HC 07: Super-resolution microscopy
Quick core formulas & sampling rules:
• Abbe / Rayleigh (diffraction limit):
o d ≈ λ / (2·NA) (Abbe) — lateral diffraction-limited resolution ~200–250 nm with visible light.
o dRayleigh = 1.22·λ / (2·NA) (Rayleigh form)
.
• Confocal approximate: dₓᵧ ≈ 0.4·λ / NA, dz ≈ 1.4·λ·n / NA² (useful for sampling & Nyquist).
• Nyquist sampling: pixel ≤ dₓᵧ / 2 ; z-step ≤ dz / 2.
• Localization precision (rule of thumb): σ ≈ s / √N, where s = standard deviation (PSF width), N = number of detected photons (note: real
expression includes pixelation, background & camera noise).
• STED scaling: effective PSF size ∝ 1 / √(ISTED) (resolution improves with depletion intensity).
Why super-resolution?
• Overcome diffraction limit (~200 nm lateral, ~500–700 nm axial) to see nanostructure: filaments, vesicles, virus (~100 nm), protein
complexes (~10–50 nm).
• Techniques trade resolution ⇄ speed ⇄ photon budget ⇄ sample constraints (live vs fixed, depth, label density).
Major families of super-resolution methods — quick comparison:
1. Single-molecule localization (PALM / FPALM / STORM / dSTORM / GSDIM / BaLM / PAINT):
• Resolution: ~10–30 nm practical (theoretically down to a few nm).
• Principle: sparsely switch on small subsets of fluorophores, localize each PSF to high precision, repeat → accumulate pointillist image.
• Best for: thin samples, TIRF or very low background, filaments & puncta, ultrastructure mapping.
• Key hardware: high-power lasers (including ~405 nm activator), high-NA (>1.4) objective, EMCCD/fast sCMOS, stable stage,
autofocus.
• Variants:
o PALM: genetically encoded photoactivatable fluorescent proteins (PA-GFP, mEos, Dendra2, etc.).
o STORM: dye pairs (e.g., Cy3–Cy5) or Alexa/ATTO dyes in special buffers (reducing + O₂ scavengers).
o dSTORM: single dye photoswitching with chemical buffer (simplified STORM).
o GSDIM / GSD: drive dyes into triplet/dark states, record stochastic returns.
o PAINT / DNA-PAINT: transient binding events (no need for photoswitchable dye); kinetics control density (excellent localization
precision, simple buffers)
.
o BaLM: use blinking/bleaching under strong excitation to build localizations (broad dye compatibility).
2. STED / GSD (RESOLFT family, PSF engineering):
• Resolution: ~30–70 nm typical (can be ~20–30 nm in optimized systems); axial & isotropic variants (iso-STED) approach ~40 nm
isotropic.
• Principle: shrink effective excitation volume by depleting fluorophores around a central zero (donut) using a depletion laser (STED) or
shelving ground state (GSD). Scan point-by-point (like confocal).
• Best for: fast point-scanned imaging of dense samples, tissue sections, 3D imaging with appropriate configs.
• Key points: STED resolution improves with depletion intensity (but higher power → more bleaching and phototoxicity). CW vs pulsed
STED; need dyes compatible with stimulated emission. GSD uses lower power and standard dyes but requires precise dye
photophysics.
• Hardware: STED depletion beam shaping (phase plate → donut), synchronized lasers, APD/PMT detectors, vibration control.
3. Structured Illumination Microscopy (SIM / 3D-SIM / SSIM):
• Resolution: ~2× improvement → ~100 nm lateral (3D-SIM gives ~100 nm lateral × ~300 nm axial).
• Principle: illuminate sample with known sinusoidal patterns (grid), acquire multiple phase/angle images, computationally demodulate
to recover higher spatial frequencies (moiré principle).
• Best for: multicolor live-cell imaging (gentler dose than localization/STED), 3D time lapses (moderately high resolution at good speed).
• Important practicalities: 2D SIM needs 9 images per plane (3 phases × 3 orientations); 3D SIM uses 15 images per plane or more.
Requires stable, bright labels and careful grid modulation depth (~70–90%). SSIM (nonlinear SIM) can exceed 2× but needs saturation
and high light doses.
1|Page
, 4. Expansion Microscopy (ExM):
• Resolution: effective improvement by physically expanding the sample, typical ~30–70 nm effective depending on expansion factor.
• Principle: anchor biomolecules to a swellable polymer, digest structural components, expand gel isotropically ⇒ image with
conventional microscope at higher effective resolution.
• Best for: fixed samples, tissue slices and organs; can be combined with STED or confocal for even greater effective resolution.
• Key steps: labeling + anchors → gelation → digestion → controlled expansion → calibration (grid beads) to determine expansion
factor & anisotropy, correct deformation.
Key concepts common to localization & RESOLFT methods:
• Localization precision depends on:
o N (photons detected): more photons → better σ (≈ s/√N).
o s (PSF width): smaller PSF helps.
o Background b and camera read noise: more background/noise worsens precision; use TIRF or very clean imaging conditions.
o Pixel size a: proper sampling (PSF spans ≥2 pixels) aids fitting.
o Label density & Nyquist: super-resolution needs high labeling density in specimen (so structures sampled densely), but sparse active
fluorophores in each frame (to avoid PSF overlap).
• Tuning the number of active fluorophores:
o Photoactivatable proteins (PALM) — control with low-power 405 nm pulses.
o Cyanine dye switching (STORM) — use activator + readout pair (Cy3→Cy5) in reducing/O₂-scavenger buffer.
o dSTORM — reduce dyes to dark radical states with thiols; 405 nm restores.
o DNA-PAINT / PAINT — control transient binding kinetics by oligo length and concentration (kinetics, not light).
o BaLM / GSDIM — exploit intrinsic blinking/photophysics with appropriate illumination/buffers.
3-D super-resolution strategies:
• Astigmatism (cylindrical lens): PSF ellipticity encodes z position (easy, fast).
• Biplane imaging: record two defocused planes → infer z by relative intensities.
• Double-helix PSF: engineered PSF with two lobes that rotate with axial position → large z range.
• Interferometric PALM (iPALM): two opposing objectives + interferometry → excellent axial localization (few nm).
• Adaptive optics: correct sample-induced aberrations for deeper tissue single-molecule imaging.
Sample prep & fluorophore choices (practical):
• Label density: maximize specific labeling density for structural continuity (but avoid steric crowding).
• Fluorophore brightness & photostability: high absorption × high quantum yield (Alexa/Alexa647/ATTO dyes often favored); for PALM
choose robust PAFPs (mEos, Dendra, PA-GFP, rsTagRFP).
• Buffers: STORM/dSTORM require oxygen scavengers (glucose oxidase/catalase or protocatechuate), reducing agents (MEA, β-
mercaptoethylamine), possibly pH control; DNA-PAINT uses simple buffers with controlled salt.
• Mounting & refractive index: match to objective immersion; avoid mismatch to limit aberration. For ExM, use gel-compatible anchors and
digestion steps.
• Controls: single-label references for cross-talk; fiduciary beads for drift correction and PSF calibration.
Instrumentation essentials (super-resolution):
• High NA objectives (≥1.4 oil) for maximal photon collection.
• Powerful, stable lasers: activation (405 nm) + readout lines (488/561/647 etc.) and for STED an additional depletion laser (tunable; phase-
shaped). AOTF/AOM for fast control.
• Detectors: EMCCD or high-sensitivity sCMOS for localization; APD/HyD for scanning STED.
• Mechanical stability & autofocus: nanometer drift ruins localization; use vibration isolation, temperature control, autofocus/coverslip lock.
• Beam shaping (STED): phase plate / SLM for donut generation; good alignment & automated bead-based correction.
• Fast computer and storage: localizations are compute-heavy; GPU acceleration often used.
2|Page
Quick core formulas & sampling rules:
• Abbe / Rayleigh (diffraction limit):
o d ≈ λ / (2·NA) (Abbe) — lateral diffraction-limited resolution ~200–250 nm with visible light.
o dRayleigh = 1.22·λ / (2·NA) (Rayleigh form)
.
• Confocal approximate: dₓᵧ ≈ 0.4·λ / NA, dz ≈ 1.4·λ·n / NA² (useful for sampling & Nyquist).
• Nyquist sampling: pixel ≤ dₓᵧ / 2 ; z-step ≤ dz / 2.
• Localization precision (rule of thumb): σ ≈ s / √N, where s = standard deviation (PSF width), N = number of detected photons (note: real
expression includes pixelation, background & camera noise).
• STED scaling: effective PSF size ∝ 1 / √(ISTED) (resolution improves with depletion intensity).
Why super-resolution?
• Overcome diffraction limit (~200 nm lateral, ~500–700 nm axial) to see nanostructure: filaments, vesicles, virus (~100 nm), protein
complexes (~10–50 nm).
• Techniques trade resolution ⇄ speed ⇄ photon budget ⇄ sample constraints (live vs fixed, depth, label density).
Major families of super-resolution methods — quick comparison:
1. Single-molecule localization (PALM / FPALM / STORM / dSTORM / GSDIM / BaLM / PAINT):
• Resolution: ~10–30 nm practical (theoretically down to a few nm).
• Principle: sparsely switch on small subsets of fluorophores, localize each PSF to high precision, repeat → accumulate pointillist image.
• Best for: thin samples, TIRF or very low background, filaments & puncta, ultrastructure mapping.
• Key hardware: high-power lasers (including ~405 nm activator), high-NA (>1.4) objective, EMCCD/fast sCMOS, stable stage,
autofocus.
• Variants:
o PALM: genetically encoded photoactivatable fluorescent proteins (PA-GFP, mEos, Dendra2, etc.).
o STORM: dye pairs (e.g., Cy3–Cy5) or Alexa/ATTO dyes in special buffers (reducing + O₂ scavengers).
o dSTORM: single dye photoswitching with chemical buffer (simplified STORM).
o GSDIM / GSD: drive dyes into triplet/dark states, record stochastic returns.
o PAINT / DNA-PAINT: transient binding events (no need for photoswitchable dye); kinetics control density (excellent localization
precision, simple buffers)
.
o BaLM: use blinking/bleaching under strong excitation to build localizations (broad dye compatibility).
2. STED / GSD (RESOLFT family, PSF engineering):
• Resolution: ~30–70 nm typical (can be ~20–30 nm in optimized systems); axial & isotropic variants (iso-STED) approach ~40 nm
isotropic.
• Principle: shrink effective excitation volume by depleting fluorophores around a central zero (donut) using a depletion laser (STED) or
shelving ground state (GSD). Scan point-by-point (like confocal).
• Best for: fast point-scanned imaging of dense samples, tissue sections, 3D imaging with appropriate configs.
• Key points: STED resolution improves with depletion intensity (but higher power → more bleaching and phototoxicity). CW vs pulsed
STED; need dyes compatible with stimulated emission. GSD uses lower power and standard dyes but requires precise dye
photophysics.
• Hardware: STED depletion beam shaping (phase plate → donut), synchronized lasers, APD/PMT detectors, vibration control.
3. Structured Illumination Microscopy (SIM / 3D-SIM / SSIM):
• Resolution: ~2× improvement → ~100 nm lateral (3D-SIM gives ~100 nm lateral × ~300 nm axial).
• Principle: illuminate sample with known sinusoidal patterns (grid), acquire multiple phase/angle images, computationally demodulate
to recover higher spatial frequencies (moiré principle).
• Best for: multicolor live-cell imaging (gentler dose than localization/STED), 3D time lapses (moderately high resolution at good speed).
• Important practicalities: 2D SIM needs 9 images per plane (3 phases × 3 orientations); 3D SIM uses 15 images per plane or more.
Requires stable, bright labels and careful grid modulation depth (~70–90%). SSIM (nonlinear SIM) can exceed 2× but needs saturation
and high light doses.
1|Page
, 4. Expansion Microscopy (ExM):
• Resolution: effective improvement by physically expanding the sample, typical ~30–70 nm effective depending on expansion factor.
• Principle: anchor biomolecules to a swellable polymer, digest structural components, expand gel isotropically ⇒ image with
conventional microscope at higher effective resolution.
• Best for: fixed samples, tissue slices and organs; can be combined with STED or confocal for even greater effective resolution.
• Key steps: labeling + anchors → gelation → digestion → controlled expansion → calibration (grid beads) to determine expansion
factor & anisotropy, correct deformation.
Key concepts common to localization & RESOLFT methods:
• Localization precision depends on:
o N (photons detected): more photons → better σ (≈ s/√N).
o s (PSF width): smaller PSF helps.
o Background b and camera read noise: more background/noise worsens precision; use TIRF or very clean imaging conditions.
o Pixel size a: proper sampling (PSF spans ≥2 pixels) aids fitting.
o Label density & Nyquist: super-resolution needs high labeling density in specimen (so structures sampled densely), but sparse active
fluorophores in each frame (to avoid PSF overlap).
• Tuning the number of active fluorophores:
o Photoactivatable proteins (PALM) — control with low-power 405 nm pulses.
o Cyanine dye switching (STORM) — use activator + readout pair (Cy3→Cy5) in reducing/O₂-scavenger buffer.
o dSTORM — reduce dyes to dark radical states with thiols; 405 nm restores.
o DNA-PAINT / PAINT — control transient binding kinetics by oligo length and concentration (kinetics, not light).
o BaLM / GSDIM — exploit intrinsic blinking/photophysics with appropriate illumination/buffers.
3-D super-resolution strategies:
• Astigmatism (cylindrical lens): PSF ellipticity encodes z position (easy, fast).
• Biplane imaging: record two defocused planes → infer z by relative intensities.
• Double-helix PSF: engineered PSF with two lobes that rotate with axial position → large z range.
• Interferometric PALM (iPALM): two opposing objectives + interferometry → excellent axial localization (few nm).
• Adaptive optics: correct sample-induced aberrations for deeper tissue single-molecule imaging.
Sample prep & fluorophore choices (practical):
• Label density: maximize specific labeling density for structural continuity (but avoid steric crowding).
• Fluorophore brightness & photostability: high absorption × high quantum yield (Alexa/Alexa647/ATTO dyes often favored); for PALM
choose robust PAFPs (mEos, Dendra, PA-GFP, rsTagRFP).
• Buffers: STORM/dSTORM require oxygen scavengers (glucose oxidase/catalase or protocatechuate), reducing agents (MEA, β-
mercaptoethylamine), possibly pH control; DNA-PAINT uses simple buffers with controlled salt.
• Mounting & refractive index: match to objective immersion; avoid mismatch to limit aberration. For ExM, use gel-compatible anchors and
digestion steps.
• Controls: single-label references for cross-talk; fiduciary beads for drift correction and PSF calibration.
Instrumentation essentials (super-resolution):
• High NA objectives (≥1.4 oil) for maximal photon collection.
• Powerful, stable lasers: activation (405 nm) + readout lines (488/561/647 etc.) and for STED an additional depletion laser (tunable; phase-
shaped). AOTF/AOM for fast control.
• Detectors: EMCCD or high-sensitivity sCMOS for localization; APD/HyD for scanning STED.
• Mechanical stability & autofocus: nanometer drift ruins localization; use vibration isolation, temperature control, autofocus/coverslip lock.
• Beam shaping (STED): phase plate / SLM for donut generation; good alignment & automated bead-based correction.
• Fast computer and storage: localizations are compute-heavy; GPU acceleration often used.
2|Page
