Can single molecule localization microscopy be used to map closely spaced RGD nanodomains?

Abstract

Cells sense and respond to nanoscale variations in the distribution of ligands to adhesion receptors. This makes single molecule localization microscopy (SMLM) an attractive tool to map the distribution of ligands on nanopatterned surfaces. We explore the use of SMLM spatial cluster analysis to detect nanodomains of the cell adhesion-stimulating tripeptide arginine-glycine-aspartic acid (RGD). These domains were formed by the phase separation of block copolymers with controllable spacing on the scale of tens of nanometers. We first determined the topology of the block copolymer with atomic force microscopy (AFM) and then imaged the localization of individual RGD peptides with direct stochastic optical reconstruction microscopy (dSTORM). To compare the data, we analyzed the dSTORM data with DBSCAN (density-based spatial clustering application with noise). The ligand distribution and polymer topology are not necessary identical since peptides may attach to the polymer outside the nanodomains and/or coupling and detection of peptides within the nanodomains is incomplete. We therefore performed simulations to explore the extent to which nanodomains could be mapped with dSTORM. We found that successful detection of nanodomains by dSTORM was influenced by the inter-domain spacing and the localization precision of individual fluorophores, and less by non-specific absorption of ligands to the substratum. For example, under our imaging conditions, DBSCAN identification of nanodomains spaced further than 50 nm apart was largely independent of background localisations, while nanodomains spaced closer than 50 nm required a localization precision of ~11 nm to correctly estimate the modal nearest neighbor distance (NDD) between nanodomains. We therefore conclude that SMLM is a promising technique to directly map the distribution and nanoscale organization of ligands and would benefit from an improved localization precision.

Topological data analysis quantifies biological nano-structure from single molecule localization microscopy

“…The localization uncertainty for each blinking event was determined using a normal distribution centred on the molecule position. Standard deviation for localization uncertainty was set using a lognormal distribution with mean 2.8 and standard deviation 0.28 (experimentally measured parameters for AlexaFluor 647) (Mollazade et al, 2017). Detection rate for blinking events was set to 70%.…”

A framework for evaluating the performance of SMLM cluster analysis algorithms

“…The simulateSTORM.r script from the RSMLM package (available at: https://github.com/JeremyPike/RSMLM) was used to generate the blinking simulations 25 . Briefly, transition between the fluorescent on-and off-state were modelled using a geometric distribution 20,25 with probability of transition to the dark state set to 0.2, generating on average 4-5 fluorescent on-states, and thus, detections per molecule. Blinking was applied to all molecules in the simulations, thus single background molecules were also prone to blinking here.…”

Topological data analysis quantifies biological nano-structure from single molecule localization microscopy

“…The localisation uncertainty for each blinking event was determined using a normal distribution centered on the molecule position. Standard deviation for localisation uncertainty was set using a log-normal distribution with mean 2.8 and standard deviation 0.28 (experimentally measured parameters for AlexaFluor647) (36). Detection rate for blinking events was set to 70%.…”