Graphical abstract

   graphic file with name fx1.jpg
   [41]Open in a new tab

Highlights

     * •
       Magnetic force direction affects the morphology and proliferation
       of MLO-Y4 osteocytes
     * •
       Magnetic direction affects ECM-integrin-CSK axis-related proteins
       in MLO-Y4 cells
     * •
       Cytoskeleton stabilization influences intracellular Wnt signaling
       in MLO-Y4 cells
     * •
       Thus, magneto-mechanical stimulation affects the function of MLO-Y4
       osteocytes
     __________________________________________________________________

   Cellular physiology; Molecular biology; Developmental biology.

Introduction

   In 1936, Kolin[42]^1 first proposed the idea of employing
   electromagnetic fields in medical research. He demonstrated that, in
   general, biological systems are greatly influenced by the application
   of external magnetic fields. Permanent and man-made alternating
   magnetic fields of various strengths surround humans in daily life,
   including geomagnetic fields, stray fields from electromagnets and
   permanent magnets, as well as electromagnetic fields near electric
   devices and electric power lines.[43]^2 Magnetic resonance imaging
   (MRI) is undoubtedly a major success in the use of man-made
   high-magnetic fields in the clinical field. As a non-invasive,
   non-ionizing, and non-destructive medical diagnostic technique, MRI is
   widely used in clinical and animal research. The three distinct fields
   formed during the use of MRI include a static magnetic field (
   [MATH: <mrow><msub><mi>B</mi><mn>0</mn></msub></mrow> :MATH]
   ), time-dependent gradient fields, and the radiofrequency transmitter
   field (
   [MATH: <mrow><msub><mi>B</mi><mn>1</mn></msub></mrow> :MATH]
   ).[44]^3^,[45]^4 High-strength magnets allow for greater
   signal-to-noise ratios, resulting in higher resolution or shorter scan
   times. The increasing demand for spatial resolution and a high
   signal-to-noise ratio drives the development of stronger magnetic field
   systems. The 10.5 T whole-body MRI scanner installed at the University
   of Minnesota’s Center for Magnetic Resonance Research (CMRR) is the
   most powerful whole-body MRI machine available for human use till
   date.[46]^5 The past few decades have seen the rapid development of a
   wide range of technologies for generating high and ultra-high magnetic
   fields.[47]^6^,[48]^7^,[49]^8 The increasing magnetic field strength is
   accompanied by an exacerbation of the field inhomogeneity problem, that
   is, the magnetic field gradient.[50]^9 The magnetic field gradient
   describes the change in external magnetic field strength with respect
   to distance (dB/dz). The spatial gradient is the key parameter of the
   [MATH: <mrow><msub><mi>B</mi><mn>0</mn></msub></mrow> :MATH]
   field, which determines the attraction, a translation force exerted on
   an object. It is measured in Gauss per centimeter (G/cm) or Tesla per
   meter (T/m) and is represented by
   [MATH: <mrow><mover accent="true"><mo>∇</mo><mo linebreak="badbreak"
   stretchy="true">→</mo></mover><msub><mi>B</mi><mn>0</mn></msub></mrow>
   :MATH]
   , with its magnitude as
   [MATH: <mrow><mo stretchy="true">|</mo><mrow><mover
   accent="true"><mo>∇</mo><mo linebreak="badbreak"
   stretchy="true">→</mo></mover><msub><mi>B</mi><mn>0</mn></msub></mrow><
   mo stretchy="true">|</mo></mrow> :MATH]
   [51]^10. Magnetic fields have different or even opposite biological
   effects at the cellular and whole-organism level. A major issue is that
   the magnetic induction intensity and the magnetic signal gradients that
   drive cells are too small and are below the significance
   threshold.[52]^11^,[53]^12

   Hall et al.[54]^13 reported that working ion channels generate weak
   magnetic fields of 10^−9–10^−5 T due to ion currents using a
   nitrogen-vacancy probe. In fact, various components within cells have
   certain magnetic properties. For example, magnetic correlation and
   magnetic moment formation are key electronic properties of oxygenated
   and deoxygenated hemoglobin. These properties, which arise from the
   strongly correlated electrons, affect hemoglobin function.[55]^14 The
   response of cells and intracellular structures to magnetic forces in an
   applied magnetic field can be elastic, viscous, or viscoelastic.[56]^15
   Magnetic forces influence orientation and sample movement in gradient
   fields.[57]^16

   Over the past 20 years, large gradient high-static magnetic fields
   (LG-HMFs) have undergone significant advances for ground-based
   simulations of weightlessness[58]^17 and cellular magnetism
   studies.[59]^18 Qiang et al.[60]^19 collaborated with Japan
   Superconductor Technology Inc. to design and manufacture
   superconducting magnets (JMTA-16T50MF) that can generate LG-HMFs. The
   specifications of the superconducting magnet are similar to those
   described by Hirose.[61]^20 If the diamagnetic material droplet is
   placed in a vertically oriented magnetic field gradient, the apparent
   gravity of the droplet can be attenuated or enhanced to different
   extents by the magnetic force. In a mouse safety study investigating
   the effect of 3.5–23.0 T static magnetic fields, the increase in spleen
   weight was more evident in groups with field intensity of 13.5 T and a
   gradient of 117.2 T/m, rather than in those with the highest field
   intensity (23.0 T).[62]^21 This observation further suggested that the
   magnetic field gradient is a key factor contributing to magnetic
   field-induced biological effects. Water and biological materials, such
   as DNA, proteins, and lipids are examples of diamagnetic
   materials.[63]^7^,[64]^22 The magnetic susceptibility of diamagnetic
   materials is represented by a small negative value. For example, for
   water at 37°C, the diamagnetic susceptibility χ is ≈ −9.05 × 10^−6 (SI
   unit).[65]^23 Most tissue constituents have a volume magnetic
   susceptibility that is similar to that of water.[66]^24 When placed in
   a gradient magnetic field, the direction of the magnetic force on the
   diamagnetic substance points from the stronger field to the weaker
   region.[67]^12^,[68]^25 The repulsive diamagnetic force on diamagnetic
   materials can counteract and superimpose with gravity in the bore of a
   superconducting solenoid and powerful “bitter”
   electromagnet.[69]^19^,[70]^25

   The magnetic force of a unit volume of material is calculated according
   to the equation:
   [MATH: <mrow><mi>F</mi><mo linebreak="badbreak">=</mo><mo
   linebreak="goodbreak">−</mo><mfrac><msub><mi>χ</mi><mi>p</mi></msub><ms
   ub><mi>μ</mi><mn>0</mn></msub></mfrac><mi>B</mi><mfrac><mrow><mi>d</mi>
   <mi>B</mi></mrow><mrow><mi>d</mi><mi>z</mi></mrow></mfrac></mrow>
   :MATH]

   where B is the magnetic flux density, μ[0] is the magnetic permeability
   of vacuum (=4π × 10^−7 H/m), χ represents volume susceptibility, and z
   is the vertical site coordinate.[71]^26^,[72]^27 The magnetic force is
   special as a mechanical stimulus as it is a function of the density of
   materials.[73]^28 Since the advent of the LG-HMF, it has been used for
   life science research, which has led to novel
   findings.[74]^27^,[75]^29^,[76]^30

   Osteocytes are the most abundant resident cells in the bone, accounting
   for 90–95% of the total cellular component in the adult
   skeleton.[77]^31 As such, osteocytes are the master regulators of bone
   homeostasis.[78]^31 Furthermore, osteocytes are endocrine cells that
   regulate phosphate metabolism in multiple tissues. Most importantly,
   osteocytes act as primary mechanosensory cells.[79]^32^,[80]^33 They
   are more responsive to mechanical stimulation than osteoblasts and
   trigger a greater calcium influx while releasing more prostaglandin
   (PGE2) and nitric oxide (NO), in addition to exhibiting more rapid
   β-catenin-mediated transcription.[81]^31 Osteocyte mechano-transduction
   is a complex but exquisite regulatory process occurring between
   different mechanosensory cells, between cells and their environment, as
   well as between neighboring cells.[82]^34 The cellular
   mechano-transduction process requires designated mechanotransducers to
   convert mechanical force into various fundamental biological
   signals.[83]^35 In general, the force applied to the cell-matrix
   adhesion first causes physical deformation of the ECM, with mechanical
   signals then transmitted from the ECM-integrin bond, to adapter
   proteins, and finally, to the cytoskeleton.[84]^36 Catch bonds have
   been mostly described for ECM-integrin interactions but were recently
   identified in adapter protein interactions, such as that between
   vinculin and actin.[85]^37 Huang et al. discovered that vinculin forms
   a force-dependent catch bond with F-actin through its tail domain, with
   lifetimes that depended on the direction of the applied force.[86]^37
   The cytoskeleton typically moves mechanical signals from the site of
   assembly at the leading edge toward the cell centers.[87]^38 Osteocytes
   respond to mechanical stimuli by initiating intracellular signaling and
   producing factors that activate both osteoclasts and osteoblasts to
   remodel bone.[88]^39 The most researched and best described pathways
   induced by mechanical stimulations are those involving NO, ATP, PGEs,
   Ca^2+, and Wnt, which are the first factors to be released by
   mechanically stressed osteocytes. Blocking one of these signaling
   pathways inhibits loading-induced bone anabolism.[89]^40^,[90]^41
   Several experimental settings have been developed to study cells under
   reduced mechanical cues, including the random positioning
   machine,[91]^42 slow-rotating wall vessels, the 3D clinostat, or
   antimagnetism levitation.[92]^43 Commercial and custom-made devices
   used in studies can load cells with a wide range of mechanical stimuli
   such as tension, compression, centrifugation, shear, and
   vibration.[93]^44 Different concepts and equipment designs produce
   mechanical stimuli of different amplitude and frequency that may lead
   to different strain axes and thus exhibit distinct effects on cellular
   reflection. Comparisons between studies should thus be made with
   caution. Therefore, it is of particular importance to study the effects
   of mechanical stimuli of the same type, size, and duration acting on
   cells.

   When the mechanical state to which the bone is subjected changes, the
   internal structure of the bone changes and adapts accordingly while
   creating an optimal load-bearing structure. Explaining the cellular and
   molecular mechanisms through which this mechanical stimulus leads to
   bone reconstruction is the main current research task in the field of
   bone cell mechanics. In this study, magneto-mechanical stimuli of the
   same size and properties but in opposite directions were used to study
   the changes in cell structure and function of osteoblasts in response
   to mechanical loading and unloading.

Results

Cells survive in 12 T LG-HMFs

   The magnet we used ([94]Figure 1A) provides a high-homogeneous static
   magnetic field (SMF) at its center. The magnetic field direction is
   vertically upward. MLO-Y4 cell colony numbers decreased under 12 T
   upward treatment, while increasing under 12 T downward treatment
   ([95]Figure 2A). To observe the effects of LG-HMFs with opposite
   orientation on the MLO-Y4 relative cell number and viability, we used
   the blood count method and CCK-8 assays, respectively. Relative cell
   number and viability decreased under 12 T upward treatment, while
   increasing under 12 T downward treatment at 24, 48, and 72 h
   ([96]Figures 2B and 2C). Flow cytometry revealed that 12 T upward
   treatment for 24 h significantly promoted cell apoptosis.
   Alternatively, 12 T downward treatment inhibited osteocyte apoptosis
   ([97]Figures 2D and 2E). Western blotting analysis was performed to
   determine the changes in the expression of apoptosis-related proteins
   Bax and cleaved caspase 3. The 12 T upward treatment significantly
   promoted the expression of Bax and cleaved caspase3 in MLO-Y4 cells
   ([98]Figure 2F), whereas the 12T downward treatment showed a decreasing
   trend, but this was not statistically significant ([99]Figure 2F). The
   12 T upward magnetic field applied for 24 h significantly altered
   osteocyte morphology, and the synapses became elongated. Conversely,
   the 12 T downward treatment had no significant effect on osteocyte
   morphology ([100]Figure 2G). By counting the cell area and number of
   dendrites, we found that they decreased after 12T upward treatment,
   whereas no effect was observed after 12T downward treatment
   ([101]Figure 2H).

Figure 1.

   [102]Figure 1
   [103]Open in a new tab

   The LG-HMFs exposure system

   (A) Superconducting magnet (JMTA-16T50MF) and the object stage of cell
   culture.

   (B) Parameters of the superconducting magnet.

   (C) Magnetic field distribution in the magnetic cavity.

   (D) B(dB/dz) represents the variation of the magnetic force with
   position.

   (E) Plot of radial apparent gravity fit at the 12 T upward position.

   (F) Plot of radial apparent gravity fit at the 12 T downward position.

   (G) Schematic diagram of the magnetic forces and gravity on the cell in
   the magnet chamber.

Figure 2.

   [104]Figure 2
   [105]Open in a new tab

   The influence of a different direction of magnetic force on growth of
   MLO-Y4 cells

   (A) The clonogenic ability of MLO-Y4 cells.

   (B) The blood cell count method calculates the number of cells; n = 3.

   (C) CCK8 assay for cell viability; n = 3.

   (D and E) Flow cytometric analysis of osteocytes to examine cell
   apoptosis.

   (F) Protein expression levels of Bax and cleaved caspase 3.

   (G) Observation of cellular morphology by crystal violet staining.
   Scale bar: 100 μm.

   (H) Analysis of the cellular area and the number of dendrites in
   osteocytes by using Image-Pro Plus software; n = 100.

   Data shown as mean ± SD. ∗p < 0.05, ∗∗ <0.01, ∗∗∗ <0.001.

DEG analysis

   Control, 12 T upward, and 12 T downward-treated cells were subjected to
   RNA sequencing ([106]Figure 9). Compared to the control, 12 T
   upward-treated MLO-Y4 cells exhibited 614 DEGs, of which 214 were
   upregulated and 400 were downregulated ([107]Figure 3A). Alternatively,
   12 T downward-treated MLO-Y4 cells exhibited 94 DEGs, of which 40 were
   upregulated and 54 were downregulated ([108]Figure 3B). Furthermore,
   there were 565 DEGs in the 12 T upward group compared to the 12 T
   downward group. Upregulated and downregulated DEGs are shown in volcano
   plots ([109]Figure 3C). The Venn diagram shows 35 common DEGs in the
   two treatment groups relative to the control ([110]Figure 3D). These
   included Wisp2, Col5a3, Oas2, Plin4, Parp11, Sp100, Uba7, Trim30a,
   Stat1, Il4ra, Hist1h1c, Oas3, Gbp7, Cmpk2, Slfn8, Trim30d, Ifi47, Rbm3,
   Usp18, Tor3a, Uap1l1, Gm6136, Gm20559, Ccl2, Herc6, Cd80, Cirbp, Cbs,
   Ifi207, Rpl9-ps6, Gm9115, Ctso, Tnfaip3, Tlr3, and Gm7666. After
   identifying the DEGs, we performed a hierarchical cluster analysis with
   the upregulated and downregulated genes illustrated in a heatmap
   ([111]Figure 3E). Gene counts were log[112]^10 transformed and
   normalized as Z score. In [113]Figure 3E, two clusters were clearly
   displayed; all control groups were in one cluster and the LG-HMF groups
   in another cluster, which indicates high-intragroup consistency and
   high-intergroup variability.

Figure 9.

   Figure 9
   [114]Open in a new tab

   Library preparation for transcriptome sequencing.

Figure 3.

   [115]Figure 3
   [116]Open in a new tab

   DEG expression profiles of osteocytes after different direction
   magnetic force treatment

   (A) Volcano plot of DEGs between 12 T upward and control groups.

   (B) Volcano plot of DEGs between 12 T downward and control groups.

   (C) Volcano plot of DEGs between 12 T upward and 12 T downward groups.

   (D) Venn diagram of DEGs.

   (E) Heatmap of cluster analysis of DEGs. Red indicates DEGs with high
   expression, and blue indicates DEGs with low expression.

Gene ontology (GO) enrichment analysis of DEGs

   We performed GO enrichment analysis on DEGs, with p < 0.05 as the
   significance threshold. Genes were enriched in terms in the biological
   process (BP), cellular component (CC), and molecular function (MF)
   categories. GO enrichment classification maps of DEGs from the 12 T
   upward ([117]Figure 4A) and 12 T downward ([118]Figure 4B) treatment
   groups were generated. DEGs in the 12 T upward treatment group were
   enriched for defense response, regulation of cytokine production, and
   mitotic E in the BP category. For CC, the highest number of genes was
   enriched for ECM. Enriched MF terms included platelet-derived growth
   factor binding, growth factor binding, and receptor regulator activity
   ([119]Figure 4A). For the 12 T downward treatment group, significant BP
   terms included toll-like receptor signaling pathway, defense response
   against other organisms, and positive regulation of nitric oxide
   synthase. Among CC terms, the ribosomal subunit had the highest number
   of enriched genes. Lastly, CCR chemokine receptor binding, sulfur
   compound binding, and glycosaminoglycan binding were the significant MF
   terms ([120]Figure 4B).

Figure 4.

   [121]Figure 4
   [122]Open in a new tab

   GO enrichment analysis of DEGs

   (A) GO enrichment classification map of DEGs of 12 T upward with
   control.

   (B) GO enrichment classification map of DEGs of 12 T downward with
   control.

Effects of LG-HMFs on the ECM-integrin-CSK

   We subjected DEGs between the 12 T upward and 12 T downward treatment
   groups to Reactome pathway analysis, with p < 0.05 as a significance
   threshold. In [123]Figure 7A, significant pathways were plotted using a
   bubble diagram and result indicated that DEGs were mainly associated
   with ECM proteoglycans, ECM organization, and integrin cell surface
   interactions. Further, 12 T upward treatment significantly inhibited
   the expression of ECM proteins LAMB1 and integrin β1. In contrast, 12 T
   downward treatment significantly promoted the expression of ECM
   proteins LAMB1 and integrin β1 ([124]Figures 5A and 5B). Under control
   and 12 T downward conditions, MLO-Y4 cells showed star-like shape, and
   their actin fibers were evenly distributed across the cell bodies. The
   cells are abundant in actin fibers, which branch or rebranch in
   different planes of the cell body, forming intercellular connections
   with surrounding bone cells. However, under the 12 T upward conditions,
   MLO-Y4 cells exhibited a relatively long cell body shape, with a
   discontinuous and loose cytoskeleton ([125]Figure 6A). We also examined
   the expression of microfilament protein β-actin and microtubulin
   protein β-tubulin, observing no significant differences except that
   12 T downward treatment could promote β-tubulin expression. However,
   12 T upward treatment significantly downregulated the expression of
   skeleton-binding protein vinculin, while 12 T downward treatment had
   the opposite effect ([126]Figure 6B).

Figure 7.

   [127]Figure 7
   [128]Open in a new tab

   Effect of magnetic force on the Wnt signaling pathway in MLO-Y4 cells

   (A) The Reactome enrichment analysis of 12 T upward and 12 T downward;
   significant pathways were plotted in the bubble diagram.

   (B) Protein interactions network map centered on wisp2.

   (C) Subcellular localizations of wisp2.

   (D) Protein expression levels of the Wnt signaling pathway (Wnt1, LRP6,
   β-catenin, and SOST).

   Data shown as mean ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001,
   ∗∗∗∗p < 0.0001; vs. control.

Figure 5.

   [129]Figure 5
   [130]Open in a new tab

   The influence of different direction of magnetic force on ECM and
   integrin

   (A) Immunoluminescent staining of ECM proteins (LAMB1 and Collagen Ι)
   and integrin ligand proteins (integrin β1). Scale bar: 75 μm.

   (B) Protein expression levels of LAMB1, Collagen Ι, and integrin β1.
   Data shown as mean ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 vs.
   control.

Figure 6.

   [131]Figure 6
   [132]Open in a new tab

   The influence of different direction of magnetic force on the
   cytoskeleton in osteocytes

   (A) Immunoluminescent staining of F-actin and β-tubulin. Scale bar:
   30 μm.

   (B) Protein expression levels of β-actin, β-tubulin, and Vinculin.

   Data shown as mean ± SD. ∗p < 0.05, ∗∗p < 0.01.

LG-HMFs regulate the Wnt pathway in MLO-Y4

   DEGs between the 12 T upward and 12 T downward treatment groups were
   subjected to pathway enrichment analysis, with significant terms
   including the Wnt/β-catenin pathway ([133]Figure 7A). Wisp2
   (Wnt1-inducible signaling pathway protein 2) was the most significant
   DEG between the 12 T upward and 12 T downward groups. We analyzed Wisp2
   using the STRING protein interaction database to obtain the protein
   networks that interact with Wisp2 ([134]Figure 7B). The GeneCards
   database indicated that Wisp2 localized to the nucleus, ECM, and CSK
   ([135]Figure 7C). The STRING database highlighted the Wnt signaling
   pathway as the most relevant to Wisp2. Wnt signaling is
   mechanosensitive and acts downstream of mechanical stimuli during
   skeletogenesis.[136]^45 Here, 12 T upward treatment significantly
   inhibited the expression of Wnt pathway-related proteins, including
   Wnt1, β-catenin, and LRP6, in MLO-Y4 cells ([137]Figure 7D).
   Sclerosteosis (SOST) has an inhibitory effect on the Wnt pathway.
   Furthermore, 12 T upward treatment for 48 h significantly promoted the
   protein expression of SOST, while inhibiting that of Wnt pathway
   members ([138]Figure 7D). To verify whether the LG-HMF affects Wnt
   pathway factor expression in light of its effect on the skeleton, we
   pretreated MLO-Y4 cells with a low concentration of paclitaxel (0.5 μM)
   for 24 h before placing them under 12 T upward magnetic field
   conditions. We discovered that cytoskeleton stabilization in the LG-HMF
   stimulated Wnt signaling ([139]Figure 7D). At the same time, colchicine
   (0.5 μM) pretreatment at a low concentration significantly inhibited
   the effect of 12 T downward treatment on Wnt signaling in MLO-Y4 cells
   ([140]Figure 7D).

Conditioned medium (CM) from LG-HMF-treated osteocytes promotes osteoclasts
differentiation

   Osteocytes are the main regulators of bone homeostasis, exerting their
   effects through paracrine signaling to coordinate the differentiation
   and function of osteoclasts and osteoblasts.[141]^46 To determine
   whether LG-HMFs influence osteocyte function, we examined the effect of
   CM on bone differentiation in pre-osteoblast RAW264.7 cells. Osteoclast
   differentiation was evaluated based on F-actin rings and TRAP staining.
   Compared with those in control CM-treated group, the area of F-actin
   rings ([142]Figure 8A) and number of TRAP-positive multinucleated cells
   ([143]Figures 8B and 8C) were increased after treatment with CM from
   12 T upward-treated osteocytes, while they decreased after treatment
   with the CM from 12 T downward-treated osteocytes. OPG and NO are
   factors released by osteoblasts to inhibit osteoclast differentiation
   and activity.[144]^47 The 12 T upward treatment inhibited OPG
   (osteoclastogenesis inhibitory factor) and NO synthesis, whereas 12 T
   downward treatment promoted it ([145]Figures 8D and 8E).

Figure 8.

   [146]Figure 8
   [147]Open in a new tab

   LG-HMFs affect the function of MLO-Y4 osteocytes

   (A) Osteoclast differentiation was evaluated by the F-actin ring
   staining.

   (B and C) Osteoclast differentiation was evaluated by the TRAP
   staining.

   (D) Detection of NO secretion in CM.

   (E) Protein expression levels of OPG.

   Data shown as mean ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001,
   ∗∗∗∗p < 0.0001; vs. control. ^#p < 0.05, ^##p < 0.01, ^###p < 0.001,
   ^####p < 0.0001; 12 T upward vs. 12 T downward. Scale bar: 200 μm.

Discussion

   Magnetic forces generated by magnetic fields act on cellular
   components, including the cell membrane, cytoplasm, cytoskeleton,
   nucleus, and intracellular motors, thus, triggering cellular signaling
   pathways. We have previously shown that simulated microgravity
   generated by LG-HMF influences the fate of MC3T3-E1
   osteoblasts.[148]^27^,[149]^48 In the current study, we used
   LG-HMF-generated magnetic forces with the same magnitude and opposite
   direction to investigate the effects of magnetic force loading in
   different directions on cell fate and the underlying mechanisms. We
   observed that the viability and relative number of osteocytes cultured
   under 12 T downward treatment for 24 h were enhanced, while the same
   parameters were decreased following 12 T upward treatment. There was no
   significant change in cell morphology, cell area, and number of cell
   dendrites in the 12 T downward treatment group compared with those in
   the geomagnetic group, and apoptosis tended to decrease, which may
   account for the increase viability of MLO-Y4 cells. Alternatively, the
   12 T upward treatment significantly promoted cell apoptosis, with
   changes in osteocyte morphology, dendrites elongation and reduced
   dendrites number, which directly led to a decrease in cell number and
   viability.

   Compared to controls, 12 T upward-treated MLO-Y4 cells exhibited 614
   DEGs, while 94 DEGs were obtained for 12 T downward-treated cells. This
   suggests that the upward magnetic force may cause more biological
   effects. We further performed GO functional enrichment analysis of the
   DEGs. GO enrichment results showed that the biological processes that
   changed after both 12 T upward and 12 T downward treatment were
   associated with the regulation of defense responses, collagen
   polymerization, extracellular matrix protein binding, integrin binding,
   and ion binding. Interestingly, the biological processes that change in
   MLO-Y4 cells after 12 T upward treatment include regulation of
   cell-cell adhesion, while the biological processes that change after
   12 T downward treatment include nitric oxide synthase biosynthesis
   processes. At the same time, DEGs between the 12 T upward and 12 T
   downward groups were subjected to analysis based on the Reactome
   database. These were associated with ECM proteoglycans, ECM
   organization, and integrin cell surface interactions. To further
   validate the transcriptome-sequencing results, we examined the
   expression of extracellular matrix proteins and integrin-related
   proteins using immunofluorescence and Western blot assays. The 12 T
   upward treatment inhibited the expression of LAMB1, collagen I, and
   integrin β1 in MLO-Y4 cells. Mechanical homeostasis involves ECM
   constituents such as collagens and laminin, which support and transmit
   loads. Integrins bridge extracellular and intracellular structures,
   while their associated linker proteins (such as talin and vinculin)
   connect receptors to the cytoskeleton.[150]^49 The microfilament
   skeleton of osteocytes was clustered at the edges of cells treated with
   12 T upward, indicative of altered microfilament distribution.
   Alternatively, 12 T downward promoted the skeletal arrangement of
   osteocytes. Our Western blot results showed that the expression of
   β-tubulin and β-actin was less affected by magnetism. However,
   vinculin, which is involved in cytochemical signaling in combination
   with F-actin, showed opposite expression trends in the 12 T upward and
   12 T downward treatment groups.

   Wisp2, the most significant DEG between 12 T upward- and 12 T
   downward-treated cells, localized to the nucleus, ECM, and
   cytoskeleton. Wisp2 was most strongly correlated with the Wnt pathway
   based on the STRING protein interactions database. Wisp2 is both a
   secretory and cytosolic protein that activates the classical Wnt
   pathway and increases β-catenin levels.[151]^50 Herein, we discovered
   that the 12 T upward treatment inhibited the Wnt1/LRP6/β-catenin
   pathway. In combination with our experimental results, cytoskeleton
   stabilizer pretreatment of MLO-Y4 cells attenuated the inhibitory
   effect of 12 T upward on the Wnt signaling pathway. Cytoskeleton
   inhibitor pretreatment of MLO-Y4 cells inhibited the effect of the 12 T
   downward treatment of the Wnt signaling pathway. Thus, cytoskeletal
   stability is important for the transduction of the Wnt signaling
   pathway in MLO-Y4 cells. Furthermore, the 12 T upward treatment
   decreased NO production and increased SOST expression, further
   synergistically inhibiting the Wnt pathway in MLO-Y4 cells. Wnt pathway
   inhibition further reduced the expression of downstream bone function
   protein OPG. NO and OPG are typical cytokines that inhibit osteoclast
   differentiation.[152]^47 Alternatively, osteoclast formation was
   significantly enhanced following the 12 T upward treatment due to the
   significantly lower NO and OPG secreted by cells. Conversely, the 12 T
   downward treatment increased NO production and decreased SOST
   expression. Higher secretion of NO and OPG suppressed osteoclast
   differentiation in the 12 T downward treatment group relative to the
   control group.

   Overall, the parameters associated with the mechanical stimulation to
   which cells were subjected in this study are more standard and
   accurate. In MLO-Y4 osteocytes, the ECM-integrin-CSK axis and
   Wnt1/LRP6/β-catenin/OPG pathway were altered by magnetic force.
   Osteocyte fate and function were altered as a result. We further
   clarified the effect of osteocytes on bone reconstruction in response
   to different mechanical stimuli. We provide a basis for further
   research on this issue in terms of cellular-molecular mechanisms that
   are of great importance to clinical practice.

Limitations of the study

   This study had some limitations. The first is the short exposure time
   of 24 h in LG-HMFs for the cells used for transcriptome sequencing to
   ensure that the cells are in optimal condition. Second, only the
   differentiation of osteoclasts based on osteocyte secretory factors and
   not that based on osteocyte secretory factors was tested for
   osteoblastes function. Finally, only the effect of magnetic force on
   the expression of ECM-integrin-CSK was examined. The information
   transfer between ECM-integrin-CSK is based on previous studies. Further
   experiments are needed to verify the transmission of magneto-mechanical
   signals in cells.

STAR★Methods

Key resources table

   REAGENT or RESOURCE SOURCE IDENTIFIER
   Antibodies
     __________________________________________________________________

   GAPDH Monoclonal antibody Proteintech RRID:[153]AB_210743;
   Cat#60004-1-Ig
   Anti-Bax antibody Abcam RRID:[154]AB_2938987;
   Cat#ab182733
   Cleaved Caspase-3 (Asp175) Antibody Cell Signaling Technologies
   RRID:[155]AB_2341188;
   Cat# 9661
   Laminin beta 1 Polyclonal antibody Proteintech RRID:[156]AB_2879288;
   Cat#23498-1-AP
   Collagen Type I Polyclonal antibody Proteintech RRID:[157]AB_2082037;
   Cat#14695-1-AP
   Integrin Beta 1 Polyclonal antibody Proteintech RRID:[158]AB_2130085;
   Cat#12594-1-AP
   β-Actin (13E5) Rabbit mAb Cell Signaling Technologies
   RRID:[159]AB_2223172;
   Cat#4970
   Anti β-Tubulin Mouse Monoclonal Antibody Proteintech
   RRID:[160]AB_2881629;
   Cat#66240-1-Ig
   WNT1 Polyclonal antibody Proteintech RRID:[161]AB_2881013;
   Cat#27935-1-AP
   LRP6 (C47E12) Rabbit mAb Cell Signaling Technologies
   RRID:[162]AB_1950408;
   Cat#3395
   Beta Catenin Polyclonal antibody Proteintech RRID:[163]AB_2086128;
   Cat#51067-2-AP
   Anti-Sclerostin antibody Abcam RRID:[164]AB_956321;
   Cat#ab63097
   Anti-Osteoprotegerin antibody Abcam RRID:[165]AB_2934183;
   Cat#ab183910
   Cy3-labeled Goat Anti-Rabbit IgG (H + L) Beyotime Biotechnology
   RRID:[166]AB_2893015;
   Cat#A0516
   Goat Anti-Mouse IgG H&L (Alexa Fluor® 488) Abcam RRID:[167]AB_2576208;
   Cat#ab150113
   HRP-labeled Goat Anti-Rabbit IgG(H + L) Beyotime Biotechnology
   RRID:[168]AB_2892644;
   Cat#A0208
   HRP-labeled Goat Anti-Mouse IgG(H + L) Beyotime Biotechnology
   RRID:[169]AB_2860575;
   Cat#A0216
     __________________________________________________________________

   Chemicals, peptides, and recombinant proteins
     __________________________________________________________________

   Murine sRANK-Ligand Peprotech Cat# 315-11-10UG
   Paclitaxel TopScience Cat# T0968
   Colchicine TopScience Cat# T0320
     __________________________________________________________________

   Deposited data
     __________________________________________________________________

   RNA-seq transcriptome data This paper NCBI-based platforms: NCBI
   Sequence Read Archive (SRA): BioProject ID PRJNA932050
     __________________________________________________________________

   Experimental models: Cell lines
     __________________________________________________________________

   MLO-Y4 University of Texas Health Science Center RRID:CVCL_M098
   RAW 264.7 ATCC RRID:CVCL_0493;
   Cat#TIB-71
     __________________________________________________________________

   Software and algorithms
     __________________________________________________________________

   GraphPad Prism 8.2.1 [170]https://www.graphpad.com/
   Python 3.8 [171]https://www.python.org/
   R-project [172]https://www.r-project.org/
   ImageJ [173]https://ImageJ.net/software/fiji/downloads
   Image-Pro Plus6.0
   [174]https://www.epixinc.com/vision_archive/imagepro.htm
     __________________________________________________________________

   Other
     __________________________________________________________________

   Collagen Type Ι BD Corning Cat#354236
   Crystal Violet Amersco Cat#0528-25G
   Triton X-100 Sinopharm Chemical ReagentCo., Ltd Cat#30188928
   Cell Counting Kit-8 Beyotime Biotechnology Cat#C0038
   Annexin V-FITC Apoptosis Detection Kit Beyotime Biotechnology
   Cat#C1062M
   ActinRed™ 555 ReadyProbes™ Thermo Fisher Scientific Cat#[175]R37112
   DAPI dihydrochloride Beyotime Biotechnology Cat#C1002
   NO assay kit Beyotime Biotechnology Cat#S0021S
   Leukocyte acid phosphatase (TRAP) Kit Sigma-Aldrich Cat# 387A-1KT
   [176]Open in a new tab

Resource availability

Lead contact

   Further information and requests for resources should be directed to
   and will be fulfilled by the Lead Contact, Peng Shang, Northwestern
   Polytechnical University, China ([177]shangpeng@nwpu.edu.cn).

Materials availability

   This study did not generate new unique reagents. Further information
   and request for resources and reagents should be directed to and will
   be fulfilled by the Lead Contact, Peng Shang, Northwestern
   Polytechnical University, China ([178]shangpeng@nwpu.edu.cn).

Experiment methods and subject details

Cell lines

   The mouse osteocyte-like cell line (MLO-Y4) used in this study was
   gifted by Prof. Jean X. Jiang (University of Texas Health Science
   Center, San Antonio, TX, USA). MLO-Y4 cells were cultured on
   collagen-coated (Collagen Type Ι, BD Corning, Cat#354236) petri dishes.
   The cells were cultured in α-MEM medium (Gibco) supplemented with 5%
   fetal bovine serum (FBS, Gibco), 5% calf serum (Every Green), 2 mM
   L-glutamine, and 1% penicillin/streptomycin (Beyotime Biotechnology,
   Cat#C0222) in a humidified 5% CO[2] and 37°C controlled by an incubator
   system (Thermo Fisher Scientific).

   Pre-osteoclast RAW 264.7 cells were purchased from the Cell Bank of
   Chinese Academy of Science (CAS; Shanghai, China), ATCC (TIB-71). RAW
   264.7 cells were cultured in α-MEM medium (Gibco) supplemented with 10%
   fetal bovine serum (FBS, Gibco), 2 mM L-glutamine, and 1%
   penicillin/streptomycin (Beyotime Biotechnology, Cat#C0222). The cells
   were cultured in a humidified 5% CO[2] and 37°C controlled incubator
   system (Thermo Fisher Scientific).

Method details

Large gradient high magnetic fields exposure system

   In this study, LG-HMFs were produced by a superconducting magnet
   (JMTA-16T50MF) ([179]Figure 1A). The LG-HMF consisted of the NbTi and
   Nb[3]Sn superconductors. To generate the 16 T magnetic field in a
   region with a diameter of 50 mm, the superconducting magnet was
   combined with reverse and main coils. The main coils consisted of two
   NbTi coils and one Nb[3]Sn coil. The main coil was designed to generate
   a high field gradient. The reverse coil consisted of a Nb[3]Sn coil
   that generates a reverse field for future high magnetic field gradients
   of approximately 150 T/m.[180]^27^,[181]^51 In addition to the space
   occupied by the temperature control unit in the magnet hole, the
   diameter of the space used for the experiment was approximately 44 mm,
   while the total length was 400 mm. We independently developed an
   experimental platform to study the cell biology effect of
   superconducting magnets, including object stages ([182]Figure 1B),
   temperature control systems, and gas control systems. The constant
   temperature circulation system (MPG-13A, Bluepard instruments) was
   composed of a temperature controller, temperature sensor, heater,
   circulation pump, and low water lever alarm device. The temperature of
   the magnetic chamber was controlled through heat conduction of
   temperature-controlled water via the magnet chamber at 37°C–37.5°C. We
   designed our original gas mixing device to mix CO[2] and air
   proportionally by adjusting the gas flow meters and then humidifying
   them through the humidity controller and passing them into the magnet
   chamber to ensure the humidity as well as the 5% CO[2] concentration
   during cell culture. The direction of the magnetic field inside the
   magnet chamber was vertically upward ([183]Figure 1C). In this study,
   cells were placed in the positions of 12.56 T, −109 T/m and 12.50 T,
   105 T/m.

   The object stage ([184]Figure 1B) contained two 12 T cell culture
   platforms that were corresponding to different magnetic field gradients
   of −109, and 109 T/m in a 5 cm diameter room temperature bore,
   respectively. The change in magnetic field strength and gradient within
   the LG-HMF exhibited continuity ([185]Figures 1D and 1E). Schematic
   diagram of the magnetic and gravity on the cell in the magnet chamber
   are illustrated in [186]Figure 1F.

Cell proliferation and viability

   For cell colony formation assays, MLO-Y4 cells were seeded in 35 mm
   petri dishes at 1000 cells/dish. After 12 h of cell adhesion on the
   plates, cells were cultured separately in a geomagnetic incubator, 12 T
   upward and 12 T downward LG-HMFs. The medium was refreshed every two
   days and colonies were observed after 10 days. Next, cells were washed
   with 1×PBS and fixed with 4% paraformaldehyde for 20 min. Subsequently,
   1 mL of 0.1% crystal violet staining solution was added to each dish
   for 30 min. The excess crystal violet staining solution was washed with
   distilled water and air-dried. Finally, we observed the results using
   the body vision microscope (SZ61, Olympus) and acquired images using a
   DSC-H5 Sony camera.

   For cell counting, MLO-Y4 cells were seeded into 35 mm petri dishes at
   1 × 10^5 cells/dish. After adherence to the plates, cells were treated
   with 12 T upward and 12 T downward LG-HMFs for 24, 48, and 72 h. After
   trypsin digestion, we counted the total number of cells using a
   hemocytometer.

   MLO-Y4 cell viability was measured using a Cell Counting Kit-8
   (Beyotime Biotechnology, Cat#C0038). MLO-Y4 cells were inoculated in
   removable 96-well plates at a density of 5 × 10^3 cells/dish. After
   adherence to the plates, cells were treated with 12 T upward and 12 T
   downward LG-HMFs for 24, 48, and 72 h. Next, 10% CCK-8 was added to the
   culture medium and incubated for 2 h at 37°C. The absorbance values
   were measured at 490 nm using an enzyme marker (Synergy HT, BioTek).

Cell apoptosis assay

   MLO-Y4 cell were seeded into 35-mm petri dishes at 1.5 × 10^5
   cells/dish. After the MLO-Y4 cells were treated with 12 T upward and
   12 T downward LG-HMFs for 24 h, cells were stained using Annexin V-FITC
   apoptosis Detection Kit (Beyotime Biotechnology, Cat#C1062M). The cell
   culture solution was aspirated into a 2-mL centrifuge tube and the
   apoptotic cells were washed once with 1 × PBS; then, an appropriate
   amount of trypsin solution was added to digest the cells. The cells
   were incubated at room temperature until they could be blown off by a
   gentle blow, the collected cell culture medium was added, cells were
   gently blown off, transferred to a centrifuge tube, centrifuged at
   1000 × g for 5 min, and the supernatant was discarded. The cells were
   washed once by gently resuspending in 1 × PBS, centrifugation at
   1000 × g for 5 min, and ultimately, the supernatant was discarded.
   Next, we added 195 μL Annexin V-FITC conjugate, 5 μL Annexin V-FITC, as
   well as 10 μL propidium iodide staining solution and mix gently.
   Subsequently, we incubated the solution for 10–20 min at room
   temperature (20°C–25°C), protected from light. The detection was
   performed using a flow cytometer (FACSCalibur, BD Bioscience). Annexin
   V^+/PI^− were considered early apoptotic cells. Annexin V^+/PI^+ were
   considered late apoptotic cells.

Cell morphology observation

   MLO-Y4 cells were seeded into 35 mm petri dishes at 1 × 10^5
   cells/dish. After adhering to the plates, cells were immediately placed
   in the LG-HMFs for 24 h. After treatment, cells were washed with
   1 × PBS and fixed with 4% paraformaldehyde for 20 min. Next, the fixed
   cells were washed with 1 × PBS, stained with 0.1% crystal violet
   solution for 30 min, and washed cleanly using ultrapure water. Finally,
   light microscopy (MD IL HC, Leica) was used to observe and image the
   cells. The cell area and number of dendrites were determined using the
   Image-Pro Plus 6.0 software.

Library construction and sequencing

   First, total RNA of MLO-Y4 cells was obtained using the MagZol reagent
   (Magen, China). The mRNA was enriched using poly-T oligo-attached
   magnetic beads. Subsequently, the obtained mRNA was randomly
   interrupted in fragmentation buffer, and the first strand of cDNA was
   synthesized using the fragmented mRNA as the template. RNaseH was used
   to degrade the RNA strand, and the second cDNA strand was synthesized
   from dNTPs as raw material under the DNA polymerase I system. The
   purified double-stranded cDNA underwent terminal repair, had an A tail
   added, and was attached to the sequencing linker. Next, we screened
   approximately 370–420 bp of cDNA with AMPureXP beads, amplified the
   sequences using PCR, and purified the PCR product again using AMPureXP
   beads to finally obtained a library. Fluorescently labeled polymerase
   was added to the sequencing flow cell, and the sequencing instrument
   passes through to capture the fluorescence signal. Data collection and
   statistical analysis were performed using a specific computer software
   (Novogene, China).[187]^52^,[188]^53^,[189]^54 ([190]Figure 9).

RNA-seq data processing and function gene annotation

   DEGs analysis between the two comparative combinations was performed
   using DESeq2 software (1.20.0). DEG analysis of two conditions were
   performed using the edgeR software package (3.22.5). p-values were
   adjusted using the Benjamini & Hochberg method. After correction,
   P-values and log[191]^2 fold change (|log[192]^2 FC|) were used as
   broad values for significant differential expression. DEGs were
   screened using adjusted P-values <0.05 and |log[193]^2 FC| > 0.5 as
   screening criteria. The acquired differentially expressed genes were
   imported into Python 3.8, data were read using numpy and pandas, and
   then the volcanos were plotted using matplotlib and seaborn. We used
   gene set enrichment analysis (GSEA) tools
   ([194]http://www.broadinstitute.org/gsea/index.jsp) to analyze the GO
   and Reactome datasets. GSEA results were plotted using the ggplot2
   plotting package in R-project 4.2.2. PPI analysis of DEGs was based on
   the STRING database ([195]https://cn.string-db.org/) of known and
   predicted protein-protein interactions. The GeneCards database
   ([196]https://www.genecards.org/) was used to obtain the localization
   of genes in cells.

Immunofluorescence staining

   MLO-Y4 cells were seeded into confocal petri dishes at a density of
   1 × 10^5 cells/dish. After adhering to the plates, cells were placed
   under the LG-HMFs for 48 h. The supernatant was removed, and 1× PBS was
   used to wash the cells once. Next, cells were fixed with 4%
   paraformaldehyde for 20 min and permeabilized with 0.2% Triton X-100
   for 5 min (Sinopharm Chemical ReagentCo., Ltd, Cat#30188928).
   Subsequently, cells were blocked in 2% BSA (2% BSA-PBS) for 30 min at
   room temperature. Next, after 3 washes with 1× PBST, cells were
   cultured overnight at 4°C with primary antibodies, including actinRed
   555 ReadyProbes Reagent (1:400 dilution in 1× PBS, Thermo Fisher
   Scientific, Cat#[197]R37112), mouse anti-β-Tubulin antibody (1:200
   dilution in 1× PBS, Proteintech, Cat#66240-1-Ig), rabbit anti-LAMB 1
   antibody (1:200 dilution in 1× PBS, Proteintech, Cat#23498-1-AP),
   rabbit anti-collagen I antibody (1:200 dilution in 1× PBS, Proteintech,
   Cat#14695-1-AP), and rabbit anti-Integrin β1 antibody (1:200 dilution
   in 1× PBS, Proteintech, Cat#12594-1-AP). The cells were washed 5 times
   with 1× PBS, then stained with Goat anti-Rabbit Alexa
   Fluor550-conjugated antibody (1:500 dilution in 1× PBS, Beyotime
   Biotechnology, Cat#A0516) and Goat anti-Mouse Alexa Fluor488-conjugated
   antibody (1:500 dilution in 1× PBS, Abcam, Cat#ab150113) for 60 min at
   room temperature in the dark. Nuclei were stained with DAPI (1:1000
   dilution in 1× PBS, Beyotime Biotechnology, Cat#C1002), followed by 5
   washing steps. The cells were imaged with using a laser scanning
   confocal microscope (SP8, Leica).

Protein expression assay

   MLO-Y4 cells were subjected to LG-HMFs for 48 h and washed with 1× PBS.
   The RIPA Lysis Buffer (Beyotime Biotechnology, Cat#P0013B) containing
   2 mM Protease inhibitor cocktail (Beyotime Biotechnology, Cat#P1005)
   and 5 mM Phosphatase inhibitor cocktail (Beyotime Biotechnology,
   Cat#P1081) was used to lyse the cells. A commercial BCA Protein Assay
   Kit (Beyotime Biotechnology, Cat# P0009) was used to examine the
   protein content. Proteins were loaded into the SDS-PAGE gels, separated
   through electrophoresis, and then transferred onto a polyvinylidene
   fluoride membranes. The polyvinylidene fluoride membranes were blocked
   with 5% nonfat-dried milk and then incubated with specific primary
   antibodies, including mouse anti-GAPDH (1:5000 dilution in 1× TBST,
   Proteintech, Cat#60004-1-Ig), Anti-Bax antibody (1:2000 dilution in 1×
   TBST, abcam, Cat# ab182733), Cleaved Caspase-3 (Asp175) Antibody
   (1:1000 dilution in 1× TBST, Cell Signaling Technologies, Cat# 9661),
   rabbit anti-LAMB 1 antibody (1:1000 dilution in 1× TBST, Proteintech,
   Cat#23498-1-AP), rabbit anti-collagen I antibody (1:2000 dilution in 1×
   TBST, Proteintech, Cat#14695-1-AP), rabbit anti-Integrin β1 antibody
   (1:4000 dilution in 1× TBST, Proteintech, Cat#12594-1-AP), rabbit
   anti-β actin antibody (1:1000 dilution in 1× TBST, Cell Signaling
   Technologies, Cat#4970), mouse anti-β-Tubulin antibody (1:20000
   dilution in 1× TBST, Proteintech, Cat#66240-1-Ig), rabbit anti-Wnt 1
   antibody (1:1000 dilution in 1× TBST, Proteintech, Cat#27935-1-AP),
   rabbit anti-LRP6 antibody (1:1000 dilution in 1× TBST, Cell Signaling
   Technology, Cat#3395), rabbit anti-β catenin antibody (1:1000 dilution
   in 1× TBST, Proteintech, Cat#51067-2-AP), rabbit anti-Sclerostin
   antibody (1:1000 dilution in 1× TBST, Abcam, Cat#ab63097), and rabbit
   anti-Osteoprotegerin antibody (1:1000 dilution in 1× TBST, Abcam,
   Cat#ab183910) overnight at 4°C. After washing the membranes five times
   with 1× TBST, they were incubated with species-specific secondary
   antibody conjugated to horseradish peroxidase (1:2000 dilution in 1×
   TBST, Beyotime Biotechnology, Cat#A0208 and Cat#A0216) for 2 h at room
   temperature. An ECL Plus Western Blotting Detection System (JS-M6P, P&Q
   Science Technology) was used to image the immunoreactive bands. The
   grayscale bands were determined using the ImageJ 1.8.0 software.

Detection of osteoclast differentiation

   MLO-Y4 cells were collected in the medium treated with control, 12 T
   upward, and 12 T downward LG-HMFs for 48 h. The medium was filtered
   through a 0.22-μm membrane to remove bacteria and mixed with α-MEM
   containing 10% fetal bovine serum at a ratio of 1:2 to obtain
   conditioned medium (CM). To induce osteoclast differentiation, Raw264.7
   cells were cultured in α-MEM containing 10% fetal bovine serum and
   50 ng/mL soluble RANKL as previously described.[198]^55 After 24 h, the
   medium was replaced with fresh CM. This procedure was repeated every
   day.

   For F-actin rings staining, until day 4, cells were stained with
   actinRed 555 ReadyProbes Reagent (1:400 dilution in 1× PBS, Thermo
   Fisher Scientific, Cat#[199]R37112) and DAPI (1:1000 dilution in 1×
   PBS, Beyotime Biotechnology, Cat#C1002). The procedure was similar to
   that of immunofluorescence staining. The osteoclast F-actin rings were
   observed and imaged using an inverted fluorescence microscope (Observer
   3, ZEISS).

   For tartrate-resistant acid phosphate (TRAP) staining, on day 4, cells
   were washed with 1× PBS and fixed with 4% paraformaldehyde for 20 min.
   Next, cells were stained with Leukocyte Acid Phosphatase Kit
   (Sigma-Aldrich, Cat#387A-1KT) according to the manufacturer’s protocol.
   The results were observed and imaged using a light microscope (CKX53,
   Olympus). The cells with a burgundy color or more than three nuclei
   were recognized as osteoclasts. The number or area of formed
   osteoclasts was calculated using the Image-Pro Plus 6.0 software.

Bioassay for NO production

   MLO-Y4 cell were seeded into 35 mm petri dishes at 1.5 × 10^5
   cells/dish. After the treatment with LG-HMFs for 48 h, the MLO-Y4 cells
   supernatant was collected. First, we diluted the standard with the
   solution used for the sample to be tested and made a standard curve
   according to manufacturer’s protocol (Beyotime Biotechnology,
   Cat#S0021S). We added standards and samples in 96-well plates at a
   volume of 50 μL/well. Next we added Griess Reagent I to each well
   (50 μL/well) followed by Griess Reagent II (50 μL/well). The absorbance
   values were measured at 540 nm using an enzyme marker (Synergy HT,
   BioTek).

Quantification and statistical analysis

   All experimental data are expressed as mean ± SD, and statistical
   analysis was performed using GraphPad Prism version 8.2.1 software for
   Windows (GraphPad Software, San Diego, California USA). Data obtained
   from western blots are representative for 3 independent experiments.
   One-way ANOVA with Tukey’s multiple-comparison method was used to
   evaluate the differences between the control, 12 T upward, and 12 T
   downward. P-values <0.05 were considered statistically significant.
   Significance was defined as follows: ∗p < 0.05, ∗∗p < 0.01,
   ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Differential expression analysis was
   compared between the two combinations using DESeq2 1.20.0 software. The
   method of Benjamini and Hochberg was used to adjust the resulting
   P-values to control for false discovery rates (FDRs). Genes with an
   FDR-corrected P-value <0.05 by DESeq2 were considered differentially
   expressed. Significance for GO and Reactome enrichment analysis was
   determined using an FDR-corrected P-value (<0.05).

Acknowledgments