Bank, www.wwpdb.org) observed in static structural analyses (24). The overall smFRET distribution, however, is very broad, indicating a wide variety of different conformations in addition to the most probable conformation. In fact, some smFRET values (e.g., those <∼0.60) represent such open conformations that they would not be observable in the full heterodimer construct due to steric hindrance, but are possible in the isolated LBD. The small amounts of protein exhibiting FRET efficiency values >1 are due to the background correction using the average value resulting in noise bringing the donor signal to <0 before FRET efficiency is calculated.

The FRET efficiency distribution in Fig. 1 is best described by seven underlying states (Fig. 2). The seven states are located at FRET efficiency values of 0.96, 0.87, 0.79, 0.68, 0.53, 0.43, and 0.28 (Table 1). To determine the number, location, and state-to-state transition points, we first reduced the noise of each smFRET trajectory with the wavelet denoising method (58,59), then analyzed the trajectories using the STaSI method (51), which was specifically developed to work with binned smFRET data (details are explained in Materials and Methods and in the Supporting Material). Based on state determination methods for single-photon counting and the MDL principle (62,63), STaSI was tested for validity with simulated trajectories with similar levels of noise to our data (see Fig. S4). Individually, each of the identified states has a relatively tight distribution, with a FRET efficiency standard deviation of ±0.04 on average, which is consistent with our measured noise level.

Figure 2.

Figure 2

The STaSI method determines the number and location of the states based on the denoised data. (a) All of the denoised smFRET trajectories were combined after local background subtraction and blink filtering, to provide a sufficiently large basis for the STaSI method to accurately locate the states, a 100× downsampled portion of which is shown. (b) The percentage assigned to each state from the total data set. FRET efficiency can be tracked horizontally from the data trajectory to the histogram. The contour of the distribution reflects the distribution of the measured FRET efficiency in Fig. 1.

Table 1.

Summary of the states found for both the cysteine-labeled NMDAR LBD and the UnAA-labeled NMDAR LBD

Cysteine states FRET efficiency 0.96 ± 0.03 0.87 ± 0.03 0.79 ± 0.03 0.68 ± 0.05 0.53 ± 0.05 0.43 ± 0.05 0.28 ± 0.04
Cysteine state (%) 27.0 21.6 15.6 13.7 12.3 6.0 3.8
UnAA states FRET efficiency 0.98 ± 0.03 0.90 ± 0.04 0.81 ± 0.04 0.65 ± 0.05 0.53 ± 0.05 0.43 ± 0.05 0.21 ± 0.03
UnAA state (%) 31.1 23.9 18.7 6.2 5.5 12.0 2.7

A comparison of the FRET response from a modified version of the NMDAR LBD containing an alternate labeling scheme shows that the FRET response is independent of labeling strategy and only occurs under normal labeling conditions (see Fig. S5). This is an important control because the traditional fluorescence labeling strategy binds the maleimide construct of the desired fluorophores to the sulfur atom of a point-mutated cysteine residue. However, inside the NMDAR LBD there exist native cysteine residues that cannot be mutated out, introducing potential alternate binding sites in the protein. The control involved an alternative method of dye attachment recently introduced by Dolino et al. (37) and Young et al. (57), which uses UnAA-containing functional groups not found in the standard 20 eukaryote amino acids to elicit greater binding site specificity. By point-mutating an UnAA at the desired labeling site, it is possible to have much higher specificity for fluorescence labeling. We used STaSI to compare the data from the ketone UnAA construct of the NMDAR LBD in glycine-saturated conditions to the data obtained from the cysteine-mutation-labeled NMDAR LBD (see Fig. S6) (37). STaSI analysis of both systems suggests seven states that match very closely to each other within the measurements and extraction errors (Table 1). Furthermore, to ensure that rotational hindrance caused by the attachment of the dyes to the protein did not affect the smFRET results, we used time-resolved rotational anisotropy to measure the rotational correlation time. The cysteine-labeled NMDAR LBD had a correlation relaxation time of 0.34 ns, indicating that the dye had sufficient rotational freedom to justify the use of 2/3 for the orientation factor in the FRET calculation (see Fig. S1) (64). These controls show that the contribution of the labeling scheme to the breadth of the smFRET distribution is minimal.

In order to show that the range of FRET efficiencies is related to conformation shift, we performed a denaturant assay to shift the state distribution. Upon the addition of the denaturant GdmCl to the NMDAR LBD solution, the average FRET efficiency decreases linearly with increasing concentration of GdmCl (Fig. 3). This shift is much more apparent when the average FRET values of all the molecules are plotted against the concentration of GdmCl (Fig. 3 b). Additionally, the shift toward lower FRET efficiencies in the presence of denaturant is reversible when denaturant concentration is lowered, as seen when the average values are replotted as a function of experimental time (Fig. 3 c). This effect is seen more clearly in the single-molecule trajectories taken for molecules before exposure to denaturant, during high concentrations of denaturant, and after denaturant concentration had been lowered (Fig. S7). The rebuilt ensemble distributions show a loss of the high FRET efficiency states in the presence of GdmCl, but the original state distribution is recovered after the denaturant is washed out, demonstrating that the conformation shift is a reversible process.

Figure 3.

Figure 3

Low-resolution raster-scanned image analysis shows how the average smFRET value from a single population of NMDAR LBD proteins changes as a function of GdmCl concentration. (a) A shift from high FRET efficiency to low FRET efficiency as denaturant concentration is increased can be seen in the histograms for the calculated FRET efficiency for identified molecules under each condition. (b) Combined average FRET efficiency for all of the areas as a function of GdmCl concentration. (c) Replotting the calculated average FRET efficiency against time elapsed for the denaturation experiment shows that by lowering the denaturant concentration the protein was able to recover, and that the denaturant’s effect on the NMDAR LBD is a reversible process; however, time-dependent photobleaching of the acceptor dye is also observed. The error bars plotted represent the distribution in the average among the different areas.

Due to the low signal/background intrinsic to the image analysis method (as compared to trajectory analysis), some populations of molecules are identified as having a FRET efficiency value of 1 or 0, because at these positions either the donor signal or the acceptor signal was indistinguishable from the background (Fig. 3 a). For statistical reasons, these values are not removed from the histograms, but do not reflect any physically meaningful structural information. Besides the effects of the denaturant concentration, there is an apparent drop in calculated FRET efficiency values occurring from the time-dependent photobleaching of the acceptor dye for labeled protein molecules. This explains why, at the two final concentrations, the average FRET values return to a slightly lower FRET efficiency than would be predicted. Regardless, the overall trend is consistent, namely that the population of lower FRET states correlates well with denaturant.

A broader explanation for the wide range of FRET values is that they reflect the equilibrium folding/unfolding landscape of the NMDAR LBD. The exact physical cause of the low FRET states is a matter for further investigation and debate, but by exploring the interactions of the states with one another from the single-molecule trajectories, and by observing the shift in the conformational distribution as a function of denaturant, we can begin to get insight into the broader topic of protein conformation dynamics, not just for the NMDAR LBD. The remainder of this work represents an effort to establish a relationship between the conformations of the NMDAR LBD in particular as related to ion-channel protein function, as well as the more general topic of equilibrium protein folding/unfolding, based on an analysis of interstate equilibrium dynamics.

The seven states can be generally grouped into either high FRET efficiency or low FRET efficiency regions. The four states with high FRET efficiency (0.96, 0.87, 0.79, and 0.68; corresponding to intercleft distances of 3.0, 3.7, 4.1, and 4.5 nm, respectively) can be further divided into two categories: the 0.96 and 0.87 FRET efficiencies, which correspond to closed-cleft conditions (24); and the 0.79 and 0.68 FRET efficiencies, which are associated with more open conformations. The division for these assignments is based on the known crystal structure of the closed cleft agonist-bound form (PDB:1PB7) and the open cleft antagonist-bound NMDAR LBD (PDB:1PBQ), the former of which shows a cleft distance of 3.4 nm and the latter of which shows cleft distances of 3.9–4.0 nm (24).

The states with low FRET efficiencies (0.53, 0.43, and 0.28; corresponding to distance of 5.0, 5.3, and 5.9 nm, respectively) indicate that the protein can visit unexpected conformations that are more open than any values observed in the full receptor for bulk ensemble studies and are expected not to be present in a functional full-length receptor. Based on the glycine-bound crystal structure (24), the furthest attainable distance for any type of motion of the LBD, and thus the maximum possible distance between the two dye attachment sites, is ∼5.0 nm. Thus, further opening of the LBD fragment may be due to the lack of steric hindrance experienced by the isolated LBD structure used in these experiments as compared to the spatially constrained LBD in the membrane-imbedded tetramer, and could also include partial unfolding. In the full-length NMDAR, neighboring segments of the tetrameric structure limit the possible distance achievable for any motion of the LBD >∼5 nm (15,16). Thus, although open conformational states of the agonist-bound LBD are predicted from the theory, in our smFRET studies the cleft can explore even more extreme conformations than would exist in the full-length protein.

There are 145 smFRET trajectories (86% of the traces; 80% of the data points) that occupy only one or two states, and therefore only contain a few transitions (Fig. 4, a–c). Of these, 78 trajectories (46% of the traces; 37% of the data points) display only a single state for the duration of observation. Due to the stochastic nature of photobleaching, many of these trajectories are shorter than the mean lifetime of observation, and hence do not fluoresce long enough to visit multiple states. These trajectories do not contribute to the transition analysis but are still valuable in recreating the overall conformational landscape. Furthermore, each of the seven STaSI-identified states is well represented within the individual static trajectories.

The remaining 24 trajectories (14% of the traces; 20% of the data points) exhibit reversible interstate dynamics (Fig. 4, d–f). The protein not only transitions back and forth between two different states, but also remains stable in both states for hundreds of milliseconds at a time, before reverting again. This reversible transition suggests that the protein is actively exploring alternate conformations and that the protein is unlikely to be simply degrading and/or completely unfolding. The reversible nature of these trajectories indicates that all of the visited states lie in equilibrium with each other, and represent real conformational states that the LBD experiences.

Our method is unable to comment on dynamics faster than could be detected by the 1-ms bin time of our collection. However, for the ranges presented, there is good agreement with the ion channel studies for the NMDAR LBD (27–29). The apparent semistable states that exist near the distance corresponding with ion channel activation could be responsible for the inability of the NMDAR to activate unless all four subunits of the full tetramer act together. Further work is needed to observe the effects that alternate agonists have on the conformational landscape of the NMDAR.

The number of transitions from each state to every other state was extracted from the smFRET time trajectories, and their transition pattern is consistent with the previous theoretical predictions (Fig. 5) (38,49). The transition counts in Fig. 5 show that the major transitions happen primarily between adjacent states, represented by the warm colors of the transitions along the diagonal. As the analysis describes motions between states that are further away from each other, the number of observed transitions decays quickly. Given that the positional placement of the acceptor and donor attachment sites are on opposite sides of the cleft, adjacent FRET states, also being adjacent conformational states, suggest that the observed motion is an ordered opening and closing of the cleft. This property agrees well with predicted models of the NMDAR LBD dynamics published by Yao et al. (38), who postulated that the protein moves along a well-ordered opening pathway between adjacent conformations. This also agrees well with the predicted clam-shell model often used to describe the motion of the glutamate receptor LBDs (35).

Figure 5.

Figure 5

Matrix of transition counts between the STaSI states. The state of the protein before the transition point defines the vertical axis, and the state that the NMDAR LBD transitions to is the horizontal axis. State transitions cannot occur between a state and itself, which nullifies the diagonal. (Color) Total number of transitions for that state pair.

The transitions have a high degree of symmetry across the diagonal in Fig. 5, indicating that the protein is undergoing equilibrium transitions, even among the lowest FRET states. These reversible transitions suggest that the low FRET efficiency states are indeed metastable states within the folding/unfolding equilibrium. Nonequilibrium unfolding, due to either unfavorable mutations or the immobilization process, could also cause transitions between FRET states to occur but would not be reversible processes. Thus, the data is strongly suggestive of interconversions among a broad range of conformations that reflect both those that would occur in the native, membrane-imbedded NMDAR and transitions to more open states within the underlying highly dynamic LBD that are otherwise constricted in the full heterodimer construct. Buried within the broad range of conformations, however, is information about the conformational flexibility within the ligand-bound form, the possible pathways for functional activation, and insight into reversible protein folding/unfolding (19,35,36,38).

Because we observe such a wide range of reversible transitions, and because conformational flexibility is a required hallmark of activation within the full NMDAR heterodimer (13), the observed transitions within isolated single NMDAR LBDs could give some insight into the activation process. Based on crystal structures, it has been hypothesized that the cleft-closure conformational change is the primary mediator of activation. In fact, among all of the possible interstate transitions, the most probable occurs between 0.87 and 0.79 FRET efficiencies (Fig. 5), just at the break between the two closed-cleft FRET states and the two open-cleft FRET states. The distance of the open-cleft conformation of the LBD from crystal structure is just above 0.82 FRET efficiency (24). Thus, it is interesting to observe that the most probable observed transition corresponds to the distance at which a cleft-closure conformational change would result in an active channel. Further experiments and theoretical modeling would provide more direction as to how the varying smFRET states could be of value in understanding the cleft-closure mechanism.

The origin of the transitions among the states with FRET <0.60, however, is not clearly relatable to function in the full membrane-imbedded receptor, but still provides valuable insight into the reversible folding/unfolding at a few key points of the cleft, as mentioned earlier. The shifting of the equilibrium of the FRET values to the lower states during high concentration of GdmCl indicates that the lower states may be associated with partial unfolding of the protein. GdmCl should weaken the bonds in the backbone structure of the protein, increasing its overall structural flexibility (65). As the flexibility of the protein increases, the hinge between the top and bottom regions of the NMDAR LBD becomes looser, and the protein moves to an overall more open conformation and therefore lower FRET states. The recovery of the FRET states with reduced denaturant concentration suggests that the structural change is reversible. When the hinge region is partially unfolded, the protein’s equilibrium would shift toward the open-cleft lower FRET states with the two arms well separated. When the hinge region reforms, the protein can stiffen again and fold back to the base structure. This model is consistent with our results, but could potentially relate to the functional issue of LBD activation and/or desensitization. Further work is required to elucidate any possible relationship.

Conclusions

We have shown that the glycine-saturated LBD of the GluN1 subunit of the NMDAR has a complex conformational landscape. Using smFRET, we were able to measure and locate conformational states that the protein adopts on a timescale of the order of milliseconds to seconds. Introducing increasing concentrations of denaturant shifts the protein’s equilibrium toward open conformations and is a reversible process. The denaturant-induced shift toward open conformations maintains reversibility, which supports the notion that we are observing equilibrium folding/unfolding transitions. The protein moves primarily from adjacent states and can migrate to both more open conformations and more closed conformations similarly, matching both theoretical calculations as well as a simple model of opening and closing of the binding cleft. The most probable transitions are consistent with cleft distances related to the opening/closing transitions in the full protein. These findings can be helpful in further studies to address the complex functionalities of the NMDAR LBD and have a possible impact on drug design for treating NMDAR-related diseases.

Author Contributions

For this article, D.R.C. designed research, performed research, contributed analytic tools, analyzed data, wrote the article, and edited the article; D.M.D. designed research, performed research, wrote the article, and edited the article; H.J. performed research, analyzed data, wrote the article, and edited the article; B.S. contributed analytic tools, analyzed data, wrote the article, and edited the article; S.R. performed research and edited the article; C.E.N. performed research, wrote the article, and edited the article; J.C. contributed analytic tools, analyzed data, and edited the article; V.J. designed research, wrote the article, and edited the article; and C.F.L. designed research, wrote the article, and edited the article.

Acknowledgments

The authors thank Lydia Kisley, Joey Tauzin, and Chad Byers as well as Professor Stephan Link and his research group for their thoughts and discussions on this project. The authors also thank Nathan Cook from the Marti Lab at Rice University for his help with the anisotropy data collection as well as Dr. Xuemei Luo from the Mass Spectrometry Core Lab at the University of Texas Medical Branch for her work with the mass spectrometry data collection and analysis. The authors thank Dr. Mike Cascio for his advice on mass spectrometry.

This work is supported by the Welch Foundation (grant No. C-1787), the National Science Foundation (grant No. CHE-1151647), and the National Institutes of Health (grant No. GM94246-01A1). C.E.N. acknowledges NIH- 2T32 GM008280-26.

Editor: Andrew Plested.

Footnotes

Jixin Chen’s present address is Department of Checmistry and Biochemistry, Ohio University, Athens, Ohio.

Supporting Materials and Methods, Supporting Results, Supporting Discussion, eight figures, and two tables are available at http://www.biophysj.org/biophysj/supplemental/S0006-3495(15)00537-8.

Contributor Information

Vasanthi Jayaraman, Email: vasanthi.jayaraman@uth.tmc.edu.

Christy F. Landes, Email: cflandes@rice.edu.

Supporting Material

Document S1. Supporting Materials and Methods, Supporting Results, Supporting Discussion, eight figures, and two tables
mmc1.pdf (1.2MB, pdf)
Document S2. Article plus Supporting Material
mmc2.pdf (2.6MB, pdf)

References

  • 1.Chen P.E., Wyllie D.J.A. Pharmacological insights obtained from structure-function studies of ionotropic glutamate receptors. Br. J. Pharmacol. 2006;147:839–853. doi: 10.1038/sj.bjp.0706689. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Johnson K.A., Conn P.J., Niswender C.M. Glutamate receptors as therapeutic targets for Parkinson’s disease. CNS Neurol. Disord. Drug Targets. 2009;8:475–491. doi: 10.2174/187152709789824606. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Madsen U., Ebert B., Krogsgaard-Larsen P. Modulation of AMPA receptor function in relation to glutamatergic abnormalities in Alzheimer’s disease. Biomed. Pharmacother. 1994;48:305–311. doi: 10.1016/0753-3322(94)90176-7. [DOI] [PubMed] [Google Scholar]
  • 4.Querfurth H.W., LaFerla F.M. Alzheimer’s disease. N. Engl. J. Med. 2010;362:329–344. doi: 10.1056/NEJMra0909142. [DOI] [PubMed] [Google Scholar]
  • 5.Sase S., Khan D., Lubec G. Modafinil improves performance in the multiple T-maze and modifies GluR1, GluR2, D2 and NR1 receptor complex levels in the C57BL/6J mouse. Amino Acids. 2012;43:2285–2292. doi: 10.1007/s00726-012-1306-y. [DOI] [PubMed] [Google Scholar]
  • 6.Zamanillo D., Sprengel R., Sakmann B. Importance of AMPA receptors for hippocampal synaptic plasticity but not for spatial learning. Science. 1999;284:1805–1811. doi: 10.1126/science.284.5421.1805. [DOI] [PubMed] [Google Scholar]
  • 7.Dingledine R., Borges K., Traynelis S.F. The glutamate receptor ion channels. Pharmacol. Rev. 1999;51:7–61. [PubMed] [Google Scholar]
  • 8.Mayer M.L. Glutamate receptor ion channels. Curr. Opin. Neurobiol. 2005;15:282–288. doi: 10.1016/j.conb.2005.05.004. [DOI] [PubMed] [Google Scholar]
  • 9.Paoletti P., Neyton J. NMDA receptor subunits: function and pharmacology. Curr. Opin. Pharmacol. 2007;7:39–47. doi: 10.1016/j.coph.2006.08.011. [DOI] [PubMed] [Google Scholar]
  • 10.Reichling D.B., MacDermott A.B. NMDA receptor-mediated calcium entry in the absence of AMPA receptor activation in rat dorsal horn neurons. Neurosci. Lett. 1996;204:17–20. doi: 10.1016/0304-3940(96)12305-3. [DOI] [PubMed] [Google Scholar]
  • 11.Sun H., Kawahara Y., Kwak S. Expression profile of AMPA receptor subunit mRNA in single adult rat brain and spinal cord neurons in situ. Neurosci. Res. 2005;52:228–234. doi: 10.1016/j.neures.2005.03.008. [DOI] [PubMed] [Google Scholar]
  • 12.Ulbrich M.H., Isacoff E.Y. Rules of engagement for NMDA receptor subunits. Proc. Natl. Acad. Sci. USA. 2008;105:14163–14168. doi: 10.1073/pnas.0802075105. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Mayer M.L., Vyklicky L., Jr., Clements J. Regulation of NMDA receptor desensitization in mouse hippocampal neurons by glycine. Nature. 1989;338:425–427. doi: 10.1038/338425a0. [DOI] [PubMed] [Google Scholar]
  • 14.Traynelis S.F., Wollmuth L.P., Dingledine R. Glutamate receptor ion channels: structure, regulation, and function. Pharmacol. Rev. 2010;62:405–496. doi: 10.1124/pr.109.002451. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Karakas E., Furukawa H. Crystal structure of a heterotetrameric NMDA receptor ion channel. Science. 2014;344:992–997. doi: 10.1126/science.1251915. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Lee C.H., Lü W., Gouaux E. NMDA receptor structures reveal subunit arrangement and pore architecture. Nature. 2014;511:191–197. doi: 10.1038/nature13548. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Mayer M.L. Glutamate receptors at atomic resolution. Nature. 2006;440:456–462. doi: 10.1038/nature04709. [DOI] [PubMed] [Google Scholar]
  • 18.Rambhadran A., Gonzalez J., Jayaraman V. Subunit arrangement in N-methyl-D-aspartate (NMDA) receptors. J. Biol. Chem. 2010;285:15296–15301. doi: 10.1074/jbc.M109.085035. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Dai J., Zhou H.-X. An NMDA receptor gating mechanism developed from MD simulations reveals molecular details underlying subunit-specific contributions. Biophys. J. 2013;104:2170–2181. doi: 10.1016/j.bpj.2013.04.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.MacDermott A.B., Mayer M.L., Barker J.L. NMDA-receptor activation increases cytoplasmic calcium concentration in cultured spinal cord neurones. Nature. 1986;321:519–522. doi: 10.1038/321519a0. [DOI] [PubMed] [Google Scholar]
  • 21.Lemoine D., Jiang R., Grutter T. Ligand-gated ion channels: new insights into neurological disorders and ligand recognition. Chem. Rev. 2012;112:6285–6318. doi: 10.1021/cr3000829. [DOI] [PubMed] [Google Scholar]
  • 22.Myers S.J., Dingledine R., Borges K. Genetic regulation of glutamate receptor ion channels. Annu. Rev. Pharmacol. Toxicol. 1999;39:221–241. doi: 10.1146/annurev.pharmtox.39.1.221. [DOI] [PubMed] [Google Scholar]
  • 23.Speranskiy K., Kurnikova M. On the binding determinants of the glutamate agonist with the glutamate receptor ligand binding domain. Biochemistry. 2005;44:11508–11517. doi: 10.1021/bi050547w. [DOI] [PubMed] [Google Scholar]
  • 24.Furukawa H., Gouaux E. Mechanisms of activation, inhibition and specificity: crystal structures of the NMDA receptor NR1 ligand-binding core. EMBO J. 2003;22:2873–2885. doi: 10.1093/emboj/cdg303. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Inanobe A., Furukawa H., Gouaux E. Mechanism of partial agonist action at the NR1 subunit of NMDA receptors. Neuron. 2005;47:71–84. doi: 10.1016/j.neuron.2005.05.022. [DOI] [PubMed] [Google Scholar]
  • 26.Ylilauri M., Pentikäinen O.T. Structural mechanism of N-methyl-D-aspartate receptor type 1 partial agonism. PLoS ONE. 2012;7:e47604. doi: 10.1371/journal.pone.0047604. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Kussius C.L., Popescu G.K. Kinetic basis of partial agonism at NMDA receptors. Nat. Neurosci. 2009;12:1114–1120. doi: 10.1038/nn.2361. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Kussius C.L., Popescu G.K. NMDA receptors with locked glutamate-binding clefts open with high efficacy. J. Neurosci. 2010;30:12474–12479. doi: 10.1523/JNEUROSCI.3337-10.2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Paganelli M.A., Kussius C.L., Popescu G.K. Role of cross-cleft contacts in NMDA receptor gating. PLoS ONE. 2013;8:e80953. doi: 10.1371/journal.pone.0080953. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Foucaud B., Laube B., Betz H. Structural model of the N-methyl-D-aspartate receptor glycine site probed by site-directed chemical coupling. J. Biol. Chem. 2003;278:24011–24017. doi: 10.1074/jbc.M300219200. [DOI] [PubMed] [Google Scholar]
  • 31.Amico-Ruvio S.A., Popescu G.K. Stationary gating of GluN1/GluN2B receptors in intact membrane patches. Biophys. J. 2010;98:1160–1169. doi: 10.1016/j.bpj.2009.12.4276. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Zhang W., Howe J.R., Popescu G.K. Distinct gating modes determine the biphasic relaxation of NMDA receptor currents. Nat. Neurosci. 2008;11:1373–1375. doi: 10.1038/nn.2214. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Abele R., Keinanen K., Madden D.R. Agonist-induced isomerization in a glutamate receptor ligand-binding domain. A kinetic and mutagenetic analysis. J. Biol. Chem. 2000;275:21355–21363. doi: 10.1074/jbc.M909883199. [DOI] [PubMed] [Google Scholar]
  • 34.McFeeters R.L., Oswald R.E. Emerging structural explanations of ionotropic glutamate receptor function. FASEB J. 2004;18:428–438. doi: 10.1096/fj.03-0873rev. [DOI] [PubMed] [Google Scholar]
  • 35.Landes C.F., Rambhadran A., Jayaraman V. Structural landscape of isolated agonist-binding domains from single AMPA receptors. Nat. Chem. Biol. 2011;7:168–173. doi: 10.1038/nchembio.523. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Ramaswamy S., Cooper D., Jayaraman V. Role of conformational dynamics in α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA) receptor partial agonism. J. Biol. Chem. 2012;287:43557–43564. doi: 10.1074/jbc.M112.371815. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Dolino D.M., Cooper D., Jayaraman V. Structural dynamics of the glycine-binding domain of the N-methyl-D-aspartate receptor. J. Biol. Chem. 2015;290:797–804. doi: 10.1074/jbc.M114.605436. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Yao Y., Belcher J., Lau A.Y. Conformational analysis of NMDA receptor GluN1, GluN2, and GluN3 ligand-binding domains reveals subtype-specific characteristics. Structure. 2013;21:1788–1799. doi: 10.1016/j.str.2013.07.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Schuler B., Eaton W.A. Protein folding studied by single-molecule FRET. Curr. Opin. Struct. Biol. 2008;18:16–26. doi: 10.1016/j.sbi.2007.12.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Postila P.A., Ylilauri M., Pentikäinen O.T. Full and partial agonism of ionotropic glutamate receptors indicated by molecular dynamics simulations. J. Chem. Inf. Model. 2011;51:1037–1047. doi: 10.1021/ci2000055. [DOI] [PubMed] [Google Scholar]
  • 41.Robertson J.W.F., Kasianowicz J.J., Banerjee S. Analytical approaches for studying transporters, channels and porins. Chem. Rev. 2012;112:6227–6249. doi: 10.1021/cr300317z. [DOI] [PubMed] [Google Scholar]
  • 42.Braselmann E., Chaney J.L., Clark P.L. Folding the proteome. Trends Biochem. Sci. 2013;38:337–344. doi: 10.1016/j.tibs.2013.05.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Baker D., Agard D.A. Kinetics versus thermodynamics in protein folding. Biochemistry. 1994;33:7505–7509. doi: 10.1021/bi00190a002. [DOI] [PubMed] [Google Scholar]
  • 44.Borgia M.B., Borgia A., Clarke J. Single-molecule fluorescence reveals sequence-specific misfolding in multidomain proteins. Nature. 2011;474:662–665. doi: 10.1038/nature10099. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Chung H.S., McHale K., Eaton W.A. Single-molecule fluorescence experiments determine protein folding transition path times. Science. 2012;335:981–984. doi: 10.1126/science.1215768. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Darugar Q., Kim H., Landes C. Human T-cell lymphotropic virus type 1 nucleocapsid protein-induced structural changes in transactivation response DNA hairpin measured by single-molecule fluorescence resonance energy transfer. J. Virol. 2008;82:12164–12171. doi: 10.1128/JVI.01158-08. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Schuler B. Single-molecule FRET of protein structure and dynamics—a primer. J. Nanobiotechnol. 2013;11(Suppl 1):S2. doi: 10.1186/1477-3155-11-S1-S2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Clements J.D., Westbrook G.L. Activation kinetics reveal the number of glutamate and glycine binding sites on the N-methyl-D-aspartate receptor. Neuron. 1991;7:605–613. doi: 10.1016/0896-6273(91)90373-8. [DOI] [PubMed] [Google Scholar]
  • 49.Dutta A., Shrivastava I.H., Bahar I. Comparative dynamics of NMDA- and AMPA-glutamate receptor N-terminal domains. Structure. 2012;20:1838–1849. doi: 10.1016/j.str.2012.08.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Cooper D., Uhm H., Landes C.F. Photobleaching lifetimes of cyanine fluorophores used for single-molecule Förster resonance energy transfer in the presence of various photoprotection systems. ChemBioChem. 2013;14:1075–1080. doi: 10.1002/cbic.201300030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Shuang B., Cooper D., Landes C.F. Fast step transition and state identification (STaSI) for discrete single-molecule data analysis. J. Phys. Chem. Lett. 2014;5:3157–3161. doi: 10.1021/jz501435p. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Vogelsang J., Kasper R., Tinnefeld P. A reducing and oxidizing system minimizes photobleaching and blinking of fluorescent dyes. Angew. Chem. Int. Ed. Engl. 2008;47:5465–5469. doi: 10.1002/anie.200801518. [DOI] [PubMed] [Google Scholar]
  • 53.Glembockyte V., Lincoln R., Cosa G. Cy3 photoprotection mediated by Ni2+ for extended single-molecule imaging: old tricks for new techniques. J. Am. Chem. Soc. 2015;137:1116–1122. doi: 10.1021/ja509923e. [DOI] [PubMed] [Google Scholar]
  • 54.Lau A.Y., Roux B. The free energy landscapes governing conformational changes in a glutamate receptor ligand-binding domain. Structure. 2007;15:1203–1214. doi: 10.1016/j.str.2007.07.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Gelman H., Gruebele M. Fast protein folding kinetics. Quart. Rev. Biophys. 2014;47:95–142. doi: 10.1017/S003358351400002X. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Brust R., Lukacs A., Meech S.R. Proteins in action: femtosecond to millisecond structural dynamics of a photoactive flavoprotein. J. Am. Chem. Soc. 2013;135:16168–16174. doi: 10.1021/ja407265p. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Young T.S., Ahmad I., Schultz P.G. An enhanced system for unnatural amino acid mutagenesis in E. coli. J. Mol. Biol. 2010;395:361–374. doi: 10.1016/j.jmb.2009.10.030. [DOI] [PubMed] [Google Scholar]
  • 58.Taylor J.N., Makarov D.E., Landes C.F. Denoising single-molecule FRET trajectories with wavelets and Bayesian inference. Biophys. J. 2010;98:164–173. doi: 10.1016/j.bpj.2009.09.047. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59.Taylor J.N., Landes C.F. Improved resolution of complex single-molecule FRET systems via wavelet shrinkage. J. Phys. Chem. B. 2011;115:1105–1114. doi: 10.1021/jp1050707. [DOI] [PubMed] [Google Scholar]
  • 60.McKinney S.A., Joo C., Ha T. Analysis of single-molecule FRET trajectories using hidden Markov modeling. Biophys. J. 2006;91:1941–1951. doi: 10.1529/biophysj.106.082487. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 61.Bronson J.E., Fei J., Wiggins C.H. Learning rates and states from biophysical time series: a Bayesian approach to model selection and single-molecule FRET data. Biophys. J. 2009;97:3196–3205. doi: 10.1016/j.bpj.2009.09.031. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 62.Watkins L.P., Yang H. Detection of intensity change points in time-resolved single-molecule measurements. J. Phys. Chem. B. 2005;109:617–628. doi: 10.1021/jp0467548. [DOI] [PubMed] [Google Scholar]
  • 63.Hanson A.J., Fu P.C.-W. Application of MDL to selected families of models. In: Grunwald P.D., Myung I.J., Pitt M.A., editors. Advances in Minimum Description Length: Theory and Applications (Neural Information Processing series) The MIT Press; Cambridge, MA: 2005. pp. 125–150. [Google Scholar]
  • 64.van der Meer B.W. Kappa-squared: from nuisance to new sense. J. Biotechnol. 2002;82:181–196. doi: 10.1016/s1389-0352(01)00037-x. [DOI] [PubMed] [Google Scholar]
  • 65.Camilloni C., Rocco A.G., Tiana G. Urea and guanidinium chloride denature protein L in different ways in molecular dynamics simulations. Biophys. J. 2008;94:4654–4661. doi: 10.1529/biophysj.107.125799. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Document S1. Supporting Materials and Methods, Supporting Results, Supporting Discussion, eight figures, and two tables
mmc1.pdf (1.2MB, pdf)
Document S2. Article plus Supporting Material
mmc2.pdf (2.6MB, pdf)

Articles from Biophysical Journal are provided here courtesy of The Biophysical Society

RESOURCES