Construction and use of an adaptive optics two-photon microscope with direct wavefront sensing

Abstract

Two-photon microscopy, combined with the appropriate optical labelling, enables the measurement and tracking of submicrometer structures within brain cells, as well as the spatiotemporal mapping of spikes in individual neurons and of neurotransmitter release in individual synapses. Yet, the spatial resolution of two-photon microscopy rapidly degrades as imaging is attempted at depths of more than a few scattering lengths into tissue, i.e., below the superficial layers that constitute the top 300-400 µm of the neocortex. To obviate this limitation, we shape the focal volume, generated by the excitation beam, by modulating the incident wavefront via guidestar-assisted adaptive optics. Here, we describe the construction, calibration and operation of a two-photon microscope that incorporates adaptive optics to restore diffraction-limited resolution at depths close to 900 µm in the mouse cortex. Our setup detects a guidestar formed by the excitation of a red-shifted dye in blood serum, used to directly measure the wavefront. We incorporate predominantly commercially available optical, optomechanical, mechanical and electronic components, and supply computer-aided design models of other customized components. The resulting adaptive optics two-photon microscope is modular and allows for expanded imaging and optical excitation capabilities. We demonstrate our methodology in the mouse neocortex by imaging the morphology of somatostatin-expressing neurons that lie 700 µm beneath the pia, calcium dynamics of layer 5b projection neurons and thalamocortical glutamate transmission to L4 neurons. The protocol requires ~30 d to complete and is suitable for users with graduate-level expertise in optics.

Figure 1 |. Concepts of adaptive optics and wavefront sensing.

a, Schematic diagram illustrating the propagating wavefront of the excitation beam in an aberration-free imaging system in which the beam focused as a diffraction limited spot. b, Aberrated wavefront as it propagates through the inhomogeneous imaging sample, resulting in an enlarged and distorted focus. c, Compensating the aberration by the phase modulator to recover the diffraction limited focus. d, Acquiring the desired wavefront for the phase modulator by direct wavefront sensing of the aberrated wavefront that generates by a point source in the sample.

Figure 2 |. Concepts of Shark-Hartmann wavefront sensing.

a, Diagram showing that the aberrated wavefront is split into segments by the lens array and focus on the EMCCD respectively. b, Diagram demonstrating that the wavefront can be estimated by calculating the slope of each segment based on the deflection of the focal spot. Δy, the distance of the deflected spot from the central axis of the microlens; f, the focal length of a microlens.

Figure 3 |. Schematic diagram of AO-TPLSM.

a, Schematic diagram of the entire system showing optics and beam path of the AO-TPLSM. DM, deformable mirror; SHWS, Shark-Hartmann wavefront sensor; TS, translational stage; M, mirror; L, lens; D, dichroic mirror; SP filter, shortpass filter; BP filter, bandpass filter; GS, galvo scanner; RS, resonant scanner; fs laser, femtosecond laser; MPPC, multi-pixel photon counters. b, Schematic diagram showing only the optics that are involved in descanning of the guide star signal.

Figure 4 |. Implementation example of AO-TPLSM.

Computer model of AO-TPLSM. Detailed technical drawings are provided in Supplementary Data 1.

Figure 5 |. Overview of the procedure.

Flowchart illustrating the essential steps for constructing, aligning, and calibrating the AO-TPLSM as well as showing the main steps for conducting an imaging experiment with AO-TPLSM.

Figure 6 |. Concepts of collimation and conjugation.

a, Diagram demonstrating that the divergence of the beam passing through a telescope can be controlled by adjusting the distance between the lens pair. b, Diagram illustrating the concepts of conjugation. The conjugate of a given plane P₀ is defined as the plane P_c where the points on P₀ are imaged. Upper panel shows a 4-f system that the front focal plane of the first lens and the back focal plane of the second lens are conjugate to each other. Middle and lower panel shows that adjusting the distance between the lenses of a telescope does not change the conjugation of the two focal planes. We applied this feature to the design the of relay between the DM and the scanner to orthogonalize the collimation and conjugation.

Figure 7 |. Conjugating the pupil plane with the scanners.

a, Diagram for finding the objective pupil plane that conjugates to the scanners. b, Real item image of the iris at the pupil plane taken through the IR viewer. The beam passing through the iris center (arrow) shows that the excitation beam is well-aligned to the back aperture of the objective. While the scanner is on, little movement of the bright ring should be observed if the iris is conjugate to the scanners.

Figure 8 |. Conjugating the DM to the pupil plane.

a, Schematics for finding the DM location along the optical axis that conjugates to the pupil plane. The objective is replaced by a camera with its sensor at the pupil plane. The deformable mirror is replaced by a mirror with scratch for the camera to image. b, Real item image of the scratched mirror we used to replace the DM. c, Image of the scratched mirror taken by the camera at the pupil plane. The sharp grid pattern indicates that the mirror is at the conjugates plane of the pupil.

Figure 9 |. Conjugating SHWS to the pupil plane.

a, Schematics showing the setup for finding the plane in the wavefront sensing path that conjugates to the pupil. The laser shutter was closed in this step. Another light source (e.g., flashlight) is needed to illuminate the paper target at the pupil plane. b, Sharp image of the target, indicating that the EMCCD sensor is conjugated to the pupil plane.

Figure 10 |. AO Calibration preparation.

a to f, Images obtained on the EMCCD before the AO Calibration. Left column, diagram showing the light source for the EMCCD images. a to c was taken with the Cy5.5 fluorescence, d to f used the femtosecond laser beam reflected by a mirror at the pupil plane. Upper row, diagram showing the position of the EMCCD along the optical axis. a and d were taken by the EMCCD at the conjugate plane (P1). c and f were taken by the SHWS of which the microlens array is at the conjugate plane (P2). b and e were taken with EMCCD at the same position as c and f (P2) but no microlens array was placed on the light path. The edge of bright disk in a was indicated by a red circle in image a, b, d, and e, showing that the bright disks in the four images are concentric. g, Merge of c and f. It is critical to the AO calibration that a, b, d, and e are concentric, and that c and f are overlapped.

Figure 11 |. AO Calibration.

a, Schematics illustrating the AO calibration. b, Flowchart of the AO calibration procedure. It corresponds to the MATLAB codes “DM_SHWS_calibration_Ctr.m” and “DM_SHWS_calibration.m”.

Figure 12 |. Configuration for System aberration calibration.

a, Schematics illustrating the configuration for system aberration calibration. b, Flowchart of the system aberration calibration procedure. It corresponds to the MATLAB codes “sensorlessWF_sys_abr_descend.m”.

Figure 13 |. Results for system aberration calibration.

Representative outcomes of system aberration calibration for the excitation beam in 930 nm (upper row), 1030 nm (middle row), and 1250 nm (bottom row). Left column, showing the changes of Zernike coefficients as a function of optimization cycles. Middle column, Zernike coefficients after 50 optimization cycles. Right column, wavefront phase map on the DM that compensates for the system aberration.

Figure 14 |. Point spread function after system aberration correction.

Point spread function measured from a 200-nm diameter fluorescent bead after the correcting the system aberration. Scale bar, 1 μm. Upper row, image of the X-Y plane with the signal profile showing the lateral resolution. Bottom row, image of the X-Z plane with the signal profile, showing the axial resolution.

Figure 15 |. System verification by correcting the aberration of the double-layer cover glass.

a to c, Images of a fiber stained with rhodamine B. Scale bar, 10 μm. a, Images taken without any AO correction. b, Images taken after correcting the system aberration. c, Images taken with both system aberration and sample aberration corrected. d, Signal profiles along the line in a to c. Scale bar, 2 μm. e, Spot patterns formed by the SHWS from the descanned guide star signal for inferring the sample aberration. f, Reconstructed wavefront that applied to the DM. g, Zernike coefficients of the reconstructed wavefront.

Figure 16 |. In vivo experiment workflow.

a, Schematics of the guide star (Cy5.5-dextran) delivery by retro-orbital injection. b, Schematic diagram of the AO correction. Spot patterns formed by the SHWS from the descanned guide star signal in the capillary, together with the sample reference pattern and AO calibration results of the DM Zernike modes, are used for calculating the DM pattern for correcting the sample aberration. The sample aberration correction pattern is added to the system aberration pattern for full AO correction.

Figure 17 |. Morphological imaging with AO correction.

a,b, SST⁺ neurons in mouse somatosensory cortex. Images were acquired at 675 – 690 μm below the pia with system (a) or full AO correction (b) using excitation wavelength λ = 1030 nm. The SST⁺ neurons were labeled with tdTomato. Scale bar, 20 μm. c, Signals profiles along the lines in a and b. Scale bar, 1 μm. d and e, Spectral power as a function of spatial frequency k for the images in a and b. f, |k|-space plot of the spatial frequency with system AO and full AO correction. g,h, Spot patterns on the SHWS (g) and reconstructed wavefront phase map on DM (h) for correcting the sample aberration in the a. g, Zernike coefficients of the reconstructed wavefront.

Figure 18 |. Functional imaging AO correction.

a, In vivo two-photon imaging of the layer 5 neurons of vS1 cortex which were labelled with jRGECO1a. Data were obtained at 630 μm below the pia with full AO correction using excitation wavelength λ = 1030 nm. b, Calcium signal in the ROIs defined by the dashed circles in panel a. c, In vivo two-photon imaging of the thalamocortical boutons which were labelled with iGluSnFR3. Data were obtained at layer 4 of vS1 cortex with full AO correction using excitation wavelength λ = 970 nm. d, Glutamate dynamics responding to 5 Hz air puff in the ROIs defined by the dashed circles in panel c. The trace is averaged from 60-s recording.