Color mapping of multitarget protein clusters

Abstract

Visualizing subcellular distributions _coof proteins and assessing their colocalization patterns are central to understanding the organization and interactions of molecular assemblies. In this issue of Structure, Kiuchi et al. ¹ introduce Protein Cluster coloring to map complex association patterns among eight endogenous proteins at the periphery of the clathrin-coated structure, revealing their multilayered complex associations upon epidermal growth factor stimulation.

A key goal in cell biology is to understand the mechanisms governing protein distribution and assembly into functional multiprotein complexes. Recent advancements in highly multiplexed single-molecule localization microscopy (SMLM) have enhanced our ability to visualize the spatial distributions of multiple proteins down to single protein resolution ^2,3. While methods to improve spatial resolution are rapidly advancing ^4,5, the quantitative potential of SMLM remains limited. In particular, labeling strategies involving antibody immobilization often lead to insufficient labeling densities due to the steric hindrance from bulky labels. Underlabeling leads to inaccuracies in characterizing the nanoscale spatial organization of densely packed protein assemblies, such as protein condensates and signalosomes.

The introduction of integrating exchangeable single-molecule localization (IRIS) by Kiuchi et al. ⁶ and subsequent innovations ^7,8 represent a promising alternative. IRIS utilizes reversible binding to bypass steric hindrance associated with antibody immobilization and enables high labeling densities by oversampling the target through cycles of binding and unbinding of the probes. For example, the authors’ earlier work demonstrated that using Lifeact probes for IRIS increased the maximum labeling density of actin filaments by a factor of 60 compared to antibody-based labeling ⁶.

With highly multiplexed IRIS and high-density labeling, a major challenge is to quantitatively determine the spatial relationships between multiple overlapping protein targets at the nanoscale. Standard colocalization analyses rely upon methods developed for conventional fluorescence images. In these approaches, single-molecule localizations are converted to pixel-based probability histograms, and colocalization is performed using pixel intensities across the entire image or within a specified region of interest, such as in Pearson correlation analysis. Colocalization analyses directly using coordinate-based SMLM data, such as Clus-Doc, ClusterViSu, Coloc-Tesseler, and coordinate-based colocalization index, have also been developed ⁹. Despite the advancements, these methods have remained limited to evaluating only two targets at a time, whereas protein associations are often more complex and multifaceted.

The new report by Kiuchi et al. ¹ has now addressed this bottleneck. The authors developed antiserum-derived antibody fragment (Fab) probes for multiplex IRIS imaging and introduced a technique termed Protein Cluster coloring (PC-coloring) to reveal multi-layered protein assemblies from eight protein targets in the clathrin-coated structure (CCS) before and after epidermal growth factor (EGF)-induced endocytosis. Previous studies have identified hotspots of clathrin-mediated endocytosis and recruitment of epidermal growth factor receptor (EGFR) upon EGF stimulation in CCSs, suggesting that CCSs are pre-formed for receptor recruitment and multiple endocytic events ¹. Building on this premise, the authors investigated the spatial relationships of CCS-associated proteins in response to EGF stimulation.

First, the authors validated a new set of Fab probes derived from antiserum to target the eight endogenous proteins for IRIS imaging: clathrin light chain (CLC), the beta subunit of adaptor protein 2 complex (AP2β2), EGFR, growth factor receptor-bound protein 2 (Grb2), transferrin receptor (TfR), FCH domain only proteins 1 and 2 (FCHo1/2), epidermal growth factor receptor pathway substrate 15 (Eps15), and intersectin-1 (ITSN1). Next, the authors introduced PC-coloring to quantitatively map complex spatial relationships among these target proteins at the EGFR recruitment sites. Briefly, since IRIS detects the target through repeat and reversible probe binding, pixel intensity in the reconstructed superresolution image directly reflects target abundance. As such, the intensity ratios among protein targets reflect unique association patterns. As a proof of principle, the authors demonstrated PC-coloring using four-target IRIS imaging of EGFR, Grb2, Eps15, and AP2β2 (Figure 2 in Kiuchi et al. ¹). From the composite superresolution image, the relative abundance of each protein was calculated by dividing its intensity by the total pixel intensity across all four targets, generating four intensity ratios per pixel, which were subsequently normalized. Principal Component Analysis (PCA) was then performed on normalized intensity ratios, reducing the high-dimensional data to a space defined by two principal components. Intensity distributions were then grouped based on their similarity in the PCA space using k-means clustering, dendrogram, and 8-connectivity analysis. The process identified eight distinct protein cluster patterns, each represented by a distinct color. The color scheme was based on the primary colors assigned to each protein (EGFR: red, Grb2: green, Eps15: cyan, and AP2β2: blue) and intermediate colors. The colored pixels were then mapped back onto their original locations in the IRIS superresolution image, generating the PC-coloring image. For validation, Pearson’s correlation analysis was performed between pairs of protein targets in the identified protein cluster patterns to assess their degree of colocalization.

Using multiplexed IRIS, the authors confirmed the presence of clusters of all eight protein targets in a single CCS after EGF stimulation. Using PC-coloring on three- or four-target IRIS imaging, the authors revealed that EGFR and Grb2 form complexes at the rim of CCS, with a stepwise increase of the Grb2-to-EGFR ratio from the outer rim to the center. Grb2 and Eps15 also colocalize in specific regions, with Eps15 forming sequential complexes with FCHo1/2 and ITSN1. The authors further identified six distinct layers at the CCS rim, including EGFR-dominant sites, EGFR-Grb2 complexes, Grb2-dominant sites, Grb2-CCS component complexes, and Eps15-FCHo1/2 and Eps15-ITSN1 complexes. These results suggest that Eps15, FCHo1/2, and ITSN1 are recruited in a layered, stepwise manner, contributing to CCS formation and endocytosis. The separation of EGFR and TfR recruitment sites further underscores distinct recruitment mechanisms.

To summarize, PC-coloring introduced by Kiuchi et al. ¹ offers a fresh approach for multitarget colocalization analysis of highly multiplexed SMLM data. Distinct colors represent spatial territories of signature patterns of protein association, delineating boundaries much like regions on a customized map. The PC-coloring framework is broadly applicable to other structures, from focal adhesions to immune synapses. As the superresolution microscopy applications move toward quantitative spatial biology, such faithful spatial relationship mapping will be key to uncovering how molecular associations, their stoichiometry, and spatial organizations collectively regulate cellular functions.