Super-Resolution Imaging of Self-Assembled Nanocarriers Using Quantitative Spectroscopic Analysis for Cluster Extraction

Abstract

Self-assembled nanocarriers have inspired a range of applications for bioimaging, diagnostics, and drug delivery. The noninvasive visualization and characterization of these nanocarriers are important to understand their structure to function relationship. However, the quantitative visualization of nanocarriers in the sample's native environment remains challenging with the use of existing technologies. Single-molecule localization microscopy (SMLM) has the potential to provide both high-resolution visualization and quantitative analysis of nanocarriers in their native environment. However, nonspecific binding of fluorescent probes used in SMLM can introduce artifacts, which imposes challenges in the quantitative analysis of SMLM images. We showed the feasibility of using spectroscopic point accumulation for imaging in nanoscale topography (sPAINT) to visualize self-assembled polymersomes (PS) with molecular specificity. Furthermore, we analyzed the unique spectral signatures of Nile Red (NR) molecules bound to the PS to reject artifacts from nonspecific NR bindings. We further developed quantitative spectroscopic analysis for cluster extraction (qSPACE) to increase the localization density by 4-fold compared to sPAINT; thus, reducing variations in PS size measurements to less than 5%. Finally, using qSPACE, we quantitatively imaged PS at various concentrations in aqueous solutions with ∼20 nm localization precision and 97% reduction in sample misidentification relative to conventional SMLM.

Figure 1.

(a) Chemical structure of NR; (b) Chemical structure and illustration of the BCP for PS assembly; (c) Illustration of the assembled PS (the green color represents the polar end of the BCP and the blue color represents the non-polar end) and the difference in the emission spectra of NR when transiently bound to the PS (yellow) and the PLL-coated glass substrate (red). Free non-fluorescent NR is shown in gray; (d) Schematic of our sPAINT experimental setup. TL: tube lens; S: slit; G: transmission grating; L: lens; EMCCD: electron multiplying charge-coupled device.

Figure 2.

(a) A representative reconstructed super-resolution image of the immobilized the PS sample and NR interactions (Scale bar: 1 μm); (b) Histogram of the λ_max of NR interactions in three ROIs containing PS as highlighted by the yellow squares numbered 1–3; (c) Histogram of the λ_max of non-specific NR interactions in the control sample. The SW used for detecting NR interactions with PS is highlighted in red; (d) Representative single-molecule spectra from NR+PS binding and non-specific NR binding (dashed lines indicate the SW). Clusters extracted from (e) the PS sample and (f) control based on DBSCAN alone; Clusters extracted from (g) the PS sample and (h) control using sPAINT (Scale bar: 500 nm).

Figure 3.

The qSPACE workflow shows the (i) detected spatial localizations with the location of the sample highlighted by the green circles. A subset of localizations containing spectroscopic information is used to create a (ii) validation map that shows clusters with the selected spectra. All detected localizations are used for (iii) spatial clustering without considering spectroscopic information. (iv) Localizations from spatial clusters, which are spatially correlated with the validation map, are recovered for further analysis while artifacts are rejected. (v) The number, size, and morphology of the extracted sample can be further analyzed.

Figure 4.

(a) Histogram of the localization densities (black dashed lines indicate the 3.5×10⁻³ nm⁻² LD threshold) for qSPACE and (b) sPAINT; (c) Comparison of the global FRC curves for sPAINT (red) and qSPACE (blue). The dashed lines are the corresponding FRC resolution for each method at the 1/7 FRC threshold; (d) Size distribution of PS (0.02 mg/mL) measured by NTA and qSPACE.

Figure 5.

Representative super-resolution reconstructions of the (a) CTRL sample with artifacts pseudo-colored in red and misidentified PS pseudo-colored in cyan. The three white arrows highlight three examples of sample misidentification; (b) LC sample (0.02 mg/mL) with artifacts pseudo-colored in red and validated PS pseudo-colored in cyan; and (c) HC sample (0.2 mg/mL) with artifacts pseudo-colored in red and validated PS pseudo-colored in cyan (Scale bar: 1 μm). The size distributions for the (d) misidentified PS in the CTRL samples; (e) validated PS in the LC samples; and (f) validated PS in the HC samples. Comparison between the total of number of PC found and the qSPACE VC in the (g) CTRL samples, (h) LC samples, and (i) HC samples.