The need for improved clinical strategies which incorporate bone grafting techniques to restore function to damaged or degenerated bone is well recognized. Tissue engineering strategies that combine porous biomaterial scaffolds with cells capable of osteogenesis or bioactive proteins have shown promise as effective bone graft substitutes. However, due to mass transport limitations, attempts to culture bone tissue-engineering constructs thicker than 1mm in vitro have typically resulted in a shell of viable cells and mineralized matrix surrounding a necrotic core. The overall goal of this project was to enhance the amount and distribution of cell-mediated mineralization throughout 3-D structural scaffolds in vitro using a perfusion bioreactor system. It was hypothesized that media perfusion would increase the amount, rate, and distribution of mineral deposited throughout cell-seeded scaffolds during in vitro culture compared with static control conditions. Using our initial bioreactor design that perfused cylindrical constructs transverse to their long axis, we found that the cell viability and proliferation of MC3T3-E1 osteoblast-like cells seeded on trabecular bone scaffolds were significantly affected by perfusion flow rate. Computational fluid dynamics (CFD) modeling techniques were employed to visualize the flow distribution of media and estimate local shear stresses within perfused constructs.^[1] As described in Chapter 4, flow field visualization revealed that transverse fluid flow through scaffolds with randomly interconnected pore structure was highly nonuniform and shunted to a few pathways of least resistance. More significantly, CFD modeling of axial perfusion through cylindrical scaffolds with a regular microarchitecture revealed a more uniform flow field distributed throughout the scaffolds. We therefore redesigned our bioreactor to provide axial perfusion and switched our base scaffold material to polycaprolactone (PCL), a polymer that could be fabricated via fused deposition modeling into a structure with a repeating lattice and uniform pores.

Comparing the effects of different culture conditions on mineralized matrix synthesis and distribution within perfused constructs has proved difficult without quantitative analysis tools. Micro-computed tomography (micro-CT) scanning and 3-D image analysis offers a non-destructive, quantitative technique for assessing mineral formation, distribution, density, as well as particle number and size. The global objective and central hypothesis were tested by evaluating mineralized matrix synthesis within axially perfused, PCL scaffolds, seeded with rat marrow stromal cells (rMSC). Although the PCL scaffolds improved the distribution of perfusion, initial studies indicated that poor cell adhesion to the polymer struts was a limitation, even at very low rates of perfusion. Therefore, in Chapter 5, we evaluated the effects of lyophilizing type I collagen into PCL scaffolds (PCL+C) or coating PCL with fibronectin (PCL+FN) on cell viability and mineralized matrix synthesis.

Micro-CT was used to detect and quantify mineralized matrix volume, spatial distribution, and density within both axially perfused and statically grown, cell seeded, 3 mm thick, PCL scaffolds (66% porosity) compared with static control constructs. Perfused constructs, which were shown to develop mineralized matrix, had not exceeded 1.5 mm in thickness.^[2] We therefore chose to test 3 mm thick constructs to more rigorously evaluate the benefits of perfusion. Significant increases in mineralized matrix deposition were detected within PCL+C constructs compared with scaffolds coated with fibronectin (PCL+FN) in both static and perfused culture conditions. More importantly, after 5 weeks of culture, continuous perfusion increased mineralized matrix deposition 3- fold within PCL+C constructs compared with static cultures. Additionally, mineral was distributed throughout the entire thickness of 3 mm thick PCL+C constructs. However, possibly due to stagnant flow at the periphery (visualized with CFD modeling), 90% of the mineral was localized to the central annulus of these constructs. In an attempt to improve fluid flow through a larger diameter of the scaffold, the lip on which the constructs rest in the bioreactor was reduced in size. These positive results in the 3 mm thick constructs directed our efforts towards increasing the length of the constructs.

As described in Chapter 6, we next evaluated the feasibility of creating constructs large enough for future implantation in a preclinical, critical size, rat segmental defect model. PCL+C scaffolds (3 mm thick), seeded with rMSCs, were stacked in the perfusion bioreactor to create constructs 3, 6 and 9 mm in length. After 5 weeks in culture, mineral was detected and quantified throughout the entire length of all construct sizes. We also quantified the mineral volume within two sequentially smaller cylindrical subregions (radially excluding the outer 500 and 1000 µm of the construct) within the scaffold to determine if mineralized matrix had formed at the center of the construct. As hypothesized, the number of mineral particles within each scaffold increased proportionally as the scaffold length was increased. In addition, the total detected mineral volume tripled as the construct length was increased from 3 to 9 mm. Therefore, increasing scaffold length did not affect the mineral volume fraction (MVF) within the full volume of each construct. However, when analyzing the two cylindrical subregions at the inner core of 6 mm constructs, the MVF was significantly increased compared with 3 and 9 mm constructs. Mineral particle analysis showed an increase in mineral particle size within 6 mm constructs compared with 3 and 9 mm constructs. A large number of pores on the top surface of each construct were filled with extracellullar matrix and completely occluded. This effect was also observed in previous perfusion experiments in Chapter 5 using PCL scaffolds with a porosity of 66%. In an attempt to mitigate this effect, subsequent perfusion experiments employed scaffolds with a larger pore diameter and greater porosity.

To evaluate bone construct development in vitro, it would be advantageous to monitor the mineral growth and distribution non-invasively during culture. Repeat scanning offers the ability to monitor mineral growth over time, but subjecting cells to multiple x-ray scans may negatively affect cell viability and mineralized matrix synthesis. Therefore, two replicate experiments, documented in Chapter 7, were performed to assess the effect of one additional x-ray scan on cell viability, distribution and matrix production. No significant differences in cell viability, mineralized matrix volume, or mineral density were detected within constructs that had been scanned during the experiment at 3 and 5 weeks, compared with those that had only been scanned at 5 weeks.

Having validated repeat scanning and shown that perfusion culture supported mineral deposition throughout 9 mm long scaffolds, additional work, explained in Chapter 7, evaluated the response of osteoblasts to time varying flow conditions. rMSCs seeded on 9 mm long, 75% porous PCL+C scaffolds were perfused at one of three flow regimes:​ 0.2 ml/min continuous flow (0.2), 0.8 ml/min continuous flow (0.8), or 0.2 ml/min continuous flow for 23 hours/day followed by one hour of culture with flow elevated to 0.8 ml/min (INT). Controls included constructs in static culture, as well as constructs placed in 6 well plates on an orbital rocker plate (RP), which was rotating at 0.5 Hz. Confocal microscopy images showed qualitative improvement in cell viability within 0.2 ml/min and INT constructs compared with all other culture groups. Constructs from all perfusion groups as well as the dynamic RP controls contained significantly more mineral at both 3 and 5 weeks compared with static controls. INT constructs consistently contained greater amounts of mineralized matrix as well as elevated rates of mineral production at both timepoints compared with all other experimental groups, however these values were statistically different from only the static and 0.8 ml/min culture groups. Visual inspection of the perfusion tubing revealed biological debris downstream of the constructs. Qualitatively, tubing for the 0.8 ml/min constructs contained the most debris, followed by the INT, and then 0.2 ml/min tubing. These results suggest that, while a minimum level of perfusion improves mineral deposition compared with static culture, flow rates above a certain level may be deleterious to matrix retention. Mineral particle analysis indicated that increases in mineral volume, from 3 to 5 weeks, within the 0.2 ml/min and INT constructs were predominantly due to growth of mineral particles that were already greater than 0.1 mm³. Increases in total mineral volume for constructs in the 0.8 ml/min group could be attributed to a large increase in the number of particles between 0.01 and 0.1 mm. Compared with perfused constructs, the principal mechanism for increasing total mineral volume within Static constructs was the generation of new mineral particles. RP construct mineral volume increases were caused be a significant increase in the total number of particles as well as increases in the size of particles greater than 0.1 mm³.

Together, these studies indicate that dynamic culture conditions enhance construct development with regards to cell viability, mineralized matrix deposition, growth rate, and distribution. The application of micro-CT image analysis techniques has provided the ability to quantitatively and nondestructively monitor the formation and growth of mineralized matrix regions within 3D tissue-engineered constructs in vitro. Furthermore, these techniques provide a rational approach to selecting perfusion culture conditions that optimize the amount and distribution of mineralized matrix production. Finally, the established perfusion bioreactor, in combination with micro-CT analysis, provides a foundation for evaluating new scaffolds and cell types that may be useful for the development of effective bone graft substitutes.