Live imaging of contracting muscles with wide-field second harmonic generation microscopy using a high power laser

: Wide-ﬁeld second harmonic generation (SHG) microscopy was developed using a high-power ( > 4 W) and high-repetition-rate (MHz range) laser oscillator to achieve fast SHG imaging over a large area (400 µm × 400 µm). The microscope was used for high spatial resolution imaging of contracting muscles in live Drosophila melanogaster larvae. Anisotropic and isotropic bands of striated muscle were distinguished, allowing accurate determination of sarcomere length and SHG intensity from individual sarcomeres. Therefore, wide-ﬁeld SHG microscopy has applications in basic contractility research and studying arrhythmias, muscular dystrophies and pharmaceutical eﬀects on the muscle contraction dynamics of sarcomeres.


Introduction
Second harmonic generation (SHG) microscopy has become a technique of choice for label-free in vivo and in situ visualization of non-centrosymmetrically ordered biological structures, including collagen fibers in the extracellular matrix and myosin filaments in the sarcomeres of striated muscles [1][2][3][4][5]. SHG image contrast relies on the frequency-doubled response of molecules to high intensity laser pulses. For optimal SHG imaging, a laser beam from a femtosecond oscillator is focused with a microscope objective lens onto a sample, and the laser beam is raster-scanned to acquire an image either via pixel by pixel signal detection with a single element photodetector (usually a photomultiplier tube or an avalanche photodiode), or a sample is imaged onto an array detector (a charge-coupled device, CCD, or a complementary metal-oxide-semiconductor, CMOS) [6][7][8][9][10]. A large field of view image acquisition rate is limited by the laser scanning speed and SHG efficiency of the sample, restricting the investigations to static or slowly changing (on the order of seconds to minutes) biological structures. This ultimately poses a challenge in performing real-time imaging on a large area of the sample. On the other hand, wide-field nonlinear microscopy enables fast large area imaging however, it requires short laser pulses with high energy to achieve wide-field illumination and attain high image contrast. Amplified laser systems with µJ pulse energies and ∼200 kHz pulse repetition rates have been used for wide-field SHG imaging [6,10]. Macias-Romero et al. revealed that indirect sample damage occurs at the pulse duration flux threshold of 12.2 mJ/cm 2 for human embryonic kidney 293 cells at 1035 nm and 200 kHz [10]. Similarly, Harzic et al. reported the pulse energy threshold for Chinese hamster ovary cells as 381 mJ/cm 2 at 1035 nm and 100 kHz [11]. Therefore, pulse energy has to be kept low due to damage thresholds, the use of higher pulse repetition rates could increase SHG signal, as long as both the pulse duration flux is below threshold, and that the heat deposition due to the average laser power is kept below the physiologically relevant thermal impairment threshold. For example, König et al. reported that adrenal chrommafin cells had indirect damage at 5.3 mJ/cm 2 pulse energy flux, and 0.44 MW/cm 2 average power flux, at 840 nm, 190 fs and 82 MHz [12]. High pulse repetition rates in the megahertz range can be achieved by employing high power oscillators, or by amplified laser systems. In this study we utilize the simpler and more cost-effective high-power laser oscillator configuration. Furthermore, when the linear and nonlinear absorption of the sample at the laser wavelength is negligible, high laser powers can be applied to the sample without inducing adverse photo-bleaching and photo-damage effects. The imaging area can be scaled up proportionally by increasing the incident laser power while maintaining the power density below the tissue damage threshold. Optical sectioning can additionally be achieved with temporal or spectral focusing [13][14][15][16]. Therefore, wide-field SHG microscopy with high power and high pulse repetition rate lasers opens new opportunities for real-time imaging of fast structural dynamics and enables live optical biopsy and histopathology imaging on large areas.
In this paper, we used a high-power and high repetition rate laser oscillator to demonstrate the use of wide-field SHG microscopy for live imaging of contracting muscle tissue. Fast contraction dynamics over a large imaging area allowing parallel analysis of sarcomeres, to our knowledge, has not been previously demonstrated. The visualization and quantification of sarcomere contractility over a large field of view opens new possibilities for contractility research, and as a testing ground for new cardiac pharmaceutical agents.

Wide-field SHG microscope
Wide-field SHG imaging was accomplished with a homebuilt high power ultrafast laser oscillator. The outline of the oscillator is illustrated in Fig. 1. The laser radiation is generated in a 5-atom%doped 5-mm-long Yb:KGW crystal oriented for propagation along the Ng-axis and polarization parallel to the Np-axis [17,18]. The crystal is pumped by a fiber-coupled diode laser that provides a maximum pump power of 40 W at a central wavelength of 980 nm (Apollo Instruments, Inc.). The pump radiation is reimaged by two plano-convex lenses with focal lengths of 35 mm and 125 mm. The mode locking is self-starting and maintained by the semiconductor saturable absorber mirror (SESAM) with 1% modulation depth and 500 fs recovery time (Batop GmbH). All the highly reflective (HR) mirrors in the cavity provide -10,000 fs 2 global dispersion in a round-trip to balance the intracavity positive dispersion and the chirp induced by the self-phase-modulation. With a 7.5% output coupler, pulses with an average output power of 5.5 W at a repetition rate of 60 MHz were obtained. The central wavelength was 1041 nm and the pulse duration was measured to be ∼400 fs.
Wide-field SHG imaging is accomplished with a homebuilt microscope (Fig. 1). The laser beam from the oscillator is collimated to ∼4 mm in diameter and relays to an achromatic doublet with a focal length of 75 mm. In order to obtain wide-field imaging, the sample is placed above the focal plane. The image size can be adjusted by axially translating the doublet. The incident power can be tuned by rotating the half wave plate placed before a polarizing beamsplitter. The SHG signal radiated in the forward direction is collected by a microscope objective lens. Two objective lenses (Zeiss) are used in this study: a 20× 0.5 numerical aperture (NA) air objective lens and a 40× 1.3 NA oil immersion objective lens. The SHG from the sample is focused onto the electron multiplying CCD (EM-CCD, Hamamatsu ImagEM X2) by using a tube lens (f = 160 mm). The fundamental laser and fluorescence photons are filtered from SHG by a BG39 filter and a 510-520 nm band-pass interference filter (CVI Laser Optics). SHG images with 512 × 512 pixels were obtained at an acquisition rate of 10 frames per second, limited only by the number of emitted SHG photons. Faster acquisition rates could be achieved by using a lower repetition rate and higher pulse energy laser to generate more SHG signal, as long as the pulse energy fluence is below the damage threshold.
A thin lithium triborate (LBO) crystal of 0.1 mm thickness attached to a thin gold grid was placed on a translation stage to characterize the SHG radiation from the microscope. The SHG signal intensity had a Gaussian distribution along both lateral (horizontal X and vertical Y) directions in the image. The lateral beam spot size on the crystal is 350 µm full width at half maximum (FWHM), and the axial depth of field is 8 µm FWHM with the 20× 0.5 NA objective as determined by the point spread function from 2 µm fluorescing beads.

Sample preparation
Drosophila melanogaster stocks were grown at 20°C on Bloomington standard cornmeal medium. For static imaging, third instar larvae were dissected along the dorsal midline in calcium-free HL3 saline solution (pH 7.2). Upon removal of internal organs and fat bodies, the body was unfolded to expose the body wall muscles and was fixed in 4% formaldehyde dissolved in 1× phosphate buffered saline (PBS, pH 7.2). The tissue was washed in fresh buffer before preparation for microscopic imaging. The fixed muscles were held between a 1 mm thick microscope slide and a 0.17 mm thick coverslip (No. 1.5), and sealed with nail polish. For live imaging, intact larvae in PBS were squeezed between a microscope slide and a coverslip.

Wide-field SHG microscopy of fixed larval muscles
We initially explored fixed larval muscles with the 20× 0.5 NA objective lens. A typical SHG image of muscle myofibrils is shown in Fig. 2(a). The image exhibits a periodic structure, originating from the striated pattern of the sarcomere anisotropic bands (A-bands). Due to the large depth of field, the myocytes located at different heights can be visualized in the same image. Optical sectioning with illumination was not employed however, it could help to reduce out of focus background signal. The structure of the myofibrils becomes readily apparent at an incident power above 1.5 W. Optimal imaging was achieved with an incident power of 3.5 W at the sample providing a signal to noise ratio of 20 dB. Figure 2(b) shows the SHG intensity profile of 15 sarcomeres along a myofibril. The heterogeneity of the intensity of the peaks is attributed to the variation in sarcomere length and structural configuration of myosin nanomotors within myofilaments in the A-bands of sarcomeres. The average sarcomere length is estimated by measuring the M-line-to-M-line distance, which is assumed to be positioned at the peaks in the SHG intensity profile. By applying Gaussian fitting to each peak, the positions of M-lines were determined, yielding an average sarcomere length of 9.1 ± 0.1 µm (mean ± standard error). In order to demonstrate the capability of wide-field SHG microscopy for higher resolution imaging, the 40× 1.3 NA oil objective was also used. Figure 2(c) shows a typical SHG image with a 213 µm × 213 µm field of view. Some myocytes with a double-band structure are visible (circled in Fig. 2(c)). The average sarcomere length is 8.83 ± 0.09 µm in the SHG profile.

Wide-field SHG microscopy of contracting larval muscles
Next, we used the wide-field SHG microscope to capture myocyte contractions of live larvae. Figure 3(a) shows the first frame of Visualization 1 that visualizes the active contractions recorded at 10 frames per second for a duration of 30 s, at an incident power of 3 W (2.1 kW/cm 2 , 0.035 mJ/cm 2 ). The viability of myocytes was verified by observing periodic muscle contraction for a duration longer than 1 minute under continuous imaging. With the laser power above 5 W (3.5 kW/cm 2 , 0.06 mJ/cm 2 ) the larva would slow its movement indicating photodamage.
Visualization 1 captures live larva contractions with individual sarcomere length resolution. Live imaging allows the possibility of studying the synchronization of sarcomere shortening and elongation during muscle contraction. Visualization 1 also reveals SHG intensity flickering that is synchronized with changes in sarcomere length. The SHG intensity increase corresponds to the elongation of the sarcomere. The variations of SHG intensity can be monitored live, while measurements of individual sarcomere length require post image processing, hindering live applications. The flickering of sarcomeres was not observed in fixed muscle samples (Fig. 2).
The correlation between sarcomere length and SHG intensity can be studied by analyzing a time series of myofibril segment images over the period of a contraction shown in Fig. 3(b). The myofibril segment is selected from Visualization 1. The variation of average sarcomere length and the corresponding variation of SHG intensity are shown in the blue line with triangles and the red line with circles, respectively, in Fig. 3(c). The average sarcomere length is calculated by using an ImageJ (NIH) plugin, SarcOptiM, which uses the fast Fourier transform method [19], and the SHG intensity is averaged for the number of distinguishable sarcomeres (i.e. averaged SHG intensity per sarcomere). During the moment of contraction labeled A in Fig. 3(c), the sarcomeres shorten [20,21] to an average length of 6.7 ± 0.1 µm, and the corresponding average SHG intensity reaches a minimum value. The following moment of relaxation results in stretching of the sarcomeres to an average length of 9.2 ± 0.1 µm (labeled B in Fig. 3(c)) [22,23], and the corresponding SHG intensity increases by more than 50%, indicating a positive correlation between the sarcomere length and the SHG intensity. The correlation of average sarcomere length with SHG intensity matches previous measurements of contracting larva myofibrils imaged with a scanning SHG microscope [5,24].
The linear correlation between sarcomere length and SHG intensity enables live visualization of sarcomere length alterations by monitoring changes in SHG intensity. The SHG intensity changes are easier to monitor compared to the tedious determination of sarcomere length via fitting analysis. A transient intensity decrease corresponds to the sarcomere shortening, while sarcomere elongation results in a transient SHG intensity increase.

Conclusion
We have constructed a high-power femtosecond oscillator and wide-field SHG microscope for live imaging of biological specimens. Live muscle contractions were imaged with the wide-field microscope by utilizing the endogenous SHG image contrast originating from the A-bands of sarcomeres. The shortening of larval myocytes is associated with decreasing SHG intensity in individual sarcomeres. Wide-field SHG microscopy of myocytes can be employed as a contractility assay platform for studying arrhythmias and testing cardiac pharmaceutical agents. Wide-field SHG microscopy can also be used for imaging the extracellular matrix collagen in various tissues for cancer diagnostics [4], fibrosis and connective tissue disorders.