Multi-dimensional super-resolution imaging enables surface hydrophobicity mapping

Abstract

Super-resolution microscopy allows biological systems to be studied at the nanoscale, but has been restricted to providing only positional information. Here, we show that it is possible to perform multi-dimensional super-resolution imaging to determine both the position and the environmental properties of single-molecule fluorescent emitters. The method presented here exploits the solvatochromic and fluorogenic properties of nile red to extract both the emission spectrum and the position of each dye molecule simultaneously enabling mapping of the hydrophobicity of biological structures. We validated this by studying synthetic lipid vesicles of known composition. We then applied both to super-resolve the hydrophobicity of amyloid aggregates implicated in neurodegenerative diseases, and the hydrophobic changes in mammalian cell membranes. Our technique is easily implemented by inserting a transmission diffraction grating into the optical path of a localization-based super-resolution microscope, enabling all the information to be extracted simultaneously from a single image plane.

Figure 1. sPAINT instrumental setup.

(a) Scheme of the sPAINT setup; single-molecule fluorescence is collected by a high NA objective lens and focused by the tube lens before passing through a blazed transmission diffraction grating. Here the fluorescence emission is divided into either the spatial region (x,y space, zeroth order diffraction) or the spectral region (λ space, first order diffraction) in the image plane and recorded on a single chip of an EMCCD camera; where, D is the grating-to-detector distance (∼20 mm), d is the grating line spacing (300 groves per mm), γ is blaze angle (8.6°) and f is the focal length of the tube lens (∼200 mm). (b) Contrast adjusted, representative raw sPAINT data of 100 nm TetraSpeck microspheres – labelled with four different fluorophores – demonstrating both the zeroth order diffraction spatial region (left; x,y) and first order diffraction spectral region (right; λ). Scale bars are 5 μm (spatial) and 100 nm (spectral). A magnified inset of TetraSpeck beads with the corresponding emission spectrum is included as an inset (red). Scale bars are 1 μm (spatial) and 20 nm (spectral). (c) Comparison of fluorescence intensity versus wavelength for a single sPAINT bead and bulk fluorimeter data. (d) Empirically determined spatial precision, (σ_xy) (n=1,000, number of beads used for calibration). (e) Empirically determined spectral precision (σ_λ) (mean±s.d., n=400 beads). (Supplementary Fig. 4). (d,e) The photon values refer to the same spatial location. (f) The chemical structure of nile red and corresponding fluorescence emission wavelength dynamic range in typical hydrophobic environments.

Figure 2. Nile Red sPAINT calibrated on synthetic 100 nm unilamellar vesicles (LUVs).

(a) Scheme of the LUV and rows from top-to-bottom; LUVs composed of DOPC lipid, SM lipid or SM/cholesterol lipid. (b) Columns from left-to-right; diffraction-limited image (D.L.), super-resolution image (S.R.), sPAINT hydrophobicity map and (c) frequency histogram of fluorescence emission peak (peak photon values above background were typically ∼750; DOPC, n=42,559 (N=542); SM, n=5,968 (N=154); SM+CL, n=48,589 (N=102); where n is the number of spectrum centres in the histogram and N is the number of LUVs. (d) NR localization rate as a function of lipid type (green line: DOPC, blue line: SM, red line SM+CL). (e) NR localization density as a function of NR concentration. Error bars show the inter-quartile range for at least 90 LUVs in each case. (f) Mean on-time per NR localization, (g) concentrations above 50 nM led to multiple NR binding events within a typical exposure time of the detector (50 ms). Scales bars are 500 nm and 20 nm in zoom.

Figure 3. sPAINT imaging of protein aggregates associated to disease.

(a) Representative sPAINT hydrophobicity image of single αS oligomers. Scale bar is 100 nm. (b) Representative sPAINT hydrophobicity image of individual αS fibrils. By examining aggregates one-at-a-time, it is possible to extract information about the variation in hydrophobicity of a single fibril (b – red box). Scale bar is 1 μm. (c) Representative sPAINT hydrophobicity image of amyloid-β oligomers. Scale bar is 100 nm. (d) Representative sPAINT hydrophobicity image of individual amyloid-β fibrils. Scale bar is 1 μm. (e) Frequency histogram of the individual sPAINT localizations that are used to generate the fibril (b – red box, n=9,996 localizations). (f) Total frequency histogram of the individual sPAINT localizations from both αS oligomers (red, n=17,619 localizations, N=539 oligomers; peak photon values above background ∼780) and αS fibrils (blue, n=120,275 localizations, N=1,528 fibrils; peak photon values above background ∼1,100) – note the shift in bulk hydrophobicity and narrowing of the distributions that are used to generate the fibril (b – red box). (g) Frequency histogram of the individual sPAINT localizations from both aβ_1-42 oligomers (red, n=6,133 localizations, N=80 oligomers; peak photon values above background ∼860) and aβ fibrils (blue n=17288 localizations, N=668 fibrils; peak photon values above background ∼710).

Figure 4. sPAINT imaging along the length of the cell plasma membrane.

(a) Representative SR image of a live neuron-like SH-SY5Y cell (inset: white-light image). Scale bar is 5 μm (n=31,744 localizations peak photon values above background ∼543) and corresponding zoom SR (bar is 0.5 μm, n=2,859 localizations) and sPAINT image (exposure time 50 ms, 3,000 frames). (b) HEK 293 cells untreated or cholesterol manipulated. Scale bar is 100 nm. (c) Frequency histogram of fluorescence emission peak for cholesterol experiments, red line is untreated cells (n=18,227 localizations; peak photon values above background ∼700), blue line is cholesterol-enriched cells (n=27,183 localizations; peak photon values above background ∼990) and green line is cholesterol-depleted cells (n=20,838 localizations; peak photon values above background ∼640). Data in (c) is from ∼10 cells per treatment. (d) Spatial-temporal information of fixed epithelial SH-SY5Y cell. Representative SR image and spatial-temporal changes in nile red emission wavelengths (Scale bars are 1 μm, n=42,099 localizations). Localization metadata generated by sPAINT include localization (x and y coordinates), time (frame number) and spectral (λ) information (Supplementary Video 1). Hydrophobicity maps were generated by applying Nadaraya–Watson kernel regression to the spectral information retrieved at every point localization and observed for a given time interval, with a temporal resolution defined by box-car averaging to ∼0.8 s.