Phalloidin-PAINT: Enhanced quantitative nanoscale imaging of F-actin

Abstract

We present phalloidin-based points accumulation for imaging in nanoscale topography (phalloidin-PAINT), enabling quantitative superresolution imaging of filamentous actin (F-actin) in the cell body and delicate membrane protrusions. We demonstrate that the intrinsic phalloidin dissociation enables PAINT superresolution microscopy in an imaging buffer containing low concentrations of dye-conjugated phalloidin. We further show enhanced single-molecule labeling by chemically promoting phalloidin dissociation. Two benefits of phalloidin-PAINT are its ability to consistently quantify F-actin at the nanoscale throughout the entire cell and its enhanced preservation of fragile cellular structures. In a proof-of-concept study, we employed phalloidin-PAINT to superresolve F-actin structures in U2OS and dendritic cells (DCs). We demonstrate more consistent F-actin quantification in the cell body and structurally delicate membrane protrusions of DCs compared with direct stochastic optical reconstruction microscopy (dSTORM). Using DC2.4 mouse DCs as the model system, we show F-actin redistribution from podosomes to actin filaments and altered prevalence of F-actin-associated membrane protrusions on the culture glass surface after lipopolysaccharide exposure. The concept of our work opens new possibilities for quantitative protein-specific PAINT using commercially available reagents.

Significance

Filamentous actin (F-actin) plays a structural role in maintaining the eukaryotic cell shape. Besides its abundance in the cell body, F-actin is identified in membrane protrusions from the cell periphery. Superresolving F-actin in mechanically fragile membrane protrusions requires gentle labeling strategies and unbiased quantification to compare F-actin distribution across the entire cell. In this work, we harnessed the intrinsic and enhanced dissociation of dye-conjugated phalloidin to achieve quantitative PAINT superresolution imaging. Using membrane protrusions from dendritic cells (DCs) as a model system, we demonstrate improved labeling of F-actin using phalloidin-PAINT and quantitatively demonstrate the spatial redistribution of F-actin after lipopolysaccharide stimulation. Phalloidin-PAINT enables across-the-cell unbiased F-actin quantification.

Introduction

The actin cytoskeleton consists of distinct filamentous actin (F-actin) arrangements, including stress fibers, cortical actin, lamellipodia, and filopodia, which regulate cellular dynamics, provide mechanical support, maintain the cell shape, and modulate cell migration. Fluorescent labeling of F-actin using dye-conjugated phalloidin has enabled extensive superresolution studies (1,2,3,4,5,6,7) using single-molecule localization microscopy (SMLM) (8,9,10,11).

F-actin underlies membrane protrusions and plays a key role in regulating their dynamics. Recently, actin-enriched elongated membrane structures have been shown to play increasingly important roles in the progression of various pathologies at the nanoscale, including viral and bacterial infections (12,13,14,15), cancer (16,17,18), and neurodegenerative diseases (19). In addition, these membrane structures carry out physiological functions, such as mitochondrial homeostasis (20), immune surveillance (21,22), endocytosis (23), and exosome release (24). In particular, their involvement in intercellular communications provides new opportunities for drug delivery and therapeutic intervention (16,25,26). Electron microscopy (16,27,28,29) and correlative light-electron microscopy techniques have been used to study membrane protrusions and the presence of F-actin within these protrusions (30). Extending SMLM techniques to study morphology and F-actin distribution in these delicate cellular structures requires tailored sample preservation methods (26). The integrating exchangeable single-molecule localization (IRIS) of F-actin and points accumulation for imaging in nanoscale topography (PAINT) using transient interactions of fluorophore-conjugated Lifeact (31) present alternative strategies to achieve superresolution imaging of F-actin with higher labeling densities (32,33). However, the binding of actin probes to F-actin depends on the precise biochemical interactions dictated by the F-actin nano-architecture (34). Lifeact, for instance, may not effectively label fragile cellular structures such as filopodia in mesenchymal cells (35). Therefore, a gentle labeling strategy with a widely used actin probe such as phalloidin, which binds F-actin at a site distinct from Lifeact, can extend existing PAINT-based SMLM techniques to study F-actin nano-architectures of fragile membrane protrusions.

Here, we present PAINT imaging of F-actin using single-molecule labeling of dye-conjugated phalloidin. We show that dye-conjugated phalloidin displays moderate dissociation from the fixed cell sample. By promoting the dissociation using a chaotropic agent, we achieved superresolution imaging and molecular quantification of F-actin in U2OS and immortalized (DC2.4) and bone marrow-derived primary mouse dendritic cells (BMDCs). We further demonstrate phalloidin-PAINT superresolution images revealing the spatial redistribution of F-actin and diminished F-actin-associated membrane fiber networks of DC2.4 cells after exposure to lipopolysaccharide (LPS). Thus, phalloidin-PAINT provides a simple and versatile technique for quantitative superresolution imaging of F-actin.

Materials and methods

Materials and reagents

DMEM (11960069-500mL), RPMI (11875093), penicillin-streptomycin (15140-122-100mL), L-glutamine (25030-081-100mL), and 1× DPBS (14190-144-500mL) were purchased from Gibco, Grand Island, NY, USA. FBS (F0926-500mL), Triton X-100 (X100-1L), MES (M3671-50G), EGTA (E3889-25G), sodium phosphate (Na₃PO₄) (342483-500G), potassium thiocyanate (KSCN) (207799-100G), magnesium chloride (MgCl₂) (M8266-100G), DMSO (D8418-500ML), glucose oxidase from Aspergillus niger (G7141-50KU), catalase from bovine liver (C40-100MG), 2-mercaptoethanol (M6250-100ML), mouse monoclonal primary antibody against a-tubulin (clone AA13, T8203), and hydrochloric acid (HCl) (258148-500ML) were ordered from Sigma-Aldrich, St. Louis, MO, USA. Alexa Fluor 546 (AF546) conjugated mouse monoclonal primary antibody against TOM20 (clone F-10, sc-17764) was purchased from Santa Cruz, Santa Cruz, CA, USA. Alexa Fluor 647 (AF647) Phalloidin (A22287), Alexa Fluor 568 (AF568) Phalloidin (A12380), eBioscience LPS (00497693), Alexa Fluor 488 (AF488) conjugated mouse monoclonal primary antibody against α-tubulin (clone DM1A, 53-4502-82), AF647 conjugated goat anti-mouse Superclonal recombinant secondary antibody (A28181), Cytiva HyClone HEPES Solution (SH3023701), and Cytiva HyClone Non-Essential Amino Acids 100× Solution (SH3023801) were purchased from Thermo Fisher Scientific, Waltham, MA, USA. Potassium chloride (KCl) (P217-500) was obtained from Fisher Scientific, Waltham, MA, USA. Paraformaldehyde (15710), and glutaraldehyde (16120) were purchased from Electron Microscopy Sciences, Hatfield, PA, USA. Eight-well chambered cover glasses (sterile, no. 1, C8-1-N) were ordered from Cellvis, Mountain View, CA, USA. Unconjugated gold colloids (15711-20) were purchased from Ted Pella, Redding, CA, USA. The following materials were utilized for preparing and seeding BMDCs: RPMI (SH30027.01) was purchased from Cytiva, Marlborough, MA, USA. 2-mercaptoethanol (31350010) was purchased from Gibco, Grand Island, NY, USA. FBS (45000-736) and penicillin-streptomycin 16777-164) were obtained from VWR, Radnor, PA, USA. GM-CSF (576306) was obtained from BioLegend, San Diego, CA, USA. A 100 × 20 mm petri dish (353,003) was purchased from Corning, Corning, NY, USA.

Buffers

Cytoskeleton buffer (10 mM MES [pH 6.1], 90 mM KCl, 3 mM MgCl₂, and 2 mM EGTA). Fixation buffer 1 (3.7% paraformaldehyde, 0.1% glutaraldehyde, 0.5% Triton-X-100 in cytoskeleton buffer). Fixation buffer 2 (3.7% paraformaldehyde, 0.1% glutaraldehyde in DPBS). Fixation buffer 3 (2% paraformaldehyde, 0.05% glutaraldehyde, 0.2 M HEPES in DPBS). Fixation buffer 4 (4% paraformaldehyde, 0.2 M HEPES in DPBS). Postfixation buffer (3.7% paraformaldehyde in DPBS). Permeabilization buffer (0.1% Triton-X-100 in DPBS). Buffer A (10 mM Tris [pH 8.0], 50 mM sodium chloride). Buffer B (10 mM Tris [pH 8.0], 50 mM sodium chloride, 10% glucose). GLOX solution (14 mg glucose oxidase, 17 mg/mL catalase in 200 μL buffer A). STORM buffer (7 μL GLOX solution, 7 μL 2-mercaptoethanol in 690 μL buffer B).

Cell culture

U2OS cells (ATCC HTB-96) were cultured in DMEM supplemented with 10% FBS, 2 mM L-glutamine, and 100 units/mL penicillin-streptomycin. DC2.4 cells (Millipore, SCC142) were cultured in RPMI supplemented with 10% FBS, 100 units/mL penicillin-streptomycin, 1× HEPES, and 1× NEAA. All cell lines were maintained at 37°C in a humidified atmosphere of 5% CO₂ and split at the confluence.

Preparation of BMDCs

The bone marrow was harvested from the femurs of freshly euthanized BALB/c mice. The collection of bone marrow from mice was approved by the Animal Care Committee (ACC) at the University of Illinois Chicago (Protocol number 21-098). Bone marrow cells were then seeded onto a petri dish (100 × 20 mm) at a concentration of 2 × 10⁵ cells/mL and cultured in RPMI-1640 medium supplemented with 10% FBS, 1% penicillin-streptomycin, 50 μM 2-mercaptoethanol, and 20 ng/mL GM-CSF. Cells were maintained at 37°C in a humidified atmosphere of 5% CO₂. Cell medium was refreshed on days 3 and 6. Differentiated BMDCs were harvested on day 8 and collected in RPMI-1640 medium supplemented with 10% FBS and 1% penicillin-streptomycin. Cells were centrifuged at 350 × g for 5 min.

Sample preparation

Cell fixation

1) For imaging cytoskeletal actin in U2OS cells, U2OS cells were seeded (approximately 5000 cells/well) in a chambered coverglass and grown in an incubator at 37°C and 5% CO₂. After 36 h of incubation, cells were fixed and permeabilized in freshly prepared fixation buffer 1 for 20 min at room temperature. Cells were maintained in PBS at 4°C until imaging. 2) For imaging cytoskeletal actin and proximal membrane protrusions in DC2.4 and BMDCs, DC2.4 cells were seeded (approximately 2500 cells/well) in a chambered coverglass and grown in an incubator at 37°C and 5% CO₂ for 24–48 h, and BMDCs (approximately 5000 cells/well) were seeded in a chambered coverglass and incubated for 16–24 h at 37°C and 5% CO₂. Following incubation, three drops of fixation buffer 2 were added into the medium of each well and incubated for 4 min at room temperature. All consequent solution exchange steps were performed with a Gilson PIPETMAN P200 pipette placed at the edge of the sample well. The chamber slides were held at an angle such that the solution slowly rose up, filling the well. The solution dispensing rate was approximately 400 μL/min. Performing solution exchanges at a slow rate is pivotal to minimizing damage to delicate structures. The medium was replaced with fixation buffer 2 and incubated for 10 min at room temperature. The sample was then incubated with the permeabilization buffer for 5 min. Permeabilization buffer was replaced with DPBS, and the sample was imaged immediately thereafter. 3) For imaging distal membrane protrusions with extremely low actin contents, the following fixation procedure was adapted from Abounit et al. (26). In brief, three drops of freshly prepared fixation buffer 3 was added into the medium of each well and incubated for 4 min at 37°C. All consequent solution exchange steps were performed with a Gilson PIPETMAN P200 pipette placed at the edge of the sample well. The chamber slides were held at an angle such that the solution slowly rose, filling the well. The solution dispensing rate was approximately 400 μL/min. Performing solution exchanges at a slow rate is pivotal to minimizing damage to delicate structures. The medium containing drops of fixation buffer 3 was removed and replaced with fixation buffer 3. The chamber slide was incubated for 20 min at 37°C and 5% CO₂. Fixation buffer 3 was replaced by fixation buffer 4 and incubated for 20 min at room temperature. Fixation buffer 4 was replaced with DPBS for 30 s, and the sample was imaged immediately after. The sample chambers were incubated with a 1:5 dilution of gold nanoparticles to be used as fiducials.

LPS treatment

Overnight grown DC2.4 cells were treated with LPS-supplemented culture medium (5 μg/mL) for 21 h in an incubator at 37°C and 5% CO₂. The cells were fixed and permeabilized as previously described and imaged immediately after.

Phalloidin dye conjugate handling

Phalloidin dye conjugates (phalloidin-AF647 and phalloidin-AF568) were constituted in 150 μL of DMSO to a concentration of 66 μM according to the manufacturer guidelines. All consequent dilutions were made in DMSO, aliquoted, and stored at −20°C to avoid repeated freeze-thaw cycles. The longevity of the diluted phalloidin-dye conjugate aliquots, especially for single-molecule imaging, can be prolonged by storing them in a sealed container with a desiccant. The final DMSO percentage (v/v) in the imaging buffer was 1–3%.

Fluorescence intensity decay analysis

Fixed and permeabilized cells were immunostained with phalloidin-AF647 (0.165 μM) overnight at 4°C in a humid chamber. Cells were washed once with DPBS. Imaging was performed in 1× DPBSand 1× DPBS supplemented with KSCN at intended concentrations (150 or 300 mM).

Phalloidin-PAINT

For phalloidin-AF647 PAINT imaging, fixed and permeabilized U2OS and DC2.4 were imaged in an imaging buffer containing phalloidin-AF647 supplemented with or without KSCN in 1× DPBS. The chaotropic salt KSCN is a denaturant at higher concentrations. The concentration of the KSCN (300 mM) was chosen to be sufficiently high to promote dissociation but within the nondenaturing concentration range (200–400 mM) (36). The phalloidin-AF647 concentration was adjusted between 1 and 3 nM to maintain a sufficient single-molecule event density. For instance, given the low density of F-actin, phalloidin-PAINT imaging of membrane protrusions used 3 nM phalloidin-AF647 to achieve sufficient single-molecule localizations. The phalloidin-AF568 was used at a concentration of 63 pM.

dSTORM imaging of DC2.4

Fixed and permeabilized DC2.4 cells were incubated with phalloidin-AF647 (0.165 μM) in 1× DPBS overnight at 4°C. The next day, the phalloidin solution was replaced with STORM buffer and imaged immediately after.

Phalloidin-PAINT after dSTORM imaging

Fixed and permeabilized U2OS cells were incubated with phalloidin-AF647 (0.33 μM) in 1× DPBS for 2 h at room temperature. The cells were washed three times with DPBS, and dSTORM imaging was performed immediately after in the STORM buffer. After the dSTORM acquisition, the sample was washed on stage three times, and phalloidin-PAINT was performed.

Evaluation of phalloidin staining loss

Fixed and permeabilized DC2.4 cells were stained with phalloidin-AF647 (0.33 μM) for 2 h at room temperature. The cells were washed three times with DPBS. Fluorescent imaging was performed in 1× DPBS buffer or STORM buffer.

Cross-linking of the phalloidin staining

Fixed and permeabilized DC2.4 cells were stained with phalloidin-AF647 (0.33 μM) for 2 h at room temperature. The cells were washed three times with DPBS. The cells were then fixed in the postfixation buffer for 15 min at room temperature. The cells were washed three times with DPBS, and fluorescence imaging was performed in 1× DPBS.

SiR actin labeling

SiR actin constituted in DMSO (stock concentration: 1 mM) was diluted in the DC growth medium to a final concentration of 1 μM. The culture medium of DCs grown on a chambered coverglass was replaced with the staining solution and incubated overnight (12 h) in an incubator at 37°C and 5% CO₂.

Immunofluorescence staining

Fixed and permeabilized cells were blocked with 5% BSA in DPBS for 30 min. For direct staining, blocking buffer was replaced with primary antibody solution (5 μg/mL, each, in the blocking buffer) and incubated at room temperature for 1 h. The samples were washed with DPBS three times and maintained in DPBS until imaging. For indirect staining, blocking buffer was replaced with the primary antibody solution (5 μg/mL, in the blocking buffer) and incubated overnight at 4°C overnight. The sample was washed three times in DPBS, and incubated with the secondary antibody (1 μg/mL) for 1 h at room temperature. The samples were washed with DPBS three times and maintained in DPBS until imaging. Cross-linking was performed in the postfixation buffer for 10 min at room temperature.

Microscopy

Fluorescence/superresolution imaging was performed at a highly inclined and laminated optical sheet (HILO)/near TIRF setting on an inverted microscope (Nikon Instruments, Eclipse Ti2E). A 100×/1.49 oil-immersion objective (CFI Apochromat TIRF 100XC) was used with a 1.5× external magnifier. Single-molecule movies were recorded using a Prime 95B sCMOS camera at 16-bit with 2 × 2-pixel binning, creating an effective pixel size of 147 nm.

Fluorescence imaging

Fluorescence imaging of DC2.4 cells for the phalloidin-AF647 labeling loss experiments, after phalloidin-AF647 cross-linking, and SiR actin labeling was performed using a 641 nm laser at an integration time of 50 ms and a laser power density of 2 W/cm². Three consecutive images were captured, followed by captures at 40 and 80 min, as intended. Fluorescence imaging of the DC2.4 cells for all other experiments was performed at an integration time of 50 ms and laser power density of 64 W/cm². Immunofluorescence imaging was performed using 488, 568, and 641 nm lasers at laser power densities 62, 50, and 34 W/cm², respectively. An integration time of 50 ms was used for all images.

Fluorescence intensity decay analysis

Movies for fluorescence intensity decay analysis were acquired with a 647 nm laser at a power density of 0.2 W/cm² and an integration time of 50 ms, with a 1 min nonilluminating interval between consecutive frames. For each analysis, 81 frames were collected.

Phalloidin-PAINT

For experiments using phalloidin-AF647, a 641 nm laser was used at a power density of 251 W/cm² and a 2 s nonilluminating interval between consecutive frames. Due to the relatively slow kinetics of phalloidin, we used an integration time of 400 ms in the single-molecule labeling experiments. Single-molecule labeling of U2OS cells included an additional periodic photobleaching step at every 20^th frame at an integration time of 200 ms and a laser power density of 1.2 kW/cm² to reduce out-of-focus fluorescence background. Single-molecule movies of 8000–30,000 frames were recorded. Experiments with phalloidin-AF568 were performed similarly using a 568 nm laser at a power density of 300 W/cm². Single-molecule movies of 10,000 frames were recorded.

dSTORM imaging

dSTORM imaging was performed under continuous illumination of 647 nm (1.046 kW/cm²). During the movie, a 405-nm laser was gradually increased from 8 to 32 W/cm² to illuminate the sample and maintain a convenient density of activated molecules. dSTORM movies of 20,000–80,000 frames were collected for DC2.4 and 100,000–120,000 for U2OS cells.

Data processing

Superresolution image reconstruction was performed using the open-source ImageJ plug-in ThunderSTORM (37) in ImageJ (version 1.54f). The camera pixel size was set to 147 nm, photoelectrons per A/D count 0.76 (Prime 95B sCMOS camera at 16 bit, serial number: A21C203010), and base level 100. Images were reconstructed using the “Wavelet filter (B-Spline)” for particle detection and filtering, “integrated Gaussian” for the PSF model, and “maximum likelihood estimation” fitting for subpixel localization of molecules. Localizations were binned into 20 × 20 nm² pixels and visualized using the normalized Gaussian method (for DC2.4 cells) or the average-shifted histogram method (for U2OS cells) in the ThunderSTORM plugin. The normalized Gaussian method, which fits a normalized symmetric 2D Gaussian function integrated over every localized molecule, was used with the standard deviation specified as 20 nm (localization uncertainty). The average-shifted histogram rendering algorithm uses a density estimation approach where the width of the histogram bin is specified as the pixel size of the superresolution image (20 nm) multiplied by the number of shifts (4 or 2) (37). Drift correction was performed using fiducial beads that were present during the entire acquisition. Events with a sigma value greater than 170 were removed in the postprocessing to reduce the diffusive appearance.

The fast Fourier transform (FFT) analysis was performed using the SciPy function, fft.fftn, in Python (version 3.12.1). Cross-sectional intensity profiles along the membrane protrusions for each technique (n = 9) were obtained in ImageJ. Their corresponding FFT values were calculated, keeping only the positive frequencies. The SciPy function, find_peaks, was then used to identify peaks to display the frequency components of the dSTORM and phalloidin-PAINT methods in frequency space. Equal threshold values were used for peak finding in the FFT.

The actin densities in DCs were analyzed using ImageJ (version 1.54f). The distinct regions of interest (ROIs) were selected based on the intensity of the image using the “threshold” option under the “adjust” menu. Subsequently, distinct actin architectures were further isolated based on the size and the circularity using the “analyze particles” tool in the “analyze” menu (Table S1). The “area” and “RawIntDen” of the ROIs were measured using the “measure” command in the “analyze” menu. “RawIntDen” is the sum of the pixel values in a selected ROI. The actin density per unit area was obtained by dividing the “RawIntDen” value by the area of an ROI.

Fluorescence intensity decay analysis

The dissociation rate assay using fluorescence measurements is modeled as a function of fluorophore photobleaching and dissociation of the ligand (phalloidin). The two processes are independent and described as (38),

$$I_{\left(t\right)}=F_{\left(0\right)}+A{e}^{-\left(k_b+k_{off}\right)t}$$ (Equation 1)

where I(t) represents the function of total intensity. A and F₀ are constant terms. F₀, k_b, and k_off account for the local fluorescence background, photobleaching rate, and dissociation rate, respectively (38). In a typical fluorescence intensity decay experiment, the ligand can be cross-linked to the substrate via a postfixation step isolating the photobleaching process. This step enables to determine the photobleaching rate first, which can be subsequently used to extract the dissociation rate (38). However, phalloidin is resistant to postfixation, and thus the dissociation rate cannot be independently determined. Therefore, the combined dissociation plus photobleaching rate $k^{\prime }$, where $k^{\prime }=k_b+k_{off}$, for phalloidin-AF647 in the presence and absence of KSCN was evaluated by fitting into Eq. 2.

$$I_{\left(t\right)}=F_{\left(0\right)}+A{e}^{-\left(k^{\prime }\right)t}$$ (Equation 2)

Intensity profiles for the 81-frame image acquisition were obtained. The combined dissociation plus photobleaching rates $k^{\prime }$ were determined by fitting the fluorescence intensity decay curves into an exponential decay model, Eq. 2, using the “cftool” of MATLAB (R2020b). A goodness of fit of at least 0.9 was maintained for all fitted data. Provided the same experimental conditions, including laser illumination density and integration time, we assume the photobleaching rate, $k_b$, to be the same across all experiments. To compare the dissociation rates, the relative enhancement in the combined dissociation plus photobleaching rate in the presence of KSCN was calculated using Eq. 3.

Relative enhancement in the combined dissociation plus photobleaching rate:

$$\frac{k_{{KSCN}^{+}}^{\prime }}{k_{{KSCN}^{-}}^{\prime }}=\frac{[k_b+k_{off|{KSCN}^{+}}]}{[k_b+k_{off|{KSCN}^{-}}]}$$ (Equation 3)

Results and discussion

Single-molecule labeling using dye-conjugated phalloidin achieved superresolution imaging of F-actin

We selected the widely used AF647 conjugated phalloidin (phalloidin-AF647) as the dye-conjugated phalloidin for our study. As phalloidin-dye conjugates are reported to show intrinsic faster dissociation from F-actin (33,38,39), we first characterized the loss of phalloidin-AF647 staining in DPBS. We stained DC2.4 cells with phalloidin-AF647 and monitored the fluorescent intensity over time. Consistent with previous reports, the fluorescence signal from phalloidin staining of DC2.4 dropped by approximately 40–50% within 40 min in DPBS (Figs. 1, a–d) (33,38). The consecutive images of the same cell exhibited no significant change in the fluorescence intensity, indicating a minimal impact from photobleaching (Fig. 1 a). Although F-actin-rich podosome structures remained distinct after 40 min, the membrane protrusions were barely visible (Figs. 1, a–c). We further characterized the fluorescence signal loss in the STORM buffer, which is used in dSTORM superresolution imaging of phalloidin-labeled F-actin. The fluorescence signal was reduced even more in STORM buffer by approximately 70–85% within 40 min (Figs. 1, e–h). STORM buffer contains 2-mercaptoethanol to facilitate the photoswitching of AF647. Thus, the additional fluorescence signal loss in the STORM buffer may be due to 2-mercaptoethanol attacking the pi system of AF647 (40). We found this to be valid as the fluorescence signal after 40 min could be partially recovered by stimulating the sample with a 405 nm laser.

Figure 1.

Intrinsic dissociation and single-molecule labeling with phalloidin-AF647 enabled superresolution imaging of F-actin. (a) Three consecutively acquired fluorescence images of the same DC2.4 cell stained with phalloidin-AF647 incubating in 1× DPBS buffer. (b) A fluorescence image of the DC2.4 cells shown in (a), acquired after 40 min. Image contrast is adjusted to that of (a). (c) Zoomed-in views of the boxed regions in (a), across (a) and (b). (d) Corresponding cross-sectional intensity profile along the lines indicated (1,2,3,4) in (c, ii). (e) Three consecutively acquired fluorescence images of the same DC2.4 cell stained with phalloidin-AF647 in the STORM buffer. Images were acquired immediately after adding the buffer. (f) A fluorescence image of the DC2.4 cells shown in (e), acquired after 40 min. Image contrast is adjusted to that of (e). (g) Zoomed-in views of the boxed regions in (e), across (e) and (f). (h) Corresponding cross-sectional intensity profile along the lines indicated (1,2,3,4) in (g, ii). (i) A representative single-molecule labeling superresolution image reconstructed from a 30,000-frame acquisition with phalloidin-AF647 on a U2OS cell and a schematic illustration of single-molecule labeling of F-actin with phalloidin-AF647 (inset). (j) Zoomed-in view of the boxed region in (i). (k) Gaussian-fitted cross-sectional profile across the F-actin fibers indicated in (j). All images are visualized using the “Fire” LUT. Scale bars, 10 μm (a, b, e, f, and i) and 5 μm (c, g, and j).

Phalloidin is a rigid bicyclic heptapeptide small-molecule actin probe with specificity and affinity to F-actin in cells (41). Given the comparable size of phalloidin (790 Da) to the conjugated small-molecule dyes (AF647 1160 Da), the conjugated dye influences its interaction with F-actin. Early reports on phalloidin conjugates report a “relative affinity” to phalloidin conjugates compared with unconjugated phalloidin (42,43,44). Cationic fluorophores such as rhodamine may introduce extra electrostatic attraction (relative affinity of 2), while anionic fluorophores such as fluorescein may repel from F-actin (relative affinity of 0.1) (42). Similarly, electrostatic repulsions from the anionic AF647 (45) might explain the observed intrinsic dissociation of phalloidin-AF647.

Next, we utilized U2OS cells as our model system to demonstrate phalloidin-based single-molecule labeling of F-actin. Our approach involves reducing the concentration of phalloidin to decrease its binding rate, enabling us to observe and record individual phalloidin binding events in a single-molecule imaging setup (46). In a proof-of-concept experiment, we performed single-molecule labeling on fixed U2OS cells with phalloidin-AF647 under HILO illumination. We introduced a nonilluminating interval of 2 s between consecutive image frames to ensure adequate detection of single-molecule events without significant spatial overlap (46). Fig. 1 i shows a superresolution image reconstructed from a 30,000-frame movie, and Fig. 1 j shows a zoomed-in region. Fig. 1 k demonstrates the superresolution achieved by phalloidin-based single-molecule labeling, resolving the adjacent F-actin fibers by approximately 88 nm at the point indicated in Fig. 1 j. While single-molecule labeling of phalloidin-AF647 successfully resolved thick F-actin stress fibers (Fig. 1 i), we observed inadequate representation of the thin actin filaments (Fig. 1 j). This limitation is likely due to the slower sampling rates resulting from the relatively slow interaction kinetics between phalloidin and F-actin. To improve the sampling rate, we destabilized the phalloidin-AF647-F-actin interaction.

Promoting phalloidin dissociation through chaotropic perturbation enhances F-actin labeling density and quantification consistency of phalloidin-PAINT

Phalloidin interacts with F-actin through hydrophobic interactions (34,47,48). We have previously demonstrated the capability of chaotropic perturbation of hydrophobic interactions to promote dissociation rates between antibody fragments and their polypeptide ligands to achieve superresolution molecular census (49). Here, we employed the chaotropic salt KSCN to further enhance the dissociation of phalloidin from F-actin (Fig. 2 a, top). We evaluated the dissociation of phalloidin-AF647 in situ in the presence of KSCN using a fluorescent intensity decay model (materials and methods, Fig. S1) (49)). Fig. 2 a (bottom) shows the relative enhancement in combined dissociation and photobleaching rates of phalloidin-AF647 in the presence of 150 and 300 mM KSCN. Ensemble characterizations revealed an approximately twofold increase in the dissociation of phalloidin-AF647 at 300 mM KSCN concentration. Next, we incubated fixed U2OS cells with phalloidin-AF647 in an imaging buffer supplemented with 300 mM KSCN (materials and methods). Fig. 2 b shows a corresponding superresolution image reconstructed from a 30,000-frame movie. The chaotropic perturbation enhanced the sampling rate, enabling us to resolve stress fibers (Figs. 2, b and c) and thin F-actin filaments (Fig. 2 c, right). Fig. 2 d demonstrates the superresolution achieved by chaotropic perturbation-enhanced single-molecule labeling with phalloidin-AF647, resolving the adjacent F-actin fibers by approximately 75 nm at the point indicated in Fig. 2 c. Given the similarity between the imaging modalities, we refer to chaotrope-assisted phalloidin-based single-molecule labeling as phalloidin-PAINT.

Figure 2.

Chaotropic perturbation enhances F-actin labeling consistency and quantification of phalloidin-PAINT. (a) Schematic illustration of chaotrope (KSCN) enhanced single-molecule labeling of F-actin (top). Relative enhancement of phalloidin-AF647 combined dissociation and photobleaching rate in the presence of KSCN (bottom). Error bars were determined by the propagation of error. (b) A representative superresolution image reconstructed from a 30,000-frame acquisition with phalloidin-AF647 in the presence of 300 mM KSCN on a U2OS cell. (c) Zoomed-in view of the boxed region in (b). A zoomed-in view of the indicated boxed region shows the thin actin fibers (right). (d) Gaussian-fitted cross-sectional profile across the F-actin fibers indicated in (c). (e) dSTORM image of a phalloidin-AF647-stained U2OS cell. (f) The phalloidin-PAINT image of the U2OS cell shown in (e) performed after the dSTORM acquisition. (g) Merged view of the boxed region shown in (e) and (f) (top). The zoomed-in view of the yellow boxed region indicated (bottom). Yellow arrowheads point to the thin actin fibers. (h) Linear regression fit of the cumulative number of events per 1000 frames in the presence (squares, blue solid line fit) and absence (circles, orange dash line fit) of 300 mM KSCN. (i) Quantitative comparison of the relative F-actin densities in stress fibers (SF) and thin fibers (TF). Error bars represent standard deviation. (j) Phalloidin-PAINT imaging on a U2OS cell immunostained for microtubules and mitochondria. Shown is a zoomed-in view of the phalloidin-PAINT superresolution image (cyan) merged with the immunofluorescence images of microtubules (DM1A-AF488, magenta) and mitochondria (F10-AF546, green) captured after the phalloidin-PAINT image acquisition. The corresponding full field of view merged image is shown in Fig. S4. (k) Immunostaining persisted through phalloidin-PAINT image acquisition. Single-channel images of F-actin (phalloidin-PAINT, cyan), microtubules (magenta), and mitochondria (green) shown in (j). (l) Fluorescence intensity decreased after 17 h in the phalloidin-PAINT imaging buffer. Cross-sectional intensity profiles across the indicated lines in (k) at time points 0 and 17 h. Blue arrowheads point to representative peaks with fluorescence intensity reduction marked. Solid line, 0 h; dashed line, 17 h. (m) Cross-linking of antibody staining prevented KSCN-induced fluorescence intensity loss of DM1A-488. Immunofluorescence images of the microtubules labeled with DM1A-AF488 and cross-linked to the substrate via a postfixation step before (0 h) and after (17 h) incubating with 300 mM KSCN. (n) Cross-sectional intensity profiles across the indicated lines in (m). Solid line, 0 h; Dashed line, 17 h. (o) Cross-linking of antibody staining prevented KSCN-induced fluorescence intensity loss of F10-AF546. Immunofluorescence images of the mitochondria labeled with F10-AF546 and cross-linked to the substrate via a postfixation step before (0 h) and after (17 h) incubating with 300 mM KSCN. (p) Cross-sectional intensity profiles across the indicated lines in (o). Solid line, 0 h; Dashed line, 17 h. (b, c, e, and f) Images are visualized using the “Fire” LUT. Scale bars, 10 μm (b), 5 μm (c, left, e, f, j, k, m, n), 2 μm (c, right), and 1 μm (g).

To colocalize the superresolution images and validate that the phalloidin-PAINT localizations align with those obtained using the standard F-actin SMLM technique, dSTORM, we performed PAINT imaging with phalloidin-AF647 on the same cell after performing dSTORM imaging (Figs. 2, e–g). The overlaid image demonstrates the successful colocalization of the phalloidin-PAINT image with the dSTORM image, represented as white regions and along the thin fibers (Fig. 2 g, arrowheads). Thus, single-molecule labeling of F-actin recapitulated the F-actin structures resolved by dSTORM imaging, showcasing the precise capture of F-actin. We evaluated the sampling rates quantitatively by fitting the cumulative number of single-molecule events per 1000 frames to a linear regression model. The slope of the plot depicted in Fig. 2 h corresponds to the average number of single-molecule events detected per frame. Notably, the presence of KSCN in the single-molecule labeling image acquisitions resulted in an approximately threefold increase in the sampling rate. Moreover, the progressive and linear accumulation of localizations suggests that F-actin was continuously probed throughout the image acquisition, maintaining a constant sampling rate over 16 h. Fig. S2 a shows the individual phalloidin-F-actin interactions under HILO illumination, manifested as distinct single-molecule events with an average signal-to-noise ratio of 4. Fig. S2 b shows a single-molecule intensity track exhibiting single-molecule binding events detected over 2000 frames acquired in the presence of KSCN. The distinct intensity profiles in Fig. S2 b represent phalloidin-AF647 probing F-actin in the cell continuously. Since phalloidin-PAINT directly probes F-actin, the intensity of the 2D probability histogram represents the local F-actin population densities. We evaluated the F-actin densities across thick stress fibers and thin fibers of the cell shown in Fig. 2 f by taking mean intensity profiles (Fig. S3). Fig. 2 i shows that the F-actin density of stress fibers of the U2OS cell analyzed is approximately 10 times higher than that of thin fibers.

We next evaluated the impact of KSCN on antibody staining in the context of multitarget imaging. We performed phalloidin-PAINT on U2OS cells that were co-immunostained for microtubules and mitochondria using AF488-conjugated mouse monoclonal primary antibody against a-tubulin (DM1A-AF488) and AF546-conjugated mouse monoclonal primary antibody against TOM20 (F10-AF546). We recorded a dual-color immunofluorescence image in the phalloidin-PAINT imaging buffer, supplemented with 300 mM KSCN, both before starting and upon concluding the phalloidin-PAINT imaging (Figs. S4 and 2 j). Both DM1A-AF488 and F10-AF546 staining were maintained through the phalloidin-PAINT acquisition (Fig. 2 k). Nevertheless, the fluorescence intensity of DM1A-AF488 decreased by approximately 61% (Fig. 2 l, top, blue arrowhead), and that of F10-AF546 decreased by approximately 35% (Fig. 2 l, bottom, blue arrowhead) following phalloidin-PAINT imaging. Considering minimal impact of photobleaching and inherent antibody dissociation (Fig. S5), the decline in fluorescence intensity observed in the immunofluorescence images with KSCN can be attributed to KSCN-induced loss of antibody staining. We further evaluated the impact of KSCN on indirect immunostaining of microtubules. U2OS cells immunostained for microtubules using primary (mouse monoclonal anti-α-tubulin antibody, AA13) and secondary (AF647-conjugated goat anti-mouse) antibodies were incubated with 300 mM KSCN in DPBS. An immunofluorescence image was captured after the addition of KSCN buffer (0 h) and after 17 h of incubation (Fig. S6 a). The immunofluorescence staining remained consistent throughout the 17 h duration in the presence of KSCN. Furthermore, in comparison with direct immunostaining systems, the decrease in fluorescence intensity was minimal (Fig. S6 b). While the antibody staining persisted for 17 h in KSCN for all evaluated systems, the fluorescence intensity reduction suggests that KSCN may have affected the antibody staining. To prevent possible antibody staining loss from KSCN, we chemically cross-linked the immunostaining to the substrate via a postfixation step. The samples were then incubated with 300 mM KSCN for 17 h, and immunofluorescence images were taken after the addition of KSCN buffer (0 h) and after 17 h. The fluorescence intensities were compared as before (Figs. 2, m–p). There was no apparent loss of fluorescence intensity after 17 h, indicating that KSCN-induced antibody staining loss could potentially be mitigated by incorporating a postfixation step. Alternatively, conducting phalloidin-PAINT as the final step in the multiplex imaging procedure can help circumvent any potential interference from KSCN.

Phalloidin-PAINT reveals F-actin-associated membrane protrusions in immortalized and primary dendritic cells

Next, we investigated the nanoscale F-actin arrangements in membrane protrusions of DC2.4 mouse dendritic cells (DCs). Phalloidin-PAINT of DC2.4 cells revealed actin polarization at the leading edge of the DC2.4 cell, manifested as F-actin-rich podosomes. In addition, phalloidin-PAINT illustrated DC2.4 membrane protrusions that were relatively rich in actin, similar to podosomes (Figs. 3, a and b). In contrast, the local actin densities in the rest of the cell body appeared much lower. In addition to resolving F-actin-associated membrane protrusions near the trailing edge of the cell, phalloidin-PAINT also revealed retraction fibers with low actin contents (Fig. S7). We confirmed the low actin content observed with phalloidin-PAINT using SiR-actin labeling on live DC2.4 cells. Fig. S8 presents the fluorescence image of SiR-actin-labeled live DC2.4 cells and representative line profiles through a DC2.4 cell and retraction fibers for comparative analysis. Notably, the intensity of the retraction fibers was at least approximately 500 times lower than the average intensity of the cell body (Fig. S8 b). We further determined the thickness of the F-actin in retraction fibers from two independent experiments to be 52 ± 6 nm (n = 82) (Figs. S7, c and d). The measured actin fiber thickness was smaller than our previously measured fiber thickness of 150 nm (21), suggesting additional luminal space to accommodate actin-associated motor proteins for cargo transfer (50).

Figure 3.

Phalloidin-PAINT superresolution imaging of F-actin in podosomes and membrane protrusions of dendritic cells. (a) A phalloidin-PAINT image of a DC2.4 cell. (b) Zoomed-in view of the boxed region in (a). (c) Phalloidin-PAINT superresolution image of membrane protrusions at the DC2.4 cell periphery (before washing steps). (d) Structures lost during the washing step. The image was obtained by subtracting the phalloidin-PAINT image after washing from before washing in (c). (e) Representative podosome structures from a dSTORM experiment (left) and a phalloidin-PAINT experiment (right). (f) Representative membrane protrusions reconstructed from a dSTORM experiment (top). Representative membrane protrusions from a phalloidin-PAINT experiment (bottom). (g) Evaluation of the labeling consistency along membrane protrusions reconstructed from dSTORM (i, top) and phalloidin-PAINT (ii, bottom). Cross-sectional intensity profile along the membrane protrusion in (f, i) (top left) and the corresponding fast Fourier transform (FFT) plot (top right). The red circles highlight detected frequency components. Cross-sectional intensity profile along the indicated membrane protrusion in (f, ii) (bottom left) and the corresponding FFT plot (bottom right). The red circles highlight detected frequency components. (h) A phalloidin-PAINT image of a primary bone marrow-derived dendritic cell (BMDC) from mouse. (i) Contrast adjusted zoomed-in view of the numbered boxed regions indicated in (h). (j) Variation of the membrane protrusion thickness moving outward from the cell body. The contrast-adjusted zoomed-in region is indicated in (j) (top). The membrane protrusion thickness was measured at FWHM along the indicated points in the image (bottom). (c and d) Images are visualized in “magenta” LUT and all other images are visualized using the “Fire” LUT. Scale bars, 5 μm (a), 2 μm (b), 3 μm (c and d), 1 μm (e, f, and j), 10 μm (h), and 0.5 μm (i).

Membrane protrusions are fragile and can be readily damaged during solution exchange steps. Standard superresolution experiments include a phalloidin staining step, and the stained samples subsequently require multiple washing steps to remove excess fluorescent phalloidin before adding the corresponding imaging buffer. To evaluate the impact of washing steps on the membrane fiber structures of DC2.4, we first performed phalloidin-PAINT on DC2.4 cells. To imitate the conventional washing steps, we washed the sample three times immediately following phalloidin-PAINT acquisition. We subsequently re-imaged the sample with phalloidin-PAINT. Fig. 3 c shows the superresolution image of the DC2.4 cell before the washing steps. We subtracted the superresolution image after the washing steps from the superresolution image before the washing steps such that the resultant image would show structures present before the washing steps but absent after the washing steps (Fig. S9). Fig. 3 d shows the resulting image showing the membrane fiber structures, suggesting that these structures were vulnerable to shear forces during the washing steps. Therefore, phalloidin-PAINT presents a gentle labeling strategy for superresolution imaging of fragile structures. While we demonstrate the influence of washing steps on membrane fiber structures of DC2.4, it could vary across cell types depending on the precise F-actin nanoarchitecture and mechanical strength of the actin-associated membrane protrusions.

Phalloidin-PAINT demonstrated consistent labeling of DC2.4 membrane protrusions. We evaluated the labeling consistency and density between dSTORM and phalloidin-PAINT on podosomes and membrane-protrusions. Podosomes, the F-actin-rich mechanosensory apparatus of DC, have been extensively characterized using superresolution methods, including dSTORM (2,51). Phalloidin-PAINT resolved podosome structures on par with dSTORM (Fig. 3 e). In contrast, phalloidin-PAINT displayed a higher labeling consistency than dSTORM on membrane protrusions (Figs. 3, f–g and S10). We evaluated the labeling consistency of membrane protrusions resolved by dSTORM and phalloidin-PAINT in the frequency space by obtaining the FFT of the cross-sectional intensity profiles (Figs. 3 g and S11). The FFT analysis shows that the spatial frequency components extend further into high frequencies for dSTORM reconstructions. In contrast, phalloidin single-molecule labeling exhibits fewer spatial frequency components (Figs. 3 g and S11, a versus b). The segregated appearance of dSTORM reconstructions could be attributed to either a loss of labeling during data collection from photobleaching and dissociation (33,38) or blinking artifacts (52). Phalloidin-PAINT is immune to these artifacts.

Membrane protrusions were also observed in primary DCs. Fig. 3 h shows phalloidin-PAINT images of a primary BMDC isolated from BALB/c mice (materials and methods). We further observed membrane protrusions extending away from the cell body with a relatively low actin density compared with the intracellular actin (Figs. 3, i and j). Figs. 3, i and j display contrast-adjusted, zoomed-in views of these membrane protrusions. In addition to a reduction in actin density, the membrane protrusion thickness decreased, moving away from the cell body. We characterized the thickness of the F-actin by generating cross-sectional profiles across the membrane protrusions (numbered in Fig. 3 j, top). The fullwidth at half-maximum measurements showed that the individual membrane protrusions originated with a thickness of approximately 100 nm gradually decreased in thickness, and became consistent at approximately 55 nm (Fig. 3 j, bottom). Our observations of DC2.4 cells and BMDCs suggest that actin-associated membrane protrusions are characteristic of immature DCs.

Phalloidin-PAINT reveals F-actin rearrangements in DC2.4 DCs after LPS exposure

LPS activates pattern recognition receptors (TLR4) to trigger DC maturation (53), and treatment of DC with LPS induces cytoskeletal rearrangements that facilitate DC migration to the lymph nodes (Fig. 4 a) (54,55). To visualize the nanoscale F-actin rearrangement in DC2.4, we stimulated DC2.4 cells with LPS for 21 h (55,56) and performed phalloidin-PAINT. Figs. 4, b, c, and S12 show representative phalloidin-PAINT images of immature DC2.4 (iDC2.4) (without LPS treatment) and LPS-treated DC2.4 (LPS-DC2.4). Consistent with the literature (54,55,57,58), LPS-DC2.4 cells showed loss of podosome structures, no apparent actin polarization, and distinct dendrite formation compared with the iDC2.4 cells (Figs. 4, c versus b). Phalloidin-PAINT also captured membrane protrusions in both conditions (Figs. 4, d and e). Furthermore, in comparison with the fine mesh-like actin cytoskeleton of the iDC2.4, the actin cytoskeleton of the LPS-DC2.4 cells appeared to be much more defined, shaping the extending dendrite morphology (Figs. 4 g versus f, and S13). The observed rearrangement of the actin cytoskeleton, along with the generation of more dendrites, the decrease in the number and the rearrangement of membrane protrusions in mature DCs likely align with their functional shift from antigen survey to antigen presentation (57).

Figure 4.

Phalloidin-PAINT reveals nanoscale F-actin rearrangements upon LPS treatment of DC2.4 cells. (a) Schematic illustration of the altered cellular morphology in response to LPS activation. (b) Phalloidin-PAINT of an immature DC2.4 cell (iDC2.4) without LPS treatment. (c) Phalloidin-PAINT image of an LPS-treated DC2.4 cell (LPS-DC2.4). (d) Zoomed-in view of the indicated region in (b) showing membrane protrusions in iDC2.4. (e) Zoomed-in view of the indicated region in (c) showing membrane protrusions in LPS-DC2.4. (f) Zoomed-in view of the indicated region in (b) showing cytoskeletal actin arrangement of iDC2.4. (g) Zoomed-in view of the indicated region in (c) showing cytoskeletal actin arrangement of LPS-DC2.4. (h) Distribution of actin density across the podosomes and membrane protrusions in iDC2.4. Yellow solid line corresponds to the F-actin density of the cell body. Error bars represent standard deviation. (i) Distribution of actin density across the F-actin-dense regions and membrane protrusions in LPS-DC2.4. Yellow solid line indicates the F-actin density of the cell body. Error bars represent standard deviation. All images are visualized using the “Fire” LUT. Scale bars, 10 μm (b and c) and 3 μm (d–g).

Phalloidin-PAINT also enables consistent quantitative assessment of local actin concentrations across the cell body and in cytoskeletal protrusions. Based on the intensity of the superresolution image, we conducted a global analysis of the actin densities of the cells depicted in Figs. 4, b and c. Our analysis revealed three distinct actin populations in the iDC2.4 and LPS-DC2.4 cells: subcellular actin-dense regions (podosomes in iDC2.4), cell body, and membrane protrusions (Figs. S14 and 4, h, i). To facilitate comparison, we assessed the density of actin-rich regions and membrane protrusions compared with that of the cell body. Of note is that the local actin density in the cell body and membrane protrusions remained relatively unchanged before and after LPS exposure (Figs. 4, h versus i), which likely reflects the enhanced quantification consistency of phalloidin-PAINT. Fig. 4 h displays the actin densities for iDC2.4 through podosomes, cytoskeleton (yellow solid line), and membrane protrusions. In the cell shown, the actin density in the podosomes was much higher (ca. 12-fold) compared with that of the cell body. In addition, proximal membrane protrusions also displayed a higher actin density (ca. 4-fold) relative to the cell body. In contrast, intracellular actin appeared to be redistributed into more pronounced cytoskeletal structures in LPS-DC2.4. Interestingly, LPS-DC2.4 cells displayed some subcellular actin-rich regions (ca. 6-fold higher) compared with the cell body. These actin-rich regions were distributed in the cell body and along the cell periphery, as evident in Fig. 4 g (arrowheads) and Fig. S14, respectively. The latter observation aligns with existing literature that cortical actin stiffens upon maturation to support effective antigen presentation to T cells (59). While membrane protrusions were much less prevalent in LPS-DC2.4, the local actin densities remained higher (Fig. 4 i, ca. 7-fold) than the cell body. In both cases of iDC2.4 and LPS-DC2.4, the relative actin density of the membrane protrusions was higher than that of the cell body, indicative of the heightened activity of these structures. These results suggest spatial redistributions of F-actin into nanoarchitecture to accommodate DC response to LPS stimulation, i.e., from antigen sampling and sensing to cell migration and antigen presentation.

In summary, we present a single-molecule labeling strategy for superresolution imaging of F-actin in fragile actin-associated membrane fiber structures using phalloidin-AF647. We have demonstrated the performance of phalloidin-PAINT on actin-associated membrane protrusions of DCs. Phalloidin-PAINT identified an average thickness of 52 nm for the actin-associated membrane protrusions in DCs. After LPS treatment, actin is redistributed from podosomes to cytoskeletal fibers. F-actin rearrangements upon LPS treatment can be transient (58). To this end, live superresolution imaging, i.e., using a complementary superresolution technique, such as structured illumination microscopy, and live-cell-compatible probes, such as SiR-actin, will offer new insights into their dynamics and related physiological functions (60,61). While phalloidin-PAINT confirmed the F-actin involvement in DC cytoskeletal protrusions, the presence and spatial distribution of the major histocompatibility complex molecules, for example, will provide additional insights related to the antigen presentation function of these membrane fiber structures. To this end, phalloidin-PAINT combining phalloidin and specific antibodies (46,62) with optimized protocols may provide multiplexed capability.

Phalloidin-PAINT is relatively slow compared with standard PAINT imaging, with a typical 30 frames per minute image acquisition. In comparison, related techniques such as IRIS (frame rate: 20 Hz) and PAINT (frame rate: 25 Hz) using the actin probe Lifeact offer faster imaging speeds. While Lifeact has demonstrated faster imaging and high labeling density in PAINT and IRIS techniques, it has been shown to be less effective in labeling delicate structures (27,35,63). To this end, phalloidin serves as a complementary actin-probe for PAINT-based F-actin superresolution imaging of membrane protrusions. However, the binding of actin probes to F-actin is intricately governed by biochemical interactions dictated by the F-actin architecture, necessitating careful consideration of actin probes for specific investigations (34). For example, phalloidin may not bind certain F-actin conformations, such as the twisted state of F-actin induced by coflin binding (64). Therefore, further studies to enhance the association and dissociation dynamics of F-actin-specific organic and protein-based labels are needed.

Ensemble level in situ characterizations revealed an approximately twofold increase in phalloidin-AF647 dissociation in the presence of 300 mM KSCN (Fig. 2 a). This enhancement contributed to the increased sampling rate for PAINT imaging (approximately threefold) (Fig. 2 h). Phalloidin staining is widely used in actin cytoskeletal superresolution imaging, and the dissociation of phalloidin-AF647 reduces the labeling density for F-actin structures in dSTORM experiments. While this phenomenon has a minor impact on F-actin-abundant architectures, such as the cytoskeleton of adherent U2OS cells and podosomes of DC2.4 cells (Figs. 2 e and 3 e), it significantly affects the labeling density of membrane protrusions with much lower abundance of F-actin (Figs. 3, f). Phalloidin-AF647 cannot be efficiently cross-linked to F-actin, i.e., using aldehyde-based chemical fixation (Fig. S15). Alternatively, labeling with phalloidin-AF647 as the last step at a high phalloidin concentration (0.5 μM) and sample maintenance in phalloidin-AF647 until imaging has been reported to improve dSTORM imaging (39).

Cryogenic electron microscopy studies of phalloidin-bound actin have revealed that the binding pocket contains a mix of hydrophobic and charged residues, with extensive hydrophobic interactions playing a significant role in their interaction (48,65). This suggests that the phalloidin-F-actin interaction is susceptible to chaotropic perturbation, which can theoretically be adapted to other dye-conjugated phalloidins. However, given the comparable size of phalloidin with organic dye molecules, the strength of its interaction can be significantly influenced by the conjugated dye. For example, AF568 -conjugated phalloidin (phalloidin-AF568) was less receptive to chaotropic perturbation by KSCN at a 300 mM concentration (Fig. S16). Phalloidin-AF568 displayed a higher binding rate toward actin compared with phalloidin-AF647, requiring a lower concentration in the imaging buffer (63 pM vs. 1 nM for phalloidin-AF647) to maintain the single-molecule sparsity over time (Figs. S16, a and b). Both Phalloidin-AF568-PAINT with and without chaotropic perturbation resolved the thick F-actin stress fibers (Figs. S16, c and d). While we observed an increase in the sampling rate by approximately 1.5-fold in the presence of KSCN (Fig. S16 e), the reconstructed superresolution image did not adequately represent thin actin filaments (Figs. S16, f and g). Although stronger chaotropic perturbations could further improve the sampling rates, the conditions must be carefully optimized. High concentrations and combinations of chaotropic salts can be denaturing and adversely affect biological systems. Alternatively, phalloidin conjugates with low relative affinity, such as fluorescein, its derivatives, or structurally similar dyes, may serve as better phalloidin-PAINT probes (45). This influence of the dye on phalloidin binding could also be exploited to engineer more transiently interacting phalloidin probes, enabling faster phalloidin-PAINT.

F-actin-associated membrane protrusions are mechanically fragile, making them challenging to preserve. As phalloidin-PAINT does not require washing steps, it mitigates the potential structural damage from shear forces. F-actin fixation followed by a short permeabilization step (materials and methods) enabled the visualization of intracellular actin, proximal membrane protrusions, and some distal retraction fibers. An adapted fixation protocol optimized for TNT structures better preserved the distal retraction fiber network (26). In addition, the morphology of the membrane protrusions is not affected by KSCN (Fig. S17). While optimized preservation protocols still need to be developed, phalloidin-PAINT presents an effective superresolution tool to study fragile membrane protrusions.

The quality of SMLM for life sciences studies depends on the labeling probes and preservation of the biological sample, particularly for nanostructures that can be easily damaged during sample preparation. Therefore, an important endeavor in advancing SMLM is to develop gentle labeling strategies that minimally perturb the biological sample. Phalloidin-PAINT demonstrates a new strategy to quantitatively super-resolve F-actin in the cell body and mechanically fragile cytoskeletal protrusions.

Author contributions

H.G. and Y.S.H. conceived and planned the experiments. H.G. performed single-molecule labeling imaging experiments and data analyses. H.G. and T.P. performed STORM imaging. T.P. and J.A. assisted in data analyses. J.B. performed FFT analysis. C.-J.C. and Z.Z. prepared BMDCs. B.S. and T.P. prepared BMDC and DC2.4 samples for imaging. H.G. and Y.S.H. wrote the manuscript.

Editor: Kathleen Trybus.