Superresolution Imaging with Single-Antibody Labeling

Abstract

We present a versatile single-molecule localization microscopy technique utilizing time-lapse imaging of single-antibody labeling. By performing single-molecule imaging in the sub-minute timescale and tuning the antibody concentration to create sparse single-molecule binding, we captured antibody labeling of subcellular targets to generate superresolution images. Single-antibody labeling enabled dual-target superresolution imaging using dye-conjugated monoclonal and polyclonal antibodies. We further demonstrate a dual-color strategy to increase the sample labeling density. Single-antibody labeling paves a new way to evaluate antibody binding for superresolution imaging in the native cellular environment.

Graphical Abstract

INTRODUCTION

Sample labeling remains critical for fluorescence single-molecule localization microscopy (SMLM). Immunofluorescence (IF) staining represents the most versatile and widely used labeling technique. To achieve fluorescent labeling for SMLM, immobilized antibodies bring to the target photoswitchable dyes, such as for dSTORM,^1,2 or DNA-oligo-based docking strands, such as for DNA-based points accumulation in nanoscale topography (DNA-PAINT).^3–7 While IF staining is adaptable and cost-effective, the antibody quality directly affects the quality of the superresolution image. Improving sample labeling often relies upon empirically adjusting the antibody concentration and other IF staining conditions, and evaluating antibody labeling remains difficult until after the superresolution image reconstruction is complete. To date, no effective technique is available for monitoring antibody binding in the cellular environment.

An alternative sample labeling strategy is through reversible binding reagents. We have previously utilized transient interactions between an antibody fragment and its hemagglutinin (HA) polypeptide ligand to circumvent the labeling artifacts due to IF staining.⁸ More recently, similar strategies to enable transient protein-based interactions for PAINT-based superresolution imaging have been reported. These techniques include universal-PAINT (u-PAINT) employing labeled ligands and antibodies to label tagged and endogenous membrane proteins, ⁹ peptide-PAINT using coiled-coil interactions,¹⁰ peptide-PAINT using docking peptides,^11,12 protein-PAINT (pPAINT) using signaling proteins to investigate T-cell signaling with multiplexing capability,¹³ fast-dissociating antibodies generated through hybridoma technology and their fragments,¹⁴ engineered fast-dissociating antibody fragments¹⁵ generated via site-specific mutations against commonly used molecular epitope tags. While these techniques represent significant technical advancements, a more versatile approach remains desirable for using readily available antibodies.

Here, we present superresolution imaging using single-antibody labeling. We observed that local interaction densities in the cellular environment effectively discriminate specific and non-specific antibody interactions.¹⁶ The key to capturing high-density antibody-antigen interaction dynamics is extending single-molecule imaging into the sub-minute timescale and adjusting antibody concentrations to around or below nM in an antibody-dependent manner. By inserting progressively increasing non-illuminating intervals (NIIs), we demonstrate the increased capture of high-density interaction using dye-conjugated monoclonal and polyclonal antibodies alone or in combination. Based on this strategy, we demonstrate a dual-target antibody-labeling using a monoclonal anti-α-tubulin primary antibody and polyclonal secondary antibody targeting the anti-Tom20 antibody. We further demonstrate a dual-color assay to enhance the labeling density of the anti-HA antibody. The simple and versatile technique enables the screening of commercial antibodies for highly multiplexed superresolution imaging.

RESULTS

Time-lapse imaging of single-antibody labeling using monoclonal antibodies achieved superresolution.

We first used the anti-HA monoclonal antibody (12CA5) as a model system. We performed SMLM using 0.5 nM of 12CA5 on a fixed U2OS cell expressing 3xHA on the N-terminus of α-tubulin (Figure 1a).⁸ Our strategy is to scale down the on-rate of the antibody by reducing its concentration so that single-antibody binding can be observed in sufficient time before a significant fraction of the HA substrate is occupied. This strategy is distinct from standard IF staining, which employs much higher antibody concentrations than the affinity constant K_D and drives the reaction toward the bound state before significant antibody dissociation occurs. Figure 1b shows a schematic for SMLM using the total internal reflection fluorescence setup (left panel) and a typical single-molecule image obtained from streaming acquisition at an imaging speed of 20 frames per second (right panel, Supplementary Movie 1). The corresponding scatter plot revealed many random interactions and a lack of microtubule morphology (Figure 1c). A likely cause is the slow on-rate of the 12CA5 antibody at 0.5 nM, and non-specific interactions quickly outnumbered specific binding with streaming acquisition. To evaluate this possibility, we adjusted the image acquisition speed so that each image frame captured sufficient single-molecule events without significant spatial overlap. We progressively increased the duration of NII, decreasing the frame rate from 20 to 0.05 frames per second. To quantitatively investigate the impact of the NII, we employed a strategy similar to a previously reported kinetic fingerprint characterization of antibody binding.¹⁶ Briefly, specific and non-specific antibody interactions are discriminated by their local interaction densities instead of single-molecule dwell times. We employed a density-based spatial clustering of applications with noise (DBSCAN) cluster analysis (methods) and characterized the black, high-density areas, and red, low-density areas. Figure 1d shows representative NII scans (0, 5, 10, 20 s) for the 12CA5 antibody, and Supplementary Movie 2 demonstrates a representative acquisition with the 20 s NII. Figure 1e demonstrates the increasing capture of high-density events with increasing NII. In addition to capturing more high-density events, Figure 1f shows that the inclusion of the NII enhances the sampling of the microtubule structure. The enhanced capture of high-density events and sampling enabled us to reconstruct a corresponding superresolution image (Figure 1g and Figure S1). Figure 1h shows super-resolved adjacent microtubules separated by approximately 83 nm. In addition, the number of single-molecule events remains constant throughout the acquisition (Figure 1i). This observation suggests that single-antibody labeling of 12CA5 remained a consistent process, and the sampling rate for the SMLM linearly scales with the number of acquired image frames.

Figure 1. Time-lapse imaging of single-antibody labeling of monoclonal hemagglutinin (HA) tag antibody achieves superresolution.

a. Schematic illustration of the hemagglutinin (HA) tag antibody system used. U2OS cells expressing 3xHA on the N-terminus of α-tubulin are used with Alexa Fluor 647 conjugated 12CA5 (12CA5-AF647) antibody. b. Schematic illustration of the SMLM with 0.5 nM 12CA5 antibody on a fixed cell (left) and the representative single-molecule image frame from a streaming acquisition (right). Yellow dashed line contours the cell. c. The reconstructed scatter plot from the streaming SMLM acquisition. d. Classification of the high-density (black) and low-density (red) events using DBSCAN for 12CA5-AF647 at increasing NIIs (2,000 frames). e. Percentage of events recognized as high-density events at each NII for 12CA5. Error bars represent standard deviation, n = 10. f. Zoomed-in view of the boxed region indicated in panel d for each NII. g. A representative superresolution image reconstructed from a 2,000 frame acquisition with 12CA5-AF647 on HA-expressing U2OS cell. NII = 20 s. h. Gaussian-fitted cross-sectional profile across the microtubules indicated in panel g. i. Detection of single molecule events per 200 frames over an acquisition of 2,000 frames with 12CA5-AF647 (left y-axis). The right y-axis (blue) represents the cumulative number of events. Scale bars: 10 μm (b and c), 2 μm (d), and 1 μm (f and g).

We next show the time-lapse technique enhanced the capture of single-antibody labeling of dye-conjugated monoclonal antibody against endogenous proteins. Figure S2 shows an NII scan of an anti-α-tubulin antibody (clone DM1A) in a fixed U2OS cell. The DM1A antibody displayed a much faster association than the 12CA5. To maintain a sufficient single-molecule event density, we further reduced the antibody concentration (0.5 nM for 12CA5 and 3 pM for DM1A). The impact from time-lapse imaging was less significant than that with the 12CA5 antibody. However, with enhanced detection of high-density events, we achieved superresolution using the DM1A antibody at 20 s NII (Figure 2a and b). Figure 2c shows the constant single-antibody labeling behavior for DM1A through the acquisition. We further show the time-lapse single-antibody labeling can be used to image a different subcellular structure by using a monoclonal antibody against the clathrin heavy chain (clone X22). We adjusted the antibody concentration to 1 nM to obtain a sufficient single-molecule event density. Figure 2d-f shows superresolution images of clathrin-coated pits of a U2OS cell at various magnifications.

Figure 2. Time-lapse imaging of single-antibody labeling with dye-conjugated monoclonal antibodies against endogenous proteins achieves superresolution.

a. A representative superresolution image acquired using Alexa Flour 488 conjugated DM1A (DM1A-AF488) on a wild-type U2OS cell. NII = 20 s. b. Gaussian-fitted cross-sectional profile across the microtubules indicated in panel a. c. Detection of single molecule events per 200 frames over an acquisition of 2,000 frames with DM1A-AF488 (left y-axis). The right y-axis (blue) represents the cumulative number of events. d. Reconstructed superresolution image of clathrin-coated pits acquired using Alexa Flour 647 conjugated X22 (X22-AF647). NII = 20 s. Dashed line outlines the cell body. Asterisk indicates the cell body. e. Zoomed-in view of the regions indicated in d. f. Zoomed-in regions of clathrin-coated pits indicated in panel e. Scale bars: 1 μm (a), 5 μm (d), 1 μm (e), and 300 nm (f).

Time-lapse imaging of single-antibody labeling using polyclonal antibodies achieved superresolution.

The applicability to polyclonal secondary antibodies expands the utility of our single-antibody labeling technique. We achieved successful capture of single-antibody labeling using 10 or 20 s NIIs using the goat anti-mouse F(ab’)₂ fragment conjugated to Alexa Flour 647 (F(ab’)₂-AF647) and goat anti-rabbit Superclonal recombinant antibody conjugated to Alexa Flour 647 (Superclonal™-AF647). We used the 12CA5 primary antibody to label the HA-expressing α-tubulin and Tom20 primary antibody to label mitochondria. We then incubated these two samples with the F(ab’)₂-AF647 and Superclonal™-AF647, respectively (Figure 3a and b, left panel). Figure 3a and b (right panel) demonstrate the corresponding superresolution images. The morphologies revealed by enhanced capture of high-density events using time-lapse imaging appear consistent with the known morphologies of microtubules and mitochondria.

Figure 3. Reconstructed superresolution images from time-lapse imaging of single-antibody labeling of polyclonal secondary antibodies.

Reconstructed superresolution image of a. F(ab’)₂-AF647 on 12CA5 in a U2OS cell expressing 3xHA-α-tubulin (NII = 20 s, 2,000 frames). b. Goat anti-rabbit Superclonal^™-AF647 on rabbit Tom20 in a wildtype U2OS cell (NII = 20 s, 2,000 frames). Scale bars: 5 μm.

Single-antibody labeling achieved multiplexed superresolution.

A significant advantage of the single-antibody labeling technique is its ease of adaptation for multiplexed superresolution imaging using different antibodies. As a proof-of-principle study, we combined the dye-conjugated monoclonal primary antibody (DM1A-AF488) with dye-conjugated polyclonal secondary antibody (goat anti-rabbit Superclonal^™-AF647). We incubated a fixed U2OS cell prelabeled by unconjugated Tom20 monoclonal (rabbit host, methods) with the imaging buffer containing DM1A-AF488 and Superclonal^™-AF647. The concentration was 3 pM for DM1A-AF488 and 125 pM for the goat anti-rabbit Superclonal^™-AF647. Time-lapse image acquisitions were performed concurrently for the two channels. Figures 4a and b show that the number of binding events remained constant for DM1A-AF488 and decreased by approximately 33.3% for the Superclonal^™-AF647 antibody. The decrease could be due to lower substrate abundance (i.e., Tom20 antibody vs. endogenous α-tubulin). Figure 4c shows reconstructed superresolution images of the microtubule and mitochondria from each channel, and Figure 4d shows the merged image. The separation of structures between the two channels indicates minimal crosstalk between the two antibodies.

Figure 4. Multiplexed superresolution achieved by single-antibody labeling.

a. Detection of single-antibody labeling events from 3,000 frames using DM1A-AF488. b. Detection of single-antibody labeling events from 3,000 frames using Superclonal^™-AF647. c. Superresolution image of microtubules (left) and mitochondria (right) in a fixed U2OS cell. d. Dual-color overlay of the superresolution images in c. Scale bars: 10 μm.

Dual-color single-antibody labeling enhances the sample labeling density.

Due to little to no reversible binding of the antibodies, single-antibody labeling can result in low sample labeling densities compared to standard PAINT techniques that rely upon reversible labeling. While increasing the antibody concentration will increase the labeling density, spatially overlapping molecules will quickly emerge, making single-emitter fitting difficult. A possible strategy to circumvent this limitation is to utilize the same antibody conjugated with different dyes and perform multiplexed imaging. To demonstrate this concept in a dual-color setting, we conjugated the 12CA5 antibody with Alexa Flour 568 (12CA5-AF568) and Alexa Flour 647 (12CA5-AF647), separately. We then incubated the HA-expressing U2OS cell with an imaging buffer containing a total concentration of 2 nM of 12CA5 with 1 nM each of 12CA5-AF568 and 12CA5-AF647, as shown in Figure 5a. Figure 5b shows a reconstructed dual-color superresolution image with zoomed-in views shown in Figure 5c and d. The enhanced labeling density enabled us to super-resolve microtubule fibers 72 nm apart (Figure 5e). The white areas indicate spatial overlap likely resulted from antibody binding to adjacent epitopes or reversible binding by different colored antibodies. The dual-color assay can also increase the labeling density of polyclonal antibodies for superresolution imaging, such as the goat anti-mouse secondary antibody F(ab’)₂ (Figure 5 f-i).

Figure 5. Dual-color single-antibody labeling enhances the sample labeling density.

a. Schematic illustration of the dual-color imaging assay with 12CA5-HA system. b. A representative dual-color superresolution image using a mixture of an equimolar 12CA5-AF568 (green) and 12CA5-AF647 (magenta) in the imaging buffer on a fixed 3xHA-α-tubulin expressing U2OS cell acquired from 3, 000 frames. NII = 20 s. c. Zoomed-in region of the boxed region is shown in panel b. d. Zoomed-in view of the boxed region indicated in panel c. e. Gaussian-fitted cross-sectional profile across the microtubules indicated in panel d. f. Schematic illustration of the dual-color imaging assay for goat anti-mouse F(ab’)₂ in a 12CA5 labeled and fixed U2OS cell expressing 3xHA-tagged α-tubulin. g. Representative dual-color superresolution image using a mixture of an equimolar F(ab’)₂-AF488 (green) and F(ab’)₂-AF647 (magenta) in the imaging buffer and acquired from 3, 000 frames. NII = 20 s. h. Zoomed-in view of the boxed region shown in panel g. i. Gaussian-fitted cross-sectional profile across the microtubules indicated in panel h. Scale bars: 10 μm (a), 1 μm (c and g), and 250 nm (d and h).

DISCUSSION

Time-lapse imaging of single-antibody labeling represents a straightforward approach to achieving multiplexed SMLM. The technique is amenable to dye-conjugated primary and secondary antibodies. Distinct advantages of our technique compared to other PAINT techniques include higher signal-to-noise ratios due to multiple dyes conjugated to individual secondary antibodies and potential signal amplification from a single primary antibody labeled by multiple polyclonal secondary antibodies. Limitations of the technique include the extended image acquisition and relatively low labeling densities due to the slow off-rate of the full-length antibodies. Our previous work showed antibody fragmentation could result in faster dissociation,⁸ while others have utilized protein engineering approaches.^14,15 Nevertheless, promoting antibody dissociation without compromising its binding specificity is likely a complex task and highly dependent upon the nature of antibody-antigen interactions.

Unlike standard IF staining procedures, which remove non-specifically bound antibodies through the washing steps, time-lapse imaging of single-antibody labeling captures both specific and non-specific events. We show that extending single-molecule imaging into the sub-minute timescale enhances the detection of high-density single-antibody binding events, which likely corresponds to specific interactions. When the protein target is low-abundant, we anticipate that our technique may capture a significant fraction of low-density events compared to high-density events. For instance, super-resolving clathrin-coated pits using time-lapse single-molecule imaging of a dye-conjugated primary antibody resulted in some less-defined structures, which may be difficult to discern from low-density labeling events (Figure 2d-f). The dwell time, which in single-molecule binding kinetics correlates to the off-rate, is not an effective parameter to discriminate specific from non-specific events, as it has been shown that non-specific events can also manifest as long-dwelling events.¹⁶ Instead, density-based clustering algorithms enhance identifying specific binding events characterized as high-density events (Figure 1d and f and Figure S2b).

Time-lapse single-molecule imaging can be readily expanded to screen for other antibodies for the single-antibody labeling technique. Table 1 summarizes the antibodies evaluated in this study. A critical task is determining the antibody concentration suitable for single-emitter fitting without significant spatial overlap. The concentrations need to be determined empirically with the NII scan (Figure 1, Figure S3, and methods), as conventional affinity characterizations do not consider the local distribution of binding targets at the nanoscale. We evaluated the number of single-molecule events in a 15 × 15 μm area for microtubule imaging across the antibody systems; 12CA5 (0.5 nM), DM1A (0.003 nM), and F(ab’)₂ (0.063 nM) (Figure S3). Our results revealed that the event density ranged from approximately 0.09 to 0.3 events/μm² which presents a starting single-molecule event density for optimizations of new antibody systems (Figure S3). In addition, the optimal concentration is highly dependent upon both the antibody and the binding target and can vary over several orders of magnitude. To this end, the binding substrate’s spatial distribution and local concentration at the nanoscale play a critical role in the single-antibody labeling technique. Of note is that the impact from local concentrations of the substrate in the cellular environment cannot be effectively captured by the standard affinity characterizations, and our single-antibody labeling technique enables such evaluations in the native cellular environment. Our study also suggests that the duration of the NII may be proportional to the effective on-rate of the antibody at a given concentration. However, we found that an NII value of 10–20 s was generally sufficient for time-lapse imaging of single-antibody labeling for all the antibody systems studied in our work (Table 1 and methods).

Table 1:

Summary of commercial antibodies used for single-antibody labeling. Antibodies are evaluated at a 20 s NII on fixed U2OS cells.

Antibody	Vendor/Catalog Number/Lot Number	Concentration (nM)
DM1A-AF488 (primary)	Invitrogen /53–4502-82/2331001	0.003
X22-AF647 (primary)1	Invitrogen/MA1–065-A647/UJ297446	1
Anti-rabbit Superclonal™ -AF6472 (Secondary)	Invitrogen/A27040/2097674	0.125
Anti-mouse F(ab’)2-AF6473 (Secondary)	Invitrogen/A21237/2358346 and 2486562	0.063

¹Resulted in less defined structures.

²Primary antibody for mitochondria labeling: Rabbit polyclonal Tom20 (4 μg/mL), Santa Cruz sc-11415, discontinued.

³Primary antibody for microtubule labeling: Mouse anti-HA (clone 12CA5, 10 μg/mL), Bio X Cell RT0268, lot number 753521J2

CONCLUSION

In summary, we demonstrate an SMLM technique utilizing time-lapse imaging of single-antibody labeling. The technique is amenable to dye-conjugated monoclonal and polyclonal antibodies alone or combined for multiplexed imaging. We also demonstrate a dual-color strategy to enhance the sampling density for SMLM. Time-lapse imaging of single-antibody labeling represents an alternative labeling strategy, which enables the screening and evaluation of antibody labeling in the native cellular environment for superresolution imaging.

Materials and Methods

Materials

DMEM (11960069–500 mL), penicillin-streptomycin (15140–122-100 mL), L-glutamine (25030–081-100 mL), DPBS (14190–144-500 mL), glycine (BP381–500), magnesium sulfate (7487–88-9–500G) and potassium chloride (P217–500) were purchased from Fisher Scientific. FBS (F0926–500 mL), puromycin (puromycin dihydrochloride; P8833–25 mg), Triton X-100 (X100–1 L), MES (M3671–50 g), EGTA (E3889–25 g), Bovine Serum Albumin (BSA) (A7906–100G), sodium bicarbonate (S6014–500 g), N,N-Dimethylformamide (DMF) (D4551–500mL), PIPES (P-7643–100G), HEPES (H3375–250G) and magnesium chloride (M8266–100 g) were purchased from MilliporeSigma. TetraSpeck microspheres (T7280), EDTA (AM9261), Nunc Lab-Tek chambered coverglass (155411), Zeba spin desalting column (89889), goat anti-rabbit Alexa Fluor^™ 647 IgG (Superclonal^™ Recombinant, A27040–1mg), F(ab’)₂-Goat anti-Mouse IgG (H+L) Cross-Adsorbed Secondary Antibody Alexa Fluor^™ 647 (A21237), F(ab’)₂-Goat anti-Mouse IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor^™ 488 (A11017), Clathrin heavy chain monoclonal Alexa Fluor^™ 647 (MA1–065-A647), Alexa Fluor 568 NHS ester (A20003) and α tubulin monoclonal antibody (DMIA) Alexa Fluor^™ 488 (53–4502-82) were purchased from Thermo Fisher Scientific. ReadyTag anti-HA (12CA5) (RT0268–25 mg) was purchased from Bio X Cell. Paraformaldehyde (15710) and glutaraldehyde (16120) were purchased from Electron Microscopy Sciences. Syringe filters (1159T78, 0.22 μm) were obtained from Denville Scientific Inc. Sodium borohydride (213462–25G) was purchased from Sigma-Aldrich. Unconjugated gold colloids (15711–20) were purchased from Ted Pella Inc.

Buffers:

The following buffers were used for sample preparation and imaging.

Cytoskeleton buffer:

MES (10 mM, pH 6.1), KCl (90 mM), Mg₂Cl (3 mM), and EGTA (2 mM). PHEM buffer (2x): PIPES (60 mM), HEPES (25 mM), EGTA (10 mM), and MgSO₄ (8 mM) in distilled water, with pH, adjusted to 7. Fixation buffer for microtubules and mitochondria: paraformaldehyde, PFA, (3.7%), glutaraldehyde, GA, (0.1%), Triton-X-100 (0.5%), in cytoskeleton buffer. Fixation buffer for clathrin: PFA, (3.7%), in PHEM buffer (1x). Post-fixation buffer: PFA (3.7%), in 1x DPBS. Blocking buffer 1: BSA, (5% in 1x DPBS, syringe-filtered), Blocking buffer 2: Triton X-100 (0.5%) in BSA (3% in DPBS, syringe-filtered).

Alexa Fluor NHS-Ester conjugations:

Alexa Fluor 568 NHS-Ester conjugation to 12CA5:

ReadyTag anti-HA (12CA5) was diluted to a final concentration of 4.8 mg/mL and adjusted to pH 8.3 by adding NaHCO₃ buffer (0.5 M, pH 8.3). The final concentration of NaHCO₃ was 0.1 M. Alexa Fluor 568 NHS ester was dissolved in DMF at 10 mg/mL. The antibody solution was added to the reactive dye solution at a 30:1 dye-to-antibody molar ratio and was incubated for 2 h at room temperature in the dark on a rotator. The dye was deactivated by adding glycine (1 M, pH 7.4) to a final concentration of 30 mM. The reaction mixture was run through a Zeba spin desalting column to purify the dye-conjugated antibody from the excess dye. The antibody-to-dye labeling ratio was obtained using NanoDrop (Figure S4).

Conjugation of Alexa Fluor 647 NHS Ester to 12CA5 is performed as previously reported.⁸

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. U2OS cells stably expressing HA on α-tubulin⁸ were cultured in DMEM supplemented with 10% FBS, 2 mM L-glutamine, 100 units/mL penicillin-streptomycin, and 0.25 μg/mL puromycin. Both cell lines were maintained at 37 °C in a humidified atmosphere of 5% CO₂ and split upon reaching confluence. For all imaging experiments, cells (approximately 5,000 cells/well) were seeded ~36 h before fixation in chambered coverglass slides.

Sample Preparation:

Fixation:

All steps were performed at room temperature. For microtubule and mitochondria imaging, wild-type U2OS cells and U2OS cells expressing 3xHA on α-tubulin were seeded, fixed, and permeabilized in a freshly prepared fixation buffer for 20 minutes. The sample was then reduced with sodium borohydride (0.1%, prepared immediately before use) for 7 minutes while shaking. For clathrin imaging, U2OS cells expressing 3xHA on α-tubulin were permeabilized with cold Triton-X-100 (0.1%) in PHEM buffer (1x) for 45 s. Then the sample was fixed with a freshly prepared fixation buffer for 20 minutes. The cells were rinsed two times with DPBS and reduced with sodium borohydride (0.1%, prepared immediately before use) for 7 minutes.

Blocking and immunostaining:

For single-antibody labeling with monoclonal antibodies, fixed cells were blocked (2 h) with blocking buffer 1 and post-fixed for 10 minutes with a freshly prepared post-fixation buffer to preserve blocking.

Single-antibody labeling with secondary antibodies: For microtubule labeling, fixed U2OS cells expressing HA were blocked (2 h) and immunostained (4 h) with 12CA5-IgG (10 μg/mL) in blocking buffer 1. For mitochondria labeling, fixed wild-type U2OS cells were blocked (30 minutes) and immunostained (1 h) with rabbit polyclonal Tom20 antibody (4 μg/mL) in blocking buffer 2. All immunostained samples were washed 3 times with DPBS and post-fixed for 10 minutes with a freshly prepared post-fixation buffer to preserve primary antibody labeling. All samples were maintained in DPBS. TetraSpeck beads (1:1000 in DPBS) or gold colloids (1:25 in ultrapure water) were used as fiducials and added before image acquisition.

Single-antibody labeling concentrations:

Antibody concentrations were adjusted for each antibody system to maintain molecular sparsity for single-emitter fitting (Figure S3). For dual-color experiments, the antibody concentrations of 12CA5 and F(ab’)₂ were adjusted to maintain a comparable molecular sparsity in each channel (Figure S5). The corresponding antibodies were diluted to the following concentrations in DPBS just before image acquisition: Single-color imaging with monoclonal antibodies, DM1A-AF488: 3 pM, 12CA5-AF647: 500 pM, and Clathrin-AF647: 1 nM. Single-color imaging with polyclonal antibodies, F(ab’)₂-AF647: 63 pM, Anti-rabbit Superclonal-AF647: 125 pM. Dual-color imaging of 3xHA-tagged α-tubulin, 12CA5-AF647: 1 nM, 12CA5-AF568: 1 nM. Dual-color imaging of immobilized 12CA5: Anti-mouse F(ab’)₂-AF647: 250 pM, Anti-mouse F(ab’)₂-AF488: 250 pM. Multiplex single-antibody imaging, DM1A-AF488: 3 pM, Anti-rabbit Superclonal-AF647: 125 pM.

Microscopy:

Fluorescence imaging was performed on an inverted microscope (Nikon Instruments, Eclipse Ti2E). A 100x/1.49 oil objective was used with a 1.5x external magnifier. Single-molecule movies were acquired using a Prime 95B sCMOS camera at 16-bit with 2×2 binning creating an effective image pixel size of 147 nm and an integration time of 50 ms. Dual-color imaging of immobilized 12CA5 with F(ab’)₂-AF488 and −647 was acquired at 12-bit with Optosplit II (emission splitter T 640 LPXR, emission filters AT 600lp and 59007m). Single-color image acquisitions (streaming and time-lapse) and multiplexed single-molecule image acquisitions were performed at a laser power density of 136 W/cm² (647 nm) and 380 W/cm² (488 nm). Dual-color imaging was performed at a laser power density of 136 W/cm² for each channel used (647, 488, and 568 nm).

In F(ab’)₂ AF488/647 imaging, a parallel periodic photobleaching step was applied in the 488 channel (power density of 384 W/cm²) at every 20^th frame to reduce overlapping events and out-of-focus background.

Streaming acquisition with 12CA5-AF647 was 30 minutes long. For time-lapse imaging, non-illuminating intervals (NIIs) were incorporated between two consecutive image frames, as intended. All superresolution images are acquired with a 20 s NII.

Image reconstructions:

Scatter Plots:

Gaussian-based single-emitter fitting was performed using a custom MATLAB code. Drift-corrected centroid locations of single molecules were projected into a 2D scatter plot.

2D probability histograms:

Superresolution image reconstruction was performed using the open-source ImageJ plug-in ThunderSTORM.¹⁷ Raw images were analyzed in the ThunderSTORM plug-in with the following camera settings: pixel size 147 nm, photoelectrons per A/D count 0.98 (Prime 95B sCMOS camera at 16-bit, serial number: A18B203004), and base level 100. Drift correction was performed using 100 nm gold nanoparticle fiducial beads that were present during the entire acquisition. Events with a sigma value greater than 170 appeared to produce a diffusive appearance in the 2D histogram and were removed in the post-processing. Microtubule and mitochondria superresolution images were visualized at a pixel size of 20 nm in the reconstructed image. Clathrin-coated pit superresolution images were visualized at a pixel size of 10 nm in the reconstructed image. The visualization method of “average shifted histogram” was used for all single-color and dual-color images. Multiplex single-antibody labeling experiments were visualized using the “Normalized Gaussian” at a forced lateral uncertainty of 20 nm.

DBSCAN analysis:

A MATLAB program was developed based on Ester et al.’s original DBSCAN algorithm using MATLAB’s Statistics and Machine Learning Toolbox.¹⁸ Raw images were analyzed in the ThunderSTORM plug-in as described above. The regions of the reconstructed image (10×10 μm) were cropped, and the x and y coordinates were exported for DBSCAN analysis. The analysis was performed using an epsilon distance value, N_Eps, of 60 nm and a core points value, p, of 5. The high-density and low-density points are color-coded in black and red, respectively.