Three-dimensional pore network characterization of reconstructed extracellular matrix

The extracellular matrix (ECM) has a fiber network that provides physical scaffolds to cells and plays important roles by regulating cellular functions. Some previous works characterized the mechanical and geometrical properties of the ECM fiber network using reconstituted collagen-I. However, the characterization of the porous structure of reconstituted collagen-I has been limited to the pore diameter measurement, and pore network extraction has not been applied to reconstituted collagen-I despite the importance of pore interconnectivity. Here, we aim to show the importance of characterizing the pore network of reconstituted collagen-I by comparing the pore networks of structures that have different fiber alignments. We show that the fiber alignment significantly changes the pore throat area but not the pore diameter. Also, we demonstrate that larger pore throats are directed in the direction of the fiber alignment, which may help in understanding the enhanced cell migration when fibers are aligned.

Figure 1

The overall process of the stochastic reconstruction is depicted schematically. (a) A 3D image of the reconstituted collagen-I is generated and (b) its statistics are extracted. (c) The statistics are used for the reconstruction. (d) By changing the statistics, it is possible to generate multiple structures.

Figure 2

Collagen-I structure is imaged using confocal reflectance microscopy. The collagen concentration is 2 mg/ml. One of the $z$-stack images with size $50\mu \mathrm{m}$ by $50\mu \mathrm{m}$ is presented in 2D. The scale bar is $5\mu \mathrm{m}$.

Figure 3

3D visualization of the collagen structure (a) imaged using confocal microscopy and (b) reconstructed using stochastic annealing. The $z$-stack images obtained from the confocal microscopy are stacked to create the 3D voxel image as shown in (a). The reconstructed structure is converted into a 3D voxel as shown in (b). The size of the voxel volume is $50\mu \mathrm{m}$ by $50\mu \mathrm{m}$ by $20\mu \mathrm{m}$ for both (a) and (b).

Figure 4

Comparison of network statistics of the imaged and reconstructed structures of the collagen network. (a) Valence number distribution and (b) fiber length distribution are shown.

Figure 5

Reconstructed structures of the collagen with (a) ${\mathrm{\Omega }}^{*}=0.6$ and (b) ${\mathrm{\Omega }}^{*}=0.87$ are presented. The structure with (a) ${\mathrm{\Omega }}^{*}=0.6$ shows isotropic direction of fibers, while the structure with (b) ${\mathrm{\Omega }}^{*}=0.87$ shows fibers aligned along the $z$ axis.

Figure 6

Schematics of identifying the pore network using pores and pore throats.

Figure 7

Geometrical properties of the pore networks of the target and reconstructed collagen networks. The pore properties include (a) pore diameter, (b) pore surface area, (c) pore volume, and (d) pore volume to area ratio. The pore throat properties include (e) pore throat area and (f) pore throat diameter.

Figure 8

The effects of the fiber alignment on (a) the pore diameter and (b) the pore throat area are presented. Structures with different mean cosines from 0.6 to 0.87 are compared. The minimum numbers of pores and pore throats are $n_{\mathrm{p}}=1300$ and $n_{\mathrm{pt}}=5100$. One asterisk indicates $p<0.05$, and two asterisks indicate $p<0.01$. The error bars represent the standard errors.

Figure 9

The angles between the $z$ axis and the normal vectors of the pore throats are shown. (a) The graphical representation of the angle, which is measured using the normal vector of the pore throat and the aligning axis. Two structures that have different mean fiber orientations, 0.6 (Isotropic) and 0.87 (Aligned), are compared. Pore throats are screened in their sizes with percentiles. Two asterisks indicate $p<0.01$, three asterisks indicate $p<0.001$, and four asterisks indicate $p<0.0001$. The error bars represent the standard error.