Cardiomyocyte sarcomere length variability: Membrane fluorescence versus second harmonic generation myosin imaging

Sarcomeres are the basic contractile elements in a striated muscle cell. The actual sarcomere length (SL), the inherent properties of individual sarcomeres, and the mechanical intercommunication between adjacent sarcomeres are important for the whole-cell contractile response, both in cardiac (Dobesh et al., 2002; de Tombe and ter Keurs, 2016) and skeletal muscles (Herzog, 2022). The ability to characterize SL under certain conditions (ambient, mechanical, etc.) makes it possible to conclude how changes in SL are linked to the functional behavior of the cardiomyocyte.

Optical imaging is the only available tool for characterizing the sarcomeric striation pattern and the corresponding average SL, which is highly informative for understanding the functional relationships between actual SL and the contractile state of the cell (Bub et al., 2010; Botcherby et al., 2013; Aguirre et al., 2014; Guo and Song, 2014; Pasqualin et al., 2016; Kobirumaki-Shimozawa et al., 2018; Varga et al., 2020; Dowrick et al., 2021). Recent studies have also focused on the existence of intracellular SL heterogeneity in striated muscle (Rassier, 2017; Haeger et al., 2020; Adkins et al., 2022) and, importantly, its alteration upon the transition of myofilaments from the inactivated to the activated state (Johnston et al., 2016; de Souza Leite et al., 2017; Moo et al., 2017; Haeger and Rassier, 2020; Lookin et al., 2022). Therefore, special care is needed to choose a proper method for SL measurements.

For many years, the use of membrane staining by fluorescent dyes (particularly of the ANEPPS family) has been an extensively used approach (Bub et al., 2010; Guo and Song, 2014; Wagner et al., 2014; Yue et al., 2017). These labeling dyes, however, only visualize membrane invaginations, the t-tubules, but not directly the internal cellular components that compose the sarcomeres. As a result, relying on a membrane dye may render proper determination of the location of the sarcomeric Z-disk somewhat uncertain. This limitation may be overcome by employing genetically encoded fluorescent sarcomeric proteins that are introduced by viral infection techniques. For example, fluorescent labeling of α-actinin has been used to visualize sarcomeric Z-disks (Shintani et al., 2014; Tsukamoto et al., 2016; Kobirumaki-Shimozawa et al., 2016; Kobirumaki-Shimozawa et al., 2018). However, both the fluorescent labeling and the viral infection delivery methods involve introduction of “non-natural” compounds into the living cell that may affect the function and structure of the cardiac myocyte.

Second-harmonic generation (SHG) microscopy is, by design, a label-free imaging modality that has been highly effective in the examination of cellular structures in intact living tissues (Recher et al., 2009; Buttgereit, 2017). In addition to high-contrast optical sections of cells and tissues, SHG imaging can also provide detailed structural information. This technique has been successfully employed to visualize sarcomeres (Garcia-Canadilla et al., 2014; Zhao et al., 2019; Varga et al., 2020; Homan et al., 2021) and, therefore, is specifically suited to the study of SL variability. SHG-based images of myocytes have revealed that the source of the SHG signal from the sarcomere is the myosin tails that assemble into a rod structure forming the central core of thick myosin filaments (Plotnikov et al., 2006).

As t-tubule visualization is often used for the determination of average SL, we aimed here to compare the characteristics of individual SL populations as obtained by the two principally different optical methods: SHG and t-tubular fluorescent labeling (ANEPPS).

All investigations conformed to European Parliament Directive 2010/63/EU and were approved by the ethics committee Comité d’éthique pour l’expérimentation animale Languedoc-Roussillon.

Intact cardiomyocytes were isolated from the heart by enzymatic digestion as previously described (Andre et al., 2010). Briefly, the heart was harvested from deeply anesthetized male rats (pentobarbital 50 mg/kg) while beating, quickly cannulated on a Langendorff apparatus, and perfused with a physiological solution at 37°C via the aorta. This solution contained (in mM) 117 NaCl, 5.7 KCl, 4.4 NaHCO₃, 1.5 KH₂PO₄, 1.7 MgCl₂, 21 HEPES, 11 glucose, and 20 taurine, pH = 7.2 adjusted with NaOH. The heart was first washed with calcium (Ca²⁺)-free extracellular physiological solution and then with an enzyme-containing solution (1.25 mg/ml collagenase type IV; Worthington) for 25–35 min. The left and right ventricles were separated and digestion was stopped with bovine sodium albumin. The tissue was then mechanically disrupted with a pipette to release the myocytes. The extracellular Ca²⁺ concentration was progressively increased to 1 mM. All subsequent experiments were performed using a modified Tyrode solution (in mM): 140 NaCl, 4 KCl, 1 MgCl₂, 20 HEPES, 0.1 CaCl₂, and 11 glucose, pH = 7.4 at room temperature. Myocytes were not electrically stimulated, and 20 mM BDM was added to all solutions to prevent spontaneous contractile activity.

The analysis of intracellular variability in SLs was performed using SHG microscopy measurements (SHG) + fluorescent measurements with membrane staining by di-4-ANEPPS (ANEPPS) for fully relaxed and non-stretched cardiomyocytes (n = 26 from N = 3 hearts).

Just prior to the start of the SHG/ANEPPS measurements, cardiomyocytes were labeled with the fluorescent dye di-4-ANEPPS as follows: 0.2 μl of ANEPPS (1 mM DMSO stock solution) was added to 1 ml of cell suspension in Tyrode solution at room temperature. Following sedimentation during the 10-min dye treatment, the supernatant was removed and replaced by 3 ml fresh dye-free Tyrode. An aliquot of cells was placed in ∼1 ml fresh modified Tyrode solution in a glass-bottomed 35-mm plastic petri dish (room temperature). SHG images were then recorded using a custom-made multiphoton microscope setup based on a Tsunami tunable Ti:Sapphire (Ti-Sa) laser (Spectra-Physics) and an upright SliceScope microscope (MPSS-1000P) furnished with a multiphoton galvanometer scan head (MP-2000) both from Scientifica (Varga et al., 2020). The Ti-Sa laser was operated in pulsed mode configuration for sample excitation, 875 nm wavelength, 80 MHz frequency, and ∼100 fs pulse duration. A Nikon CFI75 LWD-16×-W objective (NA 0.8, water immersion) was used to focus the laser beam onto the myocyte. The SHG signal was recorded by a 1.4 NA water-immersion condenser (U-AAC; Olympus) through a 482 nm long-pass dichroic mirror (86-331; Edmund Optics) and a 447 nm high-performance 60 nm band-pass filter (48-074; Edmund Optics) and then detected by an H7422P photomultiplier (Hamamatsu). Epi-fluorescence two-photon images were recorded using a 735 nm long-pass dichroic mirror, followed by sequential short-pass/long-pass filters to form a 450–750 nm bandpass filter, 1,024 × 1,024 ∼150 nm XY scan pixel size, and then detected by an R928P photomultiplier (Hamamatsu). Several consecutive scans (∼10 frames) of the same cell position were recorded to increase the signal-to-noise ratio during further off-line image after processing. In a subset of experiments, images were acquired at five separate Z-slices, ranging approximately from the top to the bottom of the isolated myocyte (cf. Fig. 4).

The analysis of variability in individual SLs was implemented by custom-made software EqapAll6 (developed by Oleg Lookin). For each cell in the SHG/ANEPPS measurements, we used the two optical images separately for SHG (to reveal the regularity of sarcomeric A-bands) and ANEPPS (to reveal the regularity of t-tubular invaginations of the cell membrane). By design, the two images were always aligned with each other, and five regions of interest (ROI) with the same positions, sizes, and angles were used for each image (Fig. 1 A). In all measurements, the ROIs were placed to cover as much area of a cell as possible, but their lengths and angles were set to follow a straight part of myofibrils. Curved myofibrils, crossing myofibrils, or myofibrils with irregular staining along their length were omitted from the selection.

The lengths of individual sarcomeres were determined using an algorithm based on analysis of the sarcomere striation profile. For this algorithm, the input data were the sarcomere striation profiles obtained for each ROI. The positions of sarcomere edges in ANEPPS measurements were found from the local maxima in the sarcomere striation profile. In contrast, the local maxima in the sarcomere striation profile in SHG measurements reflected the positions of sarcomere centers. To test if the use of local maxima or minima in the same striation profile can produce a difference in SL variability, we also compared the data of SHG/ANEPPS measurements using both local extrema (cf. Fig. 3). The individual SL sets obtained for each ROI were then combined to increase the resulting number of individual SLs prior to the analysis of their variability.

An expanded view of striation profiles obtained for the same ROI using SHG/ANEPPS measurement is shown in Fig. 1 B. Note that the local maxima in these two striation profiles are shifted relative to each other by a half sarcomere, as indicated by the arrows in Fig. 1 B. The local maxima in the SHG measurement correspond to the centers of A-bands, while the local maxima in the ANEPPS measurement index the local peak intensity of t-tubule staining that putatively corresponds to the position of sarcomeric Z-disks.

Statistical analysis was carried out with GraphPad Prism 9.3.1 (GraphPad Software). As some of the individual SL sets did not follow the normal shape, we analyzed median SL values as well as the absolute and relative measures of variability applicable to non-normal data samples. The evaluation of significance in the median SL values as well as the absolute/relative measures of SL variability between SHG and ANEPPS was implemented by the Wilcoxon matched-pairs signed-rank test. Differences were considered significant at P < 0.05. Data are presented as mean ± SD.

We analyzed the variability of individual SLs using SHG imaging to solely and directly visualize sarcomeric A-bands. For comparison, simultaneous ANEPPS fluorescence was used to visualize t-tubule structures. Median values of SL as well as absolute and relative measures of SL variability were analyzed and contrasted between these two image modalities. The averaged median SL values were not significantly different between the two optical measurements (P = 0.901; Fig. 2 A). In contrast, the absolute and relative measures of variability of individual SLs were sensitive to the method of measurement. The absolute measures of SL variability—interquartile range and median absolute deviation—were approximately twofold higher for SL measurements using ANEPPS vs. SHG (the difference is significant at P < 0.0001 for any measure; Fig. 2, B and C). For example, the median absolute deviation of SL was found to be 0.04 ± 0.02 μm in SHG vs. 0.09 ± 0.02 μm in ANEPPS measurements (Fig. 2 C). Likewise, the relative measures of SL variability—interquartile range divided by median SL value or median absolute deviation divided by median SL value—were significantly higher (∼2.0–2.2-fold) for ANEPPS vs. SHG (the difference is significant at P < 0.0001 for each measure; Fig. 2, D and E). The higher SL variability obtained by ANEPPS measurements is demonstrated by plotting superimposed SL distributions averaged for all SHG and ANEPPS measurements (Fig. 2 F). Each plot peaks at the averaged median SL and shows how much the SL values are distributed (as a percentage of the median SL value). The distribution for ANEPPS measurements is roughly two times wider than the distribution for SHG measurements.

The calculation of SL using ANEPPS measurement was based on the determination of local maxima in the intensity profiles that correspond to the peak intensities of the individual t-tubules. Instead, we can use local minima to check whether their positions correlate with the local maxima of the SHG measurement (A-band centers). We did not find any difference between these two approaches, that is, in the ANEPPS measurement, the spreads of local maxima (the physical presence of t-tubules) and local minima (the interval between two adjacent t-tubules) were quantitatively similar. The plots of the spread range of median SL values and median absolute deviation divided by median SL values, as obtained for SHG and ANEPPS measurements by using local maxima or local minima as an index for determining the local SL, are shown in Fig. 3. Median SL values were found to be insensitive to the method of optical measurement or the selection of local maxima/minima to retrieve the individual SLs (Fig. 3 A). At the same time, the SL variability was significantly larger in the ANEPPS vs. SHG measurements (Fig. 3 B) regardless of the use of local maxima or minima to index SL positions.

It should be noted as a side observation in this study that using Z-stacks in SHG measurements allows for a more precise evaluation of the 3-D structure of the sarcomeric A-band geometry than can be obtained using ANEPPS imaging. For example, we implemented Z-stacks in some of our SHG measurements and found that an A-band regularity pattern exists in most of the Z-planes (Fig. 4 A), with the SL distribution characteristics being non-discriminatory between slices. The local peaks of the A-band regularity pattern revealed a high persistence in their physical positions along the longitudinal axis of the cell in the Z-direction, that is, along the depth of the cell (Fig. 4 B). In other words, some parts of the cell showed minor transversal changes (∼0.5 µm, roughly one-fourth of the sarcomere, for nearly the full depth of the cardiomyocyte) in the longitudinal positions of A-bands. This may indicate the presence of some functional role for transversal A-band geometry. Quantification of the shifts in the X-position of each individual A-band when slicing across the cell depth may allow for the measurement of such transversal changes. The middle frame in a Z-stack can be used as the referent to evaluate such shifts in the other frames (Fig. 4 C). The more aligned A-bands are in the Z-direction, the smaller a positional shift is observed. Therefore, SHG measurements may be helpful in the precise 3-D imaging and reconstruction of the A-band “skeleton” as well as in the analysis of A-band (mis)alignment within a whole cell.

The main finding of our study is that the method of optical measurement is critical for the proper characterization of individual SL variability, while it does not affect the measured average SL. In the present study, we implemented two optical measurement modalities: SHG microscopy and confocal two-photon imaging of fluorescent membrane-specific dye (ANEPPS). The first method marks the individual positions of A-bands that are purely internal structures of the sarcomere (Garcia-Canadilla et al., 2014; Zhao et al., 2019; Varga et al., 2020), while the second method visualizes the t-tubular striation pattern known to be closely aligned to sarcomeric Z-disks (Bub et al., 2010; Aguirre et al., 2014; Guo and Song, 2014), but which is not actually part of the sarcomere (Setterberg et al., 2021). The determination of averaged SL was shown to be insensitive to these optical methods confirming that a proper averaged value of SL can be obtained using fluorescent staining of the t-tubules component of the sarcolemma. However, for subtle purposes, such as the analysis of SL variability, the two methods were found not to be equivalent. In our study, the use of ANEPPS dye for sarcomeric visualization resulted in higher SL variability as compared with SHG microscopy, which makes visible the central part of the sarcomere. This could be due to a different extent of variability of the positions of A-bands versus t-tubules in each myocyte. The peak optical density of the t-tubule is somewhat aligned to the physical “center” of the Z-disk, but its actual peak position could be misaligned to some extent with the Z-disk center because a t-tubule is not physically part of the sarcomere (the “sloppy” coupling of t-tubules to the sarcomere Z-disks). In contrast, the A-band is the core inner component of the sarcomere and, therefore, its optical position should strongly follow the actual position of the center of the sarcomere. Our result, therefore, supports that SHG-based sarcomeric A-band visualization is the gold standard for accurate analysis of individual SLs. It should be noted that in some of our SHG images, we observed, like in previous reports (Recher et al., 2009; Varga et al., 2020), a double-banded signal in the A-bands of the sarcomere; in our analysis, we avoided selection of myofibrils with such striation pattern.

A further question remains whether the spatial resolution is a crucial factor that affects the precision of SL determination. In the present study, we used the same spatial resolution as in our recent study on guinea pig myocytes, where we used transmitted light (TL) mode to acquire cell images (Lookin et al., 2022), and average SL was only nominally affected by the method of optical measurement—SHG, ANEPPS, or TL; in the latter mode, we found a slightly higher resting SL (data not shown), but this may be related to the absence of fluorescent dye in these cells. However, the use of SHG provided twofold lower measured SL variability as compared with both ANEPPS (and TL in the paper cited above), despite the equivalent spatial resolution of the optical systems. In the TL mode, the measured image is non-confocal and the sarcomeric striation derives from all myofibers along the cell depth. Therefore, imaging by conventional microscopy has inherently higher inconsistencies in the determination of an individual SL compared with confocal or SHG microscopy (Telley et al., 2006; Kobirumaki-Shimozawa et al., 2016). On the other hand, if the focal plane is properly chosen and optical phase-contrasting is used, the TL image may be as good for the SL analysis as the ANEPPS image, similar to previously reported data (Nance et al., 2015). Another reason that non-confocal TL imaging may yield similar results as the ANEPPS confocal images is an axial regular alignment of sarcomeric structures (i.e., along the z axes) of the cardiomyocyte, as we observed in the present study (cf. Fig. 4). The structural basis for this regularity is not known and was not the focus of the current study, but, as an interesting side note, this regularity may also be expected to contribute to axial stiffness of a non-contracting relaxed cardiomyocyte (Peyronnet et al., 2022). Therefore, quantitative measures of SL variability appear to be sensitive to the selection in which the sarcomeric (related) structure is visualized, rather than the selection of optical method per se (fluorescent or TL microscopy, confocal or brightfield microscopy).

The application of SHG microscopy can be further extended to the actively contracting cardiomyocytes and, especially, for the conditions where the myocyte is subjected to the physiological pre- or afterload. In our previous paper (Lookin et al., 2022), we reported that myocyte electrical activation increased SL variability as compared with the inactive state, concomitant with a reduction in overall SL. This result conformed to previously reported differences between relaxed and activated states in skeletal muscle (Moo et al., 2017; Moo and Herzog, 2018; Johnston et al., 2019; de Souza Leite and Rassier, 2020). However, the effect of sarcomere prestretch on SL variability in contracting cardiomyocytes remains to be determined. To elucidate this aspect in detail, SL variability in cardiomyocytes should be assessed with the use of SHG microscopy as a prospective tool in the determination of individual sarcomeres, in addition to the widely used fluorescent staining.

In conclusion, we found that the method of optical measurement is critical for the proper characterization of individual SL variability, while it does not affect the measured average SL. Our results, therefore, support the notion that SHG-based sarcomeric A-band visualization forms the gold standard for precise analysis of individual sarcomeres.

Henk L. Granzier served as editor.

We thank Patrice Bideaux for assistance with the rat cell isolation procedures.

The study was supported by the Centre National de la Recherche Scientifique (France; grant #IEA00401 to O. Cazorla) and National Institutes of Health (HL62426 to P. de Tombe).

The authors declare no competing financial interests.

Author contributions: O. Lookin, P. de Tombe, N. Boulali, and O. Cazorla contributed to the conception of the study, design of experiments, analysis, and interpretation of the results. P. de Tombe, N. Boulali, C. Gergely, T. Cloitre, and O. Cazorla contributed to the experimental measurements. C. Gergely and T. Cloitre provided technical assistance in SHG measurements. O. Lookin made software for data processing. The manuscript was written by O. Lookin, P. de Tombe, and O. Cazorla. All authors approved the final version of the manuscript.