Theses Doctoral

Bone mechanobiology of modeling and remodeling and the effect of hematopoietic lineage cells

Robinson, Samuel Thomas

Osteoporosis is characterized by chronic bone loss and deterioration of microarchitecture that can leave patients more susceptible to costly and debilitating fractures. A variety of treatment options have been developed that target different cells and pathways to disrupt its progression. In addition to pharmaceutical options, regular exercise is recommended, as external, mechanical loading has long been recognized as a stimulus bone can use to regulate its size and shape to meet mechanical demands. While bone cell signaling is undoubtedly multifaceted, meaningful changes in bone mass ultimately result from the actions of bone-forming osteoblasts and bone-resorbing osteoclasts. To this end, therapies are most traditionally described through their impacts on these cells, and are broadly categorized as anabolic (activating osteoblasts and having bone-building effects, such as parathyroid hormone injections and sclerostin antibody treatment), or anti-resorptive (targeting osteoclasts and slowing resorption, such as bisphosphonates and denosumab). Bone formation and resorption are rooted in two overarching processes: coupled bone remodeling (resorption followed by formation in the same space) and uncoupled bone modeling (formation or resorption occurring independently). Hematopoietic-lineage cells have an inherent, established role in bone remodeling, as descendent osteoclasts perform the resorption to initiate remodeling, but have only more recently been implicated as potential orchestrators of anabolic bone modeling in their preosteoclastic states, suggesting the extent of their differentiation may be a mechanism steer the bone response between maintenance remodeling and adaptive modeling regimes. Understanding how pharmaceutical treatments and mechanical loading work through these regimes, augment intrinsic sensing mechanisms, or tilt local signals to favor one or the other may provide valuable insight into optimizing or combining current treatments, and potentially suggest new therapeutic avenues.

We first establish a method for quantifying modeling and remodeling in vivo using image registration on weekly micro-computed tomography scans. This technique is implemented in a study to assess the independent and combined effects of daily mechanical loading and parathyroid hormone injections in mice. We found that both resulted in significant increases in bone formation through anabolic modeling and remodeling, and while the modeling effects were usually additive or independent, the remodeling response was synergistic. Additionally, while PTH tended to exert its influence indiscriminately, the loading response was more targeted and pronounced in ways that mirrored local mechanical strains. Interestingly, this held true for catabolic modeling as well, where we observed a previously unreported phenomenon of load-induced increases in catabolic modeling in areas of low strain on the endosteal surface of cortical bone.

We then began targeted interventions into the hematopoietic lineage cells, starting at their most terminally differentiated state in bone, the osteoclast. Using an injectable osteoclast maturation inhibitor, osteoprotegerin (OPG), we observed how arresting this process influenced modeling and remodeling in response to loading in normal mice, and in mice genetically modified to reduce sclerostin expression. We observed the expected reductions in catabolic modeling regardless of genotype. We also found that in sclerostin-depleted mice treated with OPG, anabolic modeling was elevated, and there was no added benefit of mechanical loading to the response in trabecular and endosteal compartments, suggesting the controlled manipulation of these factors can fully recapitulate the intrinsic mechanosensing capabilities. Since the loading response is largely modeling-based, these findings support the hypothetical determinant of the modeling/remodeling response being the preosteoclast/osteoclast ratio in these areas. In contrast, however, on the periosteal surface a pronounced load-induced anabolic modeling response persisted in all treatment conditions, suggesting the unique cell populations in the periosteum may have more robust, more finely tuned, or differentially regulated mechanosensing mechanisms.

Finally, to probe the hematopoietic lineage further upstream and address the other side of the preosteoclast/osteoclast hypothesis, we utilized a novel genetically modified mouse model that allows for inducible macrophage (preosteoclast) ablation. Modeling and remodeling dynamics in response to loading were quantified in mice with normal or depleted macrophage quantities with concurrent normal or genetically-reduced sclerostin expression. In agreement with our hypothesis linking these cells to the anabolic modeling response, macrophage ablation resulted in significantly less anabolic modeling on trabecular and endosteal surfaces, which was not recovered by mechanical loading in either wild type or sclerostin deficient mice. Again, however, the periosteal surface was unique. Macrophage ablation did not reduce anabolic modeling on the periosteal surface, and loading still significantly increased it, regardless of sclerostin expression. Thus, similar to our findings with osteoprotegerin, a unique contrast existed between macrophage/preosteoclast ablation drastically reducing anabolic modeling and nullifying any mechanoresponse on trabecular and endosteal surfaces, but not the periosteal surface.

Taken together, these studies outline and implement a novel method to quantify modeling and remodeling in response to loading and clinically relevant treatments, with an emphasis on perturbations of the hematopoietic lineage. Concurrent stimuli are used to observe and quantify overlaps and augmentations in treatment efficacies, with a focus on mechanisms related to mechanoadaptation. Future work will focus on targeted approaches to identify unique mechanosensing factors driving the periosteal response, more sophisticated data analysis tools to observe to what extent localized bone metabolism can be predicted by strain and morphology, and the protein and cellular-level dynamics that underlie our findings.

As an addendum, a novel bone morphological parameter is described. A trabecular-cortical interface surface area metric (iSAM) is quantified on a set of cadaver bone segments from clinical high-resolution peripheral quantitative computed-tomography scans (a clinical analog to micro-computed tomography). iSAM is shown to correlate with stiffness and ultimate force derived from mechanical testing of the same samples, and improve correlations gleaned from traditional morphometric parameters alone.


  • thumnail for Robinson_columbia_0054D_16082.pdf Robinson_columbia_0054D_16082.pdf application/pdf 3.56 MB Download File

More About This Work

Academic Units
Biomedical Engineering
Thesis Advisors
Guo, Xiang-Dong Edward
Ph.D., Columbia University
Published Here
October 19, 2020