A. Characterization of ANFs

Nanofibers generated from three electrospinning solutions at different concentrations were observed under a light microscope and SEM, as shown in Fig. 3. The fiber diameter and fiber distribution were analyzed by imagej. The average diameters of 3%, 5%, and 8% PLGA fibers were about 200, 300, and 420 nm, respectively. Notably, the 3% PLGA fibers were easy to break and to form beads, whereas 5% PLGA fibers showed the most concentrated diameter. Therefore, 5% PLGA was chosen to conduct further experiments.As shown in Fig. 4, the fiber density, fiber orientation distribution, and fiber CI (ability to intercept the cells) of ANFs under different electrospinning times were observed and measured by imagej. Figures 4(a) and 4(d) indicate the time-dependent pattern with the fiber density increasing from 13, 28, 84 to 211 fibers/100 μm when the time increased from 30 s, 1, 2 to 3 min, respectively. Interestingly, the corresponding CI values showed a remarkable difference. The CI of the ANFs with electrospinning time of 30 s and 1 min were 0.15% and 1.5%, respectively, whereas the electrospinning time increase has significantly increased CI values reaching 67.5% over 2 min and 87.8% over 3 min. Fiber densities with electrospinning time longer than 2 min could intercept more than 65% of cells. It should be noted that an appropriate fiber spacing is conducive to the contact of interlayer cells and, thus, excessive fiber density is not necessary.

B. Cytocompatibility of ANF/Collagen and ANF/GelMA-PEO models

1. CFs behavior in ANF/Collagen and ANF/GelMA-PEO models

The effects of ANFs and different hydrogels on cell morphology and viability were first evaluated on CFs cell line. The morphologies of seeded-CFs on monolayer-ANF are shown in Fig. 5(a). The CFs showed good attachment to the ANF, with most cells exhibiting an elongation in the direction of the aligned nanofibers just 2 h postseeding. After embedding cell-ANF with collagen or GelMA-PEO, cell viability and morphology were investigated over a 7-day culture period, using a live/dead viability assay and F-actin immunostaining. As Fig. 5(b) shows, CFs displayed a good morphology and high viability in ANF/Collagen model during days 2, 4, and 7 of culture. Precisely, more than 95% were living cells throughout the culture, indicating a significant cell proliferation rate related to the increased area covered by cells. In contrast, CFs seeded on the ANF/GelMA-PEO model were slim and short with a cell viability lower than 80%, indicating an unobvious cell proliferation. Similarly, the results of F-actin immunostaining demonstrated the difference in cell morphology of seeded-CFs on these two ANF/Gel models, where CFs in ANF/Collagen model showed better cell spreading and oriented cytoskeleton.

2. HL-1 cells in ANF/Collagen model

HL-1 cells possess similar properties to cardiomyocytes, their culture requires more stringent conditions compared to CFs. HL-1 was cultured in the bilayer-ANF/Collagen model. In order to illustrate the cellular structure of HL-1 layered on the orthogonal double layers of 3D ANF/Collagen model, live-cell staining by Calcein-AM on day 6 of culture was performed [Fig. 6(a)]. The results showed that cells directionally grew in accordance with the direction of each layer of ANFs. Moreover, the cross section of bilayer-ANF/Collagen model revealed a layered structure with spacing between layers of about 60 μm. On the other hand, the immunostaining of α-actinin and nuclei of HL-1 on day 7 of culture revealed an uneven distribution of α-actinin in the cells and difficulty to observe the sarcomere structure. This was explained by the high cell passages of HL-1, resulting in the cytoplasm containing many vacuoles.2222. P. Dias et al., PLoS One 9, E90266 (2014). https://doi.org/10.1371/journal.pone.0090266 The immunostaining results demonstrated that the 3D culture system of PLGA-ANF with collagen hydrogel could induce the expression of cardiomyocyte-related components in HL-1.

The comparison between the two hydrogels showed that collagen had better cytocompatibility than GelMA-PEO hydrogel for both CFs and HL-1 cells, in terms of cell morphology and proliferation ability. As for the structure, ANF could significantly promote the directional growth and spreading of seeded cells. The layered ANFs stack encapsulated in hydrogel could also form an anisotropic layered structure with an appropriate spacing between the layers.

C. Discussion

Many tissues possess a multilayered anisotropic microstructure, which is complicated to replicate in vitro. Based on the theory of bionics, herein, this study has proposed a 3D cell/scaffold model composed of layers of aligned nanofiber films encapsulated in hydrogels. This 3D cell/scaffold model can provide a biomimetic anisotropic structure as a 3D environment for multiple cells to promote intralayer cellular arrangement and interlayer tissue organization, thereby expressing similar features of natural tissues.

The conventional aligned electrospinning apparatus using parallel electrodes usually has collectors with a small gap area for nanofiber deposition. To address this issue, this study designed a collector with several gaps, which have an area of about 100 cm2 to allow easy and quick fabrication of transferable and aligned nanofiber films. The electrospinning parameters including solution concentration and spinning time were optimized to achieve highly aligned nanofibers with a proper fiber density. With electrospun at 12 kV and the PLGA electrospinning solution at 5%, nanofibers with a suitable fiber diameter (about 300 nm) and a good morphology could be produced. Based on the cellular culture, the appropriate fiber density of ANF was selected. The electrospinning time of 2 min generated a fiber density of 84 fibers/100 μm, which allowed more than 65% of the seeded cells to be carried by the thin fiber films. This fiber density was just suitable to prevent cell contact between the layers. Also, ANF was highly aligned as demonstrated by the quick spreading and the high alignment of seeded-CFs in the fibers' direction just 2 h after cells seeding.

Then, collagen and GelMA-PEO hydrogels were used and compared for constructing 3D ANF/Gel models. PEO was used to improve the porous structure of GelMA.2020. D. Loessner et al., Nat. Protoc. 11, 727 (2016). https://doi.org/10.1038/nprot.2016.037 The results of this study showed that the model with GelMA-PEO hydrogel did not present superior biocompatibility compared to collagen. The CFs proliferated well in the ANF/Collagen model and formed aligned and elongated cytoskeleton arrangements.

In the construction of multilayered ANF/Gel, the deflected fiber orientations of each layer were achieved by rotating the ANF frames before embedding them in the hydrogel. Accordingly, a bilayer-ANF/Collagen model seeded with HL-1 cells was fabricated, in which cells grew aligned with the directions of each ANF layer and exhibited good cell morphology and expression of the cardiac-specific protein (i.e., α-actinin). It showed that the ANF/Collagen model has good potential for constructing in vitro cardiac tissue. Through the 3D reconstruction by a confocal microscope, it was deduced that the spacing between the neighboring layers was about 60 μm. With this fiber density and interlayer spacing, the cells could not only fuse within each layer but also contact and migrate through layers, ultimately making the cell/scaffold model an integral 3D tissue. For further prospects, if more layers of 3D tissue with deflected angles need to be constructed, the transfer frame could be a circular ring marked with an angle index to precisely regulate the direction of ANF.