To study how the dynamic subcellular mechanical properties of the heart relate to the fundamental underlying process of actin-myosin cross-bridge cycling, we developed a novel atomic force microscope elastography technique for mapping spatiotemporal stiffness of isolated, spontaneously beating neonatal rat cardiomyocytes. Cells were indented repeatedly at a rate close but unequal to their contractile frequency. The resultant changes in pointwise apparent elastic modulus cycled at a predictable envelope frequency between a systolic value of 26.2 ± 5.1 kPa and a diastolic value of 7.8 ± 4.1 kPa at a representative depth of 400 nm. In cells probed along their major axis, spatiotemporal changes in systolic stiffness displayed a heterogeneous pattern, reflecting the banded sarcomeric structure of underlying myofibrils. Treatment with blebbistatin eliminated contractile activity and resulted in a uniform apparent modulus of 6.5 ± 4.8 kPa. This study represents the first quantitative dynamic mechanical mapping of beating cardiomyocytes. The technique provides a means of probing the micromechanical effects of disease processes and pharmacological treatments on beating cardiomyocytes, providing new insights and relating subcellular cardiac structure and function.
- transverse stiffness
- Young's modulus
the cellular constituents of the myocardium undergo significant structural changes throughout each cardiac cycle, driven at the molecular level by cycling of actin-myosin cross bridges. The resultant periodic alterations of heart muscle mechanics directly affect myocardial oxygen consumption, coronary blood flow, and many other physiological and pathological aspects of cardiac function (36). Dynamic mechanical properties of the myocardium have been extensively studied at the organ (37, 45) and tissue (28) levels, whereas studies of cardiac myocyte mechanics have mostly been restricted to passive or chemically altered states (19, 23, 32, 35, 48) rather than actively beating cells. For example, pressure overload-induced heart failure causes microtubule-associated changes in viscoelastic properties of passive cardiomyocytes (48), possibly contributing to diastolic chamber stiffening; however, concomitant changes in dynamic or systolic myocyte mechanics remain undetermined.
Recent developments in nanotechnology have made it possible to study isolated myocyte dynamics at the whole cell level (25, 39, 42). Such studies have largely focused on axial cell properties in relation to force generation. However, myocardial transverse properties also impact pump function (11) and active force generation (34). The few studies on transverse mechanics of intact heart muscle (22, 41), individual cardiomyocytes (23, 32, 35, 38), or isolated cardiac myofibrils (1) have used chemically stabilized conditions to avoid technical challenges with actively beating samples. Moreover, because local environmental micromechanics can fundamentally influence cell fate and function (17, 18, 26), it is important to understand how cardiomyocyte mechanical properties vary spatially in relation to the underlying myofibrillar structure, as well as temporally throughout cyclic activation and relaxation. Such spatiotemporally resolved data are also required for increasingly sophisticated multiscale computational models of the heart (24, 43).
The atomic force microscope (AFM) has been utilized to investigate subcellular micromechanics of a wide variety of cell types, including passive cardiac myocytes (23, 32, 35). However, only two AFM studies have probed dynamic properties of actively beating cultured neonatal cardiomyocytes: one demonstrated subcellular regional heterogeneity of pulse rate (14), while the other showed AFM tip resonance behavior qualitatively reflecting the periodicity of cell stiffness (47). Recent AFM studies of isolated cardiac myofibrils revealed local heterogeneities in transverse stiffness during rigor, with z-bands significantly stiffer than other sarcomeric structures (1). However, although AFM elastography has been used to correlate regional cell mechanical properties with underlying cytoskeletal structures (3), it is not known whether mechanical heterogeneity of isolated myofibrils translates to an intact cardiomyocyte and thereby to adjacent cells. If so, the micromechanical effects of such subcellular heterogeneity on cardiac structure and function could be significant, potentially impacting key processes, such as transverse force generation, electromechanical coupling, and registration of sarcomeres across multiple neighboring cardiomyocytes during development.
No prior studies have quantified subcellular mechanical properties of living cardiomyocytes in physiologically relevant, dynamic conditions. Therefore, to relate such properties to the underlying process of cross-bridge cycling, this study introduces a novel AFM-based dynamic elastography method for spatiotemporal micromechanical mapping of spontaneously beating, intact cardiac myocytes. The proposed method will open new avenues of research by enabling testing of pharmacokinetic effects of new drugs, along with dynamic mechanical properties of myocytes at subcellular levels. Furthermore, this technique can be adapted to study the effects of molecular alterations, such as age-dependent changes in cytoskeletal protein isoforms (30), on intact myocyte mechanics.
MATERIALS AND METHODS
Cell isolation and culture.
All animals received humane care and treatment compliant with the Guide for the Care and Use of Laboratory Animals (13), and protocols were approved by the Institutional Animal Care and Use Committee at Columbia University. Chemicals were purchased from Sigma-Aldrich (St. Louis, MO), unless noted otherwise. Neonatal rat cardiac myocytes were isolated from day 2 Sprague-Dawley rats (Taconic Farms, Germantown, NY), as previously described (31), and were plated on culture dishes brushed with bovine type I collagen (Inamed Biomaterials, Fremont, CA). Attached cells formed aligned myofibrils with registered sarcomeres and started spontaneous contractions at 3–4 Hz within 24–48 h of plating. They were incubated under standard culture conditions with DMEM and 10% neonatal bovine serum (Hyclone, Logan, UT) for 5–20 days before AFM indentation studies.
A Bioscope AFM (Veeco, Santa Barbara, CA), coupled with an inverted spinning disk confocal microscope (Olympus, Center Valley, PA) and temperature-controlled stage (Veeco), was used to probe micromechanical properties of contracting myocytes at 37°C. A standard pyramidal-tipped silicon nitride AFM probe (Veeco) was mounted on a fluid cell with a silicone moisture trap to prevent changes in solution osmolarity due to evaporation. Cells were visualized using the optical microscope, allowing the AFM probe to be positioned and aligned relative to a myocyte of interest (Fig. 1). Target cells (n = 8 from separate dishes), with visible contractions along the major axis, were repeatedly indented with a triangular extension-retraction waveform applied to a central region of the target cell at a frequency that was close but unequal to the contractile rate of the cell. During setup, the AFM scanner was positioned relative to the sample, such that the probe tip came into contact with the cell at a piezo position that was approximately three-fourths the distance through the full range of z-axis extension. In this way, the AFM probe came well out of contact with the cell between each cycle, resulting in a sequence of discrete indentations. Optical video microscopy images recorded at 50 Hz with a high-speed digital camera (Redlake, San Diego, CA) were synchronized with the raw AFM deflection signal using a National Instruments (Austin, Texas) data acquisition card. This allowed the timing of each probe-cell contact event to be readily determined relative to the phase of cardiomyocyte contraction (Fig. 2, A and B). The frequency of cellular contraction was quantified from bright-field images (see Supplemental Fig. 1) by applying a fast-Fourier transform to the time course of axial cell displacement, determined using a published phase correlation algorithm (4). The convolution of the AFM indentation frequency (fAFM) and the myocyte pulse rate (fcell) created a response signal with a third envelope frequency (fenvelope) equal to the magnitude of the difference between the two input frequencies, (1) Ironically, such a phenomenon is known in mathematical terms as “beating” (16). If fcell > fAFM, the convolved response signal resembled the cardiomyocyte twitch, otherwise the shape of the signal was reversed in time, as in Fig. 2B.
Repetitive indentations were performed over multiple subcellular locations, providing a spatiotemporal description of mechanical properties for the central region of the given myocyte. Some myocytes (n = 4) were treated with 100 μM blebbistatin for 20 min to arrest contraction and were then probed again. Other cells (n = 4) were probed over a 12-μm linear array with -μm step size, aligned with the major axis of contraction, to correlate dynamic cardiomyocyte mechanics with underlying sarcomeric structures. Finally, cells were fixed in 3.7% formaldehyde, rinsed with PBS, permeabilized with 0.05% Triton X, and stained for cytoskeletal elements with anti-β-tubulin antibody, rhodamine-conjugated phalloidin (Invitrogen, Carlsbad, CA), and 4,6-diamidino-2-phenylindole dilactate.
Resultant AFM force curves were analyzed, as previously described (9). Briefly, each raw deflection-extension curve was parsed into pre- and postcontact regions by fitting a bidomain linear-quadratic function used to identify the contact point by least squares optimization. The origin was reset to this contact point to obtain the force-indentation curve, and the depth-dependent apparent elastic modulus was computed using (2) where F(D) is the instantaneous indentation force obtained by multiplying probe deflection with the cantilever spring constant; Êapp(D) is the depth-dependent pointwise apparent elastic modulus (12); and finally, ϕ(D) is a depth-dependent geometric function that governs the contact radius of the indenter (7). For a blunted conical indenter that has a spherical tip defect (with a defect radius of curvature of R, cone half-angle of θ, and maximum defect width of 2b), the geometric function is defined as: (3) where contact radius, a, is determined by numerically solving: (4) Apparent elastic modulus can be related to the Young's modulus, E, of an equivalent linear elastic substrate via Êapp = E/[2(1 − ν2)], where the Poisson's ratio ν = 0.5 is often used, assuming material incompressibility of the cell. It should be noted that, in contrast to the standard Hertzian curve-fitting approach to AFM indentation analysis (44), the pointwise apparent modulus improves the sensitivity for detection of depth-dependent variations in mechanical properties (12), including an ability to clearly identify and exclude potential stiffening artifacts due to the rigid substrate (10).
During postprocessing, the timing of initial probe-cell contact for each indentation was determined in relation to the contractile phase of the myocyte (Fig. 2, A and B, and Supplemental Fig. 1). Data were pooled together from multiple contraction cycles to obtain a mean stiffness value during systolic shortening (when the major cell axis was shortest) and diastolic relaxation (when the major cell axis was longest) for each cell. Data are presented as means ± SD for n = 8 cells, unless otherwise specified. Mean values from multiple cells were compared using one-way ANOVA and Tukey's post hoc test with statistical significance accepted for P < 0.05.
RESULTS AND DISCUSSION
Temporal changes in local cardiomyocyte mechanics.
All tested myocytes contracted spontaneously (i.e., without electrical pacing) with a mean fcell of 3.6 ± 1.0 Hz at a mean culture period of 13 ± 5.8 days. When probed repeatedly at an indentation frequency close, but unequal, to the pulse rate (fAFM = 3.9 ± 1.4 Hz; fenvelope = 1.0 ± 0.6 Hz), the relative timing of each AFM indentation shifted along the cardiomyocyte twitch cycle. This resulted in a sequence of indentations mapping out the temporal changes in mechanical state at a given spatial location within the cell. For example (see Fig. 2), when fAFM (4.35 Hz) exceeded fcell (4.10 Hz), if the first indentation began at the end of a contraction, the second indentation would begin at a slightly earlier time point during the twitch cycle, and so on for subsequent indentations. Accordingly, as the cell was being indented repeatedly at the same spot, the underlying force indentation response changed with a compound frequency, fenvelope, of 0.25 Hz, according to Eq. 1 (Fig. 2B), indicating larger indentation forces during systole than during diastole (Fig. 3A). The resultant Êapp (D) values showed greater depth dependence during systole than during diastole (Fig. 3B), and, at a given indentation depth, Êapp changed cyclically, stiffening during systolic contraction and softening with diastolic relaxation (Fig. 2C). At a representative indentation depth of 400 nm, cardiomyocyte stiffness values, obtained over three to four envelope cycles or 10–15 contractions for each cell (n = 8), had a mean diastolic pointwise modulus of 7.8 ± 4.1 kPa (Fig. 4). This was significantly different from the corresponding systolic modulus of 26.2 ± 5.1 kPa (P < 0.01), indicating more than a threefold increase in the local microscale elastic properties of the cell during contraction.
Treatment of cells with blebbistatin abruptly stopped contractions within a few minutes and reduced the average cell pointwise modulus to diastolic values; mean Êapp at 400-nm depth was 6.5 ± 4.8 kPa (n = 4), which was significantly different than pretreatment systolic (P < 0.01), but not diastolic (P = 0.89), values (Fig. 4). Unlike other arresting agents, such as 2–3-butanedione monoxime, which can have complex nonspecific effects on cardiomyocyte viability and function (27), blebbistatin does not uncouple the cross bridges, nor does it affect intracellular calcium (15). Rather, it slows ATP hydrolysis within the myosin motor (29), inhibiting active force generation and preventing contraction without adversely affecting the underlying sarcomeric structure. Prevention of systolic stiffening by blebbistatin treatment strengthens the conclusion that changes in elastic modulus observed in isolated, beating cardiomyocytes were due to activation of actin-myosin cross bridges.
Effects of culture time on cardiomyocyte mechanics.
Cell culture time varied over a range of 5–20 days. Because changes in neonatal cardiomyocyte phenotype have been documented during the first weeks of culture (8), the average measured diastolic and systolic moduli for each cell were correlated with the number of days between cell isolation and testing (Fig. 5). The systolic modulus exhibited weak but significant stiffening with increased culture time (R2 = 0.55, P = 0.04). While the diastolic modulus also showed a trend toward stiffening, the correlation was not statistically significant (R2 = 0.14, P = 0.36). In this study, we observed no apparent systematic changes in myofibrillar organization in cardiomyocytes cultured for different times on aligned collagen-coated substrates. However, some studies have documented a phenotypic maturation of neonatal cardiomyocytes over a similar time course (8), whereas others have documented a postnatal isoform switch in the sarcomeric stabilization protein titin (30), which impacts longitudinal passive stiffness (20), although the effects on transverse myocyte properties remain unknown. Although beyond the scope of the present study, identification of mechanical manifestations of specific changes in cardiomyocyte molecular expression, whether due to culture conditions or disease processes, represents an important application of these methods for future investigations.
Spatiotemporal mapping of cardiomyocyte mechanics.
In myocytes (n = 4) indented along their major axis of contraction (Fig. 6A), depth-dependent modulus vs. time maps (Fig. 6B) revealed greater spatial heterogeneity during systole, as the intracellular variation in modulus (i.e., standard deviation of modulus values within a given cell) was 6.9 ± 4.4 kPa during systole vs. 2.0 ± 1.2 kPa during diastole (P < 0.01; Fig. 6C). Thus, whereas the passive myocyte is mechanically relatively soft and uniform, contraction not only increases stiffness but also induces periodic nonuniformity of elastic properties along the length of the cell, with ∼2-μm spacing between periodic stiff sites. There was a significant correlation between the spatial elasticity pattern during systole and the underlying sarcomeric actin staining (R2 = 0.52, P < 0.001; Fig. 6D), which did not correlate with the diastolic elasticity pattern (R2 = 0.05, P = 0.91, not shown). These data reflect the first in vitro observations of transverse stiffness heterogeneity in living, beating cardiomyocytes, which had previously only been observed in isolated myofibrils (1, 40).
Interestingly, systolic stiffening was first detected deep within the cell, followed by increases in Êapp at shallower indentation depths (Fig. 6B; for full view of mechanical dynamics, see Supplemental Video 1). One interpretation is that spontaneous cardiomyocyte contraction progresses from the inside out, rather than occurring uniformly throughout the cell thickness. Alternatively, because deep indentations follow shallow indentations necessarily, this could simply reflect a systematic delay in timing between the earliest and latest portions of an indentation test performed at a frequency similar to the cell pulse rate. To resolve this issue, we considered the following. First, the AFM probe was intentionally positioned relative to the cell surface, such that the actual postcontact indentation time is a relatively small fraction (10–15%) of the total advance-retract cycle. For a hypothetical cell beating at 4 Hz and indented at a similar frequency, the estimated total indentation time would, therefore, represent <15% of the beating period of 250 ms, or 37.5 ms. In comparison, systolic shortening occupied approximately one-half of the twitch cycle, or 125 ms. Thus ∼30% of cell contraction occurred as the AFM tip advanced into the cell. Second, in preliminary studies on two cells probed at 14 and 28 Hz, the high-frequency indentations revealed a systolic modulus that was not uniform with depth, but exhibited depth-dependent behavior similar to the slower compound frequency indentations (see Supplemental Fig. 2). Thus we favor the interpretation that systolic stiffening of cultured neonatal rat cardiomyocytes progresses from the inside of the cell toward the periphery. Such intracellular spatiotemporal heterogeneity could arise from spatially heterogeneous calcium handling observed in myocyte electrophysiology studies (21). By coupling AFM elastography of beating cardiomyocytes with fluorescent transfection of cytoskeletal components, live calcium imaging, and high-powered optical microscopy, molecular mechanisms responsible for such observations can be resolved. Furthermore, combining this platform with targeted modification of specific cytoskeletal components, such as switched isoforms of sarcomeric support proteins (30), or altered expression of microtubules (48) or other cross-linking proteins (46), may offer new insights into the molecular basis of systolic and diastolic heart disease.
Comparison with previous AFM studies.
Table 1 summarizes Young's modulus values reported for several rat cardiac structures measured using AFM. For comparison, a Poisson's ratio of 0.5 was used to convert our pointwise apparent modulus, Êapp, to an equivalent elastic Young's modulus, E (see materials and methods). Neonatal rat cardiac myofibrils were approximately three- to fivefold stiffer when isolated on glass (1) than when embedded within a cytoplasmic milieu of beating cardiomyocytes, although sarcomere-related mechanical heterogeneity could still be detected in the intact cell. The diastolic neonatal myocyte modulus was also 3–3.5 times softer than that for isolated passive adult cardiomyocytes (32), which may reflect more densely packed myofibrils in adult cells, as well as increased cardiomyocyte stiffness with aging. Finally, noninfarcted adult rat myocardium (6) exhibited a Young's modulus intermediate between the diastolic and systolic values obtained in the present study, possibly indicating partial contracture of the sectioned myocardial tissue sample that was reportedly tested in standard culture media without addition of blebbistatin or 2–3-butanedione monoxime.
No prior AFM studies have quantified the systolic stiffness of cardiomyocytes or myocardium. However, the mean ratio of systolic to diastolic stiffness of beating myocytes in the present study was 4.4 ± 2.7, compared with corresponding ratio values ranging from about 10 for sectioned adult rabbit myocardium (34), to nearly 40 in the beating human heart (45). The discrepancy could reflect factors such as extracellular matrix contributions, cell-to-cell interactions, or three-dimensional structural and mechanical anisotropy, none of which are present in isolated myocyte cultures. Also, because cardiomyocyte stiffness increases with age (32), development from neonate to adult may also alter mechanical properties of the intact heart (33). In principle, our AFM technique could be applied to the beating neonatal heart for direct comparison of cardiomyocyte and myocardial diastolic and systolic properties to investigate such issues. Such multiscale studies could also relate ventricular chamber stiffness to intrinsic cellular mechanical properties, which is essential for understanding the cellular and extracellular mechanisms regulating biomechanical feedback during heart failure and remodeling (5).
The method developed in this study combines AFM stiffness data from multiple cardiomyocyte contractions to construct a time-dependent mechanical profile at each location on the cell. There are limitations to this technique. In practice, the total time for obtaining adequately sized arrays was ∼2 min, which necessitates stable contractions for that given time. The technique could be modified using high-speed indentations to map cell stiffness during a single contraction. However, in proof-of-principle studies using 14- and 28-Hz indentations, cellular integrity was compromised, and only a few contractions were successfully mapped in this manner. In addition, the measured cardiomyocyte mechanical properties were a function of both indentation depth and time during the twitch cycle, which can be difficult to separate. Furthermore, finite element simulations looking at the effects of motion artifacts on AFM measurements (specifically, simultaneous lateral contraction and thickening of an incompressible material during indentation) suggested errors on the order of 10–20% in the estimated pointwise modulus, which were relatively small compared with the 300% increase in stiffness during systole. It should be noted that depth-dependent changes in diastolic and systolic moduli for cells tested at the near-pulse-rate compound frequency were comparable to measurements obtained during the preliminary high-speed indentations, suggesting that the apparent depth dependence of mechanical contraction was not simply an artifact of the technique (see Supplemental Fig. 2).
Another limitation is that, because data were pooled from multiple cells, regardless of their spontaneous pulse rate, any effects due to cell viscoelasticity (42) or frequency-dependent contractile processes (2) were hidden within the averaged stiffness values. Although the rate of deformation for these dynamic tests was not standardized, the technique can accommodate electrical pacing for rate-controlled experiments. However, then the measured properties would not be of spontaneously beating cells, which prior AFM studies suggest may differ from uniformly contracting paced cells (14). In preliminary studies where electrically paced cells were indented using the compound frequency method, no substantial difference was noted between the depth-dependent pointwise modulus of spontaneously beating and paced cells; nevertheless, these cells were not added to the analysis pool reported in this study. Finally, this method is prone to a “homogenization effect,” similar to other AFM indentation experiments, where higher indentation depths probe a larger volume of substrate, therefore losing spatial resolution. It should be noted that the technique can also be enhanced by introducing live-cell fluorescent cytoskeletal probes, which would greatly facilitate alignment of the indentation path with the sarcomeric axis, improving experimental efficiency and reducing postprocessing time.
In conclusion, we present herein the first quantitative measurements of transverse spatiotemporal stiffness mapping in isolated, spontaneously contracting neonatal rat cardiomyocytes using a new AFM dynamic elastography technique. The apparent elastic modulus varied continuously through the twitch cycle, yielding equivalent Young's moduli that ranged from ∼12 kPa in diastole to 40 kPa in systole. Subcellular elastic properties exhibited higher spatial heterogeneity during systole than diastole, with active stiffening apparently initiated from deep within the cell. Blebbistatin treatment prevented contractions completely and reduced mean depth-dependent cellular stiffness to diastolic values, consistent with activation of actomyosin cross bridges causing the increase in cellular transverse stiffness during systole. Moreover, the periodic sarcomeric structure of the underlying molecular contractile apparatus could be detected as variations in stiffness along the length of the living cell. This may provide mechanical cues to neighboring cells in intact myocardium, possibly facilitating registration of sarcomeres in adjacent cardiomyocytes or impacting transverse active force generation and other aspects of myocardial structure and function.
Funding was provided by the National Science Foundation (CAREER Award BES-0239138, K. D. Costa).
I am not aware of financial conflict(s) with the subject matter or materials discussed in this manuscript with any of the authors, or any of the authors’ academic institutions or employers.
The authors thank Dr. Jeffrey W. Holmes for helpful discussions, Drs. Eun Jung (Alice) Lee and Do Eun Kim for assistance with myocyte isolation, and Drs. Barclay Morrison, III and Benjamin S. Elkin for technical assistance with high-speed video equipment.
- Copyright © 2010 the American Physiological Society