The balanced design of bioactive materials for guiding bone repair requires a comprehensive understanding of how substrate-mediated cues regulate cell interactions with soluble signals. The deposition of a biomineral on the surface of synthetic polymers yields hybrid biomaterials with enhanced osteoconductivity while maintaining desired biodegradability. Our mineralization process, previously shown to enhance hMSC-secreted mineral (Davis et al. 2009
), entails microparticle hydrolysis, subsequent incubation in mSBF, and then scaffold formation, resulting in constructs with biomimetic apatite deposited throughout their volume. BMP-2, a protein known to direct progenitor cells toward the osteoblastic lineage, was supplied in the media for the duration of culture. These data reveal the important interactions between osteoinductive proteins such as BMP-2 with the carbonate apatite on biomineralized polymeric substrates and demonstrate the resulting osteogenic response of hMSCs in an environment designed to enhance bone formation.
Construct microarchitecture is a key mediator in progenitor cell differentiation. Several studies have reported the role of pore size and porosity on inducing hMSCs towards the osteoblastic phenotype (Mygind et al. 2007
; Whang et al. 1999
). Moreover, changes in scaffold porosity led to alterations in growth factor binding and cell response to BMP-2 (Tsuruga et al. 1997
; Whang et al. 2000
). In both nonmineralized and premineralized constructs, salt crystals of a defined size (250–425 µm) were used to create substrates with high porosities, and hence pore size should be relatively similar. The addition of hydroxyapatite nanoparticles to similar scaffolds markedly decreased porosity (He et al. 2010
). Thus, to characterize potential differences in porosity for these two substrates, we employed a volume displacement method that ultimately demonstrated similarly high porosities for both nonmineralized and premineralized scaffolds. The alkaline treatment employed to hydrolyze the polymer increases the surface area and generates nanotopographical changes on the polymer surface previously shown to increase osteoblast adhesion (Kay et al. 2002
; Smith et al. 2007
). We previously observed increased osteogenic differentiation of hMSCs on premineralized scaffolds compared to nonmineralized, alkaline-treated substrates with similar levels of cellularity (Davis et al. 2009
), indicating that NaOH-induced nanotopographical changes have a less pronounced effect on cell adhesion and differentiation in three-dimensions. However, the surface of these scaffolds is a combination of carbonated apatite and NaOH-treated domains, as mineral coverage is not completely uniform. The increased surface area produced by alkaline treatment prior to biomineralization may provide additional sites for BMP-2 binding and adsorption to the substrate, and this surface heterogeneity may contribute to these data. Osteogenic differentiation of hMSCs is also affected by differences in substrate stiffness (Engler et al. 2006
). Our previous study demonstrated no differences in compressive moduli between premineralized and nonmineralized substrates (Davis et al. 2009
In light of similar porosities and compressive moduli between substrates, differences in the expression of osteogenic markers by hMSCs seeded on premineralized and nonmineralized scaffolds are likely attributed to the combinatorial effects of soluble signals with other material properties such as wettability or increased surface area following microsphere functionalization that result in more protein binding sites. Premineralized scaffolds adsorbed more trypan blue compared to the relatively hydrophobic nonmineralized scaffolds and are likely more capable of adsorbing surrounding plasma proteins. Indeed, when specific adsorption of BMP-2 to the substrates was investigated, premineralized scaffolds adsorbed greater amounts of protein compared to nonmineralized substrates for all BMP-2 concentrations tested. Interestingly, we observed that functionalized scaffolds adsorbed similar amounts of BMP-2 as premineralized scaffolds. Functionalization of PLG scaffolds creates hydrolyzed, carboxylic acid rich surfaces. Although the predominant mechanism for BMP-2 adsorption to apatite surfaces is the interaction between BMP’s carboxylic group with surface calcium ions, functionalization of PLG likely lowers the polymer’s inherent hydrophobicity, hence creating more water-bridged hydrogen bonds (Dong et al. 2007
). Thus, functionalized polymers may offer another approach to increase BMP-2 binding without necessitating the presence for mineral.
In agreement with earlier studies, ALP activity of hMSCs was not increased when seeded on mineralized substrates versus nonmineralized scaffolds in the absence of BMP-2 (Davis et al. 2009
; Murphy et al. 2005
). However, the addition of 25 and 100 ng/mL BMP-2 to cells on premineralized substrates resulted in greater activities at certain time points, suggesting that substrate-mediated cues acted synergistically with the soluble signals at these concentrations. At 200 ng/mL BMP-2, differences in the cellular response as a function of the substrate were masked and the magnitude of ALP activity was increased for both groups, suggesting that BMP-2 became the principal mediator in progenitor differentiation. At all BMP-2 concentrations, cell-secreted calcium for hMSCs seeded on premineralized scaffolds was significantly greater than for cells cultured on nonmineralized scaffolds, suggesting that substrate composition was of more importance. At 200 ng/mL, these differences were not apparent at early time points, and it was only at this concentration that an increase in calcium deposition was observed for cells on nonmineralized substrates. These results suggest that premineralized scaffolds nucleate cell-secreted calcium more efficiently than nonmineralized substrates. Moreover, these data confirm that higher concentrations of BMP-2 can override the contribution of substrate-mediated cues, and hMSC osteogenic differentiation may not benefit from substrate mineralization at such protein levels.
Construct composition, in conjunction with a soluble osteoinductive factor, had a considerable effect on osteogenic gene expression profiles. These studies were performed with a single BMP-2 concentration to further analyze apparent differences in ALP activity among groups while using a dosage that is commonly cited in the literature as effective for inducing hMSC osteogenic differentiation. Our data demonstrate down-regulation of RUNX2
with the addition of BMP-2 at an early time point (although not significant for premineralized substrates) and a significant increase in expression for only premineralized substrates at a late time point, suggesting that RUNX2
expression is both mineral- and BMP-2 dependent. However, osterix (SP7
) expression exhibited only significant increases in expression for substrates cultured in the presence of BMP-2 with comparative results between premineralized and nonmineralized groups, suggesting that BMP-2 is the predominant osteogenic factor. Osterix is generally considered to be downstream of RUNX2
in the progenitor osteogenic differentiation pathway (Nakashima et al. 2002
), and thus, it was surprising to see a large increase in RUNX2
expression at day 21. Others have reported that osterix expression does not consistently follow the same trend as RUNX2
, as there are other pathways that act alongside, or independent of, RUNX2
to modulate osterix expression during osteogenesis (Celil et al. 2005
). Osteonectin (SPARC
) exhibited poor correlation with the presence of the osteoinductive protein. Although hMSCs on nonmineralized substrates exhibited significantly more expression at both time points, cells on each substrate failed to upregulate gene expression in response to the addition of BMP-2. Osteonectin is a glycoprotein expressed in bone undergoing active remodeling. Since premineralized substrates already possessed mineral nucleation sites due to mSBF incubation, it is possible that hMSCs seeded on nonmineralized scaffolds had more osteonectin gene expression as they attempted to remodel the polymer scaffold to nucleate mineral. Bone sialoprotein (IBSP
) expression was dependent on both substrate cues and signal induction, as cells on premineralized substrates in the presence of BMP-2 demonstrated significantly increased gene expression at day 21. hMSCs exposed to large amounts of apatite demonstrate increased deposition of bone sialoprotein in culture (Bhumiratana et al.). The addition of BMP-2 in the presence of mineral might possibly alter the timing of bone sialoprotein expression, resulting in a peak earlier in the culture period. Furthermore, premineralized and nonmineralized scaffolds have distinct affinities for BMP-2, potentially enabling differences in protein adsorption and presentation to hMSCs. However, protein affinity for both scaffolds is a dynamic variable given the increasing calcium deposition on the substrate. This potential contribution to these results merits further investigation.
The present data indicate that the combination of apatite and BMP-2 do not simply enhance the osteogenic response of hMSCs, but act through multiple pathways that may be both substrate- and growth factor-mediated. BMP-2 is a costly recombinant growth factor, and bone regeneration strategies should attempt to limit its use to therapeutic doses to control patient costs. While many have attempted to present BMP-2 from osteoconductive substrates containing bone-like minerals as a means to enhance osteogenesis, the combination of soluble signals in conjunction with substrate-mediated cues remains largely unexplored. These results suggest that the osteogenic differentiation of hMSCs is dependent on both substrate- and soluble-mediated cues, although one does not simply augment the other. Each has a more potent effect on different aspects of progenitor differentiation, and the magnitudes of these effects are dependent on BMP-2 concentration. In light of these observations, multiple signaling strategies may be necessary to achieve otimal bone regeneration.