Activating transcription factor 4 regulates osteoclast differentiation in mice

Abstract

Activating transcription factor 4 (ATF4) is a critical transcription factor for osteoblast (OBL) function and bone formation; however, a direct role in osteoclasts (OCLs) has not been established. Here, we targeted expression of ATF4 to the OCL lineage using the Trap promoter or through deletion of Atf4 in mice. OCL differentiation was drastically decreased in Atf4-/- bone marrow monocyte (BMM) cultures and bones. Coculture of Atf4-/- BMMs with WT OBLs or a high concentration of RANKL failed to restore the OCL differentiation defect. Conversely, Trap-Atf4-tg mice displayed severe osteopenia with dramatically increased osteoclastogenesis and bone resorption. We further showed that ATF4 was an upstream activator of the critical transcription factor Nfatc1 and was critical for RANKL activation of multiple MAPK pathways in OCL progenitors. Furthermore, ATF4 was crucial for M-CSF induction of RANK expression on BMMs, and lack of ATF4 caused a shift in OCL precursors to macrophages. Finally, ATF4 was largely modulated by M-CSF signaling and the PI3K/AKT pathways in BMMs. These results demonstrate that ATF4 plays a direct role in regulating OCL differentiation and suggest that it may be a therapeutic target for treating bone diseases associated with increased OCL activity.

Figure 1. OCL differentiation is dramatically diminished in Atf4^–/– BMM cultures and bones.

(A) Western blot. Whole cell extracts (20 μg) from primary BMMs were incubated with or without 1 unit calf intestinal phosphatase (CIP) at room temperature for 30 minutes. (B) IHC. Differentiated BMMs were stained with an ATF4 antibody or control IgG. (C) Tibial sections were stained for TRAP activity for 30 minutes at 37°C. TRAP activity in both metaphyseal (top) and epiphyseal (bottom) regions of tibias is shown. (D) TRAP⁺ OCLs (arrows) on trabecular surfaces of WT and Atf4^–/– tibiae. Oc.S/BS and Oc.Nb/BPm values for primary and secondary spongiosa are shown in Table 1. (E–G) WT and Atf4^–/– BMMs were maximally differentiated for 9 days, followed by TRAP staining. TRAP⁺ MNCs (F) and the number of nuclei per OCL (G) were scored. (H and I) Bone resorption pit assay on dentin slices. BMMs were differentiated on dentin slices for 9 days. (H) Bone resorption pits were stained with hematoxylin solution. (I) Pit area versus total bone area on each dentin slice was measured as described in Methods. (J) Time course of TRAP⁺ mononuclear OCL differentiation. BMMs were differentiated for the indicated times followed by TRAP staining, and percent TRAP⁺ mononuclear cells was measured. *P < 0.01 versus WT. Original magnification, ×100 (C, E, and H), ×200 (B and D).

Figure 2. ATF4 deficiency impairs OCL differentiation in a cell-autonomous manner.

(A and B) CFU-GM assay. 2 × 10⁴ BMMs from WT and Atf4^–/– mice (6 per group) were cultured in methylcellulose semisolid medium in 35-mm dishes in the presence of 1.0 ng/ml recombinant human GM-CSF for 10 days. The number of CFU-GM colonies was counted under an inverted microscope. Experiments were repeated 2 times. (C and D) OCL-OBL coculture. Primary calvarial OBLs from 3-day-old WT mice were cocultured with WT or Atf4^–/– BMMs as described in Methods. (E and F) Effects of increased RANKL. Primary BMMs (E) and purified CD11b⁺ BMMs (F) from both genotypes were differentiated in the presence of increasing concentrations of RANKL for 7 days, and the number of TRAP⁺ MNCs per well was counted. *P < 0.01 versus WT. Original magnification, ×40 (A); ×200 (C).

Figure 3. OCL-targeted transgenic overexpression of ATF4 dramatically increases OCL differentiation and bone resorption and results in a severe osteopenic phenotype.

(A) Schematic representation of a transgene construct. An 1,846-bp fragment of the mouse Trap promoter was used to drive expression of full-length mouse ATF4 cDNA. Atf4 transgene expression in RANKL-differentiated and undifferentiated BMMs, BMSCs, or calvarial OBLs was measured by quantitative real-time RT-PCR using transgene-specific primers as described in Methods. (B–D) In vitro OCL differentiation. BMMs from 4-week-old WT and Trap-Atf4-tg mice (founder no. 2360) were differentiated into OCLs for 5 days followed by (B) Western blot analysis of ATF4, NFATc1, PU.1, CSFR1, and β-actin for loading; (C) real-time RT-PCR analysis for Trap, Cat K, Mmp9, Rank, Spi1, and Csfr1 mRNAs; and (D) TRAP staining of the BMM cultures. (E) TRAP staining. Tibial sections from 4-week-old WT and Trap-Atf4-tg mice were stained for TRAP activity. (F) TRAP⁺ OCLs (arrows) on trabecular surfaces of WT and Trap-Atf4-tg tibiae. Oc.S/BS and Oc.Nb/BPm values for primary and secondary spongiosa are shown in Table 2. (G) μCT analysis. Fixed nondemineralized femurs from 3-month-old male WT and Trap-Atf4-tg mice were used for μCT analysis as previously described (41). BV/TV, Tb.N, and Tb.Sp values are shown in Table 3. n = 3–7. *P < 0.01 versus WT. Original magnification, ×100 (D and E); ×200 (F).

Figure 4. ATF4 regulates NFATc1 expression in BMM cultures and bones.

(A and B) Total RNAs and protein lysates from differentiated WT and Atf4^–/– BMMs were used for real-time RT/PCR analysis (A) and Western blot (B). (C and D) Differentiated BMMs and tibial sections were subjected to IHC staining for NFATc1. (E) WT BMMs were infected with increasing amounts of ATF4 adenovirus, then switched to differentiation medium for 72 hours, followed by Western blot for NFATc1. (F) ATF4 activates the Nfatc1 P1 promoter. COS-7 cells were transfected with 0.8-kb Nfatc1-luc or 2.8-kb mouse Runx2-luc constructs and pRL-SV40 with the indicated amounts of ATF4 expression plasmid. After 30 hours, cells were harvested for the dual luciferase assay. *P < 0.01 versus 0 μg ATF4. (G) COS-7 cells transfected with 0.8-kb Nfatc1-luc (WT) or the same plasmid containing a 4-bp substitution mutation (MT) in the putative ATF4-binding site and pRL-SV40 with or without ATF4 expression plasmid. *P < 0.05 versus β-gal; ^#P < 0.05, WT versus MT ATF4/β-gal. (H) ChIP assay. A schematic representation of the relevant region of the mouse Nfatc1 P1 promoter is shown. P1 and P2 indicate PCR primers used to analyze ChIP DNAs. RAW264.7 cells were treated with or without 50 ng/ml RANKL for 24 hours. ChIP assays were performed using antibodies against ATF4, c-Fos, or NFATc1. (I) WT and Atf4^–/– BMMs were cultured and infected with increasing amounts of retrovirus expressing caNFATc1, and switched to differentiation medium for 7 days. The number of TRAP⁺ MNCs per well was counted. *P < 0.01 versus WT. Original magnification, ×100.

Figure 5. RANKL activation of the MAPK pathways is severely compromised in Atf4^–/– OCL progenitors.

(A) Western blot. WT and Atf4^–/– BMMs were cultured in proliferation medium for 3 days and switched to 2% FBS α-MEM without M-CSF overnight, after which cells were exposed to 100 ng/ml RANKL for the indicated times. Cells were then lysed, fractionated by SDS–PAGE, and analyzed by Western blot analysis using antibodies recognizing phosphorylated and total ERK1/2, p38, JNK, and IκBα. β-Actin served as the loading control. Similar results were obtained from 3 independent experiments. (B–E) Statistical analysis of the Western blots in A. *P < 0.01, WT versus KO.

Figure 6. ATF4 is upregulated by M-CSF and PI3K/AKT and is required for M-CSF induction of RANK expression in BMMs.

(A and B) Effects of M-CSF on ATF4 in BMMs. Cells were cultured with or without 30 ng/ml M-CSF for the indicated times, followed by Western blot (A) or real-time RT-PCR (B) for ATF4. (C) Effects of various inhibitors or activators on the level of ATF4 in BMMs. Cells were cultured in M-CSF–containing medium with and without the indicated inhibitors or activators (10 μM) for 24 hours. LY, LY294002; SB, SB209580; GF, GF109203X. (D) Effect of PI3K/AKT inhibition on OCL differentiation. BMMs were seeded in proliferation medium for 3 days and treated with increasing concentrations of LY294002 for 24 hours. Inhibitor was then removed by switching cells to differentiation medium for 5 days, followed by TRAP staining. *P < 0.01 versus 0 μm. (E and F) COS-7 cells were transfected with 1.0 μg pCMV/ATF4 expression plasmid. After 24 hours, cells were treated with or without 10 μM LY294002 as well as with or without 10 μg/ml CHX (E) or 10 μM MG115 (F) for another 24 hours. (G) IHC. Purified CD11b⁺ BMMs were seeded in proliferation medium for 72 hours, followed by IHC with an anti-RANK antibody or control IgG. (H) Western blot. Primary BMMs were seeded in 35-mm dishes in proliferation medium for 72 hours. (I) Real-time RT-PCR. WT and Atf4^–/– BMMs were cultured in proliferation medium for 3 days and switched to 2% FBS α-MEM without M-CSF overnight. Cells were then treated with 10 ng/ml M-CSF for the indicated times. *P < 0.01, WT versus KO. Original magnification, ×200.

Figure 7. ATF4 deficiency increases CD11b⁺ cells in bone marrow and spleen and reduces CD3^–CD45R^–CD11b^–/loc-kit⁺CD115^hi cells in bone marrow.

(A) Flow cytometry. Splenocytes and bone marrow cells from WT and Atf4^–/– mice were stained with bio-strep-PB–conjugated CD11b, FITC-conjugated CD3, PE-conjugated CD45R, Pecy5-conjugated c-kit (CD117), and APC-conjugated CD115 antibodies and analyzed with flow cytometry as described in Methods. The percentage of CD3^–CD45R^–CD11b^–/loc-kit⁺CD115^hi cells in bone marrow cells or splenocytes was calculated by multiplying the percentages of gated populations as indicated. A representative experiment is shown; values in Results were averaged over 5 independent experiments. 1 WT and 1 Atf4^–/– mouse were used in each experiment. (B and C) BrdU staining. Purified CD11b⁺ BMMs were cultured in 8-well chambers (5 × 10⁵ cells/well) in proliferation medium for 72 hours, followed by BrdU staining as described previously (40, 41). Arrows indicate BrdU⁺ (i.e., proliferating) cells. (D and E) TUNEL staining. CD11b⁺ BMMs were treated as in B, followed by TUNEL staining as described previously (40, 41). Arrows indicate apoptotic cells. Original magnification, ×100.