Jun 23, 2026

ExoSloNano Protocol: Fluorescent nanogold labeling for in-cell cryo-electron tomography

DOI
https://dx.doi.org/10.17504/protocols.io.rm7vzw8p2vx1/v1
  • Lindsey N. Young1,
  • Anna T. Tifrea2,3,
  • Elizabeth Villa1,4
  • 1School of Biological Sciences, University of California, San Diego, La Jolla, CA, USA;
  • 2Department of Physics, University of California, San Diego, La Jolla, CA, USA;
  • 3Medical Scientist Training Program, University of California, San Diego, La Jolla, CA, USA;
  • 4Howard Hughes Medical Institute, La Jolla, CA, USA
Lindsey Goodman
Icon indicating open access to content
QR code linking to this content
DOI: https://dx.doi.org/10.17504/protocols.io.rm7vzw8p2vx1/v1
Protocol Citation: Lindsey N. Young, Anna T. Tifrea, Elizabeth Villa 2026. ExoSloNano Protocol: Fluorescent nanogold labeling for in-cell cryo-electron tomography. protocols.io https://dx.doi.org/10.17504/protocols.io.rm7vzw8p2vx1/v1
Manuscript citation:
Young, L.N., Sherrard, A., Zhou, H., Shaikh, F., Hutchings, J., Riggi, M., Narasimhan, M., Flaherty, W.A., Bennett, E.J., Rosen, M.K., Giraldez, A.J. & Villa, E. ExoSloNano: multimodal nanogold labels for identification of macromolecules in live cells and cryo-electron tomograms. Nat. Methods 23, 131 (2026). https://doi.org/10.1038/s41592-025-02928-4
License: This is an open access  protocol  distributed under the terms of the  Creative Commons Attribution License,  which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Protocol status: Working
We use this protocol and it's working
Created: June 22, 2026
Last Modified: June 23, 2026
Protocol  Integer ID: 319598
Keywords: cryo-electron tomography, cryo-FIB milling, in-cell cryo-EM, nanogold labeling, HaloTag, Streptolysin O, protein localization, CLEM, calibrating intracellular probe delivery, intracellular probe delivery, quantitative nanogold labeling of protein, molecules for cryo, nanogold signal during cryo, efficient intracellular delivery, quantitative nanogold labeling, fluorescence microscopy, probe localization by fluorescence microscopy, cell cryo, delivery of nanoparticle, mediated membrane permeabilization, electron microscopy, nanogold particle, containing nanogold particle, nanoparticle, membrane permeabilization, optimized cryo, cryo, visualization of macromolecular structure, handling nanogold signal, exoslonano protocol, vitrified cellular sample, labeling molecule, specific protein, cellular sample, exoslonano, mediated delivery, focused ion beam milling electron tomography, protein, electron tomography this protocol, macromolecular structure
Funders Acknowledgements:
L.N.Y
Grant ID: NIH NCI K00 CA223029
E.V.
Grant ID: NIH U54 AI170856
E.V.
Grant ID: Pew Scholar Award
E.V.
Grant ID: NSF DBI 1920374
E.V.
Grant ID: Howard Hughes Medical Institute
A.T.T
Grant ID: 5T32GM007198-48
Abstract
This protocol is for the EXOgenous SLO-mediated delivery of NANOparticles (“ExoSloNano”) for labeling molecules for cryo-focused ion beam milling electron tomography (cryo-FIB-ET). In-cell cryo-electron microscopy (cryo-EM) enables the visualization of macromolecular structures within their native environment. However, identification of specific proteins remains a challenge, particularly for low-abundant and unknown targets. Here, we provide a comprehensive protocol for the Exogenous Streptolysin O (SLO)-mediated delivery of Nanoparticles (ExoSloNano). This workflow enables quantitative nanogold labeling of proteins for cryo-FIB-ET in live cells.

This protocol details step-by-step procedures for optimizing SLO-mediated membrane permeabilization, calibrating intracellular probe delivery, and validating probe localization by fluorescence microscopy. Cell treatment conditions are systematically refined to achieve efficient intracellular delivery while maintaining cell viability and structural integrity. Vitrified cellular samples are then prepared on cryo-EM grids and tomograms containing nanogold particles (AuNPs) are collected using optimized cryo-FIB-ET imaging parameters. Particular emphasis is placed on quantitative matching of probe concentration to target abundance and handling nanogold signal during cryo-ET data processing using provided scripts. This protocol is applicable to both cytoplasmic and nuclear targets engineered with self-labeling tags and is compatible with standard in-cell cryo-ET workflows. Overall, the protocol provides a practical framework for molecularly specific protein labeling across length scales in live mammalian cells.
Image Attribution
Overview of ExoSloNano, Delivery of EXOgenously SLO-mediated delivery of NANOparticles for labeling molecules. Example nanogold probe design, custom synthesis with Nanoprobes, Inc. Components: 1.4 nm Nanogold (Nanoprobes, #2025-30NMOL); 10,000 MW amino-dextran spacer (conjugated product, Thermo Fisher #D22910); HaloTag Ligand (Promega, #P6751); AlexaFluor488. [fig:experimental_overview]
Materials

Materials

Buffers & Media
  • HEPES-buffered Tyrode's Solution, without calcium (see recipe)
  • Recovery media (see recipe)
  • PBS + 1 mM MgCl₂ (see recipe)
  • Calcium- and magnesium-free DPBS (Corning, #21-031-CV)
  • Dulbecco's Modified Eagle Medium (DMEM), phenol red-free
  • Fetal Bovine Serum (FBS), phenol red-free
  • OptiMEM with Sodium Pyruvate (Gibco, #31985062)
  • HBSS (Gibco, #14025092)
Reagents
  • Streptolysin O (SLO) from Streptococcus pyogenes (Sigma-Aldrich, #S5265)
  • 0.1 M TCEP, pH 7.0 (Sigma, #51805-45-9)
  • Corning Matrigel hESC-Qualified Matrix, LDEV-Free (Corning, #354228)
  • Fluorescent dextrans: CF® Dye 488 Dextran Anionic and Fixable (Biotium, #80110), 10 kDa or 40 kDa
  • Fluorescent nanogold probe
  • Propidium Iodide, 0.5 μg/mL final concentration (BD Biosciences)
  • Annexin V (BioLegend, #640912)
  • 25% Glutaraldehyde (EMS, #16200) — dilute to 2.5% in DPBS
  • 30 mM Glycine
  • ProLong Gold Antifade Mountant with DAPI (Invitrogen, #P36931)
  • Fibronectin, 50 μg/mL (or Matrigel)
  • Trypsin
Cell Culture Consumables
  • 12-well plate (GenClone, #25-106MP)
  • Cellvis 24-Well Glass-like Polymer Bottom Plate (#P24-1.5P)
  • Cellvis 12-Well Glass Bottom Plate (#P12-1.5H-N)
  • CellVis 12-Well Plate with #1.5 glass-like polymer coverslip bottom (Cellvis, #P12-1.5P) — for plate reader assay
  • iBidi μ-slide 8-Well ibiTreat (#80826-90) — for fixed cell imaging
  • 35 mm glass-bottom dish (MatTek, #P35G-1.5-20-C) — for EM grid seeding
EM Supplies
  • Quantifoil carbon-on-gold EM grids (R 1/4 Au 200-mesh)
  • Multi-well PDMS Stencil (Avéole, 4-well pattern #4W001)

Equipment

  • 37°C, 5% CO₂ incubator
  • Phase contrast microscope
  • Fluorescence microscope
  • Plate reader — configure bandpass filters for your fluorophore before the experiment
  • PELCO easiGlow Glow Discharge Cleaning System
  • Cryo-plunge device (e.g., Vitrobot or Leica EM GP2)

Software

  • Warp for pre-processing, for tomogram reconstruction, and subtomogram extraction
  • IMOD — for tomogram reconstruction and lamella height determination
  • MATLAB + Dynamo — for particle picking and subtomogram analysis (scripts available at https://github.com/villalab/exoslonano)
  • RELION-3.1.4 — for subtomogram averaging and nanogold masking
Troubleshooting
Problem
Too much cell death
Solution
Use less SLO or reduce the SLO incubation time.
Problem
Cells floating after SLO delivery
Solution
Coat the plates with Matrigel (instead of fibronectin, collagen, etc).
Problem
Insufficient probe delivery
Solution
Increase probe concentration or incubation time; alternatively, co-incubate SLO and probe.
Problem
Fluorescent dextran probe delivery less than 80% of cells
Solution
Increase SLO or probe concentration. Optional: sort cells by flow cytometry based on fluorescent probe delivery.
Problem
Incomplete nanogold randomization
Solution
Adjust the threshold cutoff from 3.3σ to a higher or lower value based on previous attempts in your data.
Before start
Always validate fluorescent dextran delivery before attempting nanogold delivery. SLO activity varies between batches and cell lines, so optimization is required. This protocol is optimized for adherent mammalian/human cell lines; buffers, cell attachment, and permeabilization conditions will need to be adapted for other systems.
Background
The original papers describing Streptolysin O (SLO) to mediate delivery of exogenous proteins/probes are Walew et al. (2001) and Teng et al. (2016). The SLO delivery protocol is based on Teng et al. (2018) for the delivery of fluorescent-based probes for super-resolution fluorescent microscopy. This protocol is a complement to the ExoSloNano methods paper which describes the necessity of delivering gold nanoparticles (AuNPs) conjugated to fluorescent probes into live cells for the identification of macromolecules in cells by cryo-electron tomography (cryo-ET) (Young et al., 2026).

Figure 1: Overview of ExoSloNano. Delivery of EXOgenously SLO-mediated delivery of NANOparticles for labeling molecules.
Fluorescent-Nanogold Probe Design
Consider the following when selecting a fluorescent dye: photostability (resistance to photobleaching), compatibility with your cryo-correlative light and electron microscopy (cryo-CLEM) set-up (existing laser lines), your cell plate reader, and the quantum yield of the fluorophore. Keep in mind that there will be auto-fluorescence from the electron microscopy (EM) grids, particularly gold grids. For a thorough breakdown of grid fluorescence, see Last et al. (2023). The second major consideration involves the fluorophore-nanogold interaction. The ability of AuNPs to quench fluorescence signal depends on particle size and the distance between the nanogold and the fluorophore. Gold nanoparticles 2–5 nm emit surface plasmon resonance around 510–520 nm; those 5–10 nm emit around 520–550 nm (Zhou et al., 2016). To mitigate quenching, a 10 kDa dextran spacer is recommended between the 1.4 nm Halo-Nanogold and the fluorophores (custom ordered from Nanoprobes; see Figure 2 for probe design). Always minimize light exposure when creating aliquots and performing experiments - keep covered with aluminum foil when possible. Options for conjugating the nanogold to your target include self-ligating enzymes such as HaloTag and SNAP-tag. These will be genetically encoded as fusion constructs to your target of interest. The HaloLigand on the nanogold bridges the target molecule and the nanogold particle. Monovalency is ideal (1 ligand per 1 nanogold). Multiple ligands on the nanogold could cause aggregation of the target molecule. After making or procuring the nanogold, determine the homogeneity (size-distribution) by preparing EM grids and imaging by cryo-EM.
Figure 2: Example nanogold probe design, custom synthesis by Nanoprobes, Inc. Components: 1.4 nm Nanogold (Nanoprobes, #2025-30NMOL); 10,000 MW amino-dextran spacer (conjugated product, Thermo Fisher #D22910); HaloTag Ligand (Promega, #P6751); AlexaFluor488.
Step 1: Optimization of SLO Concentration to Your Cell Line of Interest
When working with Streptolysin O (SLO) for cellular permeabilization, several critical factors must be considered. As extensive incubation or high concentrations of SLO will cause cell death, it is essential to minimize both time and duration of treatment. Since SLO binds to cholesterol on the plasma membrane and different cell lines have varying cholesterol levels, some cell lines will require more SLO than others for efficient pore formation. It is important to note that probe internalization is not complete - not all of the probe will enter the cell - so later quantification of probe uptake will be necessary. SLO-induced damage is multifaceted, including plasma membrane damage, loss of cytoplasmic contents, and loss of membrane potential. Membrane repair depends on calcium addition and subsequent recruitment of the ESCRT pathway to the site of damage. As ATP and GTP concentrations inside cells drop upon SLO treatment, it is crucial that recovery media contains supplemental ATP-MgSO4 and GTP-MgSO4.

This experimental protocol involves probe incubation on ice to minimize endocytosis, though this can sometimes cause increased cell detachment, particularly with HEK 293T cells. Pore resealing occurs through ESCRT-mediated exocytosis and endocytosis within approximately 10 minutes, provided there is excess calcium to trigger recruitment of ESCRT component ALIX (Corrotte et al., 2012; Jimenez et al., 2014; Keyel et al., 2011). The recovery timeline is within minutes for exocytosis/endocytosis repair of pores, while restoration of homeostasis (membrane potential, signaling changes) occurs over several hours. This technique has been successfully applied across various cell mammalian types, including COS-7 cells (Walew et al., 2001), CHO and monocytes THP-1 (Walew et al., 2001), NIH 3T3 fibroblasts (Keyel et al., 2011), U2OS and HeLa (Teng et al., 2016; Keyel et al., 2011), HEK 293T, RPE1, and iNeurons (Young et al., 2026). Recovery depends on functioning endocytosis and exocytosis. Note that SLO pores are large enough to accommodate antibodies (Teng et al., 2016).
Step 1: Optimization of SLO Concentration to Your Cell Line of Interest
Materials
  • Corning Matrigel hESC-Qualified Matrix, LDEV (Corning #35428)
  • 12-well plate (GenClone #25-106MP)
  • Streptolysin O (e.g., Sigma-Aldrich, cat. no. S5265); manufacturer states 25,000–50,000 U/ml: assume 25,000 U/ml to make dilutions
  • 0.1 M TCEP, pH adjusted to 7.0 (Sigma, cat. no. 51805-45-9)
  • Calcium- and magnesium-free DPBS (e.g., Corning, cat. no. 21-031-CV) containing 1 mM MgCl
  • Streptolysin O (e.g., Sigma-Aldrich, cat. no. S5265); manufacturer states 25,000–50,000 U/ml: ssume 25,000 U/ml to make dilutions
  • Recovery buffer (see recipe below)
  • Adherent cells, grown to 75–90% confluency, cultured in a 12-well chamber coated with Matrigel
  • Dulbecco’s Modified Eagle Medium (DMEM)
  • Phenol red–free Fetal Bovine Serum (FBS)
  • Fluorescent dextrans, e.g., CF Dye 488 Dextran Anionic and Fixable (Biotium, #80110) 10 kDa or 40 kDa
  • OptiMEM with Sodium Pyruvate (Gibco, #31985062)
  • Propidium Iodide (final concentration 0.5 g/ml, BD Biosciences)
  • Annexin V (BioLegend, #640912)
Step 1: Optimization of SLO Concentration to Your Cell Line of Interest
Equipment
  • 37°C, 5% CO2 incubator
  • Phase contrast microscope
  • Fluorescence microscope
12-Well Plate Layout
  • A1: Control(−) SLO(−) Probe
  • A2: Control(+) SLO 40 U/ml(+) Fluorescent Dextran
  • A3: Control(+) SLO 60 U/ml(+) Fluorescent Dextran
  • A4: Experimental(+) SLO 80 U/ml(+) Fluorescent Dextran
  • B1: Experimental(+) SLO 100 U/ml(+) Fluorescent Dextran
  • B2: Experimental(+) SLO 120 U/ml(+) Fluorescent Dextran
  • B3: Experimental(+) SLO 140 U/ml(+) Fluorescent Dextran
  • B4: Experimental(+) SLO 160 U/ml(+) Fluorescent Dextran
SLO Optimization Protocol

1. Coat 12-well plates with 400 μL of Corning Matrigel hESC-Qualified Matrix at 1:100 dilution overnight.
a. Coating is absolutely critical. SLO can affect the actin network, leading cells to temporarily round up and detach. Matrigel (also: Collagen, Fibronectin, Fish Gelatin) prevents this. Matrigel works best because it most closely simulates the native extracellular matrix using a complex mixture of basement membrane proteins. HEK 293T cells are particularly prone to detachment.
b. Incubate 2 hours to overnight.
c. Aspirate the excess.

2. Seed cells so they are at 70–90% confluency at the time of the experiment.

3. Activate SLO with 10 mM TCEP, pH 7.0 at 37˚C for 20 minutes.
a. Mix 9 μL of SLO with 1 μL of 100 mM TCEP, pH 7.0.
b. **Critical: SLO must be reduced with TCEP prior to use to reduce disulfide bonds.

4. Prepare dilutions of activated SLO in PBS + 1 mM MgCl2: 40, 60, 80, 100, 120, 140 U/ml.
a. Assume stock concentration of SLO is 25,000 U/ml.
b. There is batch-to-batch variation in SLO activity; re-optimize each batch.
c. SLO binds cholesterol on the plasma membrane. Different cell lines have different cholesterol levels and some will require more SLO for efficient pore formation.
d. Extensive incubation or high amounts of SLO will cause cell death - minimize time and amount.

5. Prepare 2500 μL of 1 μM CF-Dye in Tyrode’s buffer.

6. Wash cells gently with 500 μL PBS per well (gently apply pipette to the edge of the well).

7. Add 300 μL of PBS + 1 mM MgCl2 to each control well.

8. Add 300 μL of each SLO dilution to each experimental well.

9. Move plate to the incubator (37˚C) for 5–10 minutes.

10. Check cells by phase contrast microscopy at 10x or 20x.
a. By phase contrast, the nucleoli within the nucleus will look brighter and enlarged relative to pre-treated or control cells. See Figure 1 in Teng et al. (2018) to visualize changes in the nucleus following SLO treatment. After recovery, the appearance of the nucleoli will return to that of pre-treated cells.

11. Gently remove the SLO solution from the wells by pipetting slowly off the edge of the well, minimizing turbulent aspiration.

12. After 5 minutes, wash with PBS + 1 mM MgCl2, using the pipette tip against the edge of the well to dissipate the force of the liquid and gently wash the cells.

13. Add 400 μL of Tyrode’s buffer to the control wells, or 400 μL of 1 μM CF-40 kDa Dextran to the experimental wells.

14. Incubate on ice for 8 minutes (under aluminum foil to prevent photobleaching) on a rotating platform (~30 rpm).
a. Probe incubation on ice minimizes endocytosis. In some cases, incubation on ice may cause a greater number of cells to detach; however, Matrigel will reduce cell detachment even if probe incubation is on ice.

15. Remove excess probe.

16. Wash with Tyrode’s buffer, gently.

17. Gently add 500 mL to 1 mL of recovery media to each well.
a. Membrane repair is dependent on the addition of Ca and recruitment of the ESCRT pathway.
b. ATP/GTP concentrations inside the cell drop upon SLO treatment; ensure the recovery media contains ATP-MgSO4 and GTP-MgSO4.

18. Cover with aluminum foil and return to the incubator to let cells recover. Minimum recovery time is 20 minutes; recommended is 2 hours to overnight (overnight is best).
a. In minutes: exocytosis and endocytosis of pore and damaged membrane.
b. In hours: restoration of homeostasis (membrane potential, signaling and transcriptional changes).
c. See Keyel et al. (2011), Corrotte et al. (2012), and Jimenez et al. (2014) for additional explanations of the repair following SLO-induced membrane damage.

19. After recovery, aspirate recovery media.

20. Gently wash with PBS + 1 mM MgCl2.

21. Gently replace media with DMEM phenol-free / 10% FBS / 1% PenStrep.

22. (Recommended) Add Propidium Iodide (final concentration 0.5 μg/ml, BD Biosciences) to assess cell viability. This is not required to visualize fluorescent dextran delivery, but is useful for monitoring cell health.

23. Fluorescence-based imaging to visualize:
a. Dextran delivery
b. Propidium Iodide, 0.5 μg/ml final (fluoresces in dead cells)
c. Annexin V, 5 μg/ml (fluoresces in apoptotic cells)
Tyrode's Buffer
Final composition
  • 140 mM NaCl
  • 5 mM KCl
  • 1 mM MgCl₂
  • 10 mM HEPES
  • 10 mM Glucose
50 mL preparation
Volume | Stock solution
  • 7.5 mL | 1 M NaCl
  • 0.25 mL | 1 M KCl
  • 50 μL | 1 M MgCl₂
  • 500 μL | 1 M HEPES
  • 500 μL | 1 M Glucose


Recovery Media

Final composition
  • OptiMEM (Gibco, #31985062)
  • 10% FBS
  • 2 mM ATP-MgSO₄
  • 2 mM GTP-MgSO₄
  • 11 mM Glucose
  • 2 mM CaCl₂

5 mL preparation
Volume | Component
  • 4.5 mL | OptiMEM with Sodium Pyruvate
  • 500 μL | 100% FBS
  • 100 μL | 100 mM ATP-MgSO₄
  • 100 μL | 100 mM GTP-MgSO₄
  • 55 μL | 1 M Glucose
  • 10 μL | 1 M CaCl₂


PBS + 1 mM MgCl₂

Add 50 μL of 1 M MgCl₂ to 50 mL of PBS.


SLO (Streptolysin O) Handling Notes

  • SLO (Streptolysin O) from Streptococcus pyogenes (Sigma-Aldrich, S5265). Activity from manufacturer reported as 25,000–50,000 U/mL. Assume 25,000 U/mL for dilutions.
  • Dissolve 1 mg of lyophilized protein in 1 mL milliQ water. Make 9 μL aliquots and store at -80°C.
  • Thaw a fresh aliquot prior to each experiment.
  • Treat 9 μL with 1 μL of 100 mM TCEP, pH 7; final concentration is 10 mM TCEP.
  • After treating with 10 mM TCEP pH 7, recommended use is within 5 hours.
  • Manufacturer batch-to-batch variation means each batch will need to be optimized.
Step 2: Determine Target Copy Number
It is important to quantify the number of targets within the cell in order to (a) have an expectation of the number of particles per tomogram (Young & Villa, 2023) and (b) know how much nanogold to deliver. The goal is for the exogenous probe to match the copy number of the endogenous target to minimize the number of unbound probes. CRISPR knock-ins of HaloTag/SNAP/etc. at the endogenous locus are highly recommended. Expression from transient transfections can be stochastic cell to cell. Similarly, lentiviral-mediated integration is stochastic, although cell sorting can improve this. Please refer to the original publication from Cattoglio et al. (2019) for the technical details on determining cellular abundance. The critical equipment needed are a flow cytometer and the standardization cell line (U2OS CTCF-HaloTag).
Step 3: Pilot Fluorescence Microscopy of Probe Delivery
Use optimized SLO concentrations and perform the SLO protocol (Step 1), replacing the proxy fluorescent dextran with the fluorescent nanogold probe.


Reagents

  • Same as SLO protocol (Step 1)
  • Fluorescent nanogold probe
  • 2.5% Glutaraldehyde (25% Glutaraldehyde EMS #16200), diluted in DPBS
  • 30 mM Glycine
  • ProLong Gold Antifade Mountant with DNA Stain DAPI (Invitrogen #P36931)
  • Cell culture plate options:
  • Cellvis 24-Well Glass-like Polymer Bottom Plate (#P24-1.5P)
  • Cellvis 12-Well Glass Bottom Plate (#P12-1.5H-N)

Safety — Glutaraldehyde: Use caution when breaking open the ampule. Open in a fume hood and wear eye protection. Make 500 μL aliquots and store at -80°C. There is a breakpoint at the neck of the ampule. Dispose of 2.5% GA according to your EH&S guidelines.


Starting Probe Concentrations Based on Target Abundance

  • Highly abundant targets (1–10×10⁶ copies): start with 500 μL of 3–4 μM per well in a 12-well plate
  • Medium abundant targets (1–10×10⁵ copies): start with 500–800 nM per well in a 12-well plate
  • Very low to low abundant targets (1–10×10⁴ copies): start with 100–200 nM per well in a 12-well plate


Protocol

  1. Perform SLO assay as in Step 1. Let cells recover a few hours. Cover plate with aluminum foil to protect the fluorophore from light exposure.
  2. Options for fluorescence imaging:
Live cell imaging:
  1. directly in the same dish. See Kiepas et al. (2020) for optimizing live-cell fluorescence imaging to minimize phototoxicity, and Lee & Kitaoka (2018) for general fluorescence guidelines.
If fixing cells:
  1. Detach cells with ~300 μL of Trypsin (enough to cover the well).
  2. Seed cells onto a small plate such as iBidi μ-slide 8-Well ibiTreat (#80826-90).
  3. Let cells attach ~4 hours.
  4. Remove media.
  5. Fix with 2.5% Glutaraldehyde (GA) for 15 minutes.
  6. Aspirate (dispose of 2.5% GA per EH&S guidelines).
  7. Gently wash 2× with DPBS.
  8. Aspirate media.
  9. Apply 30 mM Glycine quencher for 15 minutes to quench unreacted glutaraldehyde and reduce autofluorescence (Cheung & Brown, 1982).
  10. Aspirate media.
  11. Add ProLong Gold Antifade Mountant with DAPI (Invitrogen #P36931).
  12. For room-temperature resin-embedded EM: nanogold will need to be enhanced; see Boassa (2015) for technical details.
Potential outcomes:
  1. Undersaturation (not enough probe localized to target): under-decorated mitochondria, unlabeled cytoskeletal filaments.
  2. Oversaturation will result in background labeling (excess probe and non-specific localization).
  3. Keep in mind that cells will continue to grow and divide post-SLO delivery of the nanogold probe, distributing excess probe or labeled molecules across progeny cells.
Step 4: Bulk Quantification of Fluorescent Nanoparticle Delivery

Plate Layouts

After determining how much SLO to use and the target's abundance, match the delivery of the fluorescent probe to the target's abundance. The goal is to know how many molecules of the probe are internalized per cell and to confirm this value is consistent across experiments. Pilot experiments can first be performed using a small fluorescent dextran (~10 kDa) as a proxy.


Materials

  • Same reagents as for SLO assay (Step 1)
  • Dilution series of fluorescent probe: 0, 10, 100, 250, 500 nM, 1 μM
  • HBSS (Gibco, #14025092) — maintains cell health better than PBS; keep plate outside the incubator no longer than 30–60 minutes
  • CellVis plate options:
  • CellVis 24-Well Plate with #1.5 glass-like polymer coverslip bottom (Cellvis #P24-1.5P)
  • If the signal-to-noise ratio (SNR) is too low, use a 12-well plate: CellVis 12-Well Plate (Cellvis, #P12-1.5P)


Equipment

  • Plate reader — configure bandpass filters for the excitation/emission of your fluorophore before the experiment.


Plate Layouts

12-Well Experimental Plate
  • A1: Control(−) SLO(−) Probe
  • A2: Control(−) SLO(−) Probe
  • A3: Control(+) Optimized SLO(−) Probe
  • A4: Control(+) Optimized SLO(−) Probe
  • B1: Experimental(+) Optimized SLO(+) Probe conc. 1
  • B2: Experimental(+) Optimized SLO(+) Probe conc. 1
  • B3: Experimental(+) Optimized SLO(+) Probe conc. 1
  • B4: Experimental(+) Optimized SLO(+) Probe conc. 2
  • C1: Experimental(+) Optimized SLO(+) Probe conc. 2
  • C2: Experimental(+) Optimized SLO(+) Probe conc. 2

12-Well Probe Standard Plate (must be the exact same plate type)
  • A1: 0 nM
  • A2: 10 nM
  • A3: 100 nM
  • A4: 250 nM
  • B1: 500 nM
  • B2: 1 μM


Protocol

  1. Perform the SLO assay (Step 1) using the optimized SLO concentration.
  2. Prepare enough fluorescent nanogold probe at concentration 1 for 3 wells (~1500 μL total, 400 μL/well).
  3. Prepare enough at concentration 2 for 3 wells (~1500 μL total, 400 μL/well).
  4. After SLO treatment, let cells recover a few hours or overnight.
  5. Prepare a dilution curve in HBSS (see standard plate values above): 400 μL of 0, 1, 10, 100, 250, 500 nM, and 1 μM.
  6. Prior to measuring, replace cell media with phenol-free HBSS. Keep the plate under aluminum foil.
  7. Measure fluorescence by plate reader.
  8. Convert molar concentration to number of molecules:
  9. Multiply molar concentration (e.g., 1 μM) by volume (400 μL) to obtain moles.
  10. Multiply moles by Avogadro's number (6.022×10²³) to obtain molecules.
  11. Subtract background fluorescence (A1) from probe wells to normalize the standard curve.
  12. Plot number of molecules (X) vs. normalized fluorescence intensity (Y) and fit a trendline.
  13. Average fluorescence of SLO-treated control wells without probe (A3 and A4).
  14. Normalize experimental wells (B1–B4, C1, C2) by subtracting the averaged SLO-treated control fluorescence.
  15. Why: Cells auto-fluoresce from endogenous fluorophores (NADH/NADPH, flavins). Normalizing to SLO-treated wells without probe accounts for this.
  16. Use the trendline to convert normalized fluorescence of experimental wells to number of molecules.
  17. This gives approximately the total number of probes delivered per well.
  18. Count cells in each well by trypsinizing and counting.
  19. Divide total probes per well by number of cells to obtain probes delivered per cell.
  20. Average across wells B1–B4, C1, C2. Adjust concentration if needed to match target copy number.
  21. Repeat to reproduce results and to vary concentrations to better match the target.
Step 5: Delivery of Nanogold Probe to Live Cells by SLO
  1. Using the optimized SLO and nanogold concentrations, perform the SLO protocol (Step 1) with the fluorescent nanogold probes.
  2. Let cells recover overnight in recovery media.
Step 6: Seed Cells onto EM Grids and Cryo-Fix
  1. The day after SLO delivery, assemble two imaging chambers.
  2. Apply a multi-well PDMS Stencil (Avéole, 4-well pattern #4W001) to the center of a 35 mm dish (MatTek #P35G-1.5-20-C).
  3. Repeat for chamber 2.
  4. Prepare EM grids:
  5. Plasma clean 8 Quantifoil carbon-on-gold EM grids (R 1/4 Au 200-mesh): 1 minute at 20 mA (PELCO easiGlow Glow Discharge Cleaning System).
  6. Optional: Micropattern your grids as per Swistak et al. (2021) and Sibert et al. (2021).
  7. Place each glow-discharged grid in a well of the PDMS stencil.
  8. Apply 50 μg/mL Fibronectin (or Matrigel) for 45 minutes.
  9. Detach cells from plate.
  10. Critical: Avoid excess handling of cells. Do not spin down to remove trypsin, as trypsin becomes inactivated by proteases in the FBS; excessive cell handling can be avoided by adding trypsin to detach cells and then moving them into suspension with the addition of OptiMEM media supplemented with 10% FBS. (Hebia et al., 2014).
  11. Seed cells onto the coated EM grids: apply ~125,000 cells per chamber.
  12. Allow cells to attach for 4–6 hours.
  13. For additional technical details on seeding cells on grids, see Wagner et al. (2020).
  14. Cryo-fix (vitrify) grids using a cryo-plunge device (e.g., Vitrobot or Leica EM GP2) following standard protocols.
Step 7: Cryo-CLEM
Correlative light and electron microscopy (CLEM) is helpful for:
  1. Identifying specific cells within a population and targeting those on an EM grid.
  2. Identifying where in the cell your feature of interest is located for fluorescence-guided FIB-milling.

Please see Perton et al. (2025), Rose (2025), and Berkamp et al. (2023) for detailed steps on fluorescence-guided cryo-FIB-milling. Be careful not to devitrify the sample with high laser intensities.
Step 8: Cryo-FIB Milling
For detailed step-by-step instructions on best practices for cryo-FIB milling, please see Wagner et al. (2020) and Lam & Villa (2021). Ideal cryo-lamellae are ~120–180 nm. See the data analysis section (step 10) for further explanation.
Step 9: Cryo-ET Data Collection
and alignmentCollect data at a pixel size that sufficiently covers the nanogold particle. The 1.4 nm nanogold is spherical, while the camera sensor is composed of tiles; imaging the nanogold over a tile boundary produces boundary effects with incomplete coverage (see Figure 3). A high-quality tilt series alignment is absolutely critical to prevent distortions of the nanogold. Misalignment of the 2D images will result in non-spherical, elongated representations of the nanogold in the tomogram. For a detailed analysis of benefits and trade-offs for tilt-increment schemes (1° vs. 2° vs. 3°), see Tuijtel et al. (2025). If the goal is to identify 1.4 nm nanogold at the tomographic level, collect at 1° per tilt for the best possible alignment. There will be a larger missing wedge, but with sufficient randomly distributed particles this can be mitigated through subtomogram averaging.

Figure 3: TEM data collection and alignment. Schematic showing considerations on magnification and the effects of nanogold boundary effects and tilt series alignment on nanogold representation in the tomogram. Namely, that misalignments (imperfect alignments) lead to distortions of the spherical nanogold.

Step 10: Data Analysis — Tomogram-Level Particle Picking
Particle picking from the whole tomogram based on nanogold signal requires extremely well aligned tilt series (especially if the nanogold is 1.4 nm), and that the tomograms are not >200 nm thick, as thicker tomograms will have reduced SNR. During FIB-milling, gallium ions penetrate the sample approximately 20–30 nm (Marko et al., 2007; Fukuda et al., 2019). To isolate nanogold signal directly from the cryo-tomogram, the top and bottom ~35 nm of the tomogram must be excluded and the tomogram reconstructed at low binning (~3 Å/pixel) to remove gallium-derived high-intensity pixels. The new tomogram is read into MATLAB using Dynamo's dread('your_tomo.mrc') command. A script creates a histogram of the signal intensity distribution, isolates high-intensity pixels, groups neighboring high-intensity pixels, calculates new centers, filters by volume consistent with nanogold size (accounting for undersampling), and exports the coordinates. The subtomograms will be centered at the nanogold, so the box size must be large enough to also contain the molecule of interest. The MATLAB scripts referenced below (average_slices.m, HIP.m) are available at https://github.com/villalab/exoslonano.


Protocol

  1. Pre-process tilt images with gain reference, motion correction, and CTF estimation.
  2. Perform tilt series alignment and verify alignment handedness.
  3. Reconstruct the whole tomogram with a generous number of pixels to cover the Z-height.
  4. Determine the height of the lamella in IMOD (cursor on one side, hover to the other side, press Q).
  5. Reconstruct a new tomogram at ~3 Å/pixel:
  6. Exclude gallium depositions from the top and bottom ~35 nm.
  7. 1.4 nm nanogold is 14 Å; reconstruct at 3 Å/pixel for suitable coverage.
  8. Average 10 slices to improve SNR using the average_slices.m script.
  9. Run MATLAB script HIP.m to pick high-intensity nanogold signal.
  10. Default threshold: 3.3σ above the mean (empirically determined).
  11. Generate subtomograms using coordinates from the 3 Å/pixel tomogram.
Schematic for step 10

Figure 4: TEM Data Analysis at whole tomogram level. Schematic showing the histogram-based intensity thresholding pipeline for isolating nanogold signal and exporting particle coordinates.

Step 11: Data Analysis — Subtomogram Analysis and Masking Nanogold Signal
Use this approach to aid in:
  • Confirmation of a specific target on a membrane.
  • Differentiation between homologous proteins — the presence or absence of a nanogold particle can be informative.

This protocol requires MATLAB/Dynamo and RELION-3.1.4. MATLAB scripts are available at https://github.com/villalab/exoslonano. Additional potential methods for gold removal include fidder (Burt, 2023), with a step-by-step guide provided by Garrels et al. (2024).

Protocol

  1. Pick particles by template matching, crYOLO, or manual picking (or from nanogold signal positions from Step 10).
  2. Generate subtomograms.
  3. Align and classify subtomograms by 3D classification.
  4. If present, the nanogold-labeled class should converge after 10–12 iterations.
  5. Nanogold signal must be randomized on a per-subtomogram basis to prevent re-alignment to the nanogold.
  6. Run MATLAB script randomize_subtomos.m: randomizes signal intensity of nanogold (3.3σ above the mean) to within 1 SD of the mean.
  7. Copy and edit the RELION star file to point to the modified subtomograms.
  8. Iteratively remove Euler angles (which reflect alignment to nanogold, not the target molecule):
  9. relion_star_handler --i refined_nanogold_class.star --o refined_nanogold_class_noeuler.star --remove_column rlnAngRot
  10. relion_star_handler --remove_column rlnAngTilt
  11. relion_star_handler --remove_column rlnAnglePsi
  12. Confirm randomization by running:
  13. relion_reconstruct_mpi --i particles_randomized_3,3std.star --o particles_randomized_3,3std.mrc
  14. Perform a new 3D Refinement using a 60 Å low-pass filtered reference.
  15. If the refinement has features not present in the reference, copy the star file and edit it to point to the original subtomograms.
  16. Run another 3D Refinement with a soft mask around the molecule to hide the nanogold signal.
  17. Proceed with further subtomogram analysis. For tutorials, see Pyle et al. (2022), Zivanov et al. (2022), Castaño-Díez et al. (2017), and Burt et al. (2021).
Step 11: Data Analysis — Subtomogram Analysis and Masking Nanogold Signal
242 tilt series
77336 particles
Figure 5: Nanogold randomization and subtomogram analysis. Schematic showing the pipeline for isolating nanogold signal, randomizing nanogold signal per subtomogram, exporting particle coordinates, and classifying and refining the particle of interest using subtomogram analysis.
Considerations on Target Selection
The strategies below summarize recommendations for nanogold detection depending on the scientific goal.

  • Validate a new delivery method / in silico methods (template matching): Template match to pick particles; subtomogram averaging recovers nanogold signal. Example: label the ribosome.
  • Separate isoforms (nanogold for discrimination): Template match and separate labeled from unlabeled classes. Example: histone variants within the nucleus.
  • Identify novel targets with known priors: Isolate nanogold computationally; filter based on organelle, localization, or size. Example: a receptor constrained by the membrane.
  • Identify novel targets, completely unbiased: Isolate nanogold computationally; exhaustive search for the particle through subtomogram averaging.

Limitations of using nanogold coordinates alone arise when the nanogold-to-target offset is large. Prior information facilitates more precise particle centering. Selecting particles and filtering based on the presence or absence of nanogold allows discrimination between labeled and unlabeled populations.
Troubleshooting

Troubleshooting

  • Too much cell death: Use less SLO or reduce the SLO incubation time.
  • Cells floating after SLO delivery: Coat the plates with Matrigel.
  • Insufficient probe delivery: Increase probe concentration or incubation time; alternatively, co-incubate SLO and probe.
  • Fluorescent dextran probe delivery less than 80% of cells: Increase SLO or probe concentration. Optional: sort cells by flow cytometry based on fluorescent probe delivery.
  • Incomplete nanogold randomization: Adjust the threshold cutoff from 3.3σ to a higher or lower value based on previous attempts in your data.
Nanogold Labeling Search Range
Figure 6: Nanogold labeling search range. Relying solely on nanogold coordinates alone for subtomogram analysis will require an exhaustive search range, or the use of prior information to facilitate particle centering for subtomogram analysis.
Schematic to show nanogold presence can be used for discriminating homologs through Subtomogram Analysis and Classification


Figure 7: Subtomogram analysis considerations. Example of the utility for using nanogold signal to discriminate against homologs via subtomogram analysis, specifically subtomogram averaging and classification.

Protocol references
Barad, B.A., Garrels, C., & Reichow, S.L. (2024). Tomogram Reconstruction and Gold Fiducial Removal with WarpTools, Etomo, and Fidder. protocols.io. https://doi.org/10.17504/protocols.io.6qpvr8qbblmk/v4.

Berkamp, S., Daviran, D., Smeets, M., Caignard, A., Jani, R., Sundermeyer, P., Jonker, C., Gerlach, S., Hoffmann, B., Lau, K., & Sachse, C. (2023). Correlative Light and Electron Cryo-Microscopy Workflow Combining Micropatterning, Ice
Shield, and an In-Chamber Fluorescence Light Microscope. Bio-protocol, 13(24), e4901. https://doi.org/10.21769/BioProtoc.4901.

Boassa, D. (2015). Correlative microscopy for localization of proteins in situ: Pre-embedding immuno-electron microscopy using FluoroNanogold, gold enhancement, and low-temperature resin. Methods in Molecular Biology, 1318, 173–180. https://doi.org/10.1007/978-1-4939-2742-5_17.

Burt, A., Gaifas, L., Dendooven, T., & Gutsche, I. (2021). A flexible framework for multi-particle refinement in cryo-electron tomography. PLoS Biology, 19(8), e3001319. https://doi.org/10.1371/journal.pbio.3001319.

Burt, A. (2023). fidder: Detect and erase gold fiducials in cryo-EM images. teamtomo. https://github.com/teamtomo/fidder.

Castaño-Díez, D., Kudryashev, M., & Stahlberg, H. (2017). Dynamo Catalogue: Geometrical tools and data management for particle picking in subtomogram averaging of cryo-electron tomograms. Journal of Structural Biology, 197(2), 135–144. https://doi.org/10.1016/j.jsb.2016.06.005.

Cattoglio, C., Pustova, I., Walther, N., Ho, J.J., Hantsche-Grininger, M., Inouye, C.J., Hossain, M.J., Dailey, G.M.,

Ellenberg, J., Darzacq, X., Tjian, R., & Hansen, A.S. (2019). Determining cellular CTCF and cohesin abundances to constrain 3D genome models. eLife, 8, e40164. https://doi.org/10.7554/eLife.40164.

Cheung, H.Y., & Brown, M.R.W. (1982). Evaluation of glycine as an inactivator of glutaraldehyde. Journal of Pharmacy
and Pharmacology, 34(4), 211–214. https://doi.org/10.1111/j.2042-7158.1982.tb04230.x.

Corrotte, M., Fernandes, M.C., Tam, C., & Andrews, N.W. (2012). Toxin pores endocytosed during plasma membrane repair traffic into the lumen of MVBs for degradation. Traffic, 13(3), 483–494. https://doi.org/10.1111/j.1600-0854.2011.01323.x.

Fukuda, Y., Leis, A., & Rigort, A. (2019). Preparation of vitrified cells for TEM by cryo-FIB microscopy. In Biological Field Emission Scanning Electron Microscopy, pp. 415–438. Wiley. https://doi.org/10.1002/9781118663233.ch19.

Hebia, C., Bekale, L., Chanphai, P., Agbebavi, J., & Tajmir-Riahi, H.A. (2014). Trypsin inhibitor complexes with human and bovine serum albumins: TEM and spectroscopic analysis. Journal of Photochemistry and Photobiology B: Biology, 130, 254–259. https://doi.org/10.1016/j.jphotobiol.2013.11.025.

Jimenez, A.J., Maiuri, P., Lafaurie-Janvore, J., Divoux, S., Piel, M., & Perez, F. (2014). ESCRT machinery is required for plasma membrane repair. Science, 343(6174), 1247136. https://doi.org/10.1126/science.1247136.

Keyel, P.A., Loultcheva, L., Roth, R., Salter, R.D., Watkins, S.C., Yokoyama, W.M., & Heuser, J.E. (2011). Streptolysin O clearance through sequestration into blebs that bud passively from the plasma membrane. Journal of Cell Science, 124(14), 2414–2423. https://doi.org/10.1242/jcs.076182.

Kiepas, A., Voorand, E., Mubaid, F., Siegel, P.M., & Brown, C.M. (2020). Optimizing live-cell fluorescence imaging conditions to minimize phototoxicity. Journal of Cell Science, 133(4), jcs242834. https://doi.org/10.1242/jcs.242834.

Lam, V., & Villa, E. (2021). Practical approaches for cryo-FIB milling and applications for cellular cryo-electron tomography. Methods in Molecular Biology, 2215, 49–82. https://doi.org/10.1007/978-1-0716-0966-8_3.

Last, M.G.F., Tuijtel, M.W., Voortman, L.M., & Sharp, T.H. (2023). Selecting optimal support grids for super-resolution cryogenic correlated light and electron microscopy. Scientific Reports, 13(1), 8270. https://doi.org/10.1038/s41598-023-35590-x.

Lee, J.-Y., & Kitaoka, M. (2018). A beginner's guide to rigor and reproducibility in fluorescence imaging experiments. Molecular Biology of the Cell, 29(13), 1519–1525. https://doi.org/10.1091/mbc.e17-05-0276.

Marko, M., Hsieh, C., Schalek, R., Frank, J., & Mannella, C. (2007). Focused-ion-beam thinning of frozen-hydrated biological specimens for cryo-electron microscopy. Nature Methods, 4(3), 215–217. https://doi.org/10.1038/nmeth1014.

Perton, E., Boltje, D., Jakobi, A., & Hoogenboom, J. (2025). Workflow for fluorescence-targeted lamella milling from vitrified cells with a coincident fluorescence, electron, and ion beam microscope. Bio-protocol, 15(14), e5390. https://doi.org/10.21769/BioProtoc.5390.

Pyle, E., Hutchings, J., & Zanetti, G. (2022). Strategies for picking membrane-associated particles within subtomogram averaging workflows. Faraday Discussions, 240, 101–113. https://doi.org/10.1039/D2FD00022A.

Rose, K. (2025). iFLM-guided Cryo-FIB milling. protocols.io. https://doi.org/10.17504/protocols.io.3byl4zey8vo5/v1.

Sibert, B.S., Kim, J.Y., Yang, J.E., & Wright, E.R. (2021). Micropatterning transmission electron microscopy grids to direct cell positioning within whole-cell cryo-electron tomography workflows. Journal of Visualized Experiments, (175), e62992. https://doi.org/10.3791/62992.

Swistak, L., Sartori-Rupp, A., Vos, M., & Enninga, J. (2021). Micropatterning of cells on EM grids for efficient cryo-correlative light electron microscopy. In Methods in Microbiology, vol. 48, pp. 95–110. Elsevier. https://doi.org/10.1016/bs.mim.2020.11.001.

Teng, K.W., Ishitsuka, Y., Ren, P., Youn, Y., Deng, X., Ge, P., Lee, S.H., Belmont, A.S., & Selvin, P.R. (2016). Labeling proteins inside living cells using external fluorophores for microscopy. eLife, 5, e20378. https://doi.org/10.7554/eLife.20378.

Teng, K.W., Ren, P., & Selvin, P.R. (2018). Delivery of fluorescent probes using streptolysin O for fluorescence microscopy of living cells. Current Protocols in Protein Science, 93(1), e60. https://doi.org/10.1002/cpps.60.

Tuijtel, M.W., Majtner, T., Turoňová, B., & Beck, M. (2025). Optimising the tilt-increment for in situ cryo-electron tomography. bioRxiv. https://doi.org/10.1101/2025.08.20.671201.

Wagner, F.R., Watanabe, R., Schampers, R., Singh, D., Persoon, H., Schaffer, M., Fruhstorfer, P., Plitzko, J., & Villa, E. (2020). Preparing samples from whole cells using focused-ion-beam milling for cryo-electron tomography. Nature Protocols, 15(6), 2041–2070. https://doi.org/10.1038/s41596-020-0320-x.

Walev, I., Bhakdi, S.C., Hofmann, F., Djonder, N., Valeva, A., Aktories, K., & Bhakdi, S. (2001). Delivery of proteins into living cells by reversible membrane permeabilization with streptolysin-O. Proceedings of the National Academy of Sciences, 98(6), 3185–3190. https://doi.org/10.1073/pnas.051429498.

Young, L.N., Sherrard, A., Zhou, H., Shaikh, F., Hutchings, J., Riggi, M., Narasimhan, M., Flaherty, W.A., Bennett, E.J., Rosen, M.K., Giraldez, A.J., & Villa, E. (2026). ExoSloNano: multimodal nanogold labels for identification of macromolecules in live cells and cryo-electron tomograms. Nature Methods, 23(1), 131–142. https://doi.org/10.1038/s41592-025-02928-4.

Young, L.N., & Villa, E. (2023). Bringing structure to cell biology with cryo-electron tomography. Annual Review of Biophysics, 52(1), 573–595. https://doi.org/10.1146/annurev-biophys-111622-091327.

Zhou, M., Zeng, C., Chen, Y., Zhao, S., Sfeir, M.Y., Zhu, M., & Jin, R. (2016). Evolution from the plasmon to exciton state in ligand-protected atomically precise gold nanoparticles. Nature Communications, 7, 13240. https://doi.org/10.1038/ncomms13240.

Zivanov, J., Otón, J., Ke, Z., von Kügelgen, A., Pyle, E., Qu, K., Morado, D., Castaño-Díez, D., Zanetti, G., Bharat, T.A.M., Briggs, J.A.G., & Scheres, S.H.W. (2022). A Bayesian approach to single-particle electron cryo-tomography in RELION-4.0. eLife, 11, e83724. https://doi.org/10.7554/eLife.83724.