Cell culture, transfection, and ischemic treatment

Mouse osteoblasts (MC3T3-E1) were obtained from the American Type Culture Collection (Manassas, VA, USA). The cells were cultured in minimum essential medium containing 10% fetal bovine serum (FBS) (Gibco BRL, Gaithersburg, MD, USA) and 1% penicillin/streptomycin (100 U/mL) at 37°C in a humidified 5% CO₂ incubator. The cells were transfected using Lipofectamine 3000 (Invitrogen, Life Technologies, Carlsbad, CA, USA). To grow the cells in ischemic conditions, we grew them in medium without glucose (Mediatech, Manassas, VA, USA) inside an Oxoid anaerobic jar with an AnaeroGen sachet (Thermo Fisher Scientific, Waltham, MA, USA). The AnaeroGen sachet in the sealed jar rapidly absorbed atmospheric oxygen with production of CO₂, and the oxygen concentration fell to below 1% within 30 minutes.

Primary culture

For osteoblast and osteoclast co-culture, the calvarial bones from newborn mice were cut into pieces and digested in a solution of 1 mL trypsin (Gibco BRL), 3.2 mg collagenase II (Worthington, Lakewood, NJ, USA), and 4 mL of phosphate-buffered saline (PBS) and then transferred into α-modified Eagle's minimum essential medium (α-MEM) with 10% FBS and 1% penicillin/streptomycin in 48-well plates. Bone marrow cells were flushed from the femurs of the mice and cultured overnight in MEM containing 10% FBS.

Nonadherent cells were harvested as enriched hematopoietic monocyte/macrophage progenitors. They were separated with magnetic-activated cell-sorting technology using a mature hematopoietic lineage cell depletion kit supplemented with mouse CD11b microbeads (Miltenyi Biotec, Auburn, CA, USA). When osteoblast density reached 95% confluence per well, LacZ or Cre adenovirus was administered with a multiplicity of infection of 100. CD11b⁺ bone marrow cells (1 × 10⁵ bone marrow cells per 48 wells) were gently placed on top of the osteoblasts for osteoclast differentiation in α-MEM with 10% FBS, 1% penicillin/streptomycin, 10⁻⁸ M 1,25(OH)₂D₃ (Sigma-Aldrich, St. Louis, MO, USA), and 10⁻⁶ M prostaglandin E2 (Sigma-Aldrich) for 14 days.

Animals and surgical procedures

Sirt6 transgenic mice (C57BL/6-Tg[RP23-352G18]1Coppa/J), Sirt6^fl/fl mice (B6;129-Sirt6tm1Ygu/J), and OCN-Cre mice (B6N.FVB-Tg[BGLAP-cre]1Clem/J) were obtained from Jackson Laboratory (Bar Harbor, ME, USA). Sirt6^fl/fl and heterozygous OCN-Cre mice were crossed to obtain OS6KO mice. Seven- to 8-week-old male C57BL6 mice were used in this study because juvenile ION of the femoral head occurs five times more often in males than in females.⁽²⁹⁾ Mice were given free access to water and standard mouse-chow pellets and housed under controlled temperature (22 ± 1°C) and humidity (50% to 60%) conditions with a 12-hour light–dark cycle (lights on at 7 a.m.). Osteonecrosis was surgically induced in the distal femurs of mice using a dissection microscope and a previously described technique.⁽³⁰⁾ A skin incision was made in the anterior aspect of the right knee joint, and then an external fixator was attached to both the femur and tibia. A 30G insulin needle with a 0.3 mL syringe was used for each injection. The needle was inserted into the lateral aspect of the distal femoral epiphyses. Each mouse received a single intra-osseous injection of 50 μL of adenovirus (1 × 10⁹ plaque-forming units) in PBS into the distal femoral epiphyses immediately after the surgery.^{(31, 32)} Adenoviruses expressing Sirt6 (AdSirt6), a catalytically inactive mutant Sirt6 (AdmSir6), or β-galactosidase (AdLacZ) were prepared as described previously.⁽²⁴⁾ This study protocol was approved by the Institutional Animal Care and Use Committee of Chonbuk National University (permit no. CBNU 2017–0102).

Quantitative real-time PCR with reverse-transcription analysis

Total RNA was extracted with TRIzol reagent (Invitrogen) according to the manufacturer's instructions. The RNA was then precipitated with isopropanol and dissolved in diethyl pyrocarbonate–treated distilled water. cDNA was synthesized using random hexamer primers from first-strand cDNA synthesis kits according to the manufacturer's instructions (Applied Biosystems, Foster City, CA, USA). Quantitative polymerase chain reaction (PCR) was performed in 384-well plates using an ABI Prism 7900HT Sequence Detection System (Applied Biosystems, Foster City, CA, USA). RNA from individual samples was analyzed separately. Primers used in PCR are listed in Supplemental Table S1.

Western blot and co-immunoprecipitation

Protein lysates (15 μg) were separated using 10% SDS-PAGE and transferred to PVDF membranes. After blocking with 5% skim milk for 1 hour, the blots were probed with primary antibodies. For co-immunoprecipitation, 500 μg of protein precleared with protein G-agarose was incubated with anti-Sirt6 or anti-VDR antibodies overnight and then with protein G-agarose (Thermo Fisher Scientific) for 2 hours. Blots were probed with primary antibodies against Sirt6 and VDR, and antibody signals were detected using a LAS-3000 Luminoimage analyzer (Fujifilm, Tokyo, Japan). Antibodies against the proteins listed in Supplemental Table S2 were purchased from the manufacturers indicated. The gel images of protein bands were digitized using ImageJ software (NIH, Bethesda, MD, USA; https://imagej.nih.gov/ij/).

Assessment of bone destruction by micro-CT

A SKYSCAN 1076 Micro-CT unit (SkyScan, Kontich, Belgium) was used to assess bone volume within defect sites. The X-ray source was set at 75 kV and 100 mA, with a pixel size of 8.8 mm, and 400 projections were acquired over an angular range of 180 u (angular step of 0.45 u). The area included in the computed tomography (CT) scans was the distal half of the femur from the middle margin of the femoral diaphysis to the distal femoral epiphysis. A global thresholding algorithm was applied at a constant threshold for all specimens. The threshold was the intensity (gray value) corresponding to 45% of the average intensity of the intact cortical bone in the specimens. Voxels with intensities exceeding the threshold were considered to contain mineralized tissue. All system aspects were operated using Data Viewer software (SkyScan). On stacked, reconstructed micro-CT cross-sectional images, manual regions of interest (ROIs) on irregular anatomical contours were drawn on transverse images at the middle of the femoral condyle. ROIs excluded cortical bone. The volume of interest was a stack of ROIs drawn from distal femoral epiphyses. Trabecular bone in primary spongiosa was analyzed in the distal area extending proximally 1.75 mm from the end of the primary spongiosa. Cortical bone was analyzed in the diaphyseal area extending 1 mm proximally and distally from the midpoint between the femoral ends. In this work, Tb.Th is the trabecular mean thickness, Tb.Sp is the mean distance between trabeculae, and Tb.N is the average number of trabeculae present per unit length. Tb.Th and Tb.Sp were assessed using direct three-dimensional methods, and Tb.N was calculated using the formula Tb.N = (BV/TV) / Tb.Th. Cort.Th is the cortical bone mean thickness, and Cort.Po is the cortical bone total porosity. The shape of the distal femoral epiphysis was assessed for deformity, and the degree of epiphyseal collapse was evaluated by measuring the height and width of the epiphysis in the mid-coronal plane on micro-CT images. The ratio of height to width was calculated for each epiphysis. A lower height/width ratio indicates greater deformity.⁽³⁰⁾

Measurement of RANKL and OPG

Whole blood was collected from mice through inferior vena cava punctures. Bone marrow cells and fluid obtained from femurs and tibias were isolated by centrifugation. Bone marrow extracellular fluid was obtained by flushing bone marrow cells and fluid with 1 mL of PBS, and the supernatant was harvested after centrifugation. RANKL (Abcam, Cambridge, MA, USA; catalog no. 100749) and osteoprotegerin (R&D Systems, Minneapolis, MN, USA; catalog no. MOP00) were detected using ELISA kits according to the manufacturer's instructions.

Luciferase assay

Transient transfections were performed using Lipofectamine 3000 (Invitrogen) according to the manufacturer's instructions. Briefly, MC3T3-E1 cells were transfected with plasmids encoding the Flag, Sirt6, mutant Sirt6, RANKL-luc reporter, or Renilla luciferase reporter (pRL-TK-luc). The day after transfection, the cells were treated with 10 nM 1,25(OH)₂D₃. After 24 more hours, the cells were harvested in a reporter lysis buffer. Luciferase activity was determined in whole-cell lysates using a Promega luciferase assay kit (Promega, Madison, WI, USA; catalog no. E1960) and is expressed relative to the Renilla signal.

ChIP assay

Chromatin immunoprecipitation (ChIP) assays were performed in MC3T3-E1 cells using a SimpleChIP enzymatic chromatin IP kit (Cell Signaling, Beverly, MA, USA; catalog no. 9004) according to the manufacturer's instructions. DNA samples were amplified by PCR with a RANKL primer (Supplemental Table S1) to identify the VDR-element binding site on the RANKL promoter region.

Histological methods

Resected femurs and tibias were fixed in 10% neutral buffered formalin and decalcified in 10% EDTA for 20 days or in rapid decalcifying solution (Sigma-Aldrich) for 12 hours. Tissue sections were collected from the femoral midline, which was considered the most representative area. To evaluate the histologic findings, paraffin-embedded tissue sections were stained with hematoxylin and eosin (H&E) and tartrate-resistant acid phosphatase. Safranin O staining used Fast Green and Safranin O. Immunohistochemistry used a specific HRP/DAB (ABC) detection IHC kit, and the TUNEL assay used a DeadEnd colorimetric TUNEL system (Promega). Statistical analysis of empty lacunae in H&E staining and TUNEL-positive osteocytes was conducted as described previously.⁽³⁰⁾ The staining intensity of the osteocytes was graded from 0 to 3 (0 = no expression; 1 = mild expression; 2 = moderate expression; and 3 = strong expression), and the staining area of positive osteocytes was graded from 0 to 5 (0 = no stained cells; 1 = 1% of the cells stained positive; 2 = 2% to 10% of the cells stained positive; 3 = 11% to 33% of the cells stained positive; 4 = 34% to 66% of the cells stained positive; and 5 = 67% to 100% of the cells stained positive). The immunohistochemical staining score was obtained by summing the staining intensity score and area score.^{(33, 34)} Images were acquired using a Leica DM750 microscope (Leica, Wetzlar, Germany). Additional details on the histological methods are provided in Supplemental Table S2.

Statistical analysis

Statistical analyses for the data were performed using SPSS software version 25.0 (IBM, Armonk, NY, USA). Comparisons between two groups were evaluated using an unpaired two-tailed Student's t test. To compare more than two groups, we used one-way ANOVA followed by Tukey's test for multiple comparisons. All quantified data are presented as box plots using the median value (cross), interquartile range (box), minimum/maximum (whiskers), and symbols to represent all data points. Differences with p value <0.05 are considered statistically significant.

Ischemic injury induces expression of RANKL and inflammatory cytokines in osteoblasts and increases bone resorption with a deformity in mouse distal femur

To evaluate the molecular mechanisms involved in ION, we performed ION surgery on mice to recapitulate the juvenile form of ION.⁽³⁰⁾ At 2 and 4 weeks (early stage), necrotic and inflammatory tissues were observed in bone marrow areas (Fig. 1A). TRAP⁺ cells became more numerous, and severe bone resorption was detected in the epiphyses of the distal femurs (Fig. 1A, B). Bone deformity and destruction of articular cartilage were observed at 6 weeks (late stage) (Fig. 1A, B). Because cytokines associated with osteoclastogenesis and inflammation were secreted from osteoblasts,^(13-16) we evaluated the expression of cytokines in MC3T3-E1 osteoblast cells after ischemic injury. Twenty minutes after ischemia, the expressions of RANKL, OPG, IL-1β, IL-6, and TNF-α were highly increased in osteoblast cells (Fig. 1C). Similarly, expression of the same factors increased markedly in distal femoral epiphyses after ION surgery, as shown by immunohistochemical staining and Western blotting (Fig. 1E; Supplemental Fig. S1).

Sirt6 expression is attenuated in osteoblasts and distal femur after ischemic injury

To investigate the role of sirtuins in ION, we observed their expression after ischemic injury in osteoblasts and bone tissue. Among sirtuins, Sirt5 level decreased, and Sirt6 expression almost disappeared in osteoblast cells and distal femoral epiphyses (Fig. 2A, B). These results suggest that Sirt6 plays a major role in osteoblasts after ischemic injury. To test that hypothesis, we performed Sirt6 overexpression experiments in osteoblast cells and distal femoral epiphyses after ischemic injury. Sixty minutes after ischemic injury, RANKL, IL-1β, IL-6, and TNFα expression was highly suppressed in Sirt6-overexpressed cells compared with controls (Fig. 2C). Two weeks after ION surgery and administration of adenovirus-LacZ or adenovirus-Sirt6 through intraosseous injection, transduction of epiphyseal bone tissue with Sirt6 showed consistently decreased expression of RANKL, IL-1β, IL-6, and TNFα (Fig. 2D).

Re-expression of Sirt6 prevents bone resorption with deformity

Next, we tested whether Sirt6 blocks bone resorption and deformity after ischemia-inducing surgery. When adenovirus Sirt6 was injected into the distal femoral epiphysis after ION surgery, inflammation of the bone marrow cavity was reduced, and the numbers of empty lacunae and TUNEL-positive osteocytes, which indicate apoptosis of osteocytes, decreased compared with the control group (Fig. 3A; Supplemental Fig. S2). In addition, Sirt6 re-expression lowered TRAP⁺ cell numbers and markedly increased the parameters of bone mass (BV/TV, trabecular number, and trabecular thickness) at 2, 4, and 6 weeks after ION surgery (Fig. 3A–C). By the late stage, bone deformity, as determined by the height/width ratio of the distal femoral epiphysis, and cartilage destruction were improved by injection with adenovirus-Sirt6 (Fig. 3D, E).

Deacetylating function of Sirt6 suppresses RANKL, inflammatory cytokines, and bone resorption

Sirt6 reportedly plays several roles through its ability to deacetylate.^(23-25) To determine whether the protective effect of Sirt6 on the distal femoral epiphysis after ischemic injury is due to its deacetylase function, we performed an ischemic injury experiment using mutant Sirt6 without the deacetylase function. As expected, acetylation on histone H3 lysine 9 was highly suppressed in Sirt6-treated distal femoral epiphyses after ischemic injury compared with LacZ-treated epiphyses. However, that effect disappeared in epiphyses treated with mutant Sirt6 (Fig. 4A). Furthermore, the ability of Sirt6 to reduce the levels of RANKL and other inflammatory cytokines in bone tissue and bone marrow fluid was not observed when mutant Sirt6 was administered after ischemic injury (Fig. 4A, B). Alleviation of bone resorption by Sirt6 was not found with mutant Sirt6 (Fig. 4C, D). These results suggest that the deacetylase function of Sirt6 prevents inflammation and bone resorption in the distal femoral epiphysis after ischemic injury.

Osteoblast/osteocyte Sirt6 regulates the expression of RANKL by binding directly to the VDR

To better understand whether Sirt6 in osteoblasts regulates osteoclasts directly or in relation to ischemic mechanisms, we performed osteoblast/osteoclast co-culture experiments during normoxia. First, we cultured primary osteoblasts from Sirt6^fl/fl calvaria and then infected the cells with adenovirus-LacZ or adenovirus-Cre recombinase. Next, bone marrow cells from control or OS6KO mice were seeded in the LacZ-treated or Cre-treated Sirt6^fl/fl primary calvarial osteoblasts. As shown in Fig. 5A, the number of TRAP⁺ cells in the Cre virus–induced Sirt6-knockout osteoblast group (Cre-Sirt6^fl/flOb) was significantly higher than in the LacZ virus group (LacZ-Sirt6^fl/flOb) regardless of whether the bone marrow cells came from control or OS6KO mice, indicating that Sirt6 in osteoblasts directly regulates osteoclast differentiation.

To identify which osteoclastogenic factors derived from osteoblasts are regulated by Sirt6, we evaluated the expression of osteoblast- and osteocyte-derived osteoclastogenic factors in Sirt6^fl/fl primary calvarial osteoblasts after treatment with the LacZ or Cre adenovirus. Compared with LacZ-treated cells, Cre virus–induced Sirt6-knockout primary osteoblasts showed significantly increased expression of RANKL mRNA. However, the expression of other osteoclastogenic factors, including OPG, TNFα, IL-6, M-CSF, SEMA3A, and parathyroid hormone-like peptide, were not altered in Sirt6-knockout osteoblasts (Fig. 5B). The level of RANKL protein from bone tissue increased consistently in OS6KO, but other cytokines were not changed (Fig. 5C). We next assessed the levels of RANKL, OPG, and RANKL/OPG in serum from control and OS6KO mice. The serum RANKL level and RANKL/OPG ratio were significantly increased in OS6KO mice compared with their littermate controls (Fig. 5D).

Next, we investigated the molecular mechanism by which Sirt6 regulates RANKL. Luciferase reporter assays using RANKL-Luc confirmed that Sirt6 overexpression suppressed RANKL transcriptional activity, but transduction with an adenovirus carrying mutant Sirt6 that lacked deacetylase activity had no effect on that activity (Fig. 5E). After treatment with 1,25(OH)₂D₃, which stimulates RANKL expression in osteoblasts,^{(35, 36)} RANKL transcriptional activity in MC3T3-E1 cells was highly increased. However, 1,25(OH)₂D₃-induced RANKL transcriptional activity was strongly inhibited by Sirt6 (Fig. 5E). To confirm that the inhibitory effect of Sirt6 on RANKL was specifically in response to 1,25(OH)₂D₃, we administered parathyroid hormone (PTH), another ligand that induces RANKL expression,⁽³⁷⁾ to osteoblasts overexpressing Sirt6. RANKL increased after treatment with PTH and was not suppressed by Sirt6 (Supplemental Fig. S3). Thus, the mechanism of RANKL inhibition by Sirt6 depends on vitamin D signaling. 1,25(OH)₂D₃ is known to trigger RANKL expression through signals initiated by its VDR.^{(38, 39)} Given that Sirt6 suppressed 1,25(OH)₂D₃-induced RANKL promoter activity in our study and that Sirt6 is a histone deacetylase that inhibits the expression of genes, we questioned whether Sirt6 could bind to the VDR and prevent expression of RANKL.^{(23, 40)} As shown in Fig. 5F, a co-immunoprecipitation assay determined that Sirt6 and the VDR can bind directly to each other. In addition, RANKL expression was increased by Sirt6 knockdown, but that effect was not observed after VDR knockdown (Fig. 5G). Furthermore, ChIP analysis using a Sirt6 antibody showed that endogenous Sirt6 in osteoblasts can recruit the RANKL promoter regions, but this result disappeared after VDR knockdown (Fig. 5H). These results indicate that Sirt6 regulates expression of RANKL via the VDR.

Osteoblast/osteocyte Sirt6 is vital to preventing ION-induced bone deformity

To confirm whether Sirt6 prevents ION-induced bone deformity, we performed ION surgery on Sirt6 Tg mice (mice that physiologically overexpress Sirt6) and OS6KO mice. Before the ION surgery, we verified the expression of Sirt6 in the Sirt6 Tg and OS6KO mice and analyzed their phenotypes. In Sirt6 Tg mice, the expression of Sirt6 was generally elevated in various tissues, including bone and bone marrow, but the bone volume did not differ significantly between genotypes (Supplemental Fig. S4C, D). In OS6KO mice, on the other hand, the expression of Sirt6 decreased significantly only in bone, and metaphyseal trabecular bone was significantly decreased. However, we found no change in cortical bone and epiphysis between control and OS6KO mice (Supplemental Fig. S4A, B; Fig. 6E, F). After ION surgery, as expected, the Sirt6 Tg mice were resistant to bone resorption and bone deformity compared with the wild-type (WT) mice (Fig. 6A–D). In addition, empty lacunae of osteocytes and TUNEL-positive osteocytes in the trabecular framework of the epiphysis were significantly reduced in ION-Sir6 Tg mice compared with ION-WT mice (Fig. 6D; Supplemental Fig. S5A, B). In OS6KO mice, ION surgery resulted in more severe bone loss and bone destruction than in control mice (Fig. 6E–H), and almost all osteocytes in the epiphyseal trabecular bone were found to have empty lacunae and be TUNEL positive (Fig. 6H; Supplemental Fig. S5C, D).