Cardiac Fibrotic Remodeling on a Chip with Dynamic Mechanical Stimulation

Background

Cardiac fibroblasts turn into contractile myofibroblasts in response to cytokines such as transforming growth factor-β (TGF-β) and to mechanical load, and persistent activation leads to cardiac fibrosis. The authors note that cardiac strain is cyclic at roughly 1 Hz and spans 2% to 20% depending on the pathological state, yet earlier reports on how cyclic strain affects myofibroblast activation disagree. Commercial stretchers apply only one strain magnitude at a time, a recent bioreactor reached two strains (2% and 8%) on a 2D polydimethylsiloxane (PDMS) membrane, and stiff culture substrates can activate myofibroblasts without any biochemical stimulus.

Design and approach

The PDMS microdevice stacks a nitrogen pressure chamber, a medium chamber with posts, and a cover glass. Cavities of decreasing diameter (8–5 mm) under the posts convert one nitrogen injection into graded vertical displacements, so that hydrogels on different posts receive different compression strains. CFs are encapsulated in 7% GelMA and crosslinked above the posts by 40 s of UV exposure; cell-free 7% GelMA crosslinked in this way has a Young's modulus of 9.8 ± 0.6 kPa, close to the 4 to 11 kPa reported for healthy neonatal rat heart extracellular matrix. Because the hydrogel is nearly incompressible, vertical compression is accompanied by lateral expansion that places the surrounding matrix in tension.

How it was evaluated

Primary CFs from two-day-old neonatal Sprague Dawley rats were cultured in the device under 1 Hz cyclic compression for 7 days. The authors measured proliferation (EdU labeling), spreading (F-actin staining) and mRNA expression of α-smooth muscle actin (α-SMA), collagen I (Col-I), TGF-β, fibronectin (Fn) and matrix metallopeptidase 2 (MMP-2), comparing passage 2 (P2) and passage 5 (P5) cells and 2D plates with 3D gels. In a fibrosis model, P5 CFs were pretreated with 10 ng mL−1 TGF-β1 for 24 h and then compressed with medium alone (M), with the TGF-β inhibitor Tranilast at 50 × 10−6 m (DM), or with continued TGF-β1 (TM), against an uncompressed, untreated control (C).

Key results

With air in the medium chamber, the device reached a maximum compression of 50% at 28 kPa. When the chamber was filled with medium and closed, strain was limited to 20% even at 35 kPa, whereas leaving the medium tubes open gave 20% strain at 28 kPa, and this open configuration produced 0%, 5%, 10%, 15% and 20% strain across the pillars. A frequency of 2 Hz also worked but gave smaller strains than 1 Hz. Encapsulating CFs raised the gel modulus to 13.3 ± 4.1 kPa without a significant difference from cell-free GelMA, and the modulus stayed similar over 7 days of compression at 37 °C.

Compression strain had no significant effect on CF proliferation, which increased over 7 days; cells on 2D plates grew faster on day 1, but this difference was reversed by day 7. Cell spreading increased with strain, and the spreading area at 20% strain was three times that at 0%.

P5 CFs expressed fibrotic markers at significantly higher levels than P2 CFs, and cells in 3D gels expressed lower levels than cells on stiff 2D substrates. For P2 CFs, marker expression rose with strain and reached a maximum at 20%, 4–6 times the level at 0%. P5 CFs responded differently: static 3D culture gave higher marker levels than 5% and 10% compression, and expression rose again at 15% and 20%.

In the TGF-β1 fibrosis model, the M group at 0% strain showed α-SMA, Col-I, Fn and MMP-2 expression 2.8, 2.9, 1.7 and 1.6 times the control. With Tranilast (DM), expression at 0% did not differ from the control, and only 15% and 20% strain raised α-SMA significantly (3.1 and 2.4 times). With continued TGF-β1 (TM), the corresponding values at 0% were 4.3, 5.1, 4.9 and 2.5 times the control, expression peaked at 20%, and 5% and 10% strain gave no significant increase over 0% for α-SMA and Col-I. The authors concluded that 5–10% strain maintained or attenuated myofibroblast activation, 15–20% strain promoted it, and compression acted synergistically with TGF-β without overriding it.

Open questions

The authors noted that the model lacks cardiomyocytes and their interplay with fibroblasts, and that fibrosis hallmarks and the function of the fibrotic tissue still need to be characterized in a more physiologically biomimetic model.

How it connects to other research

The eLife bioreactor cited by the authors stretched cardiac fibroblasts on a 2D PDMS membrane at two strains (2% and 8%) in combination with hypoxia, whereas this device compresses CFs inside 3D GelMA at five strain levels from one gas input and adds TGF-β1 or Tranilast to separate mechanical from biochemical cues.