Self-organizing human cardiac microchambers mediated by geometric confinement

Abstract

Tissue morphogenesis and organ formation are the consequences of biochemical and biophysical cues that lead to cellular spatial patterning in development. To model such events in vitro, we use PEG-patterned substrates to geometrically confine human pluripotent stem cell colonies and spatially present mechanical stress. Modulation of the WNT/β-catenin pathway promotes spatial patterning via geometric confinement of the cell condensation process during epithelial-mesenchymal transition, forcing cells at the perimeter to express an OCT4+ annulus, which is coincident with a region of higher cell density and E-cadherin expression. The biochemical and biophysical cues synergistically induce self-organizing lineage specification and creation of a beating human cardiac microchamber confined by the pattern geometry. These highly defined human cardiac microchambers can be used to study aspects of embryonic spatial patterning, early cardiac development and drug-induced developmental toxicity.

Figure 1. Spatial differentiation of patterned hiPSCs.

(a) Schematic of the fabrication process for PEG-patterned substrates confirmed by ToF-SIMS imaging to show the (b,c) PEG-related peak and TCPS-related peak. Scale bar, 50 μm . (d) Only the etched region can adsorb bovine serum albumin. Scale bar, 600 μm. hiPSCs were patterned as (e–g) arrays of circles with different diameters (scale bar, 600 μm), and (h–m) they maintained their pluripotency (scale bar, 200 μm). After CHIR treatment, the confocal images showed (n) cells retained OCT4 expression as an annulus adjacent to the pattern perimeter with (o,p) higher E-cadherin expression and (q,r) higher cell density. Scale bar, 50 μm. Data represent as the means with error bars s.d. with n=20 individual patterns. Statistical comparison was made between the centre and perimeter using two-sided Student's t-test. *P<0.05. (s) The heatmap of cell density for the 400-μm circle showed that the highest cell density occurred at the interface between OCT4+ and OCT4− cells. (t–v) The OCT4+ annulus was observed on all patterns. Scale bar, 200 μm. (w,x) Patterns with 200-μm diameters had highest cell density and highest OCT4+ cells compared with 400- and 600-μm diameter patterns. Data represent as the means with error bars s.d. with n=20 for each pattern size. Statistical comparison was made among different pattern sizes using one-way analysis of variance (ANOVA) with post hoc Tukey tests. *P<0.05.

Figure 2. Biophysical cues directed the spatial differentiation.

(a) Single-cell tracking showed that cells in the centre of the pattern had higher potential to migrate directionally towards the perimeter during the EMT with a higher (b) migratory velocity, motility coefficient, persistent time and percentage of ‘biased' cell migration. Data represent as the means with error bars s.d. with n=50 for five individual patterns multiplied by 10 cells within each pattern. Statistical comparison was made between centre and perimeter using two-sided Student's t-test. *P<0.05. (c) Gene expression for cells on the 200- and 600-μm diameter circle patterns, normalized to the 400-μm diameter patterns, before and after CHIR treatment. (d) EdU staining to visualize the cell proliferation on 400-μm diameter patterns. (e) More EdU+ cells were located near the pattern perimeter compared with the centre. Data represent as the means with error bars s.d. with n=20 individual patterns. Statistical comparison was made between the centre and perimeter using two-sided Student's t-test. *P<0.05. (f) Confocal images of 200-μm patterns treated with CHIR and different compounds that modulate cell mechanotransduction. (g) A proposed mechanism of OCT4+ annulus formation on the patterns relies on cell condensation during the EMT, via a combination of cell migration and proliferation, to generate a high-cell-density annulus adjacent to the pattern perimeter. Scale bars, 100 μm.

Figure 3. Generation of spatially organized 3D cardiac microchambers.

(a) 3D cardiac microchamber generated from WTC hiPSCs on a 400-μm pattern, where cardiomyocytes only appeared in the centre and myofibroblasts on the perimeter. Panels above and to the right of the main panel represent the z axis projection images at their respective x and y cross-sections. (b) Two-photon microscopy image of the void inside the 3D cardiac microchamber with the cell-free chamber shown at two chamber heights (plane a and b). The z axis projection images at their respective x and y cross-sections are shown in panels above and to the right of the main panel. (b) The image through plane a, and the panel in the upper right shows the image through plane b. The cardiomyocytes in the centre were positive for (c) cardiac troponin T, (d) sarcomeric α-actinin and (e) myosin heavy chain, whereas the myofibroblasts on the perimeter were positive for (c) SM22, (d) calponin and (e) smooth muscle actin. (f) Spatially organized cardiac microchambers were also generated on a triangle pattern with 540-μm leg length and similar area to a 400-μm circular pattern. Scale bars, 100 μm.

Figure 4. Assessment of cardiac developmental toxicity.

(a) 3D cardiac microchamber generated from WTC hiPSCs on a 600-μm circular pattern. (b) Microchamber generated from WTC hiPSCs on a 600-μm circular pattern after exposure to Thalidomide during the cardiac differentiation. Panels above and to the right of the main panel represent the z axis projection images at their respective x and y cross-sections. (c) Thalidomide exposure reduced the cardiac differentiation efficiency, (d) lowered the microchamber height, (e) reduced the microchamber half width half maximum, (f) decreased the beat frequency and (g) decreased the contractility (for example, contraction velocity). Data represent as the means with error bars s.d. with n=6 for each experimental condition. Statistical comparison was made between the control group and drug group using two-sided Student's t-test. *P<0.05. Scale bars, 100 μm.