Biophysical Society Bulletin | February 2023

Publications

Editor’s Pick Biophysical Reports Single-photon smFRET. I: Theory and conceptual basis Ayush Saurabh, Mohamadreza Fazel, Matthew Safar, Ioannis Sgouralis, Steve Pressé “smFRET is a widely used

Know the Editor Melanie Cocco

University of California, Irvine Associate Editor Biophysical Reports

Melanie Cocco

Figure 5: Learned bivariate posterior for the system state escape rate FRET efficien ies ε FRET given synthetic data. To produce this plot, w synthetic data generated using an excitation rate of λ ex = 10 ms − 1 , and escape ra & 2 ms − 1 and FRET efficiencies of 0.09 & 0.29 for the two system states, respec ground truth is shown with red dots. The FRET efficiencies estimated by o are 0 . 288 +0 . 007 − 0 . 006 and 0 . 092 +0 . 003 − 0 . 003 . Furthermore, predicted escape rates are 2 . 03 +0 . 1 − 0 . 1 0 . 98 +0 . 10 − 0 . 07 ms − 1 . The s all bias away from the ground truth is due to the finiten We have smoothed the distributions, for illustrative purposes only, using ker estimation (KDE) available through the Julia Plots package.

technique for studying kinetics of molecular complexes. However, until now, smFRET data analysis methods have required specifying a priori the dimensionality of the underly ing physical model (the exact number of kinetic parameters). Such approaches are inherently limiting given the typically unknown number of physical configurations a molecular com plex may assume. The methods presented here eliminate this requirement and allow estimating the physical model itself along with kinetic parameters, while incorporating all sources of noise in the data.” s = Kλ ex λ probe M σ ,

What are you currently working on that excites you? I have a collaboration with a pharmaceutical company to study the mechanism of action for a new class of drugs to treat sickle cell disease by binding hemoglobin and preventing fiber formation. This drug has been described as life-changing for individuals with this painful and damaging condition. Crys tallography has shown that the protein structure does not change substantially upon drug binding, leaving many ques tions about how the drug influences the protein in solution. Our industry-sponsored project will use NMR spectroscopy and other biophysical techniques to understand the effect of drug binding on protein stability and dynamics. What has been your biggest “aha” moment in science? My lab determined the structure of the Neurite Outgrowth Inhibitor (NOGO) protein. The extracellular part of this protein is active so we chose to work on that fragment. I asked my student, Jessica Schulz , to perform a measurement of residual dipolar couplings (RDCs) to enhance the structural informa tion we had on this system. The RDC experiment uses lipid bicelles in a liquid crystal form. A few weeks later, I asked Jessica about the results and she told me that the experi ment did not work since all of the protein signals disappeared every time she added the lipid bicelles. Although she inter preted this as a failure, I was very excited since it indicated that the protein could be binding to the lipid bilayer, even in the absence of the two flanking transmembrane helices. We confirmed that NOGO does associate strongly with mem branes even though it does not have a canonical lipid-binding motif. In this case, lipid interactions drive the structure of the protein to form.

budg t needed t accurat ly stimate th transition rates in e model as

where K is the total number of photons in a single photon smFRET trace (photo λ ex is the excitation rate, λ probe represents the escape rate (timescale) that we wan and M σ is the number of system states. The parameters in the numerator control of data available and the temporal resolution. On the other hand, the parame denominator are the properties of the system under investigation and represent t resolution. From experimentation, we have found a photon budget index of approxima be a safe lower threshold for keeping errors below 15% (this error cutoff is a u in parameter estimates. In the simple parametric example above, we have K = λ ex = 10 ms − 1 , and the fastest transition that we want to probe is λ probe = 2 M σ = 2 , which corresponds to a photon budget index of 1 . 08 × 10 7 . In Fig. demonstrate the reduction in errors (confidence interval size) for parameters o system as the photon budget is increased from 12500 photons to 400000 photon of those cases, we used 9000 MCMC samples to compute statistical metrics such a 49

Version of Record Published December 1, 2022 DOI: https:/doi.org/10.1016/j.bpr.2022.100089

DriveAFM Performance without compromise New and expanded functionality for bio in 2023

Speed • Ultra-low Noise • Unparalleled Stability • Automation • Nanosurf Studio Software Suite

nanosurf.com/DriveAFM

February 2023

8

T H E N E W S L E T T E R O F T H E B I O P H Y S I C A L S O C I E T Y