Identification of the factors affecting co-localization precision for quantitative multicolor localization microscopy

Abstract

This work explores the potential of multi-color Photoactivated Localization Microscopy (PALM) imaging to probe sub-diffraction limit interactions between proteins with spectrally separated labels. Using a PALM setup built around a commercial microscope axially stabilized to nm-level, we determined the ultimate registration accuracy that could be achieved (10 nm) and compared the performance of three different pairs of fluorescent proteins that can be used in dual color PALM. Fusion constructs were cloned and imaged either in vitro or at the cell plasma membrane, allowing to identify a current limit to co-localization precision of approximately 30-40 nm. We identified the better performing pair and present a concluding perspective application to a co-clustering study.

Counting molecules in single organelles with superresolution microscopy allows tracking of the endosome maturation trajectory

“…As shown in previous studies, blinking of PAFPs can be corrected by combining fluorescent bursts from a single molecule based on a cutoff time determined from the measured characteristic darktime distribution (25,27). In addition, it has been previously reported that the detectable fraction of PAFPs can be substantially smaller than 100% (29)(30)(31). Both the blink characteristics and the detectable fraction are sensitive to the experimental conditions.…”

Challenges in quantitative single molecule localization microscopy

“…In the case of PA-FPs, the fraction is even lower, due to incomplete photo-conversion. In the case of the relatively bright PA-FP mEOS2, this fraction is about 60%, and for several other PA-FPs it can be as low as 40% [25,49]. This can directly affect counting studies, as the number of counted molecules can be underestimated by the same fraction (Fig.…”

“…Other quantification measures might also be affectedfor example, in the case of SM co-localization, assuming that the used co-localization measure is linearly related to the detection efficiency, the co-localization will be underestimated by a fraction xy, where x is the detection efficiency in one channel, and y in the other, leading to an underestimate of as low as 20% for commonly used PA-FP pairs (Fig. 4c) [49]. However, in practice, the effect of limited detection efficiency on cluster (Fig.…”

“…Recent reviews on the updates on the technology and its uses can be found in [4,5]. SMLM can potentially be used for quantitative measurements [6,7], e.g., in counting the number of molecules of a protein specie [8] and stoichiometry estimation of protein complexes [9][10][11], characterizing the spatial distribution of a protein specie [12][13][14][15], estimating the co-localization or co-clustering between organelles and also single molecules (SM) [16][17][18][19], estimating the relative positions of various components in a protein complex with high precision [20,21], and estimating the diffusion coefficients by means of single particle tracking (SPT) in a dense sample [22,23]. Two basic variants of SMLM are Photo-Activated Localization Microscopy (PALM) and STochastic Optical Reconstruction Microscopy (STORM).…”

“…Since the usage of fusion proteins used in PALM provides comparatively high specificity labeling as against immunolabeling (the typical labeling technique used for STORM), and since the phenomenon of photoblinking for PA-FPs is minimal (as against the photo-switchable organic dyes used in STORM, which typically blink 10 times or more before irreversible photobleaching [24]), PALM appears to be better suited for quantitative studies, and for this reason forms the focus of this article even though many of the ideas presented are applicable to SMLM in general. Yet, quantitative analysis with PALM is plagued by several sources of errors [7,70], including that of a limited detection efficiency of label molecules in the range of 40-60% [16,25], a localization uncertainty in the order of 20-50 nm [26,27], overcounting in the range of 100% due to reappearance of label molecules due to photoblinking [15,[28][29][30], errors in labeling, a sample drift in the order of 50-100 nm [1,31] and in the case of multi-color imaging, registration errors [16].…”