**4. Constitution of muscularis propria of the esophagus**

The muscle tissue of esophagus consists of striated muscle (skeletal) in the upper third, mix‐ ture of skeletal and smooth muscle in the middle third, and pure smooth muscle in the lower third. These muscle contents arrange into endo‐circular and exo‐longitudinal sub‐bilayers to play an important role in propelling the swallowed food or fluid into the stomach via mus‐ cle peristalsis. Generating an oriented muscle architecture to mimic the tissue of muscularis externa is an important issue to restore the functions of tissue‐engineered esophagus. Many researchers studied the relationship between scaffold's chemistry and microstructure and muscle cells' phenotype. Stegemann once verified that the behaviors of smooth muscle cell (SMC) were positively correlated to the scaffold geometry (2D and 3D) [60]. Li et al. believed that the scaffold geometry played an important role in modulating SMC phenotype. They cultured SMCs and discovered that cells in 3D collagen (type I) gels had lower proliferation and higher collagen synthesis than the cells in 2D collagen substrate [61]. Chan‐Park verified that smooth muscle α‐actin of SMCs cultured in microchannels upregulated greatly, sug‐ gesting a phenotype shift from synthetic to contractile state of cells [62]. They thus believed that 3D microchannels could encourage cells to reorganize into orientation patterns because SMC have a natural self‐arrangement propensity. Moreover, the narrow space of channels around 100 μm or less helped cells to achieve more uniform orientation. We also fabricated scaffolds with circular and longitudinal microchannel patterns (**Figure 5**). Further, the scaf‐ fold surface was grafted with silk fibroin using our method of diamine aminolysis and GA crosslinking. The primary esophageal SMC was cultured in these 3D protein‐grafted chan‐ nels in order to achieve SMC phenotype regulation and *in situ* muscle formation [63]. The results confirmed that primary esophageal smooth muscle cells exhibited fine alignment in all types of microchannels while SMCs in the interval channels communicated well through the gaps (**Figure 5**).

Some researchers had considered and investigated that mechanical stimulation might be an effective way to regulate SMC phenotype. Ritchie et al. designed a system to exert mechanical forces on esophageal smooth muscle cells. They discovered that cells on the flexile polyure‐ thane membrane displayed alignment parallel to the force direction when low cyclic strains (2%) was used, but alignment perpendicular to the force direction when high strains (5 and 10%) used [64]. Cha et al. reported that muscle cells would orient according to the optimal movement of the tissue. They adopted cyclic mechanical strain (a homemade stretching chamber) on primary myofibroblasts, and promoted the cell differentiation, and further mod‐ ulated the orientation and proliferation of the differentiated smooth muscle cell. Their conclu‐ sion was that myofibroblast/scaffold hybrids with cyclic strain could be applied to organize smooth muscle cells with muscle tissue functions [65].

**Figure 5.** Overview of tubular scaffold (a, inserted) and tube wall's cross‐section structure observed under SEM (a). (a1) Scaffold's morphologies of tube lumen and outer face, containing microchannels of 100 μm width with discontinuous channel walls intermitted by 30 μm gap, and both the wall thickness and depth are 30 μm; (a2) microchannels of 200 μm width with noninterval slits. (b) SEM picture of tubular scaffold with bulge wall (b1 and b2). The height of wall and gap between wall intervals is 30 μm. Esophageal smooth muscle cells were seeded and aligned in all scaffolds' channels while cells in the interval channels communicated through the wall gaps (a11, a21 and b11, b21). The scaffold was constructed through silica mold with predetermined patterns using biodegradable poly(ester‐urethane) as the substrate material. The surface was grafted with silk fibroin via the method of diamine aminolysis and GA crosslinking. Cells of a11 and a21 were stained by H&E and cells of b11 and b21 were immuno‐fluorescently stained with anti‐α‐smooth muscle as the primary antibody.
