Novel micropatterned cardiac cell cultures with realistic ventricular microstructure

Abstract

Systematic studies of cardiac structure-function relationships to date have been hindered by the intrinsic complexity and variability of in vivo and ex vivo model systems. Thus, we set out to develop a reproducible cell culture system that can accurately replicate the realistic microstructure of native cardiac tissues. Using cell micropatterning techniques, we aligned cultured cardiomyocytes at micro- and macroscopic spatial scales to follow local directions of cardiac fibers in murine ventricular cross sections, as measured by high-resolution diffusion tensor magnetic resonance imaging. To elucidate the roles of ventricular tissue microstructure in macroscopic impulse conduction, we optically mapped membrane potentials in micropatterned cardiac cultures with realistic tissue boundaries and natural cell orientation, cardiac cultures with realistic tissue boundaries but random cell orientation, and standard isotropic monolayers. At 2 Hz pacing, both microscopic changes in cell orientation and ventricular tissue boundaries independently and synergistically increased the spatial dispersion of conduction velocity, but not the action potential duration. The realistic variations in intramural microstructure created unique spatial signatures in micro- and macroscopic impulse propagation within ventricular cross-section cultures. This novel in vitro model system is expected to help bridge the existing gap between experimental structure-function studies in standard cardiac monolayers and intact heart tissues.

Figure 1

Design of photomasks for microfabrication. (A) DTMRI-measured in-plane fiber directions of three murine ventricular cross sections, shown using the circular color map from B. (B) Angle map of a transverse ventricular cross section in MATLAB. For each pixel, the white line and pixel color denote the fiber direction and angle relative to the x axis, respectively. (C) Corresponding AS micropattern in AutoCAD, where the angle map was converted to a pixilated array of angled parallel lines. (D) Corresponding IS micropattern with identical macroscopic boundaries but solid, rather than lined, pixels. (E) Standard IC coverslips with circular boundary. White regions in C–E correspond to fibronectin-printed areas on the coverslip used to guide cell adhesion and local alignment.

Figure 2

Control of microscopic cardiomyocyte alignment. (A.1–3) Local cell alignment perpendicular to that of the surrounding area, shown for pixels of different size (square regions delineated by dashed box). Inset in A.3 shows the underlying pattern of fibronectin lines (green). (B.1–3) Local cell alignment within 205 μm pixels at different angles relative to vertical cell alignment in the surrounding area. Red, sarcomeric α-actinin; green, connexin43; blue, nuclei. White lines denote the direction of underlying fibronectin lines inside the pixels. (C) Angle differences between the direction of aligned cells and underlying fibronectin lines for different pixel sizes and angles of fibronectin lines relative to those in the surrounding area (average of N = 5 cultures). Black dashed line denotes the range of angle gradients (0–20°/pixel) exhibited by 90% of pixels from the maps shown in Fig. 1A. White dashed line denotes pixel size of 190 μm used for cell micropatterning in slice cultures.

Figure 3

Formation of realistic cardiac microstructure in AS cultures. (A–C) Plated cells were found to attach (A), spread and align along the underlying fibronectin lines (B), and by day 6 (C) form confluent cardiac fibers. (D) Composite image of the entire micropatterned slice culture. (E) Close-up of four adjacent pixels delineated by dashed lines, along with the underlying fibronectin pattern (green, inset). Note abrupt changes in cardiac fiber directions in neighboring pixels without loss of cell confluence.

Figure 4

Validation of micropatterning accuracy in AS cultures. DTMRI angle maps (first column), resulting cell culture angle maps (second column), angle difference maps (third column), and corresponding angle difference histograms (fourth column) shown for (A) middle-transverse, (B) apical-transverse, and (C) coronal ventricular sections from Fig. 1A. Circular color map refers to columns 1 and 2; linear color map refers to columns 3 and 4. Note the excellent agreement in microscopic (fiber directions) and macroscopic (tissue boundaries) structural features between cell cultures and corresponding DTMRI-measured ventricular cross sections.

Figure 5

Isochrone maps of impulse propagation in cell cultures. Ensemble average isochrone maps of 2 Hz activation in (A) ICs with stimuli applied at the center, as well as (B) isotropic and (C) AS cultures with stimuli applied at each of the four pacing sites. Pulse symbols denote pacing sites. Underlying quivers in AS cultures in C denote cardiac fiber directions. Note the predominantly circular, uniformly spaced activation patterns in all isotropic cultures (IC and IS) and elongated, lobed activation patterns in AS cultures. Movies of action potential propagation in IS and AS cultures are included in the Supporting Material (Movies S1 and S2).

Figure 6

CV maps during impulse propagation in cell cultures. Ensemble average CV maps in (A) ICs with stimuli applied at the center, as well as (B) IS and (C) AS cultures with stimuli applied at each of the four pacing sites. Pulse symbols denote pacing sites. Black arrows indicate regions of collision-induced increases in CV. Note the generally uniform spatial CV distribution in isotropic cultures (IC and IS) and halo of faster conduction around the LV in anisotropic cultures (AS) as well as faster conduction in the RV free wall. (D) CV histogram distributions (% of mean) and (E) corresponding CV dispersions (N = 21, 22, and 20 cultures with 21, 46, and 78 recordings, respectively). Asterisk denotes significant difference with p < 0.05. (F) CV magnitude profiles along the dashed lines in B.2 and C.2 (F.1), and in B.3 and C.3 (F.2) from endo- to epicardium. Local fiber directions around the dashed lines are shown in Fig. S3, G and H.

Figure 7

APD maps during impulse propagation in cell cultures. Ensemble average APD maps in (A) ICs with stimuli applied at the center, as well as (B) IS and (C) AS cultures with stimuli applied at each of the four pacing sites. Pulse symbols denote pacing sites. Representative action potential traces shown below A demonstrate decrease in APD with distance from the pacing site. (D) APD histogram distributions (% of mean) and (E) corresponding APD dispersions.

Figure 8

Dependence of local CV on angle (Φ) between direction of propagating wave and underlying cell alignment. (A.1–4) Distinct Φ maps result from pacing at each of the four sites in AS cultures. Data combined for all pixels and all pacing sites reveal significant effects of Φ on the magnitude of local CVs (B) but not APDs (C). For each Φ value (shown in 1° increments), the circle represents an average of all local CV(Φ) or APD(Φ) values. (D) CV versus Φ plots in subregions denoted by boxes in A.2 and A.3. Note that the strong dependence of local CV on the directions of propagation relative to the underlying cell alignment in one subregion exhibiting uniformly anisotropic conduction (D.1, slope = −0.15 cm/s/°, p < 0.001) can be overshadowed by wavefront collision (D.2, slope = −0.01 cm/s/°, p < 0.05). Local fiber directions in these subregions are shown in Fig. S3G. (E) Histogram showing the average distribution of Φ compiled for all pacing sites.