Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Jul 18.
Published in final edited form as: Cell Chem Biol. 2019 Apr 25;26(7):926–935.e6. doi: 10.1016/j.chembiol.2019.03.009

Discovery of a small molecule promoting mouse and human osteoblast differentiation via activation of p38 MAPK-beta

Brandoch Cook 1,*, Ruhina Rafiq 1,2, Heejin Lee 1,2, Kelly M Banks 1,2, Mohammed El-Debs 3, Jeanne Chiaravalli 4, J Fraser Glickman 4, Bhaskar C Das 5, Shuibing Chen 1,*, Todd Evans 1,*,^
PMCID: PMC6642001  NIHMSID: NIHMS1524567  PMID: 31031140

Graphical Abstract

graphic file with name nihms-1524567-f0001.jpg

Summary

Disorders of bone healing and remodeling are indications with an unmet need for effective pharmacological modulators. We used a high throughput screen to identify activators of the bone marker alkaline phosphatase (ALP), and discovered 6,8-dimethyl-3-(4-phenyl-1H-imidazol-5-yl)quinolin-2(1H)-one (DIPQUO). DIPQUO markedly promotes osteoblast differentiation, including expression of Runx2, Osterix, and Osteocalcin. Treatment of human mesenchymal stem cells with DIPQUO results in osteogenic differentiation including a significant increase in calcium matrix deposition. DIPQUO stimulates ossification of emerging vertebral primordia in developing zebrafish larvae, and increases caudal fin osteogenic differentiation during adult zebrafish fin regeneration. The stimulatory effect of DIPQUO on osteoblast differentiation and maturation was shown to be dependent on the p38 MAPK pathway. Inhibition of p38 MAPK signaling or specific knockdown of the p38-beta isoform attenuates DIPQUO induction of ALP, suggesting that DIPQUO mediates osteogenesis through activation of p38-beta, and is a promising lead candidate for development of bone therapeutics.

eTOC Blurb

Cook et al. carried out a screen and identified a small molecule that can promote bone formation, and determined that it functions by activation of a specific regulatory kinase, p38-beta. This compound is a promising candidate to develop a new bone healing drug.

Introduction

Bone fracture is the most common impact injury requiring emergent medical care. Of over 6 million fractures annually in the United States, at least 5-10% do not properly resolve. Bone non-unions and other failures of healing are often caused and exacerbated by contributing factors, such as osteoporosis which can itself be affected by lifestyle factors including obesity and poor diet. These factors result in a substantial cost burden both in terms of palliative care and lost productivity. Recombinant human bone morphogenetic proteins (BMPs) including BMP-2 and BMP-7 have been approved for therapeutic use in long bone non-unions; however, wide-ranging and poorly understood effects of growth factor treatments, deleterious side effects revealed in patient studies, and expenses associated with production and scalability, limit their common application in clinical settings (Carragee et al., 2011; Fu et al., 2013; Vaccaro et al., 2008). Additionally, therapeutics developed to block osteoporotic bone resorption (Cosman et al., 2016) have recently been abandoned due to unacceptable risks (Mullard, 2016). Therefore, fracture healing is largely accomplished through a combination of mechanical intervention and natural repair over time, and effective osteoporosis therapeutics are still in the nascent stages. There is consequently an unmet need for pharmaceutically relevant compounds that can stimulate or accelerate bone regeneration and healing.

Bone differentiation and mineralization can be modeled in vitro using various cell culture platforms. The murine myoblast cell line C2C12 is bipotential and can be directed toward either muscle or bone progenitor fates (Fux et al., 2004; Katagiri et al., 1994), with the latter being assayed via expression of osteogenic markers, including alkaline phosphatase (ALP). Additionally, primary mesenchymal stem cells can be derived from stem and progenitor populations, and driven toward adipogenic, chondrogenic, or osteogenic fates using permissive cytokines (Huang et al., 2007; Jaiswal et al., 1997). The course of osteoblastogenic differentiation can be dissected in a stepwise manner, with early expression of the master regulator Runx2 controlling differentiation events associated with expression of Osterix and ALP. Activation of this program precipitates expression of later differentiation markers such as Osteocalcin (OCN), and finally signatures of terminal osteoblast differentiation that include increased expression of Sclerostin (Sost) and dentin matrix acidic phosphoprotein 1 (Dmp1), as well as extracellular matrix deposition and calcium release that can be measured with vital stains.

The C2C12 cell line provides a useful screening platform because of its bipotentiality, robust culture capacity, and adaptability to scalable and automated quantitative assays. For example, chemical screening studies with C2C12 cells identified potential novel functions for known bioactive compounds in repair of underlying defects in muscular dystrophies (Bosnakovski et al., 2014; Cabrera et al., 2012; Moorwood et al., 2011). A recent C2C12-based screen biased for BMP-2 synergists identified the FDA-approved immuno-suppressor FK-506 as a promoter of osteogenic differentiation (Darcy et al., 2012). We previously employed an Id2 promoter-driven transcriptional reporter assay to screen for BMP pathway modulators (Feng et al., 2016) and identified PD407824 (PD), a known cell-cycle checkpoint regulator, which can synergize with BMP and activate osteoblast differentiation. Specifically, PD inhibits Chk1/2 activation and functions in combination with BMP4 to augment pathway-restricted SMAD signaling and promote osteoblast-like differentiation. Currently identified small molecules that promote bone formation do so by synergizing with BMP proteins, while identification of an independent chemical bone-differentiation inducer has remained elusive.

In the present study, we performed with C2C12 cells a high-throughput screen of over 47,000 compounds, and identified a small molecule activator of ALP, 6,8-dimethyl-3-(4-phenyl-1H-imidazol-5-yl)quinolin-2(1H)-one (DIPQUO), which promotes and accelerates osteoblast differentiation and maturation in vitro and in vivo. Moreover, DIPQUO functions mechanistically to promote activation of the beta isoform of p38 MAP kinase, which places it in a unique niche as a research tool for models of skeletogenesis and as a lead hit candidate to optimize for potential therapeutic discovery.

Results

A high throughput chemical screen identifies DIPQUO, a small molecule that promotes activation of early osteogenesis marker alkaline phosphatase.

To identify small molecule activators of osteoblast differentiation, we measured ALP activity using a fluorescent emission assay as a reporter for enzymatic digestion of ALP substrate in lysates derived from C2C12 myoblasts. ALP is an established marker for conversion of the normally myogenic-biased C2C12 cells to the osteogenic lineage (Chen et al., 2004), and BMPs are known robust activators of ALP in C2C12 (Fux et al., 2004; Katagiri et al., 1994). Therefore, prior to screening, the assay was calibrated using recombinant human BMP4 protein as a positive control. Compounds were robotically deposited onto 384-well plates and then overlaid by C2C12 cells for a 4-day culture period, followed by lysis and fluorescent substrate assays (Fig. 1A, Supp. Table 1). To ensure uniformity, C2C12 cells were maintained in normal culture medium containing 10% fetal bovine serum, and lysis and substrate addition achieved using a microplate multidrop device. The primary screen (Supp. Table 2) encompassed greater than 47,000 small molecules tested at a final concentration of 10 μM. Primary hit candidates were identified in three separate subsets by setting the following thresholds: 1) raw fluorescence ratio (RFU) >800 (Fig. 1B); 2) Z score >10 (Fig. 1B); and 3) normalized percent activation (Malo et al., 2006) (NPA) >5 (not shown), yielding a total of 52 compounds, with an approximately 0.1% hit rate. When primary hits were ranked according to RFU and Z score values (Supp. Tables 1 and 3), the clear top candidate was 6,8-dimethyl-3-(4-phenyl-1H-imidazol-5-yl)quinolin-2(1H)-one (DIPQUO; Fig. 1C, note the highest peak in Fig. 1B). This was confirmed by secondary screening in a concentration response experiment of the initial hits, and by a final validation screen of four potential candidates (Supp. Table 4). Purity of DIPQUO was assayed by high-pressure liquid chromatography and mass spectrometry (Supp. Table 1; Supp. Appendix), and EC50 in C2C12 cells was measured to be 6.27 μM (Fig. 1C). Additionally, DIPQUO was re-synthesized by a commercial supplier (ChemBridge Corp.) and carbon and proton nuclear magnetic resonance (NMR) spectra were obtained to confirm the identity and purity (Supp. Appendix).

Figure 1. A high throughput screen for activators of ALP expression identifies a lead hit molecule DIPQUO.

Figure 1.

Open in a new tab

A) Screening workflow, with library compounds first plated onto 384-well clear-bottom plates. Next, C2C12 murine myoblasts were seeded at a density of 2000/well and cultured for 4 days. Cells were lysed and analyzed for enzymatic digestion of fluorescent ALP substrate by automated measurement of fluorescence Ex/Em 450/580 nm. B) Raw readings of 450/580 nm fluorescence for >47k library compounds identified DIPQUO (labeled arrows) as clearly the strongest activator of ALP. This was confirmed by additional algorithmic comparison of fluorescence Z scores (B) and calculation of normalized percent activation (not shown). C) Molecular structure of DIPQUO; measurement of EC50 was performed in triplicate using purchased re-synthesized powder (n=3). D) Confirmation of activation of ALP expression by DIQPUO. C2C12 myoblasts were treated for 2 d with 10 μM DIPQUO and stained for ALP expression using alkaline naphthol and hematoxylin. Vehicle (DMSO) and inert structural analogs BT344 and BT345 were used as negative controls. Scale bar = 200 μm. E) Flow cytometric analysis of ALP-positive C2C12 cell population. See also supplementary Fig. S1, supplementary Tables 1-6, and supplementary DataS1.

Following screening, the re-synthesized DIPQUO was tested in complementary cell-based assays. In C2C12 cells, DIPQUO treatment was found to rapidly stimulate ALP expression within 2 days, as visualized by the foci of bright purple staining shown in the bottom right panel of Fig. 1D. We carried out a structure activity relationship (SAR) study using several related compounds from the screening library and by synthesis of a small set of additional related compounds designed to probe either of two pharmacophores of which DIPQUO is comprised, imidazole and quinolinone (Supp. Fig. S1 and Supp. Table 5 for relevant structures), which demonstrated specificity for DIPQUO in the ALP assay (Supp. Table 5). We therefore expanded the SAR study to probe a library of 154 structural chemical analogs assembled by ChemBridge Corp. from available screening library compounds. Analogs contained modifications around the quinolinone and imidazole moieties. Compounds were assigned unique identifiers and deposited on screening plates as described above. Compounds were tested for activation of ALP in C2C12 cells using conditions identical to the original screen, with the following exceptions: 1) analogs were tested in duplicate at final concentrations of 1, 5, and 10 μM, and 2) DIPQUO, instead of recombinant BMP4, was used as a positive control. Supp. Table 6, for purposes of brevity, shows only data for 10 μM samples. Although there was no analog treatment that resulted in significant activation of ALP compared to DIPQUO, all compounds with greater than 10% activation were re-tested in ALP staining assays (data not shown). All were confirmed to be inert, demonstrating striking specificity for DIPQUO osteogenic activity. Flow cytometry assays confirmed activation of ALP expression in > 30% of the cell culture by day 2, which is significantly higher than the baseline of approximately 4% in DMSO-treated cells (Fig. 1E).

There was a marked difference in staining between DIPQUO-treated cultures and those treated with DMSO or inert structural analog control molecules BT344 and BT345. In contrast to the screening strategy, which quantified day 4 substrate fluorescence, abbreviated 2-day treatment of C2C12 cells was optimal to resolve cell staining. DIPQUO treatment resulted in rapid rearrangement of cellular architecture from fibroblast-like to a cuboidal phenotype (Fig. 2A) that is a morphological hallmark of post-mitotic osteoblasts (Rutkovskiy et al., 2016). Continued treatment with 10 μM DIPQUO resulted in attrition of the C2C12 culture (Fig. 2B), which was partially rescued by addition of the osteoblast survival factor insulin-like growth factor 1 (IGF-1; Fig. 2C).

Figure 2. IGF-1 protects cell survival during DIPQUO-induced osteogenesis.

Figure 2.

Open in a new tab

(A) DIPQUO potentiates a rapid rearrangement of cellular morphology to a cuboidal phenotype. (B) Cell culture attrition over successive days can be partially rescued by addition of human IGF-1 protein (C), which functions as an osteoblast survival factor. Scale bar equals 200 μm for 20X ALP staining images, and 100 μm for 40X phase-contrast images in (A) and 400 μm for ALP staining images in (B) and (C).

DIPQUO differentiates human multipotent progenitors toward mature osteoblast fate.

The robust and rapid activation of ALP in C2C12 cells after DIPQUO treatment prompted a more thorough investigation into osteoblast differentiation. Progenitor populations are known to acquire successively narrower commitment toward terminal osteoblast fate in a stepwise manner characterized by progressive expression of early, transitional, and finally mature osteoblast markers (Beederman et al., 2013; Rutkovskiy et al., 2016). A subset of these markers was measured by quantitative RT-PCR, which showed that DIPQUO treatment of C2C12 cells resulted in significant up-regulation of the master osteoblast regulator Runx2 and its immediate effector Osterix (Osx; Fig. 3A). Notably, expression of markers associated with progressive differentiation and maturation, including ALP and Osteocalcin (OCN), respectively, were substantially increased by day 2 of DIPQUO treatment (Fig. 3A). These later-stage transcripts are not normally highly expressed during early stages of directed differentiation. Although ALP expression levels were higher in BMP-treated controls, the relative CT values were less variable in DIPQUO-treated samples (Fig. 3A). Moreover, changes in expression of later, maturation-associated transcripts OCN and Osteoactivin (OA) were much higher with DIPQUO treatment than in BMP-treated positive controls, suggesting a more robust and direct effect on osteoblast differentiation by DIPQUO. Finally, broad investigation of transcriptional programs via RNA-seq and Ingenuity Pathway Analysis revealed several bone morphological and functional gene sets whose expression patterns correlated strongly with osteogenic activity (Fig. 3B).

Figure 3. DIPQUO promotes osteoblast differentiation.

Figure 3.

Open in a new tab

A) Quantitative RT-PCR analysis of relative expression of transcripts indicative of osteoblast differentiation, arranged from early differentiation markers through later ones denoting progress to osteoblast maturation. DIPQUO treatment was compared to BMP-treated cells and values were normalized to DMSO-treated controls using Gapdh as a reference control (Runx2 n=6; Dlx5 n=3; Osx, ALP, and OA n=5; OCN n=4). Error bar is +/− SEM and * denotes p <.05, and ** p <.01. B) DIPQUO impacts regulation of gene sets involved in osteoblast differentiation and maturation. Ingenuity Pathway Analysis of day 2 RNA-seq reads identified several subsets of genes that align with osteogenic differentiation and maturation processes (table). An expression heat map reveals strong agreement between separate biological replicates that show a trend for up-regulation by DIPQUO treatment of osteogenic genes (orange) and down-regulation of anti-osteogenic genes (blue), compared to DMSO-treated controls. See also supplementary Fig. S2.

To extend these observations to human osteoblast maturation, a quantitative assay was used to analyze mineralization in differentiating primary human mesenchymal stem cells (hMSCs). Bone marrow-derived hMSCs were cultured in unbiased growth medium for at least two passages and then cultured in osteogenic medium for 12 to 21 days and stained with alizarin red to identify a time window in which spontaneous mineralization first occurred (Fig. 4A). Subsequently, hMSCs in osteogenic medium were treated continuously with DIPQUO from day 12 to day 18, stained and compared to equivalent DMSO-treated controls, and then normalized to total cell numbers (Fig. 4B). DIPQUO-treated samples incorporated alizarin red in a molar ratio approximately 5 times greater than DMSO-treated samples (Fig. 4C-D). Therefore, DIPQUO showed osteogenic activity in both mouse and human model systems.

Figure 4. DIPQUO promotes osteoblast differentiation and mineralization in human mesenchymal cells.

Figure 4.

Open in a new tab

A). Timing of spontaneous mineralization in hMSCs cultured under conditions permissive for osteogenic differentiation in untreated controls. Cells were stained with 2% alizarin red on the day indicated with respect to maintenance in osteogenic medium. Representative images are shown. Scale bar equals 400 μm. B) Schematic of in vitro mineralization assay. Human mesenchymal stem cells were cultured in complete growth medium and then transferred to osteogenic medium containing known osteoblastogenic cytokines and compounds. After 12 days of differentiation, cells were treated with DMSO or 10 μM DIQPUO for an additional 6 days. At day 18, cells were fixed and stained with 2% alizarin red. C) Representative images of day 18 alizarin red-stained DMSO- and DIPQUO-treated differentiation cultures. Scale bar = 400 μm. D) Stain was harvested at 85° C with acetic acid and quantified by reading absorbance at 405 nm in flat-bottom 96-well plates. n=3, and ** denotes p <.01.

To address the specificity of DIPQUO for stimulating an osteogenic versus osteoclastic program, the RAW 264.7 murine macrophage cell line was assayed for osteoclast differentiation by staining for tartrate-resistant acid phosphatase (TRAP) to distinguish multinucleated osteoclasts from macrophages. Receptor activator of nuclear factor κB ligand (RANKL) was used as a positive control to stimulate osteoclast differentiation, and was found to promote both TRAP staining (large, light purple cells) and expression of the osteoclast marker genes Cathepsin K (CTSK) and matrix metalloprotease 9 (MMP9, Supp. Fig. S2). DIPQUO treatment resulted in comparatively few TRAP-positive cells, and even these were much smaller than those induced by RANKL (red arrows, Supp. Fig. S2). CTSK and MMP9 were up-regulated albeit to a lower level compared to RANKL-induction. Thus, DIPQUO appears to be primarily supportive of osteogenesis.

DIPQUO promotes and accelerates bone mineralization in vivo.

To address whether the observed effects of DIPQUO on osteoblast differentiation and maturation could extend to an in vivo bone model, we utilized the zebrafish, Danio rerio. First, zebrafish were used as a model system to examine developmental ossification via direct vertebral specification through conversion of the notochord sheath (Inohaya et al., 2007; Laue et al., 2008). During zebrafish larval stages, the extent and pattern of ossification observable at discrete developmental time points are susceptible to perturbation by cytokines or genetic modulation. The contribution of extrinsic factors to notochord ossification and patterning can be measured by alizarin red staining in a manner analogous to its application in gauging osteoblast maturation in cultured cells. Accordingly, a 24-hour pulse of DIPQUO treatment was found to accelerate and accentuate mineralization of incipient vertebral primordia by 9 days post-fertilization, in comparison to controls treated either with DMSO vehicle or with inert structural analog compounds (Fig. 5A and 5B).

Figure 5. DIPQUO stimulates ossification and osteoblast differentiation in zebrafish developmental and regenerative models.

Figure 5.

Open in a new tab

A) Schematic of treatment regime in larval zebrafish notochord ossification model. Zebrafish larvae at 7 days post-fertilization (dpf) were treated for 24 hours with DMSO, 15 μM inert analogs BT344 or BT345, or 15 μM DIPQUO, which was washed out at 8 dpf. Larvae were then maintained for an additional 24 hours and fixed and stained with alizarin red at 9 dpf. B) Representative images of alizarin red-stained DMSO-, analog- and DIPQUO-treated larvae at 9 dpf. DIPQUO-treated larvae displayed an accelerated and accentuated staining pattern marking ossified vertebral primordia that emerge from the developing notochord (arrows in bottom right panel). For each condition, the image to the right is expanded from the black inset box. Larvae were scored for whether >2 vertebral primordial were stained; representative images are shown with ratios reflecting the number out of total with representative staining. Independent clutches of larvae were treated and stained at least in duplicate for every condition. Scale bar = 100 μm. C) The distal portion of caudal fin was removed on a diagonal from adult fish (dotted line) and allowed to regenerate at 32° C. After 26 hours, the fin regenerate was removed more proximally, fixed and analyzed by immunofluorescence for osteoblast markers. Black square represents area of detail shown in (D). Scale bar = 100 μm. D) Representative images of Sp7 (green) and Col10a1 (red) immunocytochemical staining of osteoblasts radiating from amputation sites. White arrows mark Sp7+/Col10a1+ double-positive cells. Scale bar = 20 μm. E) In DIPQUO-treated samples, numbers of double-positive cells were increased significantly. In this experiment n=10 animals across 3 biological replicate experiments and *** denotes p< 0.001.

We next tested the regenerative capabilities of zebrafish, which have a robust capacity to replace and renew organs and tissues derived from all three germ layers. Regenerating fin joints are reported to constitute a pre-osteoblast niche from which OSX-expressing (Sp7+) osteoblasts radiate de novo (Ando et al., 2017). Simultaneously, osteoblasts and osteogenic hypertrophic chondrocytes in both early and later stages of differentiation express Collagen10 (col10a1) (Huycke et al., 2012). In a zebrafish regeneration model system, in which the distal portion of the caudal fin was amputated and allowed to regenerate (Fig. 5C), the number of Sp7+/Col10a1+ cells emanating from new fin ray joints was significantly increased in tissue derived from DIPQUO-treated fish compared to control DMSO-treated fish (Fig. 5D and 5E). Taken together, these findings demonstrate a strong activity for DIPQUO as a stimulator and enhancer of osteogenic differentiation and maturation in vivo.

DIPQUO leads to an isoform-specific activation of p38 MAPK signaling

The unbiased approach used to identify DIPQUO provides little information on which signaling pathways are impacted as downstream effectors to mediate osteogenesis. Therefore, we investigated the activation status of several pathways, focusing particularly on effectors of the TGF-beta superfamily including relevant branches of the mitogen activated protein kinase (MAPK) pathway. In C2C12 cells, DIPQUO selectively activated p38 MAPK signaling while it suppressed the p54 isoform of the c-terminal Jun kinase family (JNK; Fig. 6A). Treatment with inert analog compounds did not alter p38 MAPK or JNK signaling (data not shown). There was no change observed in other pathways of interest, including phosphoinositol-3 kinase (PI3K/Akt), extracellular signal-regulated kinase (ERK), TGF-beta/SMAD2/3, and notably BMP/SMAD1/5/9 (Fig. 6A). In BRITER cells, a murine transformed osteoblast line that allows tamoxifen-inducible repression of BMP-2 and BMP-4 expression (Yadav et al., 2012), DIPQUO maintained the ability to activate p38 MAPK even under conditions of BMP repression (Fig. 6B).

Figure 6. DIPQUO functions mechanistically through activation of p38 MAPK signaling.

Figure 6.

Open in a new tab

A) Serum-starved C2C12 myoblasts were treated with DMSO or DIPQUO for 4 hours and whole cell extracts were examined by western blotting for a panel of TGF-beta, BMP and MAPK signaling effector molecules commonly modulated during osteoblast differentiation. p38 MAPK phosphorylation was found to be significantly activated by DIPQUO while JNK p54 activity was suppressed (n=3). B) In BRITER cells, which allow CRE-ER inducible suppression of BMP2/BMP4 expression, western blotting showed significant activation of p38 MAPK under conditions of both BMP suppression and exogenous addition of BMP4. Additionally, SMAD1/5 were activated by exogenous BMP4 both in the absence and presence of DIPQUO (n=3). In all experiments * denotes p <.05, ** p <.01, and *** p <.001. C) Western blotting analysis of p38 MAPK upstream activators in BRITER cells, at time points indicated after initial DIPQUO treatment (MKK3/6 n=4; TAK-1 n=3). Shaded key for experimental treatments corresponds to parts A-C. For these and all subsequent western blotting analyses, protein levels were quantified and normalized relative to DMSO-treated controls. D) Inhibition of p38 MAPK signaling with 10 μM SCIO469 attenuated DIPQUO activation of ALP expression, and inhibition of BMP receptor II activation with 1 μM LDN193189 attenuated BMP activation of ALP expression in C2C12 cells. However, BMP inhibition did not block DIPQUO activation of ALP nor did p38 MAPK inhibition block BMP activation of ALP. Scale bar = 400 μm. E) Western blotting analysis of isoform-specific individual knock-down of p38-α and –β after 48 hr of siRNA transfection in C2C12 cells (n=3). F) Attenuation of ALP expression in DIPQUO-treated C2C12 cells specifically by p38-β knock-down (n=3). Representative image of C2C12 cells transfected with siRNA for 24 hrs, then treated with 10 μM DIPQUO for 2 days and stained for ALP expression. Scale bar = 400 μm. G) ALP activity assay in equivalent samples depleted of p38-α and –β isoforms shows specific requirement for p38-β to mediate stimulatory effect of 10 μM DIPQUO (n=3). H) Isoform-specific immunoprecipitation demonstrates significant decrease in p38-α and increase in p38-β activities induced by 10 μM DIPQUO treatment (n=3). See also supplementary Fig. S3.

In differentiating osteoblasts, p38 MAPK activation is controlled by a MAP kinase cascade initiated through the MAP kinase kinases MKK3 and MKK6 (Greenblatt et al., 2010; Thouverey and Caverzasio, 2015). Accordingly, in BRITER cells DIPQUO stimulated rapid phosphorylation of MKK3/6 (Fig. 6C). Upstream control of MKK3/6 activation did not appear to function through TAK1 (Fig. 6C), or through any of several other tested putative activators, including apoptosis signal-regulating kinase (ASK1), mixed-lineage protein kinase 3 (MLK3), mitogen activated protein kinase kinase kinase 3 (MEKK3), tumor progression locus 2 (TPL2), and TNF receptor associated factor 6 (TRAF6; data not shown). The optimal dosages and time courses of treatment to impact p38 MAPK and JNK signaling did not exactly correspond, although there was overlap at 5-10 μM treatment of 6-8 hours duration (Supp. Fig. S3A-B). Over-expression of MKK3 and/or MKK6 was not sufficient to replicate the effects of DIPQUO (Supp. Fig. S3C-E). Although DIPQUO suppressed JNK signaling via its p54 isoform, chemical suppression of JNK signaling with the commercial inhibitor SP600125, either with or without p38 MAPK activation by U46619, was not adequate to induce a differentiation phenotype in C2C12 cells, as measured by ALP staining (data not shown). Treatment with 10 μM JNK Inhibitor V was found to both activate p38 MAPK and suppress JNK p54, but this condition also did not lead to ALP-positive staining in C2C12 cultures (Supp. Fig. S3F). Neither did DIPQUO stimulate activation of a luciferase reporter driven by the AP-1 response element (Supp. Fig. S3G). However, chemical inhibition of p38 MAPK with SCIO469 attenuated ALP expression in DIPQUO-treated cells (Fig. 6D). This effect was found to be specific, as inhibition of BMP signaling attenuated BMP-driven, but not DIPQUO-driven, differentiation (Fig. 6D).

There are four separate p38 isoforms: alpha, beta, gamma, and delta. The respective roles of the alpha and beta isoforms in bone differentiation have been dissected to the extent that p38-beta is known to be involved specifically in skeletogenesis (Greenblatt et al., 2010), while p38-alpha has roles that are both wide-ranging and highly specific, for instance in dentition (Greenblatt et al., 2015). Accordingly, we used siRNA-mediated knockdown to probe the specificity of DIPQUO to block the activity of one or the other isoform (Fig. 6E). We found that knock-down specifically of the beta isoform, but not the alpha isoform, attenuated ALP expression in DIPQUO-treated C2C12 cells (Fig. 6F), and resulted in an almost total block of ALP enzymatic activity (Fig. 6G). To confirm the biological significance of this observation, the alpha and beta isoforms were immunoprecipitated from C2C12 cells and analyzed for relative activity levels. The beta isoform was significantly activated after DIPQUO treatment, while p38-alpha was suppressed (Fig. 6H). In sum, these results suggest that p38 MAPK activation is necessary for DIQPUO-driven osteoblast differentiation, and that DIQPUO functions specifically in a manner that leads to activation of p38-beta.

Discussion

Through chemical screening, we report the discovery of a small molecule, 6,8-dimethyl-3-(4-phenyl-1H-imidazol-5-yl)quinolin-2(1H)-one (DIPQUO) that promotes osteoblast differentiation and maturation in murine and human progenitor cells. Furthermore, DIPQUO stimulates developmental ossification and regenerative production of differentiating zebrafish osteoblasts in vivo. It should be noted that these are normal physiological processes that are accelerated or enhanced by DIPQUO treatment. Mechanistically, DIPQUO functions to activate p38 MAPK signaling as an intracellular effector, specifically through the p38-beta isoform, although the direct interaction target is unknown. It is important to clarify that DIPQUO has not been shown to be a p38-beta "activator," but likely targets one or more unknown proteins that result in p38-beta activation. However, p38-beta is an attractive starting point for drug discovery, given the phenotypic specificity for defects in skeletogenesis in murine models of p38-beta deficiency. DIPQUO therefore has strong potential both as a research tool and as a possible lead pipeline molecule to be tested in preclinical models of bone repair and remodeling dysfunction.

Several signaling pathways contribute to developmental control of osteogenic programs, including those regulated by BMP, Wnt, Notch, and hedgehog ligands (Chen et al., 2012; Kim et al., 2013; Rodda and McMahon, 2006). Additionally, diverse extracellular ligands impact osteoblast differentiation, including BMPs, parathyroid hormone (PTH), fibroblast growth factors (FGFs), and noncanonical WNTs, all of which converge on MAPK cascade-driven mechanisms (Chen et al., 2012; Lin and Hankenson, 2011). Roles in early osteoblast differentiation have largely been ascribed to p38 MAPK (Rey et al., 2007; Thouverey and Caverzasio, 2015), while later roles have been identified for JNK-mediated MAPK signaling (Matsuguchi et al., 2009). The relative contributions of p38 MAPK alpha and beta isoforms to osteoblast biology have been dissected in murine genetic loss-of-function models. Although p38-alpha deletion results in pleiotropic defects that include deficits in skeletogenesis and dentition (Greenblatt et al., 2015; Greenblatt et al., 2010), p38-beta-deficient mice are phenotypically normal with the exception of a skeletal deficit in bone mineral density (Greenblatt et al., 2010). Lacking in this analysis, however, is a dedicated p38 MAPK isoform-specific activator that can be used to probe models of biological function in different cell culture and in vivo systems, and also to investigate putative uses as an ameliorative agent in pre-clinical models of bone repair, regeneration, and dysfunction. Although DIPQUO stimulates MKK3/6-directed activation of p38 MAPK signaling, the panel of known MAPKKK activators that can initiate a p38 signaling cascade in differentiating osteoblasts that were tested did not yield an obvious candidate. Therefore, we hypothesize that DIPQUO maintains an affinity for an unresolved target that has not previously been appreciated to have a role in control of p38 MAPK signaling in bone biology. Impending identification of cell surface or protein kinase target(s) will impact future medicinal chemistry-driven strategies to optimize DIPQUO and potential structural analogs for solubility, efficacy, and toxicity.

The osteogenic effect of DIPQUO is robust, significantly enhancing differentiation and calcium deposition in multipotent bone progenitors and developing notochord, and stimulating emergence of new osteoblasts in regenerating tissue. However, when tested in a cell-based model of bone resorption using osteoclast markers as surrogates, we found that while there is a modest increase in phenotypic osteoclasts, there is a significant up-regulation of two osteoclast differentiation markers. We hypothesize that these findings may make DIPQUO a more attractive drug candidate to stimulate appropriate physiological bone remodeling, which requires a balance between osteoblast-driven building and osteoclast-driven resorption programs. Strikingly, recent studies have associated bone fracture and fragility with long-term use of approved resorption-blocking osteoporosis therapies (Drieling et al., 2016; Lloyd et al., 2017; Ma et al., 2017; Saita et al., 2015). As a result, we propose DIPQUO as a potential candidate for bone regenerative therapies.

STAR Methods

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Todd Evans (tre2003@med.cornell.edu).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

The mouse myoblast cell line C2C12 was purchased from ATCC and cultured in Dulbecco’s Modified Eagle Medium (DMEM) with 10% FBS. Primary bone marrow-derived hMSCs (Lonza) were maintained and passaged in complete hMSC growth medium (Lonza). The RAW 264.7 macrophage cell line was purchased from ATCC. The BRITER cell line was kindly provided by Dr. Amitabha Bandyopadhyay (ITT Kanpur) and maintained in DMEM with 10% FBS. For AP-1 reporter assays, 293T HEK cells were used. The sex of the cell lines is not available. Zebrafish were a hybrid strain from crossing AB and Tub strains. Larvae were used prior to sex determination. For adults, both males and females were used without any apparent difference. All animal work was carried out according to an IACUC-approved protocol.

METHOD DETAILS

High Throughput Screening

For the primary screen, library compounds were distributed onto 384-well microplates (Greiner Bio-One 781091) at a final concentration of 10 μM in culture medium using a Perkin Elmer Janus automated workstation and WinPrep Version 4.8.3.315. Positive (rhBMP4, 1 ng/ml, R&D Systems) and negative controls (DMSO, 0.2%) were manually added to each plate by multichannel pipet. C2C12 cells were seeded onto compounds and controls at a density of 2000/well using a Thermo Multidrop Combi reagent dispenser and cultured for 4 days at 37° C, 5% CO2. Culture medium was aspirated using a BioTek EL406 plate washer and cells were lysed in RIPA buffer (Millipore 20-188) at ambient temperature for 10 minutes. 10 μl fluorescent alkaline phosphatase substrate (AttoPhos, Promega S1000) was added to lysates and incubated an additional 10 minutes before measuring the fluorescence (Excitation 450 nm/Emission 580 nm) on a BioTek Synergy Neo2 stacking microplate reader using Gen5 software. Plates were processed and relative fluorescence raw values measured in stacks of 10-20. The data were processed using Collaborative Drug Discovery web-based software (CDD Vault) to determine Z score calculated on DMSO control wells and percent activation normalized against positive and negative controls (NPA). Cherry-picked hit candidates were re-tested in concentration response experiments, from which 4 candidates were chosen and purchased for further testing from either ChemBridge or Enamine, based on availability. DIPQUO (ChemBridge 16707928) emerged as the top candidate. For all subsequent experiments, DIPQUO was used that had been re-synthesized by ChemBridge Corp. Proton and carbon nuclear magnetic resonance spectral analysis was performed on re-synthesized compound (Supp. DataS1), and additional structure-activity relationship analysis was performed on several structural analogs that were modified around the imidazole and quinolinone moieties, again using the AttoPhos assay (Supp. Table 5). For detailed procedures used to design and synthesize analog compounds, refer to DataS1. We also purchased a library of 154 structural chemical analogs assembled by ChemBridge Corp. from available screening library compounds. Analogs contained modifications around the quinolinone and imidazole moieties, and are shown in Supp. Table 6. Compounds were tested for activation of ALP in C2C12 cells using conditions identical to the original screen, with the following exceptions: 1) analogs were tested in duplicate at final concentrations of 1, 5, and 10 μM, and 2) DIPQUO, instead of recombinant BMP4, was used as a positive control. In-depth description of the primary screen, including a complete list of primary hits, is presented in Supplementary Tables 1-4.

Cell Culture and Staining Assays

C2C12 cells were treated for 2 days with 10 μM DIPQUO or structural analogs, 1 ng/ml rhBMP4, or with inhibitors as noted in figure legends, and then fixed briefly in 70% acetone/10% formaldehyde/20% citrate. Staining was achieved using the leukocyte alkaline phosphatase kit (Sigma 86R-1KT) according to the manufacturer’s instructions. Expression of ALP was confirmed and quantified by flow cytometric analysis, using an Accuri C6 flow cytometer. Briefly, control- or DIPQUO-treated C2C12 cells were detached and harvested on day 2 using PBS-based enzyme-free cell detachment solution (ThermoFisher). 2×105 cells were incubated on ice for 30 minutes with 10 μl APC-conjugated ALPL antibody (R&D Systems), and washed in ice-cold PBS before analysis. Live cells were gated and analyzed in CFlow Plus software and then data was converted to FlowJo to achieve publication-level resolution. For siRNA experiments, gene-specific oligonucleotides were obtained for mouse p38-alpha (Cell Signaling) and p38-beta (Santa Cruz). Signal Silence scramble siRNA control oligonucleotide was purchased from Cell Signaling. C2C12 cells at 70% confluency were transfected in 12-well plates using Lipofectamine RNAiMax reagent (Invitrogen) according to manufacturer’s instructions. Transfected cells were treated 24 hours later with DMSO or 10 μM DIPQUO, and siRNA transfection was repeated after 48 hours without changing culture media. After an additional 3 days, cells were either fixed and stained for ALP expression or analyzed for ALP activity using the AttoPhos Substrate kit. Hemagglutinin and FLAG epitope-tagged MKK3 (pMT2-HA-MKK3) and MKK6 (pcDNA3-FLAG-MKK6) constructs were obtained from Addgene and transfected into C2C12 cells using Lipofectamine LTX with Plus Reagent (ThermoFisher) according to manufacturer’s instructions and cell extracts analyzed for protein expression and activity as described below.

For mineralization studies, primary bone marrow-derived hMSCs (Lonza) were maintained and passaged in complete hMSC growth medium (Lonza). Cells were then switched to osteogenic medium (Lonza PT-3002) for 12 days, after which they were treated with DMSO or 10 μM DIPQUO for an additional 6 days. On day 18, cells were washed in PBS, fixed in ice-cold 70% ethanol for 60 minutes, then incubated for 60 minutes in 2% alizarin red solution, pH 4.2. Excess stain was washed away with distilled water. Staining was quantified using the Osteogenesis Quantitation kit (Millipore ECM815) following manufacturer’s instructions. Briefly, cultures were incubated in 10% acetic acid for 30 minutes, then scraped and heated to 85° for 10 minutes, placed on ice, and neutralized with NH4OH. Absorbance was measured at 405 nm using an EMax Plus microplate reader and SoftMax Pro 7.0 software. Molar values corresponding to alizarin red incorporation were obtained in reference to a standard curve generated using serial dilutions of alizarin red in assay buffer, and final values were obtained by normalizing to cell number in each sample.

The BRITER cell line was kindly provided by Dr. Amitabha Bandyopadhyay (ITT Kanpur). Cells were maintained in DMEM with 10% FBS. For analysis of DIPQUO effects independent of BMP signaling, BMP-2 and BMP-4 knockdown was achieved by treating cells overnight with 1 μM 4-hydroxytamoxifen (4-OHT), followed by continued maintenance in 1 μM 4-OHT. Cells were serum-starved for at least 6 hours before treatment with recombinant BMP protein or DIPQUO as noted in figure legends.

The RAW 264.7 macrophage cell line was purchased from ATCC and maintained in DMEM with 10% FBS. Cells were treated with 10 μM DIPQUO or 50 ng/ml RANKL (Sigma) for 4 days. Cells were fixed in 70% acetone/10% formaldehyde/20% citrate and TRAP staining was achieved using the Leukocyte Acid Phosphatase kit (Sigma 387A-1KT).

For AP-1 reporter assays, 293T HEK cells were seeded at 0.25×106 cells/well of a gelatin-coated 24-well plate one day before transfection. Plasmid transfections were performed using Lipofectamine LTX (Invitrogen) according to manufacturer’s protocol. Briefly, 293T HEK cells were co-transfected with pGL4.44 AP1 [luc2P/AP1 RE/Hygro] reporter plasmid (Promega) and SV40Renilla plasmid as a transfection control. 24 hours after transfection, cells were incubated in serum free DMEM media for 24 hours. The following day, cells were treated with 10ng/ml PMA (Phorbol 12-myristate 13-acetate; Tocris) or indicated concentration of DIPQUO in serum free DMEM for 7 hours prior to lysis with 1X Passive Lysis buffer (Promega). Luciferase expression was measured using Dual-Glo Luciferase Assay (Promega).

Gene Expression Analysis

For quantitative RT-PCR analysis of gene expression in C2C12 cultures, cells were treated for 2 days with 10 μM DIPQUO and harvested into Trizol reagent (Invitrogen). One microgram of RNA was reverse transcribed using the VILO-RT kit (Invitrogen) to generate cDNA, which was diluted 1:25 in RNase-free H2O for qPCR with Sybr green using the Roche 480 II LightCycler (LightCycler 480 1.5.0.39 software) and the 2−ΔΔCT method (Livak and Schmittgen, 2001). For analysis of osteoclast gene expression, RAW 264.7 cells were treated with DMSO, 50 ng/ml RANKL, or 10 μM DIPQUO for 4 days, then harvested into Trizol reagent and processed as above. Mouse qPCR primers are as follows.

Runx2: F(CGGCCCTCCCTGAACTCT); R(TGCCTGCCTGGGATCTGTA);

Dlx5: F(GCCCCTACCACCAGTACG); R(TCACCATCCTCACCTCTGG);

Osterix: F(AGCGACCACTTGAGCAAACAT); R(GCGGCTGATTGGCTTCTTCT);

ALP: F(AACCCAGACACAAGCATTCC); R(GAGACATTTTCCCGTTCACC);

Osteocalcin: F(GCAGCTTGGTGCACACCTAG); R(GGAGCTGCTGTGACATCCATAC);

Osteoactivin: F(TCTGAACCGAGCCCTGACATC); R(AGCAGTAGCGGCCATGTGAAG);

CTSK: F(AGGCATTGACTCTGAAGATGCT); R(TCCCCACAGGAATCTCTCTG);

MMP9: F(GCGGACATTGTCATCCAGTTTG); R(CGTCGTCGAAATGGGCATC);

Gapdh F(CTAACATCAAATGGGGTGAGG); R(CGGAGATGATGACCCTTTTG).

RNA-seq studies were carried out with the assistance of the Weill Cornell Genomics Core Facility, using the Illumina HiSeq4000 next-generation sequencer to generate reads from cDNA libraries generated from three biological replicates of day 2 DMSO- or DIPQUO-treated C2C12 cells. Gene sets were clustered by biological/disease function using Ingenuity Pathway Analysis (Qiagen Bioinformatics). Heatmaps with hierarchical clustering were generated in R using the CRAN package for a subset of genes involved in bone morphogenesis. Following normalization of the RNA Sequencing counts in DeSeq, z-scores were computed across samples within each gene for use in the heatmap.

Western Blotting

Whole cell extracts were collected from C2C12 or BRITER cells in complete lysis buffer (20 mM Tris, 150 mM NaCl, 50 mM NaF, 1% NP40 substitute, HALT protease inhibitor cocktail (ThermoScientific). Proteins were resolved by electrophoresis on pre-cast 10% NuPage Bis-Tris gels (Invitrogen) and transferred to PVDF membranes (Bio-Rad). Membranes were blocked in 5% BSA-TBS-0.5% Tween-20 for 15 minutes, then incubated at 4° overnight with primary antibodies. Antibodies used were: rabbit anti-phospho-p38 MAPK (cat. no. 9211), anti-p38 MAPK XP (8690), anti-phospho-SMAD1/5 (9516), anti-SMAD1 XP (6944), anti-phospho-SMAD2/3 (8828), anti-SMAD2/3 XP (8685), anti-phospho-JNK (4668), anti-SAPK/JNK (9252), anti-phospho-Akt XP (4060), pan anti-Akt (4691), anti-phospho ERK p42/p44 (4377), anti-ERK p42/p44 (9102), anti-phospho-MKK3/6 (12280), anti-MKK3 (8535), anti-phospho-TAK1 (4531), anti-TAK1 (5206), anti-HA (3274), and anti-FLAG (14793); all from Cell Signaling); and mouse anti-p38α (cat. no. 33-1300), anti-p38β (33-8700; both ThermoFisher) and anti-β-actin (Sigma A1978). Proteins were visualized with HRP-conjugated secondary antibodies (Bio-Rad) with WestPico (ThermoFisher) or Immobilon (Millipore) chemiluminescence reagents. Images were obtained and analyzed for relative densitometric relationships on a LI-COR C-DiGit scanner using Image Studio software.

Zebrafish Studies

Animals studies were performed according to protocols approved by the WCMC IACUC. Wildtype (AB/TU hybrid) zebrafish were maintained at 28.5° C. Larval fish were treated from 7 dpf to 8 dpf in tank water with DMSO or with a 24-hour pulse of 15 μM DIPQUO or inert analog BT344 or BT345, and were fixed at 9 dpf in 4% paraformaldehyde overnight rocking at 4° C. Fixed larvae were washed in PBS-0.1% Tween-20 (PBST), followed by 50% ethanol/50% PBST. Larvae were transferred to staining solution (66.5% ethanol, 100 mM MgCl2, 0.02% alizarin red) and incubated for 40 hours, rocking at room temperature in the dark. Larvae were washed in H2O + 0.1% Tween-20, and excess stain removed by bleaching for approximately 10 minutes in the dark with a 1:1 mixture of 3% H2O2 and 2% KOH. Images of staining were obtained using Nikon NIS Elements-BR software version 4.6.00. For the fin regeneration study, adult fish were anaesthetized in tricaine, and the distal portion of the caudal fin was excised. Amputees were allowed to recover in 300 ml tank water, to which was added either 90 μl DMSO or DIPQUO to final concentration of 15 μM. Fish were maintained in this fashion overnight at 32° C to optimize fin tissue regrowth. After 26 hours, fish were again anaesthetized and the caudal fin was re-amputated more proximally to ensure inclusion of the original amputation site. Tissue was fixed overnight rocking at 4° C in 4% paraformaldehyde, washed several times in PBST, and then blocked at room temperature for 2 hours in PBST-0.2% BSA. Primary antibodies (rabbit anti-Sp7 and mouse anti-col10a1, Abcam ab94744 and ab49945 respectively) were incubated 1:250 and 1:100 in PBST-0.2% BSA overnight at 4° C, washed several times in PBST-0.2% BSA, and then incubated overnight at 4° C in secondary antibodies (goat anti-rabbit Alexa 488 and anti-mouse Alexa 568 IgG, ThermoFisher A-11008 and A-11004). Finally, samples were washed several times in PBST and then mounted on slides in 80% glycerol with 2.5% DABCO (1,4-diazabicyclo[2.2.2]octane, Sigma) to preserve brightness. Images were acquired on a Zeiss LSM 800 confocal microscope (ZEN v2.3 software) and Sp7+/col10a1+ cells quantified using ImageJ.

QUANTIFICATION AND STATISTICAL ANALYSIS

Related to Figure 1 and associated Supplementary Tables 1-6. In chemical screening protocols, screens were performed either in duplicate or triplicate as noted in text and figure legends. For this purpose, n denotes biological replicates. Dose-response curves were either calculated directly within the CDD Vault algorithm, or were compiled using Prism 7.0. Fluorescence Z scores and normalized percent activation were calculated within CDD Vault using the equations noted in Supplementary Table 2.

Related to Figure 3 and associated Supplementary Figure S2. For all quantitative RT-PCR calculations, at least three independent biological replicates were used, with each biological replicate consisting of three replicates. In this case, n denotes the number of biological replicates. 2−ΔΔCT calculations were made directly within the Roche 480 Lightcycler software, and median values normalized to DMSO-treated controls using Gapdh as a reference control were used in student’s t-test calculations of mean, SEM, and p value. For RNA sequencing data, gene sets were clustered by biological/disease function using Ingenuity Pathway Analysis (Qiagen Bioinformatics). Heatmaps with hierarchical clustering were generated in R using the CRAN package for a subset of genes involved in bone morphogenesis. Following normalization of the RNA Sequencing counts in DeSeq, z-scores were computed across samples within each gene for use in the heatmap. Statistical analysis of RNA sequencing data was performed within the R CRAN software package.

Related to Figure 4. For alizarin red staining of differentiated human mesenchymal cell cultures, n = 3 biological replicates were stained and processed according to kit instructions. Spectrophotometer readings at OD405 of standards and samples in SoftMax Pro 7.0 were subsequently used in Microsoft Excel to generate a standard concentration curve and calculate relative molar amounts of incorporated stain for DIPQUO-treated cultures compared to DMSO-treated controls, normalized to total cell number. The student’s t-test was used for calculations of mean, SEM, and p value.

Related to Figure 5. For calculation of the effect of DIPQUO on emergence of Sp7+/Col10a1+ progenitor cells in zebrafish caudal fin, 10 different replicates were used across 3 independent experiments using 5-6 animals each. Not all replicates were used because processing damaged a considerable proportion of available samples. In this case n = 10. Individual double-positive cells were counted using Image J, and statistical analysis was executed using Prism 7 software.

Related to Figure 6 and associated Supplementary Figure S3. For calculations of relative phosphoprotein levels, densitometry measurements of individual bands were recorded using Image Studio software. Raw numbers were normalized from at least 3 biological replicates to those for total protein, and student’s t-test was used to derive mean, SEM, and p value.

DATA AND SOFTWARE AVAILABILITY

The accession number for the sequencing data reported in this paper is NCBI GEO:GSE125052.

Supplementary Material

1
2

Data S1. Mass spectrometry, high pressure liquid chromatpgraphy, and carbon/proton nuclear magnetic resonance spectroscopy analyses (Refers to Fig. 1).

3

Supplementary Table 3. Summary of primary screen positives and validated compounds (Refers to Fig. 1). Provided as an Excel file.

4

Supplementary Table 6. Structure-activity relationship analysis of chemical analogs (Refers to Fig. 1). Provided as an Excel file.

Significance.

There are currently few therapeutics useful for promoting bone formation following bone fracture or degeneration. This study used an unbiased high throughput screen to identify a small molecule compound, DIPQUO, that activates an osteogenic program in mouse and human cells, and promotes bone formation during zebrafish development and regeneration. Although the direct target is not known, functional activity is associated with activation specifically of the beta isoform of the p38 MAPK, known from mouse studies to be important for skeletogenesis. DIPQUO is a promising lead candidate for development of bone therapeutics.

Highlights.

  • Over 47,000 small molecules were screened for activation of ALP in C2C12 cells.

  • DIPQUO was identified, also promoting osteogenesis in human MSCs and zebrafish.

  • Functional activity is associated specifically with activation of p38-beta.

  • DIPQUO is a promising lead candidate for development of bone therapeutics.

Acknowledgments

We are thankful to Dr. Amitabha Bandyopadhyay (ITT Kanpur) for generously providing the BRITER cell line, and to Matthew Greenblatt (WCM) for helpful discussions. This work was supported by National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases grant K01-DK096031 (B.C.) National Institutes of Health, National Heart, Lung, and Blood Institute grant R37 HL56182 (T.E.) and a shared facility contract from the New York State Department of Health to T.E. and S.C. (NYSTEM C029156).

Footnotes

Declaration of Interests

The authors declare no competing financial interests.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Ando K, Shibata E, Hans S, Brand M, and Kawakami A (2017). Osteoblast Production by Reserved Progenitor Cells in Zebrafish Bone Regeneration and Maintenance. Dev Cell 43, 643–650 e643. [DOI] [PubMed] [Google Scholar]
  2. Beederman M, Lamplot JD, Nan G, Wang J, Liu X, Yin L, Li R, Shui W, Zhang H, Kim SH, et al. (2013). BMP signaling in mesenchymal stem cell differentiation and bone formation. J Biomed Sci Eng 6, 32–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bosnakovski D, Choi SH, Strasser JM, Toso EA, Walters MA, and Kyba M (2014). High-throughput screening identifies inhibitors of DUX4-induced myoblast toxicity. Skelet Muscle 4, 4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Cabrera PV, Pang M, Marshall JL, Kung R, Nelson SF, Stalnaker SH, Wells L, Crosbie-Watson RH, and Baum LG (2012). High throughput screening for compounds that alter muscle cell glycosylation identifies new role for N-glycans in regulating sarcolemmal protein abundance and laminin binding. J Biol Chem 287, 22759–22770. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Carragee EJ, Hurwitz EL, and Weiner BK (2011). A critical review of recombinant human bone morphogenetic protein-2 trials in spinal surgery: emerging safety concerns and lessons learned. Spine J 11, 471–491. [DOI] [PubMed] [Google Scholar]
  6. Chen G, Deng C, and Li YP (2012). TGF-beta and BMP signaling in osteoblast differentiation and bone formation. Int J Biol Sci 8, 272–288. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Chen S, Zhang Q, Wu X, Schultz PG, and Ding S (2004). Dedifferentiation of lineage-committed cells by a small molecule. J Am Chem Soc 126, 410–411. [DOI] [PubMed] [Google Scholar]
  8. Cosman F, Crittenden DB, Adachi JD, Binkley N, Czerwinski E, Ferrari S, Hofbauer LC, Lau E, Lewiecki EM, Miyauchi A, et al. (2016). Romosozumab Treatment in Postmenopausal Women with Osteoporosis. N Engl J Med 375, 1532–1543. [DOI] [PubMed] [Google Scholar]
  9. Darcy A, Meltzer M, Miller J, Lee S, Chappell S, Ver Donck K, and Montano M (2012). A novel library screen identifies immunosuppressors that promote osteoblast differentiation. Bone 50, 1294–1303. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Drieling RL, LaCroix AZ, Beresford SA, Boudreau DM, Kooperberg C, Chlebowski RT, Gass M, Crandall CJ, Womack CR, and Heckbert SR (2016). Long-term oral bisphosphonate use in relation to fracture risk in postmenopausal women with breast cancer: findings from the Women's Health Initiative. Menopause 23, 1168–1175. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Feng L, Cook B, Tsai SY, Zhou T, LaFlamme B, Evans T, and Chen S (2016). Discovery of a Small-Molecule BMP Sensitizer for Human Embryonic Stem Cell Differentiation. Cell Rep 15, 2063–2075. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Fu R, Selph S, McDonagh M, Peterson K, Tiwari A, Chou R, and Helfand M (2013). Effectiveness and harms of recombinant human bone morphogenetic protein-2 in spine fusion: a systematic review and meta-analysis. Ann Intern Med 158, 890–902. [DOI] [PubMed] [Google Scholar]
  13. Fux C, Mitta B, Kramer BP, and Fussenegger M (2004). Dual-regulated expression of C/EBP-alpha and BMP-2 enables differential differentiation of C2C12 cells into adipocytes and osteoblasts. Nucleic Acids Res 32, e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Greenblatt MB, Kim JM, Oh H, Park KH, Choo MK, Sano Y, Tye CE, Skobe Z, Davis RJ, Park JM, et al. (2015). p38alpha MAPK is required for tooth morphogenesis and enamel secretion. J Biol Chem 290, 284–295. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Greenblatt MB, Shim JH, Zou W, Sitara D, Schweitzer M, Hu D, Lotinun S, Sano Y, Baron R, Park JM, et al. (2010). The p38 MAPK pathway is essential for skeletogenesis and bone homeostasis in mice. J Clin Invest 120, 2457–2473. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Huang Z, Nelson ER, Smith RL, and Goodman SB (2007). The sequential expression profiles of growth factors from osteoprogenitors [correction of osteroprogenitors] to osteoblasts in vitro. Tissue Eng 13, 2311–2320. [DOI] [PubMed] [Google Scholar]
  17. Huycke TR, Eames BF, and Kimmel CB (2012). Hedgehog-dependent proliferation drives modular growth during morphogenesis of a dermal bone. Development 139, 2371–2380. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Inohaya K, Takano Y, and Kudo A (2007). The teleost intervertebral region acts as a growth center of the centrum: in vivo visualization of osteoblasts and their progenitors in transgenic fish. Dev Dyn 236, 3031–3046. [DOI] [PubMed] [Google Scholar]
  19. Jaiswal N, Haynesworth SE, Caplan AI, and Bruder SP (1997). Osteogenic differentiation of purified, culture-expanded human mesenchymal stem cells in vitro. J Cell Biochem 64, 295–312. [PubMed] [Google Scholar]
  20. Katagiri T, Yamaguchi A, Komaki M, Abe E, Takahashi N, Ikeda T, Rosen V, Wozney JM, Fujisawa-Sehara A, and Suda T (1994). Bone morphogenetic protein-2 converts the differentiation pathway of C2C12 myoblasts into the osteoblast lineage. J Cell Biol 127, 1755–1766. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Kim JH, Liu X, Wang J, Chen X, Zhang H, Kim SH, Cui J, Li R, Zhang W, Kong Y, et al. (2013). Wnt signaling in bone formation and its therapeutic potential for bone diseases. Ther Adv Musculoskelet Dis 5, 13–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Laue K, Janicke M, Plaster N, Sonntag C, and Hammerschmidt M (2008). Restriction of retinoic acid activity by Cyp26b1 is required for proper timing and patterning of osteogenesis during zebrafish development. Development 135, 3775–3787. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Lin GL, and Hankenson KD (2011). Integration of BMP, Wnt, and notch signaling pathways in osteoblast differentiation. J Cell Biochem 112, 3491–3501. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Livak KJ, and Schmittgen TD (2001). Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods 25, 402–408. [DOI] [PubMed] [Google Scholar]
  25. Lloyd AA, Gludovatz B, Riedel C, Luengo EA, Saiyed R, Marty E, Lorich DG, Lane JM, Ritchie RO, Busse B, et al. (2017). Atypical fracture with long-term bisphosphonate therapy is associated with altered cortical composition and reduced fracture resistance. Proc Natl Acad Sci U S A 114, 8722–8727. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Ma S, Goh EL, Jin A, Bhattacharya R, Boughton OR, Patel B, Karunaratne A, Vo NT, Atwood R, Cobb JP, et al. (2017). Long-term effects of bisphosphonate therapy: perforations, microcracks and mechanical properties. Sci Rep 7, 43399. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Malo N, Hanley JA, Cerquozzi S, Pelletier J, and Nadon R (2006). Statistical practice in high-throughput screening data analysis. Nat Biotechnol 24, 167–175. [DOI] [PubMed] [Google Scholar]
  28. Matsuguchi T, Chiba N, Bandow K, Kakimoto K, Masuda A, and Ohnishi T (2009). JNK activity is essential for Atf4 expression and late-stage osteoblast differentiation. J Bone Miner Res 24, 398–410. [DOI] [PubMed] [Google Scholar]
  29. Molnar A, Theodoras AM, Zon LI, Kyriakis JM (1997). Cdc42Hs, but not Rac1, inhibits serum-stimulated cell cycle progression at G1/S through a mechanism requiring p38/RK. J Biol Chem 272, 13229–13235. [DOI] [PubMed] [Google Scholar]
  30. Moorwood C, Lozynska O, Suri N, Napper AD, Diamond SL, and Khurana TS (2011). Drug discovery for Duchenne muscular dystrophy via utrophin promoter activation screening. PLoS One 6, e26169. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Mullard A (2016). Merck &Co. drops osteoporosis drug odanacatib. Nat Rev Drug Discov 15, 669. [DOI] [PubMed] [Google Scholar]
  32. Raingeaud J, Whitmarsh AJ, Barrett T, Derijard B, Davis RJ (1996). MKK3- and MKK6-regulated gene expression is mediated by the p38 mitogen-activated protein kinase signal transduction pathway. Mol Cell Biol 16, 1247–1255. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Rey A, Manen D, Rizzoli R, Ferrari SL, and Caverzasio J (2007). Evidences for a role of p38 MAP kinase in the stimulation of alkaline phosphatase and matrix mineralization induced by parathyroid hormone in osteoblastic cells. Bone 41, 59–67. [DOI] [PubMed] [Google Scholar]
  34. Rodda SJ, and McMahon AP (2006). Distinct roles for Hedgehog and canonical Wnt signaling in specification, differentiation and maintenance of osteoblast progenitors. Development 133, 3231–3244. [DOI] [PubMed] [Google Scholar]
  35. Rutkovskiy A, Stenslokken KO, and Vaage IJ (2016). Osteoblast Differentiation at a Glance. Med Sci Monit Basic Res 22, 95–106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Saita Y, Ishijima M, and Kaneko K (2015). Atypical femoral fractures and bisphosphonate use: current evidence and clinical implications. Ther Adv Chronic Dis 6, 185–193. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Thouverey C, and Caverzasio J (2015). Focus on the p38 MAPK signaling pathway in bone development and maintenance. Bonekey Rep 4, 711. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Vaccaro AR, Whang PG, Patel T, Phillips FM, Anderson DG, Albert TJ, Hilibrand AS, Brower RS, Kurd MF, Appannagari A, et al. (2008). The safety and efficacy of OP-1 (rhBMP-7) as a replacement for iliac crest autograft for posterolateral lumbar arthrodesis: minimum 4-year follow-up of a pilot study. Spine J 8, 457–465. [DOI] [PubMed] [Google Scholar]
  39. Yadav PS, Prashar P, and Bandyopadhyay A (2012). BRITER: a BMP responsive osteoblast reporter cell line. PLoS One 7, e37134. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1524567-supplement-1.pdf (2MB, pdf)
2

Data S1. Mass spectrometry, high pressure liquid chromatpgraphy, and carbon/proton nuclear magnetic resonance spectroscopy analyses (Refers to Fig. 1).

NIHMS1524567-supplement-2.pdf (5.3MB, pdf)
3

Supplementary Table 3. Summary of primary screen positives and validated compounds (Refers to Fig. 1). Provided as an Excel file.

NIHMS1524567-supplement-3.xlsx (172.3KB, xlsx)
4

Supplementary Table 6. Structure-activity relationship analysis of chemical analogs (Refers to Fig. 1). Provided as an Excel file.

NIHMS1524567-supplement-4.xlsx (749.1KB, xlsx)

Data Availability Statement

The accession number for the sequencing data reported in this paper is NCBI GEO:GSE125052.

RESOURCES