Regulation of bone homeostasis by MERTK and TYRO3

Abstract

The fine equilibrium of bone homeostasis is maintained by bone-forming osteoblasts and bone-resorbing osteoclasts. Here, we show that TAM receptors MERTK and TYRO3 exert reciprocal effects in osteoblast biology: Osteoblast-targeted deletion of MERTK promotes increased bone mass in healthy mice and mice with cancer-induced bone loss, whereas knockout of TYRO3 in osteoblasts shows the opposite phenotype. Functionally, the interaction of MERTK with its ligand PROS1 negatively regulates osteoblast differentiation via inducing the VAV2-RHOA-ROCK axis leading to increased cell contractility and motility while TYRO3 antagonizes this effect. Consequently, pharmacologic MERTK blockade by the small molecule inhibitor R992 increases osteoblast numbers and bone formation in mice. Furthermore, R992 counteracts cancer-induced bone loss, reduces bone metastasis and prolongs survival in preclinical models of multiple myeloma, breast- and lung cancer. In summary, MERTK and TYRO3 represent potent regulators of bone homeostasis with cell-type specific functions and MERTK blockade represents an osteoanabolic therapy with implications in cancer and beyond.

Fig. 1. Negative regulation of bone formation by TAM receptor MERTK.

a mRNA expression analysis of TAM receptor family Mertk, Tyro3, Axl, Gas6, and Pros1 in primary murine osteoblast cultures after 0, 7, 14, and 21 days (n = 3 biological replicates) (Mertk expression d0 vs. d7: p = 0.0002; d0 vs. d14: p = 0.0042; d0 vs. d21: p = 0.0172), (Tyro3 expression d0 vs. d7: p = 0.009; d0 vs. d14: p = 0.9604; d0 vs. d21: p = 0.2928), (Axl expression d0 vs. d7: p = 0.0161; d0 vs. d14: p = 0.0152; d0 vs. d21: p = 0.0331), (Gas6 expression d0 vs. d7: p = 0.0011; d0 vs. d14: p = 0.0002; d0 vs. d21: p = 0.001), (Pros1 expression d0 vs. d7: p = 0.0157; d0 vs. d14: p = 0.0181 d0 vs. d21: p = 0.054). b, c Microcomputed tomography (μCT) of the metaphyseal proximal region of tibias from 8-week-old Mertk^flox/flox and Col1a1-cre⁺;Mertk^flox/flox female mice (top, longitudinal view of cortical and cancellous bone; bottom, longitudinal view of cancellous bone) (b). Quantification of bone volume (BV/TV) of cancellous bone determined by μCT analysis (c) (Mertk^flox/flox, n = 9; Col1a1-cre⁺;Mertk^flox/flox, n = 14). d, e Representative pictures of Calcein Demeclocycline labeling of Mertk^flox/flox and Col1a1-cre⁺;Mertk^flox/flox female mice (d). Bone formation rate of Col1a1-cre⁺;Mertk^flox/flox mice after 8 weeks (e) (Mertk^flox/flox, n = 5; Col1a1-cre⁺;Mertk^flox/flox, n = 8). f–h Representative pictures (f) and histomorphometric analysis of osteoblast (g), and osteoclast number (h) by TRAP/Hematoxylin staining in femur from Col1a1-cre⁺;Mertk^flox/flox mice. Green arrows pointing to osteoblasts visible as cuboidal or polygonal mononuclear cells on the endosteal bone surface (Mertk^flox/flox, n = 5; Col1a1-cre⁺;Mertk^flox/flox, n = 8). i Alizarin Red staining of ex vivo calvarial cell osteoblast culture from Col1a1-cre⁺;Mertk^flox/flox mice. j Analysis of MERTK protein in osteoblast cultures from Mertk^flox/flox mice treated with recombinant CRE recombinase. ß-ACTIN run on a separate gel. k Alizarin Red staining of MERTK KO calvarial cell cultures treated with PROS1 (100 nM). l, m RT-qPCR analysis of Alpl (l) and Bglap (m) mRNA expression in MERTK KO osteoblasts (n = 3 biological replicates). n Alizarin Red staining of wild-type calvarial cell cultures treated with PROS1 (100 nM) and MERTK-inhibitor R992 (200 nM). o, p RT-qPCR analysis of Alpl (o) and Bglap (p) mRNA expression (n = 3 biological replicates). Data were means ± SEMs. Statistical significance was determined by a two-tailed unpaired t-test.

Fig. 2. Inhibition of bone formation by MERTK is mediated by PROS1-MERTK-VAV2-RHOA-ROCK axis.

a–d Confocal imaging of pRLC and F-actin staining of MERTK KO osteoblasts on glass coverslips treated with TAM receptor ligand PROS1 (100 nM) (a). F-actin intensity (n = 3, mean of 50 measurements in three fields) (b), stress fiber-containing cells (n = 3, mean of 100 measurements in three fields) (c), and pRLC intensity (n = 3, mean of 50 measurements in three fields) (d) was quantified. e Immunoblot of wild-type osteoblasts treated with different concentrations of PROS1 (0, 50, 100, 200 nM) showing activated GTPγs-bound RHOA, total-RHOA, and ß-ACTIN. ß-ACTIN run on a separate gel. f Immunoblot of wild-type osteoblasts treated with PROS1 (100 nM) and MERTK-inhibitor R992 (200 nM) showing activated GTPγs-bound RHOA, total-RHOA, phosphorylated VAV2, an association of MERTK and VAV2 and ß-ACTIN. ß-ACTIN run on a separate gel. g Alizarin Red staining of wild-type calvarial cells treated with PROS1 (100 nM) and ROCK-inhibitor Y27632 (10 μM). h RT-qPCR analysis of Alpl mRNA expression (n = 3 biological replicates). i Calvarial cells were cultured in an osteogenic medium with exogenous addition of PROS1 and Y27632. Cultures were stained by F-actin staining. The top view shows representative pictures of the cell morphology of calvarial cell cultures after 5 days. The bottom view shows representative pictures of single-cell analysis on fibronectin substrate (representative image section). j The percentage of stress fiber-containing cells was quantified (n = 3, mean of 100 cells in three fields). Data were means ± SEMs. Statistical significance was determined by a two-tailed unpaired t-test.

Fig. 3. TAM receptor TYRO3 promotes osteoblastogenesis and bone formation.

a, b Microcomputed tomography (μCT) of the metaphyseal proximal region of tibias from 8-week-old Tyro3^flox/flox and Col1a1-cre⁺;Tyro3^flox/flox female mice (top, longitudinal view of cortical and cancellous bone; bottom, longitudinal view of cancellous bone) (a). Quantification of bone volume (b) of cancellous bone determined by μCT analysis (Tyro3^flox/flox, n = 8; Col1a1-cre⁺;Tyro3^flox/flox, n = 10). c, d Representative pictures of Calcein Demeclocycline labeling of Tyro3^flox/flox and Col1a1-cre⁺;Tyro3^flox/flox female mice (c). Bone formation rate of Col1a1-cre⁺;Tyro3^flox/flox mice after 8 weeks (Tyro3^flox/flox, n = 5; Col1a1-cre⁺;Tyro3^flox/flox, n = 8) (d). e–g Representative pictures (e) of histomorphometric analysis of osteoblast (f) and osteoclast number (g) by TRAP/Hematoxylin staining in femur from Col1a1-cre⁺;Tyro3^flox/flox mice. Green arrows pointing to osteoblasts visible as cuboidal or polygonal mononuclear cells on the endosteal bone surface (Tyro3^flox/flox, n = 8; Col1a1-cre⁺;Tyro3^flox/flox, n = 10). h Alizarin Red staining of ex vivo calvarial cell osteoblast culture from Col1a1-cre⁺;Tyro3^flox/flox mice. i Analysis of TYRO3 protein in osteoblast cultures from Tyro3^flox/flox mice treated with recombinant CRE recombinase. j Alizarin Red staining of Tyro3 KO calvarial cell cultures treated with PROS1 (100 nM). k, l RT-qPCR analysis of Alpl (k) and Bglap (l) mRNA expression in Tyro3 KO osteoblasts (n = 3 biological replicates). m–p Confocal imaging of pRLC and F-actin staining of TYRO3 KO osteoblasts on glass coverslips treated with TAM receptor ligand PROS1 (100 nM) (m). F-actin intensity (n = 3, mean of 50 measurements in three fields) (n), stress fiber-containing cells (n = 3, mean of 100 measurements in three fields) (o), and pRLC intensity (n = 3, mean of 50 measurements in three fields) (p) was quantified. Data were means ± SEMs. Statistical significance was determined by a two-tailed unpaired t-test.

Fig. 4. The MERTK-inhibitor R992 induces bone formation in healthy mice.

a Structure and chemical formula of R992: 3-(Butylamino)−5-((1r,4r)−4-hydroxycyclohexyl)−8-((4-methylpiperazin-1-yl)methyl)pyrimido[4,5-c]isoquinolin-6(5H)-one (C27H38N6O2). b Illustration of R992 docked in MERTK and the H-bonds R992 forms with three residues in the ATP pocket (MET-647, PRO-672, and ASP-741). c R992 on-target and off-target activity. d, e Immunoblots of MERTK phosphorylation and AKT signaling in human (d) and murine (e) cells. f μCT of the metaphyseal proximal region of tibias of 10-week-old healthy C57BL/6J mice treated after 2 weeks of treatment with vehicle or R992. g Quantification of trabecular bone volume (BV/TV) of the proximal tibia determined by μCT analysis (Vehicle n = 7 and R992 n = 7). h, i Representative pictures (h) and analysis (i) of bone formation rate by Calcein and Demeclocycline double labeling (Vehicle n = 7 and R992 n = 7). j, k Representative pictures of Alizarin red staining of calvarial cells cultures treated with different doses of R992 on day 21 (j). The number of mineralized nodules was quantified (k) (n = 3 biological replicates). l Analysis of osteoblast differentiation marker Alpl, Runx2, and Osx on day 7 in R992-treated calvarial cell cultures in comparison to control treated cultures (n = 3 biological replicates). Data were means ± SEM. Statistical significance was determined by a two-tailed unpaired t-test.

Fig. 5. Pharmacologic MERTK blockade inhibits multiple myeloma tumor progression and bone disease.

a–k After intraosseous femoral injection of U266 myeloma cell line, NSG mice were treated with vehicle or 60 mg/kg R992 BID. a Assessment of Igλ light chain paraprotein in the peripheral blood by ELISA (week 3 n = 8/6, week 5 n = 5/4, and week 8 n = 5/5). b Bone marrow CD138⁺ myeloma plasma cell infiltration in the contralateral leg by FACS (n = 5/4). c μCT analysis of bone loss by calculating the change in bone volume (BV/TV) in relation to healthy control mice (n = 7/8). d 3D reconstructions of microcomputed tomography (μCT) recordings of the metaphyseal proximal region of the femur of healthy and myeloma-bearing mice. e Representative pictures of TRAP Hematoxylin staining (osteoclasts stained in red) showing myeloma infiltration in the bone marrow (#1), osteoblasts visible as cuboidal or polygonal mononuclear cells on the endosteal bone surface (arrows in green) (#2), osteolysis (#3), histologically detectable signs of bone repair in osteolysis zones induced by R992 (arrows in green pointing to osteoblasts, arrows in red pointing to osteoclasts) (#4), and detection of TRAP-positive mononuclear cells indicating osteoclast precursor cell recruitment (#5) (arrows in yellow) (n = 5). f Histomorphometric analysis of osteoblast number per bone perimeter (n = 7/8). g Determination of serum bone formation marker P1NP by ELISA (n = 8/8). h Histomorphometric analysis of osteoclast number per bone perimeter. i Determination of serum osteoclast marker TRAP5b by ELISA (n = 8/8). j Histomorphometric analysis of the number of TRAP⁺ mononuclear cells in the bone marrow (n = 7/8). k Kaplan–Meier curve showing overall survival (n = 13/13, p = 0.030 Mantel–Cox test). Median survival was 73 vs. 88 days. Data were means ± SEM. Statistical significance was determined by a two-tailed unpaired t-test unless otherwise stated.

Fig. 6. Pharmacologic MERTK blockade inhibits tumor progression and cancer-induced bone loss in breast- and lung cancer bone metastasis.

a–n After intracardiac injection of luciferase⁺ H460 and MDA-MB-231 cells, NSG mice were treated with vehicle or 60 mg/kg R992 BID. a Representative images from bioluminescence imaging indicating the amount of tumor load. b, c Luminoscore (whole-body radiance intensity) in the H460 (b) and MDA-MB-231 (c) mouse model (H460 model: Vehicle n = 6 and R992 n = 8; MDA-MB-231 model: Vehicle n = 10 and R992 n = 8). d μCT analysis of bone loss by calculating the change in bone volume (BV/TV) of tumor-bearing mice in relation to healthy age-matched control mice (H460 model: healthy controls n = 5, vehicle n = 14, and R992 n = 10; MDA-MB-231 model: healthy controls n = 8, vehicle n = 20, and R992 n = 16). e 3D reconstructions of microcomputed tomography (μCT) recordings of the metaphyseal proximal region of the tibia of healthy and tumor-bearing mice treated with vehicle or R992. f Representative pictures of TRAP/Hematoxylin staining of bone metastasis of MDA-MB-231 injected mice. Osteoclasts are stained in red, green arrows pointing to osteoblasts, and yellow arrows pointing to intratumoral TRAP-positive mononuclear cells. g Histomorphometric analysis of osteoblast number (H460: vehicle n = 14 and n = 10; MDA-MB-231: vehicle n = 20 and R992 n = 16). h Determination of serum bone formation marker P1NP by ELISA (H460: vehicle n = 7 and R992 n = 5; MDA-MB-231: vehicle n = 9 and R992 n = 8). i Histomorphometric analysis of osteoclast number per bone perimeter (H460: vehicle n = 14 and R992 n = 10; MDA-MB-231: vehicle n = 20 and R992 n = 16). j Determination of serum osteoclast marker TRAP5b by ELISA (H460: vehicle n = 7 and R992 n = 5; MDA-MB-231: vehicle n = 10 and R992 n = 8). k, l Histomorphometric analysis of the number of TRAP⁺ mononuclear cells in the bone marrow in the H460 model (vehicle n = 14 and R992 n = 10) (k) and in the MDA-MB-231 model (vehicle n = 12 and R992 n = 7) (l). m, n Kaplan–Meier curves of overall survival in the H460 (m) and MDA-MB-231 (n) model (H460: vehicle n = 6 and R992 n = 8; p = 0.0463 Mantel–Cox test; Median survival 16 vs. 19 days) (MDA-MB-231: vehicle n = 5 and R992 n = 5; p = 0.0391 Mantel–Cox test; Median survival 55 vs. 60 days). Data were mean ± SEM. Statistical significance was determined by a two-tailed unpaired t-test unless otherwise stated.

Fig. 7. Mertk deficient mice display reduced bone loss in a syngeneic breast cancer bone metastasis mouse model.

a–k Luciferase⁺ EO771 breast cancer cells were injected intracardially in Mertk^flox/flox and Col1a1-cre⁺;Mertk^flox/flox mice. a, b Analysis of tumor load in the tibia by Bioluminescence Imaging after 3 (n = 12/16) (a) and 7 (b) (n = 12/16) days. c–h μCT 3D reconstructions (c) and analysis of BV (n = 20/22) (d), BV/TV (n = 20/22) (e), Tb.N (n = 20/22) (f), Tb.Th (n = 20/22) (g), and Tb.Sp (n = 20/22) (h) of trabecular bone of metaphyseal proximal region of the tibia. i Representative pictures of TRAP/Hematoxylin staining of bone metastasis of EO771 injected mice. j, k Histomorphometric analysis of osteoblast number (Ob.N/B.Pm) (n = 12/13) (j) and osteoclast number (Oc.N/B.Pm) (n = 12/13) (k). l–v Luciferase⁺ EO771 breast cancer cells were injected intracardially in Col1a1-cre⁺;Tyro3^flox/flox mice. l, m Analysis of tumor load in the tibia by Bioluminescence Imaging after 3 (n = 12/16) (l) and 7 (n = 12/16) (m) days. n–s μCT 3D reconstructions (n) and analysis of BV (n = 12/16) (o), BV/TV (n = 12/16) (p), Tb.N (n = 12/16) (q), Tb.Th (n = 12/16) (r), and Tb.Sp (n = 12/16) (s) of trabecular bone of metaphyseal proximal region of the tibia. t Representative pictures of TRAP/Hematoxylin staining of bone metastasis of EO771 injected mice. u, v Histomorphometric analysis of osteoblast number (Ob.N/B.Pm) (n = 12/16) (u) and osteoclast number (Oc.N/B.Pm) (n = 12/16) (v). Data were means ± SEMs. Statistical significance was determined by a two-tailed unpaired t-test unless otherwise stated.

Fig. 8. Schematic outline of the mechanism showing how the MERTK-PROS1 axis exerts osteopenia via the VAV2-RHOA pathway.

Inhibition of this axis led to the normalization of bone homeostasis by increasing osteoblast function.