Brillouin Gain Microscopy

Optical imaging with mechanical contrast is critical for material and biological discovery since it allows contactless light-radiation force-excitation within the sample, as opposed to traditional mechanical imaging. Whilst optical microscopy based on stimulated Brillouin scattering (SBS) enables mechanical imaging of materials and living biological systems with high spectrospatial resolution, its temporal resolution remains limited. Here, we develop Brillouin gain microscopy (BGM) with a 200-fold higher temporal resolution by detecting the Brillouin gain at a mechanically contrasting frequency in the sample with high sensitivity. Using BGM, we demonstrate mechanical imaging of materials, living organisms and cells at high spectro-spatiotemporal resolution.


Introduction
Optical mechanical imaging techniques are complementary to atomic force microscopy (AFM), which has traditionally been applied for mechanical research and discovery in many fields including biology and material sciences.Whilst AFM allows for two-dimensional mechanical measurements on the sample surface at nanometer scale 1,2 , optical mechanical microscopy facilitates cross-sectional imaging inside the sample along all three dimensions at (sub)micrometer scale.Among the established optical techniques used for imaging materials and living biological systems with mechanical contrast are optical coherence elastography (OCE) [3][4][5] and second harmonic generation (SHG) microscopy [6][7][8][9] .OCE is based on the application of a mechanical load to the sample and measurement of the subsequent deformation by speckle tracking or phase-sensitive detection.Whereas OCE achieves ~5-15 m resolution and impressively fast acquisition times (several seconds to tens of minutes, depending on the imaged field-of-view and pixel density), it involves intensive image acquisition and processing, and the use of specialized instrumentation for external mechanical stimulation and force sensing.SHG microscopy circumvents the need for an external mechanical load (i.e., is all-optical) while offering high spatiotemporal resolution, however, its applicability is limited to a small number of structural proteins.Stimulated Brillouin scattering (SBS) microscopy is emerging as a promising method for alloptical mechanical imaging with subcellular spatial resolution, high mechanical specificity (~100-MHz spectral resolution), and high sensitivity (fractional intensity of ~ − in 100 s in biological settings) 10 .In SBS microscopy, the interaction of the excitation light at the pump and Stokes frequencies, P and S, with a propagating coherent acoustic wave (acoustic phonon) at a frequency B = P − S of acoustic vibration in the sample results in an increase of the Stokes beam intensity (Fig. 1, A and B).By scanning the difference frequency (t)= P − S(t) over  in time, the so-called stimulated Brillouin gain (SBG) spectrum is obtained (Fig. 1B in gray line).To measure the SBG at , we use a high frequency phase-sensitive detection scheme by modulating the pump beam intensity at 1.1 MHz and detecting the SBG in the Stokes beam with a lock-in amplifier.Suppression of the unwanted pump back-reflections from the sample into the detector is accomplished by cleaning the optical noise in the forward-going, unmodulated pump beam with a Bragg grating and blocking of the back-reflected, and the modulated pump beam by an atomic filter (Fig. 1C).For a known refractive index of the sample, the Brillouin frequency shift, the Brillouin linewidth, and the peak Brillouin gain of the SBG spectrum are directly related to the acoustic speed, acoustic attenuation, and mass density in the sample.Using this acoustic information, the complex longitudinal modulus of the material constituents in the sample can be determined at gigahertz frequencies.Spontaneous Brillouin scattering microscopy based on light diffraction from thermal acoustic phonons can also be used for all-optical mechanical imaging, but it trades-off acquisition time for spectral and/or spatial resolution [11][12][13][14][15][16][17][18][19][20][21] .Despite the substantial improvements enabled by SBS microscopy for optical mechanical imaging, its temporal resolution is still limited to a few tens of milliseconds under biocompatible excitation power.
In general, Brillouin microscopes are supposed to measure the entire Brillouin spectrum in each image pixel to extract B and therefore to obtain mechanical information about the sample.In SBS microscopy, this requires the sweeping of the Stokes frequency S⎯a timeconsuming process that significantly limits the temporal resolution.For achieving high speed imaging, we develop here Brillouin gain microscopy (BGM), which measures the Brillouin gain at a single, mechanically contrasting frequency of the sample C similar to stimulated Raman scattering (SRS) microscopy [22][23][24] .For example, in a sample with a single material constituent, C can be chosen to be at the Brillouin frequency shift of the material B (Fig. 1B).Due to the detection of unwanted pump back-reflections from the sample, there is a significant drift in the baseline level of the SBG spectra acquired by SBS microscopy at different locations in the sample (Fig. 1D).This baseline drift fundamentally impedes the accurate measurement of the Brillouin gain at C.We solve this problem in BGM by designing an optimized optical filter based on a greatly simplified, free-running, temperaturecontrolled solid etalon (Methods, Note S1, and Fig. S1) that improves the suppression of the integrated optical noise power in the forward-going, unmodulated pump beam by ~20 dB with an insertion loss of ~0.75 dB (Fig. 1E).This optimization is necessary to straighten the baseline with minimal pump power loss, enabling the accurate and precise measurement of the Brillouin gain at C (Fig. S2).While our previous SBS microscope could not accurately measure the Brillouin gain at C/2 = B water /2 = 5.03 GHz across a glass-water interface, the new Brillouin gain microscopy (BGM) system successfully acquired this data with temporal resolution of 100 s⎯200-fold smaller than the pixel-dwell time required to measure a whole SBG spectrum (Fig. 1F and Fig. S3).Effectively corrected baselines were also obtained in more complex samples, including living C. elegans nematodes and NIH/3T3 cells (Fig. S4).The overall power on the sample was ~228 mW.Owing to the significantly reduced light exposure during data acquisition, phototoxicity is substantially lower than with our previous SBS microscope 10 .
As the first application, we used BGM for material imaging of phosphate-buffered saline (PBS) on polyacrylamide (PAA) gel (Fig. 2A and Methods).The SBG spectrum of PBS has a peak at B PBS /2 = 5.06 GHz, whereas PAA has a Brillouin peak at B PAA /2 = 5.39 GHz (Fig. 2B), attributable to the stiffer nature of PAA compared with PBS. Figure 2, C and D show the BGM images obtained by the acoustic vibration frequencies of the PBS and PAA, respectively (B PBS and B PAA ).Contrast using B PBS highlights the top layer of the sample (PBS) with positive contrast and the bottom layer (PAA) with negative contrast.The opposite occurs by tuning in to B PAA .The merged image of the PBS and PAA channels clearly shows the two layers of the material system (Fig. 2E).While the use of either B PBS or B PAA provides a significant difference between the materials, setting the mechanically contrasting frequency C to B PAA offers a higher contrast between PBS and PAA than B PBS (Fig. 2F).This result is reasoned by the asymmetric aperture-induced spectral broadening of Brillouin peaks 25 .
Next, we present in vivo BGM imaging of the head of a C. elegans nematode, which serves as an important multicellular model organism in biological research (Fig. 3 and Methods).
The pharynx of the nematode exhibits a Brillouin peak at B pharynx /2 = 5.53 GHz, whereas the surrounding tissue of the nematode's head has a Brillouin peak at B surrounding tissue /2 = 5.3 GHz, implying that the pharynx is stiffer than the surrounding tissue probably due to the muscular nature of the pharynx 10 (Fig. 3A).The tissue around the pharynx can be seen with positive contrast and the pharynx with negative contrast when tuned in to the 5.3-GHz acoustic vibration frequency (Fig. 3B).The corpus, isthmus, and terminal bulb of the pharynx are clearly exposed together with the pharyngeal lumen and grinder.Contrast coming from the pharynx channel (5.53 GHz) is also available (Fig. 3C) although strong back-reflected pump signals off interfaces of the pharynx and the surrounding tissue appear to somewhat increase the SBG artificially at these locations.Figure 3D illustrates the dual-color overlay of the surrounding tissue and pharynx channels showing the stiffer pharynx in cyan and the softer surrounding tissue in red.Another useful mechanically contrasting frequency C is situated near the middle of the slope of the SBG spectrum of the pharynx (C/2 = 5.7 GHz as depicted in Fig. 3A).At this frequency, changes in the Brillouin frequency shift are efficiently converted to SBG variations owing to mechanically different constituents of the sample.The resultant BGM image shows high contrast between the pharynx and the surrounding tissue regions (Fig. 3E), as explained by the greater visibility between the mean SBGs of these two regions at 5.7 GHz compared to 5.3 GHz and 5.53 GHz (Fig. 3F).
Another ability of BGM is to image biological cells with mechanical contrast at subcellular resolution despite the closely overlapping Brillouin bands of different components in the cell (Fig. 4). Figure 4A displays the SBG spectra of the cell nucleoplasm and nucleolus in a living NIH/3T3 cell (Methods), which is a widely used fibroblast cell line in biological studies.
These cell components show Brillouin peaks at B nucleoplasm /2 = 5.25 GHz and B nucleolus /2 = 5.4 GHz (i.e., only 150 MHz frequency difference), suggesting a stiffer nucleolus.By tuning in to B nucleoplasm , contrast from regions outside the nucleoli appeared so that the nucleoli could be identified (Fig. 4B).The nucleoli are hardly seen when tuned in to the nucleolus Brillouin peak at B nucleolus due to its significant overlap with the nucleoplasm band (Fig. 4, A and C).The dual-color overlay image (Fig. 4D) shows the two nucleoli, the nucleoplasm (light gray region), and the cytoplasm (dark gray region).Figure 4E shows that high contrast is available between these three regions when the mechanically contrasting frequency C is tuned near the middle of the slope of the SBG spectrum of the nucleolus (C/2 = 5.7 GHz as illustrated in Fig. 4A).Whereas the contrast coming from the acoustic vibration frequencies 5.25 GHz, 5.4 GHz, and 5.7 GHz yields statistically different cell regions, visibility is best at 5.7 GHz (Fig. 4F).
BGM overcomes the frequency-space-time tradeoff challenge in all-optical mechanical imaging to provide SBS microscopy with intrinsic mechanical contrast at high spectrospatiotemporal resolution.BGM can now be used in various applications in the material and biological sciences, including those involving living organisms and cells.As the lowfrequency counterpart of SRS microscopy, BGM is likely to become an increasingly powerful tool in material and biological research and discovery.

Temperature-controlled solid etalon
The solid etalon (LightMachinery) was mounted in a custom-made aluminum mount, placed on a 5-axis stage (Thorlabs).A thermoelectric cooler was positioned on top of the mount, above a conductive tape, with a heat sink on top of it (Fig. S1A).The mount was covered with insulating pads.A thermistor probe measured the temperature of the etalon directly through a hole on the side of the mount.The pump beam traveled through the etalon twice using a prism (Foctek Photonics) and a PID temperature controller (Thorlabs) kept the temperature of the etalon stable.The device was covered by an environmentally insulating box.The transmission response and frequency stability of the two-pass etalon was measured showing a transmission peak spectral separation, the so-called free spectral range (FSR), of 15.14 GHz, a peak linewidth of 228 MHz in full-width at half-maximum, resulting in a finesse of ~66, and frequency stability better than ~13 MHz over one hour (Fig. S1, B and C).
The measured maximum transmission of the two-pass etalon was ~82%.

Preparation of the two-layer sample of phosphate-buffered saline (PBS) and polyacrylamide (PAA) gel
We produced two-layer samples consisting of a PBS layer overlying a PAA gel layer (Fig. 2).
The PAA-PBS sample holder comprised two 0.15-mm thick glass coverslips, 25 mm and 18 mm in diameter, spaced at 0.36 mm using adhesive spacers (Grace Bio-Labs SecureSeal).A ~120-m thick PAA gel layer was prepared on the bottom, 25-mm diameter coverslip by mixing 450 μl of 40% acrylamide (Bio-Rad), 200 μl of 2% bis-acrylamide (Bio-Rad), and 350 μl PBS (Biological Industries).3 µl tetramethylethylenediamine (Merck) and 10 µl of 10% ammonium persulfate (Merck) were added to catalyze the acrylamide and bisacrylamide polymerization.A rain-X-coated coverslip was placed on top of the PAA mixture to flatten the surface of the gel.After the gel was solidified, the rain-X-coated coverslip was removed.
The gel-coated coverslip was kept soaked in PBS at 4˚C.For BGM imaging, the space between the two glass coverslips of the sample holder was filled with PBS, and ultraviolet glue was applied to the coverslip edges.

Preparation of C. elegans nematode samples
Wild-type (N2) C. elegans nematodes were grown on nematode growth medium (NGM) plates seeded with Escherichia coli OP50-1 at 15 °C; 30-60 embryos, laid at 15 °C, were picked, transferred to new plates, and grown at 25 °C for the duration of the experiment.We determined the developmental stage of young adult nematodes using a light stereoscope.For BGM imaging, we first prepared two 0.25-mm-thick agar pads (5%) mixed with 10 mM sodium azide solution (NaN3) to anesthetize the nematodes.Then, the agar pads were mounted on two 0.15-mm-thick round glass coverslips, 25 mm and 18 mm in diameter, and 10-15 young adult nematodes were sandwiched between the agar-padded coverslips.Ten microliters of an M9 contact buffer were added between the agar pads.To fix the entire sample and to avoid dehydration, ultraviolet glue was applied to the edge of the smaller coverslip, and a thin layer of Vaseline sealed the gap between the two coverslips.

Preparation of NIH/3T3 cell samples
NIH/3T3 cells (ATCC) were maintained using standard tissue culture procedures in a humidified incubator at 37°C with 5% CO2 and atmospheric oxygen.They were grown in a

Fig. 2 . 5 Fig. 3 .
Fig. 2. BGM material imaging.(A) A double-layered material system consisting of a PBS layer on a