Direct-Write Assembly of 3D Silk/Hydroxyapatite Scaffolds for Bone Co-Cultures

Three-dimensional (3D) microperiodic scaffolds composed of a gradient array of silk/hydroxyapatite (HA) filaments are fabricated by direct-write assembly. This approach allows for precise control over the printed filament size and location within each scaffold. These 3D silk/HA scaffolds are used to support the growth of co-cultures of human bone marrow derived mesenchymal stem cells (hMSCs) and human mammary microvascular endothelial cells (hMMECs) to assess in vitro formation of bone-like tissue. Histology reveals that the hMSCs and hMMECs form intricate networks of extracellular matrix within these 3D scaffolds. Vascular-like structures are only observed in co-cultures of hMSCs and hMMECs. When the spacing between silk/HA filaments is approximately 400 μm, the vascular-like structures follow the scaffold morphology. Direct writing of 3D scaffolds offers a novel route for investigating the effects of scaffold architecture and cell composition on human tissue formation.

There is a close connection between bone healing and blood vessel growth. In a fracture, damage to blood vessels induce a cascade of events, including the formation of a hemotoma and the recruitment of stem cells and fibroblasts to the wound site that trigger bone formation.^[1] Bone formation requires adequate vascularization and inhibition of angiogenesis results in decreased bone mass.^[2] In implant studies, bone vascular ingrowth in vivo is usually too slow to ensure the survival of graft and growth of new bone.^[3] To understand the interaction between angiogenesis and osteogenesis as well as to construct pre-vascularized bone constructs in vitro for bone repair, a systematic approach to the interplay between vasculargenesis and osteogenesis is needed.

The properties of biomaterials play an important role in controlling cell attachment, alignment, proliferation, differentiation, and, ultimately, functional tissue generation. Besides chemical composition, which has been widely investigated, the physical properties of biomaterial scaffolding, such as geometry, pore size, porosity and connectivity, have significant effects on tissue regeneration.^[4–6] For example, in osteogenesis, previous studies showed that honeycomb-shaped hydroxyapatite initiated higher bone formation than porous particles and porous blocks.^[7] Pore size did not significantly not affect osteogenesis in vitro, however new bone formation required larger pore sizes (>350 μm) for cell migration and angiogenesis in vivo.^[5] As with osteogenesis, the size and shape of pores can affect vascularization. Smaller pores (5–20 μm) favor endothelial cell growth in vitro in 3D scaffolds,^[8] while in vivo studies demonstrate that larger pores result in better vascularization. It was also found that 300–400 μm was the critical pore size above which caplillaries could be observed in newly formed bone.^[9–12] Optimizing pore size is therefore essential for tissue studies in vitro and in vivo, and biomaterial scaffolds with controlled porosity gradients would be useful for the study of vascularized bone. Scaffolds with gradient pore sizes allow a side-by-side comparison of the impact of pore sizes, and also provide the possibility for formation of multiple types of tissues and tissue interfaces in one system, such as skeletal tissues composed of both vascularized bone and non-vascularized cartilage.

To investigate the effect of biomaterials architecture, including pore size and shape, on the bone tissue engineering from various materials, accurate control of scaffold fabrication is required. To date, many approaches have been reported, from porogen leaching and electrospun matrices, to micromolding and related techniques. One promising method relies on directly writing the biomaterial of interest at the microscale.^{[13, 14]} In this approach, a micronozzle is attached to a computer-controlled translation stage that moves in programmable 3D pattern and deposits the biomaterial ink only in desired locations. Various materials including poly(2-hydroxyethyl methacrylate) (pHEMA),^[15] polyacrylic acid, (PAA), polyethylenimine (PEI)^[13] have been used as ink solution in the direct writing technique.

Silk fibroin is a biomaterial that exhibits good biocompatibility, impressive mechanical strength, controllable degradation, versatility in processing and modifications.^[16–19] The use of silk as a supporting biomaterial for bone formation is documented both in vitro and in vivo.^[20–24] Hydroxyapatite (HA) is a naturally occurring mineral in bone commonly termed apatite. It is commonly used for bone repair or ingrowth. Pre-mineralized silk scaffolds have been studied for bone formation ^{[21, 25, 26]} and the results suggested that mineralized porous scaffolds enhance osteogenesis from human bone marrow derived mesenchymal stem cells (hMSCs), resulting in better calcium deposition as a measure of more mature bone outcomes. However, increased mineralization also results in smaller pores and subsequently reduced transport, which can reduce cell ingrowth and function in the center of scaffolds, resulting in necrosis over time. Hence, further studies are needed to fully optimize pore sizes or porosity gradients for bone related tissue outcomes.

Here, we create a new silk fibroin ink filled with HA particles to print 3D scaffolds with gradient pore spacings ranging from 200 μm to 750 μm. Specifically, orthogonal periodic layers composed of silk/HA filaments (200 μm in diameter) are patterned in arrays with increasing spacing between each filament, to probe a variety of pore geometries on osteogenesis and vasculargenesis in one system. We first independently assessed osteogenesis with hMSCs and angiogenesis with hMMECs on these 3D silk/HA scaffolds, and then assessed co-cultures of these two cell types.

We first focused on designing a silk/HA ink with tailored rheological properties for printing 3D scaffolds with controlled pore architectures. This ink is created combining a silk fibroin solution with a concentrated HA suspension, whose respective rheological properties are shown in Figure 1. The silk solution is exhibits nearly Newtonian behavior with a low shear viscosity of ~ 8 Pa•s and a plateau shear elastic modulus of roughly 10 Pa. By contrast, the HA suspension is a viscoelastic colloidal gel that exhibits a low shear apparent viscosity of ~ 600 Pa•s, strong shear thinning behavior and a plateau shear elastic modulus of nearly 10³ Pa. Once combined, the silk/HA ink possesses an even higher apparent viscosity of ~ 10⁴ Pa•s under low shear conditions and a plateau shear elastic modulus of nearly 10⁵ Pa (Figure 1). Notably, this ink flows readily when extruded through a micronozzle (200 μm in diameter) under an applied pressure, yet quickly regains elasticity to maintain its filamentary shape upon exiting the nozzle.

Using this ink, silk/HA scaffolds are patterned by direct-write assembly in the form of a 3D lattice composed of interconnected silk/HA filaments with an interpenetrating gradient network of pore channels, whose spacing increases from 200 μm to 750 μm in increments of 50 μm across each printed layer. As a consequence, the scaffolds also contain a gradient in pore volume that increases from approximately 50 to 80%, respectively. An optical image of a representative silk/HA scaffold is shown in Figure 2a. Individual silk/HA filaments, imaged by scanning electron microscopy, span across underlying layers with little to no deformation (Figure 2b). Interestingly, the silk fibroin appears to promote bonding at each filament intersection. A higher magnification image of the surface of an individual filament (Figure 2c) reveals the presence of HA particles and the silk proteins that bind them together. The root-mean-square, surface roughness of the silk/HA filaments is 467 ± 61 nm, as determined by AFM (Figure 2d).

The silk/HA scaffolds have sufficient mechanical integrity to maintain their structure during the printing process, however it is necessary to strengthen them for cell culture experiments. Towards this objective, the scaffolds are immersed in a methanol-rich solution to promote the transformation of the silk proteins to the B-sheet conformation. Individual silk/HA filaments printed into a methanol-rich reservoir and dried exhibit an elastic module of 223 ± 9 MPa, which is three orders of magnitude higher than the shear elastic modulus of the silk/HA ink (0.1 MPa) and within an order of magnitude of methanol-treated electrospun silk-PEO blends (1300 MPa). ^[27]

The attachment and alignment of hMSCs were assessed on single layer of the scaffolds after 3 days and compared to cells grown on TCP (Figure S1). At day 3, hMSCs attached on both TCP and the silk/HA print surfaces. The cells showed star shapes and random orientations on the TCP (Figure S1b), while on the silk/HA scaffolds the hMSCS aligned along the direction of the rods and grew parallel within the spacings (Figure S1a).

Figure 3 shows that pores with characteristic sizes ranging from 200 to 750 μm are filled with cells and ECM after 6 weeks. Parallel cell strips formed between the patterned filaments and these strips have the same size as pores in which they are located within the 3D scaffolding. H&E staining confirmed cell attachment on silk/HA filaments. Cells grow on the printed filament surfaces and form tissue-like structures between them (Figure 3). More interestingly, organized swirl-like structures were observed within the ECM in the squares between adjacent filaments (Figure 3c), which resemble cell orientation in canals of human long bone. To further evaluate the osteogenesis of hMSCs, bone specific markers including collagen type I, osteocalcin and mineralization are assessed by immunostaining, von Kosa and SEM. Figure 4 showed that after 6 weeks under osteogenic conditions, collagen type I accumulated throughout the scaffold within all pore spacings. Figure S2 confirmed cell growth and tissue formation based on H&E staining and further indicated mineral deposition by cells. However, no positive staining was found by Von Kossa or for osteocalcin. In future work, longer term cultures, hypoxia and the inclusion of BMP-2 may provide a more detailed assessment of osteogenesis, as has been shown in other studies ^{[7, 19, 28]}.

The morphology of endothelial cells in the silk/HA scaffolds is characterized by confocal microscopy. The cells were cobblestone shaped after seeding in the scaffolds (data not shown). At day 4, the cells filled the pores between adjacent silk/HA rods within the scaffold and attached to them from the smallest (200 μm) to largest (750 μm) sizes (Figure 5a, b). The first sign of small aggregates of aligned cells without identifiable lumens were observed within pores of all sizes. Moreover, in the medium pore size (400 μm), a complex interconnecting network of features with lumen-like structures are starting to form, observed both by phase-contrast mode (Figure 5d) and by live staining of the cells (Figure 5c).

The lumen-like structures formed by the endothelial cells have been shown previously in vitro in angiogenesis studies with different scaffolds, such as collagen, fibronectin, hyaluronan or Matrigel ^[29–31]. However, most in vitro work was performed on two-dimensional supports or hydrogels rather than the more rigid 3D scaffolds investigated here. We found network formation in these scaffolds within 10 days with a preference for a 400 μm pore size. In vivo studies have shown that vascular tissue advance into substrates faster as the pore diameter increased ^[3], with a diameter of 250–300 μm found most effective for new vessel ingrowth among the pore sizes ranging from 20 μm to 300 μm.^[28] However, there are no prior studies regarding the effect of pore sizes above 300 μm. hMSCs have been shown to support the formation of stable vascular tubes.^{[1, 2, 14]} We observed similar results; tubular structures formed within all pore sizes evaluated. Further, cross-shape structures formed only at 400 μm spacing. Histology staining for CD 31 confirmed that the cross-shape structure had endothelial characteristics. However, more data is needed to determine if these features consist of a single cell type.

After 2 weeks culture, network-like structures are observed within co-cultures of hMMECs with hMSCs, features that are not present in monocultures of hMSCs (data not shown). In the hMMECS and hMSCS co-cultures, cross-shaped structures form only in regions with 400 μm pore sizes (red arrows in Figure 6a). To further characterize the structures, the endothelial specific marker CD31 was detected by immunostaining and observed on the cross shape body, providing evidence for a lumen-like feature of endothelial cells (Figure 6b). For other pores sizes, for example 750 μm, random network structures were observed by positive staining for CD 31 (Figure 6c).

3D HA/silk scaffolds with gradient pore size ranging from 200 μm to 750 μm were constructed by direct-write assembly. These silk/HA scaffolds offer significant advantages. First, this patterning technique provides both versatility and the posibility of control over the scaffold composition, size, and architecture. These advantages can facilitate the study of angiogenesis and osteogenesis to optimize scaffold features. Second, by combining HA, a good matrix for hMSCs osteogenesis, with silk to promote endothelial cell growth and migration, we have created scaffolds capable of supporting both stem cell and endothelial cell functions to allow new tissue formation and bone remodeling with vascular inputs within a single construct environment.

The time of culture is important with respect to the formation of bone and vascular-like structures; hence, long-term studies using these 3D silk/HA scaffolds would be useful. One challenge identified in this work is that the number of cells present within features of varying pore size limits options for analysis. For example, within a given scaffold feature, few cells can be easily recovered for more extensive qPCR and related analyses. We therefore view these biomaterial constructs as a valuable initial screening tool to identify the most suitable pore size ranges/morphologies with which to pursue more detailed studies of tissue-related outcomes.