Generation of osteoinductive grafts by three-dimensional perfusion culture of human bone marrow cells into porous ceramic scaffolds

The main aims of this thesis were (i) to identify and develop a system that could be
reproducibly used to streamline manufacture of osteoinductive grafts based on human bone marrow
stromal cells (BMSC) in the context of regenerative medicine, (ii) to characterize the developed
system in order to identify key elements responsible for its reproducible and efficient performance,
and (iii) to extend its use to a sheep cell source, thus opening the way to test the osteoinductivity of
orthotopic implants in a large animal model.
Bone Marrow Stromal Cells (BMSC), which are typically defined by their capacity to adhere
on plastic [1] and form a fibroblastic colony (CFU-f) [2], represent a very low fraction (approximately
0.01%) among the nucleated cells of the bone marrow. Therefore, to obtain a sufficient number of
cells for bone tissue engineering applications, BMSC are typically first selected and expanded in
monolayer (2D) prior to loading into 3D scaffolds. However, 2D-expansion causes BMSC to
progressively lose their early progenitor properties and differentiation potential [3-5], and to decrease
their capability to form colonies and to induce bone tissue formation upon ectopic implantation [3],
placing several potential limits on their clinical utility. To bypass the process of 2D-expansion and its
associated limitations, we used an innovative bioreactor-based approach to seed, expand, and
differentiate BMSC directly in a 3D ceramic scaffold [6]. Nucleated cells, freshly isolated from a bone
marrow aspirate, were introduced into the bioreactor system and perfused through the pores of 3D
ceramics for five days, then further cultured under perfusion for an additional two weeks. Using the
developed procedure, BMSC could be seeded and extensively expanded within the 3D environment of
the ceramic pores. Interestingly, we found that the 3D-generated constructs contained both
hemopoietic cells and BMSC, whose relative fractions could be modulated by appropriate media
supplements, and that a consistent fraction of expanded BMSC was clonogenic. In contrast, following
the typical 2D-expansion, cells of the hemopoietic lineage could not be maintained, and, consistently
with previous studies, only a minor fraction of expanded BMSC was still clonogenic. When constructs
were ectopically implanted in nude mice, those engineered in the bioreactor reproducibly generated
bone tissue that was uniformly distributed throughout the scaffold volume and filled up to 60% of the
ceramic pores. In marked contrast, when similar numbers of 2D-expanded BMSC were loaded into
ceramic scaffolds and implanted, bone was infrequently generated, and even in the most
osteoinductive constructs, it was localized to peripheral regions, filling only 10% of the ceramic pore
volume [6].
Considering the need of reproducibility or at least of predictability in the osteoinductive ability
of the constructs for their standardized clinical use, in order to validate the possibility of extending the
use of the developed bioreactor-based approach for generating osteoinductive grafts of clinically
relevant size, we then investigated whether a minimum cell density was required for the reproducible
bone tissue formation. Based on the established association between the higher clonogenicity of
BMSC expanded in the 3D-system and the more reproducible and extensive osteoinductivity of the
resulting constructs, as compared to those based on 2D-expanded BMSC, we demonstrated that
presence or absence of bone in the constructs following ectopic implantation is related not to the total
number of implanted BMSC, but to the number of CFU-f present in the construct at the time of
implantation. In particular, we identified an apparent threshold in the amount of CFU-f discriminating
between osteoinductive and not osteoinductive constructs.
The developed bioreactor-based approach has been validated in a heterotopic model. Before
envisioning a clinical trial in human, a study in a large animal model is needed to validate the safety
and the surgical feasibility of the overall procedure. Thus, in the perspective of testing our novel
approach for repairing experimental bone defects in a sheep model, it was first necessary to validate
our system using ovine BMSC. We demonstrated that osteoinductive constructs can be generated by
perfusing 3D ceramic scaffolds with the nucleated cell fraction of ovine bone marrow aspirates [7].
Ongoing studies in the context of an EU-funded Project are aimed at testing the capability of the
generated constructs to repair large bone defects in sheep (i.e. defects around titanium implants
inserted into trabecular bone of the proximal humerus, and postero-lateral spinal fusion in lumbarregion).