3D superresolution microscopy by supercritical angle detection

Joran Deschamps; Markus Mund; Jonas Ries

doi:10.1364/OE.22.029081

1. Introduction

Single-molecule localization-based superresolution microscopy (localization microscopy: Photoactivated Localization Microscopy, PALM [1]; Stochastic Optical Reconstruction Microscopy, STORM [2]; Fluorescence Photoactivation Localization Microscopy, FPALM [3]) has been a major breakthrough for cell biology, since it improves the resolution of fluorescence microscopy by an order of magnitude, reaching a lateral resolution comparable to that of electron microscopy. However, obtaining structural information on multi-protein complexes in the cellular context often requires an axial resolution on a similar scale.

Currently, two general approaches allow for three-dimensional resolution in localization microscopy. On one hand, the z position of single emitters can be determined from the shape of the point spread function (PSF) either by imaging two focal planes (bi-plane [4]) or after PSF engineering (e.g. astigmatic PSF [5], double helix PSF [6], self-bending PSF [7]). Nevertheless, the axial resolution is usually worse than the lateral one, which itself is compromised by the modification of the PSF. On the other hand, the z position can be extracted from relative intensities of single-molecule images when employing an interferometric detection scheme (iPALM [8], 4Pi-SMS [9]), reaching unprecedented z resolution. Unfortunately, such interferometric microscopes are very difficult to built and cumbersome to use.

Here we present a fundamentally different way of achieving 3D resolution in localization microscopy without significant loss in lateral resolution, called supercritical angle localization microscopy (SALM), which is based on the principle of surface-generated fluorescence. Surface-generated fluorescence occurs when a fluorophore is in the vicinity of an interface with different refractive indices, such as the water – coverslip interface. This fluorescence is emitted into large angles above the critical angle and is therefore also called supercritical angle fluorescence [10] (Fig. 1(a)). Its intensity depends approximately exponentially on the distance of the fluorophore from the interface (Fig. 1(b)). Supercritical angle and far-field undercritical angle fluorescence are radially separated in the back focal plane (conjugate plane) of a high numerical aperture (NA) microscope objective and can thus be easily detected individually (Fig. 1(c)).

Fig. 1 a: A fluorophore in close vicinity to the water – glass interface can couple its fluorescence directly into the glass, which is emitted at supercritical angles. A polar plot for the emitted intensity in dependence on the emission angle is shown for fluorophores at a z position z = 0 and z = 100 nm above the coverslip (Eqs. (3), (4)). Dashed lines indicate 90° and the critical angle Θ_c b: The fraction of supercritical angle fluorescence of the total fluorescence depends strongly on the distance of the fluorophore from the interface (Eqs. (3), (4)). c: Image of the objective back focal plane (conjugate plane) shows radial separation of super- and undercritical angle fluorescence. d: Schematic of the microscope setup.

Download Full Size | PDF

This principle has been exploited previously in conventional fluorescence microscopy. For example, blocking of undercritical angle fluorescence leads to optical sectioning in microscopy [11–13] and fluorescence correlation spectroscopy [14], with a confinement similar to total internal reflection fluorescence (TIRF) microscopy. By separately imaging supercritical angle emission and undercritical angle emission, objects such as beads or biological membranes can be axially localized with nanometer accuracy in sample-scanning microscopy [15] and wide-field microscopy [16].

Here we apply a detection scheme that separates super- from undercritical angle fluorescence to infer the axial position of single fluorophores, and thereby achieve three-dimensional superresolution microscopy with a simple to implement and robust to operate optical setup.

2. Supercritical angle localization microscopy

Localization microscopy relies on stochastic blinking of single fluorescent molecules. As their brightness fluctuates dramatically, measuring the number of photons in the supercritical angle region alone does not give sufficient information on the emitter’s axial position. However, simultaneous imaging of supercritical and undercritical angle emission in two channels in a ratiometric approach allows normalizing the supercritical to the undercritical angle emission and thus overcomes this problem.

The intuitive way of separately imaging supercritical and undercritical angle emission is to use a circular mask in a conjugate plane. This approach, known as supercritical angle fluorescence microscopy [13], suffers from diffraction at the edge of the mask. The result is a largely inflated point spread function, which makes it difficult to estimate the number of emitted photons. Therefore, we use virtual supercritical angle fluorescence microscopy (vSAF [17]) for localization microscopy (Fig. 1(d)). The fluorescence collected by the objective is laterally constricted in an image plane by a slit and subsequently split by a 50:50 beam splitter. In one channel, total fluorescence is detected, which strongly depends on the axial position of the fluorophore due to the contribution of supercritical angle fluorescence. In the second channel, a ring aperture in the conjugate plane (image of the back focal plane) blocks supercritical angle fluorescence. Thus, only undercritical angle emission is detected, which is only marginally dependent on the fluorophore’s axial position. The two channels are then simultaneously imaged onto two parts of an emCCD camera.

After calibration, we calculate the axial position of the emitter from the relative intensities in the two channels. By additionally localizing the single molecules in the image plane, we achieve three-dimensional superresolution. As the emission light is split in two channels, the lateral localization precision is reduced by a factor of $\sqrt{2}$ in each channel. However, averaging the position of the fluorophores in both channels restores the lateral localization precision to that of a 2D one-channel setup.

3. Theoretical resolution

We determine the maximum theoretical resolution that can be achieved with SALM, by relating the error in $z$ with the error in determining accurate intensities. The latter can be estimated using the Cramer-Rao Lower Bound (CRLB [18]).

In localization microscopy, the $x$ and $y$ position of a single emitter is determined directly by fitting single-molecule images. Its z coordinate however has to be inferred indirectly by measuring a property $f (z)$ that depends on $z$ .

The precision with which $z$ can be determined depends on the error $δ f (z)$ associated with the measurement of $f (z)$ as well as on how strongly $f (z)$ changes with $z$ , i.e. the partial derivative of $f (z)$ with respect to $z$ , $\partial f / \partial z$ :

δ z = \frac{\partial z}{\partial f} δ f = {(\frac{\partial f}{\partial z})}^{- 1} δ f .

In our implementation of SALM,

f (z)

is given by the total intensity

I_{t o t} (z)

, normalized by the intensity emitted into undercritical angles

I_{U}

:

f (z) = \frac{I_{tot} (z)}{I_{U}} = \frac{I_{S} (z) + I_{U}}{I_{U}} = \frac{I_{S} (z)}{I_{U}} + 1.

For freely rotating molecules, relative intensities

I_{S}^{r} (z)

and

I_{U}^{r}

can be calculated numerically [14,19]:

I_{S}^{r} (z) = \int_{0}^{v_{NA}} \frac{2 (n_{1}^{2} + n_{2}^{2}) v ƛ \sqrt{n_{2}^{2} - n_{1}^{2} - v^{2} ƛ^{2}} (n_{1}^{2} + v^{2} ƛ^{2})}{3 (n_{2}^{2} - n_{1}^{2}) (n_{1}^{4} + (n_{1}^{2} + n_{2}^{2}) v^{2} ƛ^{2})} e^{- 2 v z} d v .

Here

λ

is the wavelength of the emitted light,

ƛ = λ / 2 π

,

n_{1}^{}

and

n_{2}^{}

are the respective refractive indices of the buffer and the glass coverslip,

v_{N A} = ƛ^{- 1} {(N A^{2} - n_{1}^{2})}^{1 / 2}

, and the numerical aperture of the objective is defined as

N A = n_{2}^{} \sin Θ_{N A}

.

I_{U}^{r} = \int_{0}^{w_{0}} \frac{2 Q ƛ w}{3} (\frac{1}{{(w ƛ + Q)}^{2}} + \frac{n_{1}^{2} n_{2}^{2}}{{(n_{1}^{2} Q + n_{2}^{2} w ƛ)}^{2}}) d w .

with

Q = {(n_{2}^{2} - n_{1}^{2} + w^{2} ƛ^{2})}^{1 / 2}

and

w_{0} = n_{1}^{} / ƛ

.

From Eqs. (3) and (4) we can thus numerically calculate $\partial f / \partial z$ :

\frac{\partial f}{\partial z} = \frac{1}{I_{U}^{r}} \frac{\partial I_{S}^{r} (z)}{\partial z} .

Using Gaussian error propagation,

δ f

can be calculated from the errors

δ I_{t o t}

and

δ I_{U}

in determining the intensities

I_{t o t} (z)

and

I_{U}

:

δ f^{2} = {(\frac{\partial f}{\partial I_{tot}} δ I_{tot})}^{2} + {(\frac{\partial f}{\partial I_{U}} δ I_{U})}^{2} = {(\frac{δ I_{tot}}{I_{U}})}^{2} + {(\frac{I_{tot}}{I_{U}^{2}} δ I_{U})}^{2} .

If measured in units of photons,

δ I_{t o t}

and

δ I_{U}

can be approximately determined as follows [20]:

δ I_{tot}^{2} = I_{tot} + 4 π I_{BG} s^{2} / a^{2}, δ I_{U}^{2} = I_{U} + 4 π I_{BG} s^{2} / a^{2} .

Here

s

is the standard deviation of the PSF,

a

is the pixel size and

I_{BG}

is the background per pixel, measured in photons.

For fluorophores with a low quantum yield, supercritical angle emission does not compete with undercritical angle emission. Then

I_{U} = \frac{N}{2}, I_{tot} = \frac{N}{2} (1 + \frac{I_{S}^{t}}{I_{S}^{t} + I_{U}^{t}}) .

Here,

N

is the number of photons detected when the fluorophore is far away from the coverslip. In the limit of a high quantum yield,

I_{U} = \frac{N}{2} (1 - \frac{I_{S}^{t}}{I_{S}^{t} + I_{U}^{t}}), I_{tot} = \frac{N}{2} .

From Eqs. (1), (5) and (6) we can calculate

δ z (z)

(Fig. 2(a)).

Fig. 2 Theoretical resolution of SALM. a: Detection of total emission versus undercritical angle emission as employed in this manuscript. Shown are the axial and lateral localization precisions in dependence on the fluorophore’s distance z from the interface for a total of N = 500 or N = 5000 detected photons, without background and with a background of BG = 20 photons per pixel. b: Separate detection of super- and undercritical angle emission. This represents the maximum achievable resolution in SALM and requires extensions as discussed in the outlook section.

Download Full Size | PDF

In an analogous way we can calculate the axial resolution, if a circular aperture is used instead of a ring aperture to block all undercritical light in one channel (Fig. 2(b)). Here, the theoretical resolution is better, since a direct quantification of the supercritical angle emission is much more accurate than measuring it on top of the undercritical angle emission. However, diffraction caused by the circular aperture is not taken into account here, which impedes an accurate measurement of the supercritical angle fluorescence [17].

4. Results

4.1. Calibration

Optical aberrations as well as imperfect shape and alignment of the apertures complicate the use of a theoretical description (section 3) to directly calculate absolute z positions from the intensities in both channels. Therefore, we chose to experimentally calibrate the relative intensities in both channels using fluorescent objects with known axial positions.

To this end, we embedded 40 nm fluorescent beads in an agarose gel at different heights above the coverslip. For each field of view, we first acquired a z-stack with an astigmatic lens in the beam path to precisely determine the axial positions of the beads (see methods). Afterwards we measured total and undercritical angle emission for the same beads. The calibration curve is obtained by plotting $I_{t o t} / I_{U}$ vs. $z_{astig}$ for all beads and approximating the data with an exponential (Fig. 3). This curve is then used to determine z positions of single fluorophores from their relative intensities in both channels.

Fig. 3 Calibration plot showing the ratio of total and undercritical angle fluorescence $I_{t o t} / I_{U}$ vs. bead position $z_{astig}$ determined from an astigmatic z-stack. The exponential fit is used in the following to calculate axial positions of single fluorophores from their intensity ratio in the two channels.

Download Full Size | PDF

4.2. Validation using DNA origami tetrahedra

To validate the calibration and to characterize the single-molecule localization precision, we made use of DNA origami technology that allows placing a defined number of dyes in a specific three-dimensional arrangement [21]. Here we imaged recently developed DNA origami tetrahedra [22]. These commercially available tetrahedra (GATTAquant GmbH) have a well-defined three-dimensional geometry with a side length of 100 nm (Fig. 4(a)) and feature 12 single-stranded DNA docking strands on each of their four vertices that can be imaged using the DNA-PAINT method [23,24]. To this end, short fluorescently labeled DNA imager strands complementary to the single-stranded DNA docking strands were added into the buffer solution. Binding of imager strands results in their immobilization and a bright single-molecule localization event (Fig. 4(b)). Figure 4(c) shows typical three-dimensional images of the tetrahedra obtained with SALM. For further analysis, only complete and isolated tetrahedra were selected (Fig. 4(d)), and the four corners were automatically identified by k-means clustering of the single molecule localizations. From the centers of these clusters we calculated the height of the tetrahedra (Fig. 4(e)) by averaging the z positions of the three bottom clusters and subtracting it from the z position of the top cluster. The average height of (80.5 ± 14.5) nm (mean ± SD, N = 113) is in good agreement with the expected height of the tetrahedra of 82 nm. The spread in heights could be due to a fraction of slightly tilted tetrahedra (visible in Fig. 4(d), where some tetrahedra show deviations from the predicted geometry depicted in Fig. 4(a)) and due to slight deformation of the tetrahedra resulting from limited rigidity (compare Fig. 3 in Iinuma et al. [22]). The localization precision (standard deviation of the localizations at each corner) was σ_x = (11.2 ± 4.7) nm in the lateral and σ_z = (26.4 ± 8.4) nm in the axial direction. This is only an upper bound for the localization precision due to imperfect immobilization of the tetrahedra.

Fig. 4 Validation of SALM with DNA-PAINT on DNA origami tetrahedra. a: DNA origami tetrahedron in side view (top) and top view (bottom panel). b: Principle of DNA-PAINT. The corners of the tetrahedra are coupled to DNA docking strands. Complementary imager DNA strands conjugated with Atto655 can transiently bind, thereby get immobilized and can be detected as single-molecule localizations (with permission from Iinuma et al. [22]). c: Small field of view containing several tetrahedra. d: x-z and x-y projections of individual tetrahedra. e: Histogram of measured heights of N = 113 tetrahedra. Dotted line: nominal height (82 nm) of the tetrahedra.

Download Full Size | PDF

4.3. Application to biological samples

To demonstrate the applicability of SALM to biological samples we imaged immunostained clathrin-coated pits (Fig. 5(a)) and microtubules (Fig. 5(b)) using the dSTORM approach [25]. The expected three-dimensional geometry is clearly visible in the x-y and x-z projections. The quality is comparable to what we observed with PSF-engineering based 3D localization microscopy. Note that the fluorophores in dSTORM emit fewer photons than in DNA-PAINT, which leads to a lower localization precision here compared to Fig. 4.

Fig. 5 SALM measurements on immunostained U2OS cells, z positions are color-coded. a: Clathrin-coated pits. Inset: sections through an individual clathrin-coated pit reveals its half-spherical shape. b: Microtubules. Inset: side-views (x-z reconstructions) of areas as denoted with boxes.

Download Full Size | PDF

5. Discussion and outlook

We have demonstrated that supercritical angle emission can be used to obtain three-dimensional position information in localization microscopy in a straightforward and robust manner. However, the axial resolution currently does not surpass that of established methods based on PSF engineering and is far away from the theoretical limit. This is due to several factors, discussed in the following together with possible solutions that we intend to implement in the future.

Supercritical angle emission could be quantified much more precisely if measured directly, and not on top of the undercritical angle emission (see Fig. 2). Unfortunately, a direct measurement of only the supercritical angle emission by blocking all undercritical angle emission with a circular aperture is not feasible with our current setup because of diffraction at the circular aperture. However, new high-NA objectives (up to NA = 1.70) will have a much larger supercritical area in the back focal plane and considerably reduced diffraction. Residual spherical aberrations, currently present in such objectives, can be reduced with adaptive optics [26]. We furthermore intend to implement new fitting routines using refined PSF models to extract more precise brightness information even in case of a PSF deformed by diffraction or aberrations.

Inherently, using an aperture to block parts of the emission light leads to photon loss. This can be circumvented by splitting under- and supercritical angle fluorescence with an elliptical mirror at an intermediate image of the back focal plane that exclusively reflects out undercritical angle fluorescence.

With these measures, we hope to approach the theoretical resolution and to realize 3D localization microscopy with isotropic resolution, comparable to that of interferometric localization microscopy, albeit with a robust and much simpler setup.

One major limitation of this technique is its limited axial range. It is perfectly suited for studying samples that can typically be imaged by TIRF microscopy, such as membranes or membrane associated proteins. Isotropic resolution throughout thick samples can be realized using physical sections.

A general prerequisite for localization microscopy is the free rotation of detected fluorophores, but in biological samples images of fixed fluorophores show complex asymmetric patterns which can cause localization errors [27] and complicate the evaluation of z positions. Similarly for SALM the amount of supercritical angle emission depends on the orientation of single fluorophores. However, our dual-channel setup will easily allow to test fluorophore rotation by employing a polarizing beam splitter.

6. Conclusion

In summary, we presented a new approach for robust and simple three-dimensional superresolution microscopy. Our proof-of-principle experiments demonstrated the high potential of the method for DNA origami model structures as well as biological samples. We discussed how additional technical improvements will allow SALM to approach isotropic three-dimensional resolution in localization microscopy without compromising the lateral resolution. Due to the simple optical design of SALM, incorporation into existing setups will allow many laboratories to profit from this approach and to reach molecular resolution in the cellular context.

7. Methods

7.1 Microscope setup

Single-molecule microscopy was performed with a custom-built microscope (Fig. 1). 405 nm, 488 nm, 561 nm and 640 nm lasers originating from a single-mode fiber of a laser combiner (Toptica) were focused onto the back focal plane of a Nikon objective (60x, NA = 1.49) for total internal reflection (TIRF) or epi illumination. Emission light was filtered by a 700/100 or 613/73 filter (Chroma), laterally confined in an intermediate image plane, split with a non-polarizing beam splitter and imaged onto two halves of an emCCD camera (Andor). In one channel, a custom-made ring aperture was placed in a back focal plane image (conjugate plane) and adjusted by imaging this plane with an additional lens. Images were acquired with Micromanager [28]. For astigmatic imaging, a cylindrical lens (f = 500 mm, Thorlabs) was placed in front of the camera.

A focus stabilization was implemented by an electronic feedback loop based on the total internal reflection of a red laser at the coverslip and its detection by a quadrant photodiode; the z stability was better than ± 10 nm over several hours. The lateral drift was typically smaller than 50 nm/h; a drift-correction was implemented based on image-correlation [29].

7.2 Calibration sample

40 nm fluorescent beads (F-8770, Life technologies) were diluted 1:20000 in 50 µL of a warm, liquid 1% (w/w) solution of low melting point agarose (Sigma) in water. The solution was pipetted onto a clean glass coverslip. The gel was thinned to ~1 µm by centrifugation of the coverslip at 2500 g and imaged in water.

7.3 DNA-PAINT measurements on DNA origami tetrahedra

We used commercial DNA origami tetrahedra (GATTAquant GmbH, Braunschweig, Germany), which were prepared and measured as reported earlier [22,24]. In short, biotinylated DNA origami tetrahedra were immobilized on BSA-biotin-neutravidin coated coverslips. Docking strand sequences were used as previously [30] and consist of 10 nucleotides binding sequence and 1 nucleotide as spacer (TTAAATGCCCG). 3 nM of imaging DNA strands coupled to Atto655 (CGGGCATTTA-Atto655) were added to the imaging buffer comprised of 5 mM Tris pH 8, 10 mM MgCl₂, 1 mM EDTA, 0.05% Tween-20. Single-molecule imaging was performed under total internal reflection excitation with a frame rate of 2.5 Hz.

7.4 Biological samples

To visualize clathrin-coated pits, cells were fixed with 3% Formaldehyde/PBS at room temperature for 10 minutes, quickly washed and reduced by 0.1% NaBH₄/PBS for 7 minutes. Blocking and permeabilization was performed with 3% BSA/PBS containing 0.5% Triton X-100 for 30 minutes. Cells were immunostained with 1:300 mouse anti-clathrin heavy chain (cloneX22) (ab2731, abcam), 1:300 rabbit anti-clathrin light chain (sc-28276, Santa Cruz Biotechnology) and 1:300 mouse anti-clathrin heavy chain (msCHC5.9) (BM295, Acris) in 2% BSA/PBS for 2 hours to overnight. After washing, they were incubated with 1:200 anti-mouse (A21236, Molecular Probes) and anti-rabbit secondary antibodies conjugated to Alexa Fluor 647 (A31573, Molecular Probes) in 2% BSA/PBS for 2-3 hours.

To visualize microtubules, U2OS cells were prefixed and permeabilized with 0.3% glutaraldehyde in cytoskeleton buffer (10 mM MES pH 6.1, 150 mM NaCl, 5 mM EGTA, 5 mM glucose, 5 mM MgCl₂) with 0.25% TX-100 for 1-2 minutes, and subsequently fixed with 2% glutaraldehyde in cytoskeleton buffer for 10 minutes. After a brief wash with PBS, samples were reduced with 0.1% NaBH₄/PBS for 7 minutes to remove autofluorescence, washed 3 times with PBS, incubated with 1:500 mouse anti-alpha tubulin primary antibody (Neomarker, #MS581) in 3% BSA/PBS for 30 min, washed with PBS and stained with 1:500 of a secondary goat anti-mouse antibody conjugated to Alexa Fluor 647 (A21236 Molecular Probes) in 3% BSA/PBS for 2-3 hours.

The samples were imaged in 50 mM Tris pH 8, 10 mM NaCl, 10% glucose, 35 mM cysteamine, 0.5 mg/mL glucose oxidase and 40 µg/mL catalase. We used an exposure time of 30 ms and an EM gain of 200. Imaging laser intensity at 640 nm was 2-5 kW/cm², the 405 nm activation laser intensity was automatically adjusted to ensure a constant number of localizations per frame. Typically 100000 frames were recorded.

7.5 Data analysis

All data analysis was performed with custom written Matlab software.

7.5.1 Localization analysis

Localization analysis was performed as described previously [29,31]. Briefly, pixel counts were converted to photons by subtracting the constant offset and multiplying by the inverse gain of the camera. Approximate locations of bright spots in each image were determined by smoothing, non-maximum suppression and thresholding. Selected regions of interest were fitted by a pixelized Gaussian function of free width and a homogeneous photon background with a maximum likelihood estimator (MLE) for Poisson distributed data using a freely available, fast GPU fitting routine [32] on a GeForce GTX760 (Nvidia). Lateral drift was corrected based on the imaged features: blocks of typically 10% of the total number of frames were used to reconstruct one superresolution image. Displacements among all reconstructed images were determined by image correlation and fitting of the maximum with an elliptical Gaussian. Displacements corresponding to each time point were averaged using a robust estimator, interpolated by a spline and used to correct the position of each localization. From the variation of the spline, we estimate that the residual error for the corrected positions was about 2 nm.

To evaluate relative intensities in both SALM channels, localizations in the two channels originating from one molecule had to be linked. This was achieved by calculating an approximate shift between the two channels by cross-correlating the images and linking the brightest molecules in each frame, which were then used to calculate an affine transformation between the two channels. This transformation was used to map the positions of channel 2 onto channel 1. Regions of interest around localizations in both channels were cut out and fitted again with a Gaussian function assuming a constant width of the PSF to determine the total number of photons per localization and to calculate $f (z) = I_{t o t} (z) / I_{U}$ .

7.5.2 Calibration

Single bead images of the astigmatic z-stack were fitted with an elliptical Gaussian, resulting in the widths of the PSF in x and y, namely $P S F_{x} (z)$ and $P S F_{y} (z)$ . Localizations in different frames belonging to the same bead were associated by proximity. For each bead, $P S F_{x} (z)$ and $P S F_{y} (z)$ were fitted with a quadratic function in a range between their minima. The z position of the bead was chosen as the crossing point of the parabolas (position of symmetric PSF), after correcting for the refractive index mismatch [33].

Localizations from the same field of view, measured with SALM, were associated to the beads of the corresponding astigmatic z-stack and $I_{t o t} (z) / I_{U}$ vs. $z_{astig}$ was fitted with an exponential function. Since the offset for the $z_{astig}$ scale is not absolute, each data set was shifted along the $z_{astig}$ axis until the exponential fits overlapped at a defined $I_{t o t} (z) / I_{U}$ value. The total data was finally fitted again to an exponential with a robust non-linear least squares fit to obtain the final calibration curve.

7.5.3 Tetrahedra

Locations of complete tetrahedra were manually selected in a reconstructed image. Localizations corresponding to the four corners were associated automatically by k-means clustering of the 3D coordinates into 4 clusters. For each cluster, the centroid, as well as the standard deviations of the positions in the lateral direction ( $σ_{x}$ ) and the axial direction ( $σ_{z}$ ) were calculated. The height of the tetrahedra was calculated as the distance in z between the top corner and the average z position of the three bottom corners.

Acknowledgments

We thank Philip Tinnefeld (Technische Universität Braunschweig) and Jürgen Schmied (GATTAquant GmbH) for providing the DNA origami tetrahedra and for help with DNA paint, Ulf Matti for preparing the biological samples and technical support and Jan Ellenberg, Ingmar Schön, Philip Tinnefeld and Jürgen Schmied for critical reading of the manuscript. This work was supported by the European Molecular Biology Laboratory (J. D., M. M., J. R.), the EMBL International PhD Programme (M. M.) and the Deutsche Forschungs Gemeinschaft (DFG RI 2380/2).

References and links

1. E. Betzig, G. H. Patterson, R. Sougrat, O. W. Lindwasser, S. Olenych, J. S. Bonifacino, M. W. Davidson, J. Lippincott-Schwartz, and H. F. Hess, “Imaging intracellular fluorescent proteins at nanometer resolution,” Science 313(5793), 1642–1645 (2006). [CrossRef] [PubMed]

2. M. J. Rust, M. Bates, and X. Zhuang, “Sub-diffraction-limit imaging by stochastic optical reconstruction microscopy (STORM),” Nat. Methods 3(10), 793–795 (2006). [CrossRef] [PubMed]

3. S. T. Hess, T. P. K. Girirajan, and M. D. Mason, “Ultra-high resolution imaging by fluorescence photoactivation localization microscopy,” Biophys. J. 91(11), 4258–4272 (2006). [CrossRef] [PubMed]

4. M. F. Juette, T. J. Gould, M. D. Lessard, M. J. Mlodzianoski, B. S. Nagpure, B. T. Bennett, S. T. Hess, and J. Bewersdorf, “Three-dimensional sub-100 nm resolution fluorescence microscopy of thick samples,” Nat. Methods 5(6), 527–529 (2008). [CrossRef] [PubMed]

5. B. Huang, W. Wang, M. Bates, and X. Zhuang, “Three-dimensional super-resolution imaging by stochastic optical reconstruction microscopy,” Science 319(5864), 810–813 (2008). [CrossRef] [PubMed]

6. S. R. P. Pavani, M. A. Thompson, J. S. Biteen, S. J. Lord, N. Liu, R. J. Twieg, R. Piestun, and W. E. Moerner, “Three-dimensional, single-molecule fluorescence imaging beyond the diffraction limit by using a double-helix point spread function,” Proc. Natl. Acad. Sci. U.S.A. 106(9), 2995–2999 (2009). [CrossRef] [PubMed]

7. S Jia, J. C. Vaughan, and X. Zhuang, “Isotropic three-dimensional super-resolution imaging with a self-bending point spread function,” Nature Photon 8, 302-303 (2014).

8. G. Shtengel, J. A. Galbraith, C. G. Galbraith, J. Lippincott-Schwartz, J. M. Gillette, S. Manley, R. Sougrat, C. M. Waterman, P. Kanchanawong, M. W. Davidson, R. D. Fetter, and H. F. Hess, “Interferometric fluorescent super-resolution microscopy resolves 3D cellular ultrastructure,” Proc. Natl. Acad. Sci. U.S.A. 106(9), 3125–3130 (2009). [CrossRef] [PubMed]

9. D. Aquino, A. Schönle, C. Geisler, C. V. Middendorff, C. A. Wurm, Y. Okamura, T. Lang, S. W. Hell, and A. Egner, “Two-color nanoscopy of three-dimensional volumes by 4Pi detection of stochastically switched fluorophores,” Nat. Methods 8(4), 353–359 (2011). [CrossRef] [PubMed]

10. T. Ruckstuhl and D. Verdes, “Supercritical angle fluorescence (SAF) microscopy,” Opt. Express 12(18), 4246–4254 (2004). [CrossRef] [PubMed]

11. D. Axelrod, “Selective imaging of surface fluorescence with very high aperture microscope objectives,” J. Biomed. Opt. 6(1), 6–13 (2001). [CrossRef] [PubMed]

12. T. Ruckstuhl, J. Enderlein, S. Jung, and S. Seeger, “Forbidden light detection from single molecules,” Anal. Chem. 72(9), 2117–2123 (2000). [CrossRef] [PubMed]

13. T. Barroca, K. Balaa, J. Delahaye, S. Lévêque-Fort, and E. Fort, “Full-field supercritical angle fluorescence microscopy for live cell imaging,” Opt. Lett. 36(16), 3051–3053 (2011). [CrossRef] [PubMed]

14. J. Ries, T. Ruckstuhl, D. Verdes, and P. Schwille, “Supercritical angle fluorescence correlation spectroscopy,” Biophys. J. 94(1), 221–229 (2008). [CrossRef] [PubMed]

15. C. M. Winterflood, T. Ruckstuhl, D. Verdes, and S. Seeger, “Nanometer axial resolution by three-dimensional supercritical angle fluorescence microscopy,” Phys. Rev. Lett. 105(10), 108103 (2010). [CrossRef] [PubMed]

16. D. Axelrod, “Evanescent excitation and emission in fluorescence microscopy,” Biophys. J. 104(7), 1401–1409 (2013). [CrossRef] [PubMed]

17. T. Barroca, K. Balaa, S. Lévêque-Fort, and E. Fort, “Full-field near-field optical microscope for cell imaging,” Phys. Rev. Lett. 108(21), 218101 (2012). [CrossRef] [PubMed]

18. B. Rieger and S. Stallinga, “The lateral and axial localization uncertainty in super-resolution light microscopy,” ChemPhysChem 15(4), 664–670 (2014). [CrossRef] [PubMed]

19. J. Enderlein, T. Ruckstuhl, and S. Seeger, “Highly efficient optical detection of surface-generated fluorescence,” Appl. Opt. 38(4), 724–732 (1999). [CrossRef] [PubMed]

20. R. E. Thompson, D. R. Larson, and W. W. Webb, “Precise nanometer localization analysis for individual fluorescent probes,” Biophys. J. 82(5), 2775–2783 (2002). [CrossRef] [PubMed]

21. J. J. Schmied, C. Forthmann, E. Pibiri, B. Lalkens, P. Nickels, T. Liedl, and P. Tinnefeld, “DNA origami nanopillars as standards for three-dimensional superresolution microscopy,” Nano Lett. 13(2), 781–785 (2013). [CrossRef] [PubMed]

22. R. Iinuma, Y. Ke, R. Jungmann, T. Schlichthaerle, J. B. Woehrstein, and P. Yin, “Polyhedra self-assembled from DNA tripods and characterized with 3D DNA-PAINT,” Science 344(6179), 65–69 (2014). [CrossRef] [PubMed]

23. R. Jungmann, C. Steinhauer, M. Scheible, A. Kuzyk, P. Tinnefeld, and F. C. Simmel, “Single-molecule kinetics and super-resolution microscopy by fluorescence imaging of transient binding on DNA origami,” Nano Lett. 10(11), 4756–4761 (2010). [CrossRef] [PubMed]

24. R. Jungmann, M. S. Avendaño, J. B. Woehrstein, M. Dai, W. M. Shih, and P. Yin, “Multiplexed 3D cellular super-resolution imaging with DNA-PAINT and Exchange-PAINT,” Nat. Methods 11(3), 313–318 (2014). [CrossRef] [PubMed]

25. M. Heilemann, S. van de Linde, M. Schüttpelz, R. Kasper, B. Seefeldt, A. Mukherjee, P. Tinnefeld, and M. Sauer, “Subdiffraction-resolution fluorescence imaging with conventional fluorescent probes,” Angew. Chem. Int. Ed. Engl. 47(33), 6172–6176 (2008). [CrossRef] [PubMed]

26. I. Izeddin, M. El Beheiry, J. Andilla, D. Ciepielewski, X. Darzacq, and M. Dahan, “PSF shaping using adaptive optics for three-dimensional single-molecule super-resolution imaging and tracking,” Opt. Express 20(5), 4957–4967 (2012). [CrossRef] [PubMed]

27. M. D. Lew, M. P. Backlund, and W. E. Moerner, “Rotational mobility of single molecules affects localization accuracy in super-resolution fluorescence microscopy,” Nano Lett. 13(9), 3967–3972 (2013). [CrossRef] [PubMed]

28. A. Edelstein, N. Amodaj, K. Hoover, R. Vale, and N. Stuurman, “Computer control of microscopes using µManager,” Curr Protoc Mol Biol Chapter 14, Unit14.20 (2010).

29. I. Schoen, J. Ries, E. Klotzsch, H. Ewers, and V. Vogel, “Binding-activated localization microscopy of DNA structures,” Nano Lett. 11(9), 4008–4011 (2011). [CrossRef] [PubMed]

30. M. Raab, J. J. Schmied, I. Jusuk, C. Forthmann, and P. Tinnefeld, “Fluorescence microscopy with 6 nm resolution on DNA origami,” ChemPhysChem 15(12), 2431–2435 (2014). [CrossRef] [PubMed]

31. J. Ries, C. Kaplan, E. Platonova, H. Eghlidi, and H. Ewers, “A simple, versatile method for GFP-based super-resolution microscopy via nanobodies,” Nat. Methods 9(6), 582–584 (2012). [CrossRef] [PubMed]

32. C. S. Smith, N. Joseph, B. Rieger, and K. A. Lidke, “Fast, single-molecule localization that achieves theoretically minimum uncertainty,” Nat. Methods 7(5), 373–375 (2010). [CrossRef] [PubMed]

33. B. Huang, S. A. Jones, B. Brandenburg, and X. Zhuang, “Whole-cell 3D STORM reveals interactions between cellular structures with nanometer-scale resolution,” Nat. Methods 5(12), 1047–1052 (2008). [CrossRef] [PubMed]

3D superresolution microscopy by supercritical angle detection

Abstract

1. Introduction

2. Supercritical angle localization microscopy

3. Theoretical resolution

4. Results

4.1. Calibration

4.2. Validation using DNA origami tetrahedra

4.3. Application to biological samples

5. Discussion and outlook

6. Conclusion

7. Methods

7.1 Microscope setup

7.2 Calibration sample

7.3 DNA-PAINT measurements on DNA origami tetrahedra

7.4 Biological samples

7.5 Data analysis

7.5.1 Localization analysis

7.5.2 Calibration

7.5.3 Tetrahedra

Acknowledgments

References and links

Cited By

Figures (5)

Equations (9)

Optics Express