Genome-Wide Chromatin Landscape Transitions Identify Novel Pathways in Early Commitment to Osteoblast Differentiation.

Bethtrice Thompson1,2, Lyuba Varticovski,1,* Songjoon Baek1, and Gordon L. Hager1 

1Laboratory of Receptor Biology and Gene Expression, NCI, NIH, Bethesda, MD

2Center for Cancer Research and Therapeutic Development, Clark Atlanta University, Atlanta, GA

* Corresponding author:

Short Title: Dnase I Hypersensitivity Changes During Osteoblast Differentiation



Bone continuously undergoes remodeling by a tightly regulated process that involves osteoblast differentiation from Mesenchymal Stem Cells (MSC). However, commitment of MSC to osteoblastic lineage is a poorly understood process. Chromatin organization functions as a molecular gatekeeper of DNA functions. Detection of sites that are hypersensitive to Dnase I has been used for detailed examination of changes in response to hormones and differentiation cues. To investigate the early steps in commitment of MSC to osteoblasts, we used a model human temperature-sensitive cell line, hFOB. When shifted to non-permissive temperature, these cells undergo “spontaneous” differentiation that takes several weeks, a process that is greatly accelerated by osteogenic induction media. We performed Dnase I hypersensitivity assays combined with deep sequencing to identify genome-wide potential regulatory events in cells undergoing early steps of commitment to osteoblasts. Massive reorganization of chromatin occurred within hours of differentiation. Whereas ~30% of unique DHS sites were located in the promoters, the majority was outside of the promoters, designated as enhancers. Many of them were at novel genomic sites and need to be confirmed experimentally. We developed a novel method for identification of cellular networks based solely on DHS enhancers signature correlated to gene expression. The analysis of enhancers that were unique to differentiating cells led to identification of bone developmental program encompassing 147 genes that directly or indirectly participate in osteogenesis. Identification of these pathways provided an unprecedented view of genomic regulation during early steps of differentiation and changes related to WNT, AP-1 and other pathways may have therapeutic implications.

Key words: Chromatin, Mesenchymal Stem Cell, Osteogenesis

PMID: 26890492



Bone is a dynamic tissue characterized by continuous turnover throughout life. The stages of osteoblast formation from mesenchymal stem cells have been traditionally defined by sequential expression of known cell growth and differentiation-related transcription factors (TF) such as Runx2, Osterix/Sp7, and others.  Although studies that included microarray analysis, RNA-seq and chromatin immunoprecipitation using known DNA binding proteins (ChIP-seq) provided useful information on a subset of genes involved in this process, they have not provided broad information on all regulatory elements involved in MSC differentiation. Technologies developed in recent years allow characterization of accessible sites in the entire genome by techniques such as DNase hypersensitivity assay. In contrast to ChIP-seq, which provides a targeted view that is limited to known DNA-binding proteins and covers only a fraction of sites, detection of sites hypersensitive to DNase I, followed by fragment isolation and deep sequencing (DHS-seq) provides an unbiased view of all sites in DNA that are accessible to TFs and enhancers that participate in this process.  This approach is particularly useful in differentiation, because it identifies time-wise changes in all regulatory elements that ultimately regulate gene expression.

We made a detailed study using DHS-seq paired with microarray analysis of early mesenchymal stem cells transition to osteoblast in a human temperature-sensitive cell line.  Massive chromatin reorganization with over 30,000 unique sites was evident within hours of differentiation, prior to major shifts in gene expression, thereby providing an unprecedented view of genomic regulation during early steps of osteogenic commitment.  However, only a fraction of modified DHS (less than 10%) was mapped to the promoter regions, indicating TF engagement in this process. Using HOMER de novo motif discovery analysis we searched for specific TF motifs unique for each time point. Not surprisingly, we found preferential modifications at sites that regulate cell cycle, RUNX and other osteoblast-related TF.

Unexpectedly, the vast majority of chromatin modifications occurred at sites more than 2.5Kb of transcription start sites, and thus represent enhancers. Most were distant enhancers in regions up or downstream of gene bodies.  But how to understand the functional outcome of these chromatin modifications? There are no currently used methods to correlate modifications in distant enhancers to biologically relevant networks.  We then developed a method for such analysis by assigning each enhancer to the nearest transcription start site and used Ingenuity Pathway Analysis (IPA) to establish a signature that maps specific genes to their corresponding pathways.  To determine the significance of the enhancers signature, we overlaid these genes with expression data. The results were surprisingly rewarding and identified tissue/organ formation and mesenchymal stem cells differentiation as the two top networks. A subset of 147 genes specifically encompassed bone formation with a p value <10-19.

In addition, the cell model we chose for this analysis can differentiate spontaneously, albeit slowly, at non-permissive temperature. The process is greatly accelerated by osteogenic induction media thus permitting comparison of spontaneous differentiation with the effect of osteogenic induction media that contains dexamethasone, a potent glucocorticoid (GC).

GC-induced osteoporosis is the most common form of secondary osteoporosis. GCs have anti-inflammatory effects and are frequently used in clinically to treat inflammatory diseases including osteoarthritis. Under normal conditions, humans have pulsatile (ultradian) natural release GC with almost complete normalization in chromatin landscape between the cycles (Stavreva el al, 2009 and 2013). This pattern is disrupted by addition of long-acting steroids, such a dexamethasone. However, in the cell culture, GC are required for inducing osteogenic differentiation. We then asked whether the adverse effects of GC observed clinically could be detected in a tissue culture model? Comparing cells exposed to dexamethasone to those which differentiated spontaneously for 48 hours we observed changes in DHS that included several members of the WNT family.  Ingenuity pathway analysis of our data showed a link to many known and previously not identified genes linked to osteogenesis (Figure 1A). Overlaying this gene set with expression showed downregulation of many WNT family members by 48 hours which was not evident at earlier times (Figure 1B).  Furthermore, there was increased expression of DKK1 and DKK2, known inhibitors of the WNT pathway.  DKK1 was also recently linked to osteopenia and its promoter has GC-responsive elements.  Thus, our observations provide mechanisms supporting therapeutic interventions to combat long-term deleterious effect of GC on bone formation.