Color-neutral and reversible tissue transparency enables longitudinal deep-tissue imaging in live mice

Abstract

Light scattering in biological tissue presents a significant challenge for deep in vivo imaging. Our previous work demonstrated the ability to achieve optical transparency in live mice using intensely absorbing dye molecules, which created transparency in the red spectrum while blocking shorter-wavelength photons. In this paper, we extend this capability to achieve optical transparency across the entire visible spectrum by employing molecules with strong absorption in the ultraviolet spectrum and sharp absorption edges that rapidly decline upon entering the visible spectrum. This color-neutral and reversible tissue transparency method enables optical transparency for imaging commonly used fluorophores in the green and yellow spectra. Notably, this approach facilitates tissue transparency for structural and functional imaging of the live mouse brain labeled with yellow fluorescent protein and GCaMP through the scalp and skull. We show that this method enables longitudinal imaging of the same brain regions in awake mice over multiple days during development. Histological analyses of the skin and systemic toxicology studies indicate minimal acute or chronic damage to the skin or body using this approach. This color-neutral and reversible tissue transparency technique opens opportunities for noninvasive deep-tissue optical imaging, enabling long-term visualization of cellular structures and dynamic activity with high spatiotemporal resolution and chronic tracking capabilities.

Keywords: Kramers–Kronig relations; deep tissue imaging; optical transparency; two-photon microscopy.

Fig. 1.

Optical characterizations of ampyrone. (A) Comparison of the mass extinction coefficients of phenazone, ampyrone, and tartrazine solutions as a function of wavelength, highlighting the distinct positions of their absorption edges. A 100 mg/mL aqueous solution was used for each spectrum. (B) The real RI (n) of an aqueous solution of ampyrone as a function of concentration and wavelength. (C) The real RI (n) of an aqueous solution of ampyrone as a function of concentration at selected wavelengths. (D) The imaginary RI (k) of an aqueous solution of ampyrone as a function of concentration and wavelength. (E) Comparison of k for various index-matching agents in aqueous solutions at 310 nm as a function of concentration. (F) Comparison of the real RI (n) for the same index-matching agents shown in E as a function of concentration at 310 nm. The Dextran data had to be linearly extrapolated up to 50% w/w since high concentrations are not achievable due to its solubility limit. Data in C, E, and F are shown as mean ± SD from three wavelength-dependent measurements.

Fig. 2.

Resolution characterization of scattering phantoms. (A) Phantoms of 1 μm silica beads in 5-mm-thick scattering phantoms composed of agarose hydrogel with increasing concentrations of ampyrone. (Scale bar, 5 mm.) (B) 1951 USAF resolution test target images through 1 μm silica beads in 2-mm-thick scattering phantoms. (Scale bar, 1.2 mm.) (C) Same test target images as in B, but zoomed in to display the smallest features. (Scale bar, 296 μm.) (D) Modulation transfer functions (MTFs) for all concentrations of ampyrone used in B and C with the same color coding by concentration. Each subplot shows the MTFs at different concentrations for a single wavelength.

Fig. 3.

Achieving optical transparency in ex vivo mouse skin. (A) Mouse abdominal skin before (Top) and after (Bottom) soaking in ampyrone solutions at various concentrations. (B) Transmission spectra through the same four mouse skin samples as in A after soaking. (C) The ratio of the transmission after soaking to before soaking for the same four mouse skin samples. (D) The area change after soaking for the same four skin samples is shown as a function of concentration. (Scale bar, 5 mm.)

Fig. 4.

Achieving optical transparency in the live mouse abdomen. (A) Mouse abdomen before treatment with ampyrone solution. (B) Mouse abdomen shown after treatment with ampyrone solution. (C) The same mouse abdomen in A and B after dissection. (Scale bar, 5 mm in A–C.) (D) Images showing the abdominal area of the same mouse before treatment, after achieving a transparent window with ampyrone, after reversing the transparency effect, and hair regrowth on the subsequent days. (Scale bar, 5 mm in D.)

Fig. 5.

Longitudinal neuron structural imaging through transparent scalp. (A) 3D reconstruction of two-photon excited YFP-H fluorescence in the live mouse cortex before treatment with ampyrone. Only signals in the scalp can be seen due to the scattering of the scalp. (B) 3D reconstruction of two-photon excited YFP-H fluorescence of the same region as A after achieving scalp transparency with ampyrone. The scalp, skull, and brain (to the imageable depth) are labeled in the figure. Their approximate thicknesses are 150 μm for the scalp, 100 μm for the skull, and 400 μm for the imageable depth of the brain, respectively. (C and D) Images of the same brain region before treatment with ampyrone at 300 μm and 400 μm beneath the cortical surface, respectively. (E and F) Images of the same brain region after achieving scalp transparency with ampyrone at 300 μm and 400 μm beneath the cortical surface, respectively. (G–J) 3D reconstruction of two-photon excited YFP-H fluorescence of the same region in layer 1 of the primary visual area (V1) through the transparent window in the scalp on P22-P25. Yellow and blue colors indicate YFP and SHG, respectively. (K–N) YFP images of the same region in layer 1 of V1 at a depth of 250 μm below the surface of the scalp on P22-25. The same vascular landmarks, appearing as dark linear features, are present in all images, confirming longitudinal imaging in the same mouse brain. (Scale bar, 100 μm in A–F and 200 μm in G–N.)

Fig. 6.

Functional neuron activity imaging with GCaMP through transparent scalp. (A) 3D reconstruction of two-photon excited GCaMP8m fluorescence in the mouse cortex before treatment with ampyrone. Only signals in the scalp can be seen due to the scattering of the scalp. (B) 3D reconstruction of the same GCaMP8m-labeled mouse cortex after achieving transparency in the scalp. (C) Image from the same ROI as A at 170 μm depth from the surface of the scalp. (D) Image from the same ROI as B at 170 μm depth showing dense cell bodies. (E) A representative time-frame image of the cortex of a GCaMP6f-labeled mouse at 200 μm depth from the surface of the scalp. Representative neurons are labeled with white circles. (F) Dynamic time traces of GCaMP6f fluorescence intensity corresponding to the labeled neurons in E. Red lines indicate the stimulus applied to the mouse. (Scale bar, 100 μm in A and B, 200 μm in C and D, and 400 μm in E.) The different strain, larger FOV, use of awake mice, and lower laser power used for functional imaging in E are responsible for the lower resolution of cell bodies in E compared to D.

Fig. 7.

Proteomics results of mouse abdominal skin. (A) Schematic of protein extraction from mouse abdominal skin following a 72 h recovery period post treatment with ampyrone. (B and C) Volcano plot depicting the female (B) and male (C) mouse proteomics results with potential apoptotic markers highlighted. Dashed, horizontal lines correspond to the 0.05 P-value threshold for significance and vertical lines correspond to the thresholds for fold changes. The fold change is defined as (final value – original value)/(original value).