2.1.

Multiphoton Microscopy and Immersion Media

Data were obtained with a two-photon excited fluorescence (2PEF) microscope optimized for intravital imaging (Bergamo II Series; Thorlabs, Newton, NJ, United States). This system was configured in an upright fashion for live animal preparations. Image translation was facilitated by the movement of the microscope platform with motorized control along cartesian axes, piezo motor control along the optical axis, and an articulating body with up to $\pm 45\mathrm{deg}$ rotation of the optical axis relative to gravity. Femtosecond laser pulses were generated at 80 MHz from a tunable solid-state ytterbium-doped fiber laser with a usable power band between 680 and 1300 nm and with dispersion compensation (Insight X3; Newport Spectra-Physics, Milpitas, CA, United States). The beam was attenuated by rotating a $\lambda /2$ wave-plate relative to a polarizer, then routed onto the microscope platform and scanned by hybrid resonant-galvanometric mirrors, and relay-imaged to the back aperture of a 1.1 NA, 25× water-immersion objective with a working distance of 2 mm and coverslip correction collar to compensate for depth-dependent aberration (CFI75 Apochromat 25XC W; Nikon, Tokyo, Japan). 2PEF signal was collected by the same objective, reflected by a primary dichroic mirror (705 nm long-pass; Thorlabs), split into red and green channels with a secondary dichroic mirror (562 nm long-pass; Semrock, Rochester, NY, United States), spectrally cleaned with additional band-pass filters (Semrock), and relayed to photomultiplier tubes (PMTs; PMT2100 Series; Thorlabs). Microscope control, laser scanning, signal processing, and data acquisition were conducted using ThorImage LS (Thorlabs). Immersion media used in this study included water with a refractive index of 1.333 and surgical-grade HG. HG formulations used for experiments were 1% (w/v) sodium hyaluronate with molecular weights (MWs) of 1.8 or 2.5 million Da and zero-shear viscosities of $\sim 105\mathrm{Pa}$·s3 (Ophthalin, Zeiss, Dublin, CA, United States; Provisc, Alcon, Fort Worth, TX, United States, respectively), with refractive indices of 1.335 to 1.346, depending on the excitation wavelength.^42,46,47

2.2.

Animals

This study was carried out in strict accordance with National Institutes of Health regulations and the Institutional Animal Care and Use Committee at the University of California, San Francisco, United States. C57BL/6 wildtype mice (Jackson Laboratory, Bar Harbor, ME, United States) were used for deep neurovascular imaging and CX3CR1^eGFP/+ mice were used for timelapse imaging of retinal microglia expressing enhanced green fluorescent protein (EGFP).^48,49

2.3.

Intravital Imaging

For cortical imaging, C57BL/6 mice were anesthetized with 2% to 4% isoflurane in oxygen, stabilized with a custom stereotax, and implanted with cranial windows as previously described.^9,33 The vasculature was labeled with intravenous administration of Texas Red-dextran ($50\mu \mathrm{L}$, 1% w/v, MW = 70,000, Sigma-Aldrich, St. Louis, MO, United States) in saline. 2PEF was excited with femtosecond laser pulses centered at 750 nm and spectrally cleaned with a bandpass filter centered at 607 nm. Image stacks were acquired using either water- or HG-immersion. All other imaging parameters were kept identical including laser powers, PMT gain, $900\text{-}\mu \mathrm{m}$ stacks collected in $2\text{-}\mu \mathrm{m}$ steps, and in the same cortical region within the same animal.

For ocular imaging, ${\mathrm{CX}3\mathrm{CR}1}^{\mathrm{eGFP}/+}$ mice were anesthetized with 2% to 4% isoflurane in oxygen, stabilized with a custom stereotax, and the upper eyelid was retracted with a suture (6-0, Silk, Ethicon, Raritan, NJ, United States). A second suture was placed at the superior limbus and used to introduce the globe to expose the superior sclera. For water-immersion imaging, a round 5-mm-diameter coverslip was placed on the eye to create a retention surface and water was added between the coverslip and objective as needed. HG-immersion imaging was performed with and without the coverslip, and the gel was administered once at the beginning of an imaging session. EGFP fluorescence was excited with femtosecond laser pulses centered at 930 nm and spectrally cleaned with a bandpass filter centered at 525 nm. Timelapse image stacks were acquired in 15 s intervals using either water or HG immersion. The image focus was $\sim 120\mu \mathrm{m}$ beneath the ocular surface, and serial stacks of $30\text{-}\mu \mathrm{m}$ depth were collected in $2\text{-}\mu \mathrm{m}$ steps to encompass selected microglial cells in the retina. All optical parameters were kept the same between experiments and across all timepoints including laser powers and PMT gain.

HG was administered into each preparation using a syringe and cannula to avoid air bubbles. After imaging, care was taken to gently clean the objective lens with methanol and lens paper to remove HG before drying. Importantly, once dry, HG formed a persistent residue that was more difficult to clean than typical immersion media. If HG did dry on the objective lens, the tip of the lens was submerged in water for several minutes before being gently wiped once with methanol and lens paper. This process was repeated several times until the HG residue was fully removed. Care was taken to avoid rubbing which might scratch the lens.

2.4.

Resolution and Signal-to-Background Analyses

To evaluate depth-dependent resolution, fluorescent $0.1\mu \mathrm{m}$ beads (TetraSpeck Microspheres; ThermoFisher, Waltham, MA, United States) were suspended in 0.6% agarose gel and $30\mu \mathrm{m}$ image stacks were acquired in $0.5\mu \mathrm{m}$ steps at depths of 100, 600, and $1100\mu \mathrm{m}$ below the surface of the agarose gel, using either water or HG immersion. HG immersion media was 1% sodium hyaluronate with an approximate refractive index of 1.335 at 1040 nm excitation wavelength.⁴² Image stacks were repeated using different correction collar settings on the objective lens to empirically compensate for depth- and media-dependent spherical aberration. Other optical parameters were kept identical including laser powers and excitation centered at 1040 nm, PMT gain, and region of bead suspension. Axial and lateral point-spread-functions (PSFs) of the beads were analyzed using MATLAB code as previously described^50–52 where axial and lateral spatial resolutions were defined as the average full-width half-maximum of the axial and lateral PSFs of captured beads, respectively.

Depth- and time-dependent signal-to-background ratios (SBRs) were calculated for intravital imaging and compared between water- and HG-immersion media. For cortical imaging, SBR was determined at varying depths using $40\text{-}\mu \mathrm{m}$ projections centered at 100, 300, 500, 700, and $900\mu \mathrm{m}$ beneath the cortical surface. Mean pixel intensity in five regions-of-interest (ROIs) selecting true vascular signal was divided by mean pixel intensity from five ROIs selecting background signal to determine mean SBR at each depth. For retinal timelapse imaging, $30\text{-}\mu \mathrm{m}$ projections were centered on the inner retinal layer to capture microglial cells. The mean pixel intensity from five ROIs selecting EGFP signal in the cell bodies of microglia was divided by the mean pixel intensity from five ROIs selecting background signal to determine mean SBRs at each timepoint. ROIs were standardized to circles of 10 pixels in diameter.

3.1.

Depth-Dependent SBR in Mouse Cortical Vasculature

Intravital 2PEF imaging of the mouse cortex revealed equivalent depth-dependent SBR when imaging with water- or HG-immersion media (Fig. 1). The objective’s correction collar was set to 0.17 mm corresponding to the coverslip thickness of the cranial window. SBR decreased with imaging depth without a significant difference between imaging with water or HG, indicating that HG and water are interchangeable as immersion media without degradation of quality when imaging cortical vasculature.

Fig. 1

Comparison of depth-dependent 2PEF image quality in mouse cortex with water- or HG-immersion media. (a) $900\mu \mathrm{m}$ stack of the mouse cortical vasculature imaged with water immersion. (b) $900\mu \mathrm{m}$ stack of the mouse cortical vasculature imaged with HG immersion. (c) Average SBR calculated from optical sections at depths of 100, 300, 500, 700, and $900\mu \mathrm{m}$ with water or HG immersion (error bars = ±1 SD). (d) Representative images of the cortex at depths of 100, 300, 500, 700, and $900\mu \mathrm{m}$ with water immersion. (e) Representative images of the cortex at depths of 100, 300, 500, 700, and $900\mu \mathrm{m}$ using HG immersion. $\text{Scale bars}=100\mu \mathrm{m}$.

3.2.

Time-Dependent SBR Ratio in Mouse Retinal Microglia

Timelapse imaging with water demonstrated a complete drop-off in SBR within 30 min [Fig. 2(a) and 2(c), Video 1], whereas timelapse imaging with HG had a gradual drop-off in SBR over the course of 1 h where microglia and their subcellular processes were still visible at the final timepoint [Figs. 2(b)–2(c), Video 2]. Immersion media was not replenished in this experiment, and the decrease in signal was likely attributable to wicking and evaporative loss of immersion fluid. SBR with water immersion was slightly higher than HG immersion at the beginning of timelapse imaging. Notably, the correction collar of the objective lens was set to 0.17 mm per the manufacturer’s specifications, corresponding to the glass coverslip thickness when imaging under water immersion. We therefore sought to quantitatively evaluate image quality with either water or HG immersion at various imaging depths and to determine if this could be optimized by adjusting the correction collar on the microscope objective.

Fig. 2

Comparison of time-dependent 2PEF image quality in mouse retina with water- or HG-immersion media. (a) Timelapse imaging of retinal microglia with water immersion, demonstrating a drop-off in SBR to zero within 30 min. (b) Timelapse imaging of retinal microglia with HG immersion, demonstrating gradual drop-off in SBR over the course of 1 h with discernable microglia at the final timepoint. (c) Average SBR calculated from images at serial timepoints using water- or HG-immersion media (error bars = ± 1 SD). $\text{Scale bars}=40\mu \mathrm{m}$ (Video 1, MP4, 7.26 MB [URL: https://doi.org/10.1117/1.NPh.12.S1.S14602.s1]; Video 2, MP4, 14.1 MB [URL: https://doi.org/10.1117/1.NPh.12.S1.S14602.s2]).

3.3.

Depth-Dependent Resolution

Peak lateral and axial resolutions were equivalent using water and HG at all depths with optimal correction collar settings. Correction collar settings are reported in units of mm corresponding to the manufacturer’s intended use with varying cover glass thicknesses. For water immersion, peak resolutions were achieved with a correction collar setting of 0.07 mm. Lateral resolutions in water peaked at 0.48, 0.51, and $0.55\mu \mathrm{m}$ at depths of 100, 600, and $1100\mu \mathrm{m}$, respectively. Axial resolutions in water peaked at 1.57, 1.54, and $1.41\mu \mathrm{m}$, at depths of 100, 600, and $1100\mu \mathrm{m}$, respectively (Fig. 3). For HG immersion, peak resolutions were achieved with a correction collar setting of 0.09 mm. Lateral resolutions in HG peaked at 0.49, 0.51, and $0.54\mu \mathrm{m}$ at depths of 100, 600, and $1100\mu \mathrm{m}$, respectively. Axial resolutions in HG peaked at 1.6, 1.42, and $1.37\mu \mathrm{m}$, at depths of 100, 600, and $1100\mu \mathrm{m}$, respectively (Fig. 3). These results indicate that the replacement of water with HG as an immersion medium enables equivalent imaging resolution when optimized with a small adjustment in correction collar settings.

Fig. 3

Comparison of lateral and axial resolution at varying depths and objective correction collar settings. PSFs of $0.1\mu \mathrm{m}$ beads were analyzed to obtain lateral and axial resolutions at depths of 100, 600, and $1100\mu \mathrm{m}$ and with different correction collar settings. Peak resolutions were equivalent for HG and water at every depth when optimized with the correction collar. Correction collar settings are reported in units of mm corresponding to the manufacturer’s intended use with varying cover glass thicknesses (error bars = ± 1 SD).

3.4.

Stability of Water and Hyaluronan Gel

HG demonstrated enhanced stability in configurations that are traditionally challenging with water. When applied on glass slides, HG remained in its initial position without moving even when oriented 90 deg from horizontal, whereas water consistently drifted to the edge of the slides at oblique angles [Figs. 4(a) and 4(b)] (Fig. S1 in the Supplementary Material) In an inverted orientation, HG showed no displacement and only a slight reduction in volume due to evaporation after 30 min, whereas water dripped or evaporated quickly from the slide [Figs. 4(c) and 4(d)] (Fig. S2 in the Supplementary Material). HG was easily maintained between a glass slide and a water-immersion objective separated by a distance of 3 mm and oriented 90 deg from horizontal, a condition not possible with water [Fig. 4(e)] (Fig. S3 in the Supplementary Material). On mouse fur, HG was stable and retained its cohesive structure, whereas water was prone to wicking and absorption [Fig. 4(f)]. Figure 4(g) presents an example of HG application for ocular imaging at an oblique angle with limited surface area and surrounding fur, a preparation unamenable to water. Figure 4(h) illustrates the high stability and cohesion of HG on the tip of a metal screw with limited surface area.

Fig. 4

Comparative stability of HG and water. Adhesion of a droplet of (a) water and (b) HG to a glass slide at 0, 60, and 90 deg from 3 mm, respectively. Adhesion of (c) water and (d) HG to a glass slide inverted to gravity for up to 30 min. Adhesion of (e) water and HG between a glass slide and water-immersion objective separated by 3 mm. (f) Wicking of water and HG on mouse fur, demonstrating resistance of HG to wicking. (g) Example of HG usage with an intravital imaging preparation unamenable to water. (h) Comparison of water and HG cohesion and stability on geometry with a small surface area.