The target genes of human miR-1260b were predicted using the database TargetScan (total context + + score <−0.7) (https://www.targetscan.org/vert_80/) and mirDIP (Integrated Score > 0.6 and score class was set to “very high” (top 1%)) (https://ophid.utoronto.ca/mirDIP/).

C57BL/6NCrSlc mice (female, 8-week-old) were purchased from Japan SLC (Hamamatsu, Japan) and used under an institutionally approved animal research protocol (protocol #A21-131-2; Kyushu University).

Maxillae were removed and fixed in 4% paraformaldehyde for 24 h. Thereafter, samples were demineralized using ethylenediaminetetraacetic acid solution (Osteosoft; Merck, Darmstadt, Germany) for 72 h and embedded in paraffin. Standard hematoxylin and eosin staining and dual-color immunofluorescence analysis using specific primary antibodies for mouse anti-ATF6β antibody (1:100, 15794-1-AP; Proteintech) were performed, as previously described (Nakao et al., 2021). Images were captured using a confocal laser-scanning microscope (Carl Zeiss LSM 700; Oberkochen, Germany) and ZEN 2012 software.

Linear PEI-NPs (in vivo-jetPEI) were purchased from Polyplus-Transfection SA (Illkirch-Graffenstaden, France). miRCURY LNA microRNA mimic for hsa-miR-1260b or control miRNA (negative control: cel-miR-39-3p; Qiagen, Hilden, Germany) was dissolved in 5% glucose solution at a concentration of 50 μM. Stealth™ RNAi duplexes against mouse ATF6β, a mixture of three different siRNAs (MSS236243, MSS236245 and MSS273801) or Stealth™ RNAi negative control duplex (Invitrogen Life Technologies, Carlsbad, CA, United States) were dissolved in 5% glucose solution at a concentration of 20 μM. Solutions containing miRNA or siRNA and PEI-NPs were mixed and incubated for 15 min at room temperature. A total of 10 μl of miR-PEI-NPs (miR: 125 pmol) or siRNA-PEI-NPs (siRNA: 50 pmol) was injected. The distribution of miRNA in mice was tracked using Cy-3 labeling miRNA. miR-1260b was labeled with the Label IT siRNA Tracker Cy3 Kit without any transfection reagent (Mirus Bio LLC., WI, United States).

Experimental periodontitis was induced as previously described (Nakao et al., 2021). Mice were randomly divided into the following three groups: 1) placebo (in vivo jetPEI only), 2) miR-control, and 2) miR-1260b mimic. To induce periodontal bone loss in mice, a 5-0 silk ligature (Akiyama Medical MFC Co., Tokyo, Japan) was tied around the right maxillary second molar. After ligation, the placebo or miR-PEI-NPs were injected into the palatal gingiva of the maxillary second molar using a 33-gauge needle Hamilton syringe (Hamilton Company. NV, United States).

The samples were scanned using a Micro-computed tomography (micro-CT) imaging system (ScanXmate; Comscan, Kanagawa, Japan). The measurement conditions for micro-CT included tube voltage (kV), electrical current (A), image voxel size, and slice thickness. After scanning, three-dimensional (3D) images were reconstructed using TRI/3D-BON software (Ratoc System Engineering, Tokyo, Japan). To evaluate alveolar bone resorption, bone resorption volumes were superimposed on the non-ligated and ligated sides. The differences between the non-ligated and ligated sides were measured by following the selection of a 3D region of interest as described below. Crown buccal-palatal widths were defined as the buccal-palatal landmark. M1’s palatal root apex and M2 were assigned as the mesial-distal landmarks. The cement–enamel junction and M1 and M2 palatal root apexes were defined as tooth axial landmarks (Supplementary Figure S1A). Moreover, (non-ligated–ligated) bone volume/non-ligated bone volume (%) was taken as the degree of bone resorption. The alveolar bone loss of each group was also defined as the sum of distances from the eight sites as indicated in Supplementary Figure S1B.

Recombinant human macrophage colony-stimulating factor (M-CSF), receptor activator of NF-κB ligand (RANKL) and mouse RANKL were purchased from BioLegend (San Diego, CA, United States). Tunicamycin was obtained from Cayman Chemical Co. (Ann Arbor, MI, United States).

Human CD14⁺ peripheral blood-derived monocytes (PBMCs) were purchased from Lonza (Basel, Switzerland). The cells were cultured in the Roswell Park Memorial Institute 1,640 medium (Nacalai Tesque, Kyoto, Japan) supplemented with 10% heat-inactivated fetal bovine serum, 2 mM glutamine (Nacalai Tesque), 1% sodium pyruvate (Nacalai Tesque), 1% non-essential amino acids (Nacalai Tesque), and 25 ng/ml M-CSF. Human periodontal ligament (PDL) cells (Lonza) were grown in complete fibroblast medium (FM) using the FibroLife S2 Comp Kit (Kurabo Industries Ltd., Osaka, Japan), and cells at the second to third passages were used for subsequent experiments. The mouse osteoclast precursor clone RAW-D cells, which were derived from RAW264 macrophage-like cell line were kindly gifted by Prof. Toshio Kukita and Prof. Takayoshi Yamaza (Kyushu University, Japan) and cultured as previously described (Watanabe et al., 2004) (Kukita et al., 2004). RAW-D cells were stimulated with 50 ng/ml RANKL to differentiate into osteoclasts.

Quantitative RT-PCR (qRT-PCR) was performed as previously described (Watanabe et al., 2022). Total RNA was isolated from the cells using ISOGENⅡ (Nippon Gene, Tokyo, Japan), and first-strand cDNA was synthesized using PrimeScript RT Master Mix (Takara Bio, Otsu, Japan). qRT-PCR was performed using the Luna Universal qPCR Master Mix (NEW ENGLAND BioLabs Inc.) on a StepOnePlus Real-Time System (Applied Biosystems, Carlsbad, CA, United States) under the following conditions: 95°C for 1 min, 40 cycles of 95°C for 15 s, and 60°C for 30 s. Primer sequences used in this study are listed in Supplementary Table S1 .

Western blotting was performed as previously described (Watanabe et al., 2022). Cells were washed with phosphate-buffered saline and lysed in Passive Lysis 5× buffer (Promega, Madison, WI, United States). Nuclear proteins were extracted using a LysoPure Nuclear and Cytoplasmic Extractor Kit (FUJIFILM, Tokyo, Japan). Protein samples were separated on polyacrylamide gels and transferred to polyvinylidene difluoride membranes. The membranes were incubated with the appropriate primary antibodies, anti-ATF6β (1:1000; 15794-1-AP, Proteintech, Rosemont, IL, United States), anti-RANKL (1:1000; 12A668, Novus Biologicals, United States), anti-CHOP (1:1000, L63F7, Cell Signaling Technology), anti-Grp78 (1:1000, C50B12, Cell Signaling Technology), anti-cleaved caspase-3 (1:1000, 5A1E, Cell Signaling Technology), anti-JNK (1:1000, 56G8, Cell Signaling Technology), anti-Phospho-JNK (Thr183/Tyr185) (1:1000, 81E11, Cell Signaling Technology), anti-Lamin B1 (1:1000, 1298-1-AP, Proteintech) and anti-β-actin (1:1000; 13E5, Cell Signaling Technology) antibodies, and secondary antibodies, anti-rabbit IgG (1:2000; Cell Signaling Technology) and anti-mouse IgG (1:2000; Cell Signaling Technology). Blotted membranes were visualized using a densitometry technique on Amersham ImageQuant 800 (Cytiva, Tokyo, Japan) and quantified using Multi Gauge 3.1 software (FUJIFILM, Tokyo, Japan).

Transfection of miRNA mimics and siRNAs for PDL cells was performed using the Lipofectamine RNAiMAX Transfection Reagent (Thermo Fisher Scientific. MA, United States) for 24 h according to our reverse transfection protocol (Fukuda et al., 2013). For knockdown analysis, Stealth™ RNAi duplexes against human ATF6β, a mixture of three different siRNAs (HSS102271, HSS102272 and HSS102273) and mouse ATF6β, a mixture of three different siRNAs (MSS236243, MSS236245 and MSS273801) were obtained from the Invitrogen Corporation (Invitrogen Life Technologies). Stealth™ RNAi negative control duplex (Medium GC Duplex, Invitrogen Life Technologies) was used as a control. For miRNA mimic transfection, PDL cells were transfected with 20 pmol miRCURY LNA microRNA mimic for hsa-miR-1260b (Qiagen) or control miRNA (negative control: cel-miR-39-3p; Qiagen). RAW-D cells were transfected with mouse ATF6β siRNAs or control siRNA using Hiperfect Transfection Reagent (Qiagen), according to the manufacturer’s protocol.

The fragment of ATF6β 3′-UTR (WT) and its corresponding mutated sequence (MUT) designed by converting the miR-1260b binding sequence 5′-GGUGGGA to 5′-ACTCAA, were synthesized and cloned into the pmirGLO dual-luciferase miRNA Target Expression Vector (Promega, Madison, WI, United States), termed ATF6β-3′-UTR-WT (5′-AAA​CTA​GCG​GCC​GCT​AGA​GGG​TGG​T) and ATF6β-3′-UTR-MUT (5′-AAA​CTA​GCG​GCC​GCT​AGT​AAC​TCA​T), respectively. Reporter plasmids containing the WT or MUT ATF6β 3′-UTR (0.5 μg) and miR-1250b mimic or control miRNA (20 pmol) were co-transfected into PDL cells using Lipofectamine 3000 (Thermo Fisher Scientific. MA, United States). After 24 h of incubation, cells were lysed. Firefly and Renilla luciferase activities were detected using a dual-luciferase assay system (Promega, WI, United States).

PDL cells (5 × 10³ cells) in 100 µl of FM were seeded into a 96-well culture plate in triplicate with or without 0.5 μg/ml of tunicamycin. The number of seeded cells was limited to ensure that cell growth would continue until the end of incubation period, without reaching cell confluence. Next, 10 µl of WST-8 solution (Cell Count Reagent SF™: Nakarai Tesuque, Kyoto, Japan) were added to each well (including the control wells), at 3, 6, 12, 24 and 48 h. Cells were incubated for an additional 1 h at 37°C, then absorbance at 450 nm was measured, with a reference reading at 650 nm.

CD14⁺ PBMCs (5 × 10⁴) were placed in a 96-well plate. The cells were incubated with RANKL (30 ng/ml) and M-CSF (25 ng/ml) for 2 weeks with or without tunicamycin treatment for the first 3 days. The medium was replaced every 3 days. Cultured cells were fixed with 10% formalin for 5 min and then with ethanol–acetone (50:50 vol/vol) for 1 min at room temperature. Tartrate-resistant acid phosphatase (TRAP) staining was performed using a TRAP/ALP Stain Kit (FUJIFILM Wako). Photographs were taken using BZ8000 (Keyence Co., Osaka, Japan). The number of TRAP-positive cells was counted for each number of nuclei (7 or 10 ≤ nuclei) (Piper et al., 1992). Areas of each TRAP-positive cell were measured using the ImageJ software (version 1.43; National Institute of Health, Bethesda, MD, United States).

CD14⁺ PBMCs were cultured at a density of 1 × 10⁵ cells/well in a bone resorption assay plate 48 (PG Research, Tokyo, Japan) coated with calcium phosphate (CaP-coated). The cells were incubated as described above. After 14 days, the CaP-coated plate was treated with 5% sodium hypochlorite (Sigma-Aldrich, St Louis, Missouri, United States) for 5 min, according to the manufacturer’s instructions. The resorption pit areas were analyzed using the ImageJ software.

Data are presented as the mean ± standard deviation. Data analyses were performed using GraphPad Prism 9 (GraphPad Software Inc., La Jolla, CA, United States). One- or two-way analysis of variance followed by correction for multiple comparisons with Tukey’s post hoc test were used to compare three or more groups to analyze the statistical significance. Other statistical comparisons of data from the two groups were performed using the Student’s t-test. Statistical significance was set at p < 0.05.

To explore the most critical gene for biological regulation by miR-1260b in periodontal tissue, miRDIP (https://ophid.utoronto.ca/mirDIP/) and TargetScan (https://www.targetscan.org/vert_80/) databases were analyzed. First, we selected the top five overlapping TargetScan (total context + + score <−0.7) and mirDIP (Integrated Score >0.6) genes (Figure 1A). Considering that ATF6β is an ER stress response-associated gene and emerging evidence has indicated the involvement of ER stress in periodontal disease, we focused on ATF6β in this study. In a mouse periodontitis model, the expression of ATF6β mRNA was upregulated in ligated-gingival tissue (Figure 1B). Immunohistochemical staining revealed that ATF6β was mainly expressed in both the PDL and subgingival connective tissue, and the accumulated expression of ATF6β was observed along the alveolar bone resorption area (Figure 1C). Enhanced protein expression of ATF6β in the inflammatory periodontium was confirmed by western blotting analysis (Figures 1D,E).

To investigate the inhibitory and therapeutic potential of miR-1260b against ligature-induced periodontal bone loss in vivo, a miR-1260b mimic was injected into the interdental region of second molar (M2) using PEI-NPs (Figure 2A). Relative alveolar bone resorption was determined using a split-mouth experimental design, in which one side of the maxilla (upper jaw) was ligated and locally injected with miR mimic-PEI NPs, whereas the other side was not ligated. Thus, each animal served as a control. Fluorescent imaging revealed that Cy3-labeled miR-1260b mimic-PEI-NPs were retained in the interdental gingiva for 7 days (Figure 2B). qRT-PCR further confirmed the local administration-induced upregulation of miR-1260b expression in the gingival tissue (Figure 2C). Compared with the non-ligated site, severe alveolar bone loss was observed around the ligated M2 in both the mock- and miR-Ctrl-injected groups. In contrast, local injection of the miR-1260b mimic significantly reduced the bone resorption on day 7 after ligation (Figure 3). To gain further insight into miR-1260b-mediated inhibition of ATF6β, the expression level of ATF6β in periodontal tissue was monitored. qRT-PCR analysis revealed that treatment with miR-1260b significantly inhibited the ligation-induced expression of ATF6β mRNA (Figure 4A). Consistently, immunofluorescence staining also demonstrated the downregulated expression of ATF6β by the miR-1260b mimic on day 7 after ligation (Figures 4B,C). To validate that miR-1260b-mediated downregulation of ATF6β is responsible for the inhibition of periodontal bone loss, ATF 6β siRNA-PEI NPs was injected in mice periodontal model. qRT-PCR and immunofluorescence staining confirmed the knockdown of ATF6β expression in mice periodontal tissue (Figures 4D–F). Micro-CT analysis demonstrated that ATF 6β siRNA-PEI NPs significantly reduced alveolar bone resorption than Ctrl-siRNA-injected groups (Figures 4G–I). On the other hand, there was no difference between Ctrl and ATF6β siRNA injected groups for BV/TV (Figure 4J), indicating that the inhibitory effect for bone loss by miR-1260b is stronger than ATF6β siRNA. Collectively, these results suggest that the miR-1260b-mediated downregulation of ATF6β is responsible for the inhibition of ligation-induced alveolar bone loss.

Next, we explored the molecular mechanism of miR-1260b-mediated ATF6β suppression in the inhibition of bone resorption. Putative binding sites for miR-1260b within the 3′-UTR sequence of ATF6β were identified using TargetScan (Figure 5A). As the PDL–alveolar bone interface plays a critical role in periodontal bone homeostasis (Kanzaki et al., 2001), we performed a dual-luciferase reporter assay to confirm the physical binding of miR-1260b to ATF6β mRNA in primary PDL cells. Inserts harboring the predicted ATF6β WT (ATF6β-WT) and MUT (ATF6β -Mut) sequences were cloned into a luciferase reporter vector. Co-transfection of ATF6β-WT with the miR-1260b mimic in PDL cells showed a significant decrease in the relative luciferase activity compared to that in the negative control miRNA mimic (NC-miR mimic) group. This was not observed in the experiments using ATF6β-Mut (Figure 5B). This confirmed the preferential binding of miR-1260b to the 3′-UTR of ATF6β. Next, we assessed the miR-1260b-mediated downregulation of ATF6β expression under ER stress. Treatment of PDL cells with the ER stressor, tunicamycin, significantly increased the expression of ATF6β mRNA (Figure 5C), whereas the induction of ATF6β was diminished via transfection with the miR-1260b mimic. ATF6β is proteolytically cleaved during the ER stress response to release cleaved ATF6β (Thuerauf et al., 2007). As a transcription factor, the cleaved active form of ATF6β is translocated to the nucleus. Immunofluorescence staining captured by confocal microscopy demonstrated that tunicamycin-induced potent nuclear accumulation of ATF6β was diminished by miR-1260b (Figure 5D). Western blotting further confirmed that tunicamycin-mediated ER stress increased the expression of cleaved ATF6β in the nuclear fraction of PDL cells, which was significantly inhibited by miR-1260b (Figures 5E,F). We subsequently examined whether ER stress-induced cleavage of ATF6β was involved in RANKL expression. ATF6β knockdown successfully inhibited the nuclear expression of cleaved ATF6β (Figures 6A,B), thereby abrogating tunicamycin-induced RANKL expression in PDL cells (Figures 6H,I). Tumnicamycin not only enhanced the expression of other ER stress-associated marker proteins for Grp78 (Figure 6C) and CHOP (Figure 6D), but also apoptosis marker for cleaved caspase-3 (Figure 6E). Knockdown of ATF6β down-regulated the expression of CHOP and cleaved caspase-3, while it up-regulated the expression of Grp78. Furthermore, tunicamyci-mediated phosphorylation of JNK was attenuated by ATF6β knockdown (Figure 6F). The treatment of PDL cells with 0.5 mg/ml of tunicamycin inhibited cell growth (Figure 6F). These results indicate that the ER stress-induced RANKL expression is regulated by ATF6β, and miR-1260b targets ATF6β to mediate RANKL inhibition in PDL cells.

We further evaluated the effects of RANKL regulation on osteoclastogenesis in PDL cells. TRAP staining revealed that incubation of human CD14⁺ monocytes with cell culture supernatant from tunicamycin-stimulated PDL cells enhanced the number and size of TRAP⁺ multinucleated mature osteoclasts. In contrast, incubation with miR-1260b mimic and ATF6β siRNA-transfected cell culture supernatant significantly inhibited osteoclast differentiation (Figures 7A–D). Bone resorption assays were performed to assess the activity of differentiated osteoclasts. Consistent with the results of TARP staining, osteoclasts differentiated from the cell culture supernatants of ER stress-induced PDL cells exhibited high bone resorption activity, which was diminished by incubation with miR-1260b mimic and ATF6β siRNA transfectants (Figures 7E,F).