Hydrogen Sulfide Epigenetically Mitigates Bone Loss through OPG/RANKL Regulation During Hyperhomocysteinemia in Mice

Abstract

Hydrogen sulfide (H₂S) is a novel gasotransmitter produced endogenously in mammalian cells, which works by mediating diverse physiological functions. An imbalance in H₂S metabolism is associated with defective bone homeostasis. However, it is unknown whether H₂S plays any epigenetic role in bone loss induced by hyperhomocysteinemia (HHcy). We demonstrate that diet-induced HHcy, a mouse model of metabolite induced osteoporosis, alters homocysteine metabolism by decreasing plasma levels of H₂S. Treatment with NaHS (H₂S donor), normalizes the plasma level of H₂S and further alleviates HHcy induced trabecular bone loss and mechanical strength. Mechanistic studies have shown that DNMT1 expression is higher in the HHcy condition. The data show that activated phospho-JNK binds to the DNMT1 promoter and causes epigenetic DNA hyper-methylation of the OPG gene. This leads to activation of RANKL expression and mediates osteoclastogenesis. However, administration of NaHS could prevent HHcy induced bone loss. Therefore, H₂S could be used as a novel therapy for HHcy mediated bone loss.

Graphical Abstract

Introduction

Age-related skeletal diseases, such as osteoporosis, are characterized by low bone mass and microstructural deterioration in bone tissue, which leads to bone fragility, porosity and increased risk of fracture [1–3]. Furthermore, bone healing and recovery are markedly delayed, as has been confirmed in osteoporotic animals [4, 5]. Bone defects in patients caused by severe inflammation, accidental trauma and high-fat diet induced skeletal abnormalities are major challenges to treat, in clinical orthopaedics [6, 7]. The Western diet is rich in methionine, an amino acid found in red meat, fish, dairy and egg products, which can be detrimental to organs including the heart, liver, and kidneys [8, 9]. Methionine is a sulfur-containing amino acid and it is essential for the synthesis of cysteine and S-adenosyl methionine, which are both required for methylation reactions [10]. Excess dietary intake of methionine leads to alternation of methylation levels, induction of anemia, vascular damage and growth retardation [11–14]. In rabbits fed a high methionine containing diet, oxidative stress was augmented by a disturbance in the methionine-homocysteine (Hcy) metabolic pathway, in turn accelerating serum levels of Hcy, a condition known as hyperhomocysteinemia (HHcy) [15]. Previously our laboratory reported that HHcy in mice, induces oxidative stress-mediated alteration of cerebrovascular remodeling and neurodegeneration via epigenetic changes [16]. At present, several studies have reported the positive relation between oxidative stress and the osteoporotic condition [17]. For instance, in osteoporotic postmenopausal women, decreased bone mineral density was shown to be associated with higher Hcy levels [18]. In the HHcy mouse model, mice fed a high methionine diet (HMD) for 3 months showed increased femoral neck fragility by 18%, through deteriorating bone mechanical properties [19]. The direct adverse effect of oxidative stress by HHcy has been reported in bone and in serum, but the mechanism is still poorly understood.

Bone mass is maintained through a dynamic equilibrium between bone formation by osteoblasts (osteogenesis) and bone resorption by osteoclasts, which are produced and proliferate through a complex signalling process known as osteoclastogenesis [20]. Recent evidence suggests that Hcy mediated oxidative imbalance enhances bone resorption by promoting osteoclast activity [21–23] and induces a reduction in osteoblast activity [24]. As a result, disequilibrium between these cell types leads to skeletal diseases, such as osteoporosis. Therefore, therapeutic strategies must be aimed at preventing Hcy associated oxidative imbalance and disequilibrium in cell types to combat destructive bone diseases.

Hydrogen sulfide (H₂S) is an easily recognizable colourless gas with odorous characteristics similar to that of a rotten egg [25, 26]. Physiologically, H₂S is an endogenously released gasotransmitter, which is known to regulate signaling pathways, and it is generated in the vasculature, heart, kidney, brain, nervous system, lower gastrointestinal tract, skeletal muscle and duly noted to be generated in bone [26–28]. Abnormal H₂S production is linked to different pathophysiological conditions such as hypertension, atherosclerosis, diabetes and Alzheimer’s disease [29–33]. H₂S is mainly produced by two pyridoxal-5′-phosphate (PLP)-dependent enzymes, cystathionine-β-synthase (CBS) and cystathione-γ-lyase (CSE) in the transsulfuration pathway of homocysteine metabolism [34, 35]. CBS is predominantly expressed in bone marrow mesenchymal stem cells (BMMSCs) but is also found in the central nervous system, ileum, uterus, liver, kidney, placenta, as well as in pancreatic islets. CSE is abundantly expressed in the heart, kidney, liver, placenta, uterus and vascular smooth muscle. In addition, it has been well documented that CSE is the most relevant H₂S-producing enzyme in the cardiovascular system [28, 36–37]. Patients with CBS deficiency in turn suffer from the clinical condition HHcy, which is a multifactorial disorder encompassing mental deficiency, thrombosis, atherosclerosis and osteoporosis [38]. Our laboratory previously reported that HHcy (in a mouse model of CBS+/− deficiency) causes neurovascular dysfunction and memory loss [16]. Epidemiological and clinical data suggest that HHcy is a detrimental factor in the onset of bone loss and fracture [39–41]. Increased prevalence of osteoporotic pathophysiology in both these human and mouse models of HHcy are still understudied. A recent report also suggests that there is a notably decreased level of serum H₂S in CBS deficient mice in parallel with an increased level of serum Hcy [28]. Therefore, preventing HHcy induced bone loss may be a novel therapeutic strategy for treating destructive bone diseases.

Epigenetic mechanisms, particularly DNA methylation, are known to contribute in the controlling gene transcription and, ultimately, protein expression [42–43]. Therefore, alteration in DNA methylation by hyperhomocysteinemic means, tends to disrupt gene expression [16]. The molecular mechanisms by which HHcy causes these DNA methylation alterations, however, is not yet understood. Increased oxidation of Hcy via sulfhydryl groups could mediate epigenetic DNA methylation by changing DNA methyltransferase (DNMT) activity. DNMTs are involved in epigenetic changes in the target gene promoter or globally on methylated cytosine bases within various chromatin complexes, and are classified as DNA methyl transferases 1 (DNMT1; maintain methylation) and DNA methyltransferases 3a and 3b (DNMT-3a and 3b; regulate de novo methylation) [16]. These epigenetic changes may affect the physiologic and pathologic processes of Hcy-mediated destructive bone disease such as osteoporosis. It is still unknown whether DNA methylation regulates the expression of genes involved in bone homeostasis in hyperhomocysteinemic mice. Therefore, we hypothesized to understand the influence of DNA methylation on RANKL and OPG expression in mouse bone marrow (BM) cells and its possible involvement in the expression of the osteoporotic phenotype.

The previous report also suggests that H₂S could be a novel regulator in bone formation in the OVX or CBS^+/− mouse model. However, the protective role of H₂S in HMD enhanced HHcy induced bone loss is not well studied. In the present study, we addressed this issue and assessed the potential role of H₂S against HMD enhanced HHcy induced bone loss. We demonstrate that HHcy enhances oxidative stress and further alters epigenetic changes of the RANKL/OPG promoter homeostasis through c-Jun/JNK signaling in BMMSCs. This, in turn, activates paracrine signaling of osteoclast maturation and bone loss leading to the osteoporotic phenotype in mice. Pre-treatment with H₂S can prevent the HHcy induced osteoporotic phenotype, and thus displays an osteoprotective property.

Materials and Methods

Animals and Experimental Design

All experiments were conducted in female C57BL/6J (wild type, WT), mice starting at 12 weeks old. The animal procedures were carefully reviewed and approved by the Institutional Animal Care and Use Committee of The University of Louisville and followed the animal care and guidelines of The National Institutes of Health. Under the treatment regimen employed to develop the diet-induced HHcy condition, mice were fed with a methionine enriched (1.2%), low folate (0.08mg/Kg), low vitamin B6 (0.01mg/Kg) and B12 (10.4 μg/Kg) diet (Harlan Laboratories, Cat No.TD.97345) label as HMD diet for 6 weeks. Otherwise, control mice were fed standard chow. All mice were allowed water ad libitum. The female mice (12–weeks-old) were recruited and kept in seven different groups (n =7). To study the beneficial effects of H₂S on bone homeostasis in HHcy mice, sodium hydrogen sulfide (NaHS) was intraperitoneally (i.p) injected for the 6 week period. The mouse groups were:

• Wild-type C57BJ/L6 mice (WT)

• Wild-type mice fed with HMD (WT+HMD, or HHcy)

• NaHS-supplemented wild-type mice (WT+NaHS)

• NaHS-supplemented HHcy mice (HHcy + NaHS)

• SP600125 (JNK inhibitor)-supplemented HHcy mice (HHcy + SP600125)

• 5-Azacytidine-supplemented HHcy mice (HHcy + 5-Aza)

• N-acetyl cysteine (NAC)-supplemented HHcy mice (HHcy + NAC)

Drug Preparation and Administration

In vivo

NaHS (H₂S precursor), SP600125 (JNK inhibitor), 5-Azacytidine (DNA methyltransferase inhibitor), and NAC (anti-oxidant) were dissolved in 0.9% normal saline. HHcy mice were treated with NaHS (10 mg/kg/day) given daily through intraperitoneal route for a period of 6 weeks. SP600125 (15 mg/kg body weight), 5-Azacytidine (0.5 mg/kg body weight) and NAC (75 mg/kg body weight) treatments were given 3 times/week through intraperitoneal route for the same period of 6 weeks. Animals of the control group received normal saline through i.p injections. Biochemical, molecular, and immunohistochemical analyses were performed 24 hours after the last NaHS treatment or its vehicle injection in the separate groups.

In vitro

The BMMSCs used for in vitro study have been described in previous studies [28] and were given the following treatments: (1) WT-BMMSCs, (2) HHcy-BMMSCs, (3) WT-BMMSCs + NaHS (100 μM), (4) HHcy-BMMSCs + NaHS (100 μM). Cells were given treatment for 2-weeks and harvested for further experimentation.

Isolation of Mouse BMMSCs

Mouse BMMSCs were isolated as previously described [28] with little modification. Mouse bone marrow cells were flushed out from the bone cavities of femurs and tibias of 12-week old female mice by centrifugation (3000 rpm for 10 min at room temperature) and collected with 2% heat-inactivated fetal bovine serum (FBS; ATCC) in alpha minimum essential medium (α-MEM; Invitrogen). The cells were washed thrice with PBS and suspended in α-MEM 15% FBS. Cells were seeded at 1.5 ×10⁶ into 12-well culture plates (Corning) and initially incubated for 48 hr at 37°C in a 5% CO₂ incubation chamber. The non-adherent cells were washed twice with PBS. The adherent cells were further cultured for 21 days. The BMMSCs were cultured with α-MEM supplemented with 15% FBS, 2 mM L-glutamine (Invitrogen), 100 U/ml penicillin (Invitrogen), 100mg/ml streptomycin (Invitrogen) and 50mg/ml Amphotericin B (Invitrogen). The BMMSCs were characterized by flow cytometry analysis with specific markers such as CD73 and CD44 (Biolegends, San Diego, CA) (Supplementary Fig. S1).

In Vitro Mineralization Staining Assay

In vitro mineralization was performed as previously described [35]. For the in vitro Alizarin red-stain (ARS; Sigma Aldrich) assay, BMMSCs were seeded at a concentration of 5 × 10⁴ in 12-well culture plates under osteogenic induction medium (α-MEM+15% FBS supplemented with 2 mM β-glycerophosphate, 100 nM dexamethasone, and 50 μg/mL ascorbic acid) with or without NaHS for 21 days. At the end of the experiment, cells were washed with PBS, fixed with 70% ethanol for 35 minutes and were stained with 2% ARS, pH 4.2, for 20 minutes at room temperature. Briefly, absorbance was read at 510 nm calorimetrically for quantification of mineralized nodules. For the Von Kossa measurement, the BMMSCs were washed with PBS and fixed with 70% ethanol. The cells were then rinsed with distilled water and thereafter incubated under UV light in the presence of a 5% silver nitrate solution. After 20 minutes, the cells were washed with distilled water and incubated with a 5% sodium thiosulfate solution. All images were taken on a phase contrast microscope.

Alkaline Phosphatase (ALP) Assay

ALP activity and staining were carried out to study the osteogenic potential of BMMSCs, according to our previously published protocol [35]. Briefly, BMMSCs were cultured under osteogenic medium at a density of 5 × 10⁴ cells/well in 12-well culture plates. After 7 days of osteogenic induction, the cells were fixed with 70% ethanol and stained with ALP (Sigma) and the images were then photographed. The total ALP activity in the well was measured at 405 nm with a microplate reader and expressed as nmoles PNPP/mg of protein.

Osteoclast assay

BMMSCs were plated at 5 × 10⁴ cells/well in 24-well culture plates in α-MEM and were then treated with RANKL (30 ng/mL) and M-CSF (10 ng/mL) for 5 days. The media were changed every 48 hrs. After 5 days of culture, cells were xed with 4% paraformaldehyde (PFA) and stained with tartrate-resistant acid phosphatase (TRAP; Sigma). TRAP-positive cells were imaged under a phase-contrast microscope and counted. Similarly, osteoclast activity was determined by measuring TRAP5b activity in plasma using an ELISA kit from MyBioSource as per the manufacturer’s instructions.

Measurement of OPG, sRANKL and p-JNK/JNK activity

This assay was used to determine the amount of RANKL and OPG in the bone marrow plasma of the different experimental mice. BM plasma-derived RANKL and OPG was measured using commercial mouse RANKL (ab100749) and OPG (ab100733) sandwich ELISA kits from Abcam, as per the manufacturer’s instructions. To determine JNK protein phosphorylation and the pathway activation in BM cells in the experimental mice, we used JNK (Thr183/Tyr185) In-Cell ELISA kit (ab126424) from Abcam, as per the manufacturer’s instructions.

Homocysteine measurement assay

This assay was used to measure the total homocysteine content in experimental mouse plasma using a homocysteine assay kit from Crystal Chem, as per the manufacturer’s instructions.

Measurement of Glutathione peroxidase activity

This assay was used to measure all of the glutathione dependent peroxidases in experimental mouse plasma using a Glutathione Peroxidase assay kit from Abcam as per the manufacturer’s instructions.

Hydrogen sulfide measurement assay

BM plasma was used to measure the endogenous level of H₂S, as previously described [35]. The bone femurs were collected and flash-frozen with liquid nitrogen. They were then homogenized in ice-cold 50 mmol/L potassium phosphate buffer, pH 8.0. The homogenates were centrifuged (10,000g; 12 minutes; 4°C) and the supernatants were collected (75μL) and mixed with 0.25 mL Zn acetate (1%) and 0.45 mL water for 30 minutes at room temperature. Trichloroacetic acid (10%; 0.25 mL) was then added and the homogenates were again centrifuged (12,000g; 15 minutes; 4°C). The clear supernatant was isolated and mixed with N,N-dimethyl-p-phenylenediamine sulfate (20 mmol/L; 133μL) in 7.2 mol/L HCl and FeCl3 (30 mmol/L; 133μL) in 1.2 mol/L HCl. After 30 minutes, absorbance was measured at 670nm with a microplate reader. The calibration curve of H₂S was obtained by using varying concentrations of NaHS solution (0 to 320 μmol/L NaHS).

BMMSCs proliferation assay

A MTT cell proliferation assay was performed, as per previously described [31]. BMMSCs were collected from bone cavities from the experimental mice and seeded onto 96-well culture plates and cell proliferation was quanti ed after 24-hour stimulation, expressed as relative intensity.

Intracellular ROS analysis by flow cytometry

For the measurement of intracellular ROS, BMMSCs (5 × 10⁴ cells) were collected by flushing tibias with PBS and staining them with 5 mM of dichlorodihydrofluorescein diacetate (CM-H2DCFDA, Invitrogen, Carlsbad, CA, USA) for 30 minutes at 37°C. Stained cells were analyzed using a flow cytometer (BD Accuri^™ C6 Plus). Average intensity was analyzed using the BD Accuri C6 Software.

Lipid peroxidation assay

This assay was used to measure the total lipid peroxide in the form of malondialdehyde (MDA) content in the plasma using a lipid peroxidation (MDA) assay kit from Abcam as per the manufacturer’s instructions.

Western blot analysis

The femur bone extracts (50 μg) were loaded on a sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and run at a constant 110 volts. Separated proteins in the gels were transferred to polyvinylidene difluoride membranes using an electrotransfer apparatus (Bio-Rad) run at a constant 120 milliamps overnight. Following the transfer, membranes were blocked with 5 % non-fat dry milk in TBS-T solution for 1 h. Then membranes were incubated overnight with the desired primary antibody at 4 °C. After washing with TBS-T, the membranes were further probed with a secondary antibody [horseradish peroxidase-conjugated] for 120 minutes at room temperature. The membranes were developed with ECL Western blotting detection system (GE Healthcare, Piscataway, NJ, USA) and imaged in the gel documentation system (Bio-Rad). Band density was normalized with a loading control using Image Lab densitometry software (Bio-Rad).

Gene expression by quantitative Real-Time PCR (qRT-PCR)

qRT-PCR was performed to study the quantitative gene expression profile of mRNA transcript levels of the different genes. Total RNA was isolated with TRIzol® reagent (Invitrogen, Grand Island, NY, USA) from the isolated femur bone tissue according to the manufacturer’s instructions. Quantification and purity of total RNA were both assessed using the Nanodrop-1000 instrumentation (Thermo Scientific, Waltham, MA, USA). Total RNA (2 μg) was reverse-transcribed to _CDNA, as previously described [31]. Briefly, cDNA was prepared in a thermocycler (Bio-Rad) with the following cycle: 42 °C for 58 min, 70 °C for 18 min, and 4 °C at the end. qRT-PCR was performed for the desired genes in a final reaction volume of 20 μl by using the LightCycler® 96 Instrument (Roche). Relative quantification of PCR products was collected with the comparative CT method, comparing PCR products to the loading control mRNA expression of GAPDH. The list of primers used in this study is shown in Table 1.

Table 1.

Sequences of PCR primers used for real time quantitative PCR and quantitative methylation specific PCR.

Gene	Primer Sequences (5′ → 3′)
Mouse Runx2	FP: TTTAGGGCGCATTCCTCATC RP: TGTCCTTGTGGATTAAAAGGACTTG
Mouse Osteocalcin	FP:GCGCTCTGTCTCTCTGACCT RP: ACCTTATTGCCCTCCTGCTT
Mouse DNMT1	FP:AGGGAAAAGGGAAGGGCAAG RP: CCAGAAAACACATCCAGGGTCC
Mouse DNMT-3a	FP:CCATAAAGCAGGGCAAAGACC RP:AGTGGACTGGGAAACCAAATACC
Mouse Nfatc1	FP:GAGACAGACATCCGGAGGAAGA RP:GTGGGATGTGAACACGGAAGA
Mouse Ocstamp	FP:TGTAGCCTGGGCTCAGATGT RP:GTTGGTTGAGGACGTAGAGG
Mouse Oscar	FP:CTCTTCAAAAGTGGCCTTGACA RP:GGAAGAACTCAGCCCGCTCAA
Mouse Cathepsin K	FP:GGATGAAATCTCTCGGCGTTT RP:GGTTATGGGCAGAGATTTGCTT
Mouse RANK	FP:ACTGAGGAGGCCACCCAAGGA RP:TGAAGAGGACCAGAACGATGAG
Mouse GAPDH	FP: TGCACCACCAACTGCTTTGC RP: GGCATGGACTGTAGTCATGAG
Gene	qMSP Primer sequences
Mouse RANKL	Left M primer: CGTCGGGTTAGTCGAGATTAC Right M primer: AAACCGATACAAAAAACGCG Left U primer: ATGTGTTGGGTTAGTTGAGATTATG Right U primer: AAACCAATACAAAAAACACAAA
Mouse OPG	Left M primer: TTTTGTTTTGATTATTTTTATACGG Right M primer: TCTAAAATTCCAACTTACACCACG Left U primer: TTTTGTTTTGATTATTTTTATATGG Right U primer: CTAAAATTCCAACTTACACCACAC

DNMT activity assay

From whole bone homogenate, nuclei were isolated using the EpiQuik nuclear extraction kit (Epigentek, Farmingdale, NY). Furthermore, DNMT activity was measured to study the total methylation in the isolated genomic DNA by using EpiQuik DNMT activity/inhibitor assay ultra-kit (Epigentek) according to the manufacturer’s instructions.

5-mC ELISA

Quantification of 5-methylcytosine (5-mC) levels was used to study the global methylation profile in the genomic DNA, as described in the previously published report [16]. The 5-mC levels were determined by using the 5-mC DNA ELISA kit (Zymo Research, Irvine, CA) according to the manufacturer’s instructions.

DNA bioinformatics study and RANKL/OPG methylation analysis by qMSP

In order to identify potential CpG islands in the RANKL and OPG genes, we performed a bioinformatics analysis by using USCS Human Genome Browser public database (http://genome.ucsc.edu) and CpG Island Explorer software (cpgie.sourceforge.net). Quantitative methylation-specific PCR (qMSP) is based on the amplification of bisulfite converted DNA to show the expression of methylated specific regions particularly in CpG island regions. Primers, which targeted CpG-rich regions within the RANKL and OPG regions, were designed with MethPrimer: designing primers for methylation PCRs, available from the Li Lab [44]. From each region of the CpG islands, a pair of primers: bisulfite-treated methylated DNA and unmethylated bisulfite-treated DNA were selected (Table 1). The genomic DNA was isolated which was followed by bisulfite conversion using Zymo EZ DNA Methylation-direct kit (Zymo Research) as per the manufacturer’s instructions. Following purification of bisulfite DNA, a qMSP reaction was performed to amplify the bisulfite DNA. Each PCR product was loaded onto a 1.5% agarose gel, and band intensities were normalized to unmethylated PCR products.

Chromatin Immunoprecipitation Assay (ChIP)

Chromatin preparation in BMMSCs was carried out using the ChromaFlash Chromatin Extraction Kit (Epigentek, Farmingdale, NY, USA) as described by the manufacturer’s instructions. ChromatinImmunoprecipitation Assay was performed as previously described [45]. Briefly, the immunoprecipitated DNA (through anti-JNK antibody) was amplified by PCR reaction using GoTaq® Hot Start Green Master Mix (Promega, USA) and PCR instrumentation S1000^™ Thermal Cycler-BIORAD. Primers were designed for the DNMT1 promoter. The primer pairs used to amplify sequences surroundings predicted JNK binding sites at the DNMT1 locus: DNMT1_forward primer: GAAAGTTTAAGGCCGGGCAC, DNMT1_reverse primer: GATCACTGC-AGCCTCTACCT.

Dynamic histomorphometric analysis

Experimental mice femurs were collected and fixed in 10% neutral buffered formalin for 2 days and decalcified in 14% ethylenediamine tetraacetic acid (EDTA) for 2 weeks. Then, the femurs were embedded in paraffin and cut into 4μm slices using a Leica RM2125 RTS Microtome. The slices were stained with H&E solutions according to the manufacturer’s instructions, and the images were acquired using a confocal microscope (Leica DM4000B, Germany) with 10× magnification. Five representative images were analyzed by using NIH Image J software and the results were shown as the percentage of trabecular bone per total bone area.

3-point bending test of bone

Bone femurs were dissected and fixed in 10% neutral buffered formalin for 2 days and stored in −80°C. This method was used for physical measurements and mechanical testing by the 3-point bending test. All bone specimens were tested under a load applied at a constant rate of 20 mm/min. This test predominantly measured cortical bone strength parameters including maximum load (ultimate strength) and stiffness (slope of the linear portion of the curve representing elastic deformation) by using the Bone Strength Tester TK-252C (Muromachi Kikai Co. Ltd., Tokyo, Japan).

MicroCT analysis of bone

Micro-CT analyses of excised femur bones were carried out using the SkyScan 1076 μ-CT scanner (Aartselaar, Belgium). The bones were first fixed in 10% neutral buffered formalin and stored in 70% ethanol. After scanning, bone reconstructions were carried out by Sky Scan Nrecon software. The trabecular bone region of the distal femurs was extracted by drawing ellipsoid contours with the CT analyzer software. 3-D images of the trabecular region of bones were drawn by OsteoQuant software by selecting ROI and 2-D drawn by Data-viewer software. The trabecular bone volume/tissue volume BV/TV (%), trabecular number (Tb.N) (mm⁻¹), trabecular separation (Tb.Sp) (μm) and trabecular thickness (Tb.Th) (μm) were obtained.

Statistical analysis

Data analyses and graphical presentations were performed with GraphPad Prism, version 5.00 (GraphPad Software, Inc., La Jolla, CA). Data are represented as mean values ± standard error mean (SEM). The experimental groups were compared by one-way analysis of variance (ANOVA) in combination with Tukey’s multiple comparison test. The significance of differences between groups was determined using Two-tailed, unpaired Student’s T-test. P < 0.05 was considered statistically significant.

Results

Diet-induced HHcy alters homocysteine metabolism and H₂S production in BM cells

It is well known that H₂S is physiologically produced from CBS and CSE enzymes through the transsulfuration pathway [46]. To investigate whether diet-induced HHcy alters normal Hcy metabolism and systemic H₂S production in bone, 12-week-old wild-type (WT) female mice were fed with HMD. Six weeks post-feeding, total BM cells were collected and cultured for 14 days. mRNA expressions of CBS and CSE in total BM cells and BMMSCs were significantly decreased in the HHcy condition versus in WT mice (Fig. 1A–D). BMMSCs were characterized by flow cytometry analysis as shown in Fig. S1. Similarly, protein western blot analysis showed that both CBS and CSE levels were lowered in the HHcy condition in comparison to WT mice (Fig. 1E, F). Furthermore, we tested the CBS enzymatic activity in the experimental mice. The results show that CBS enzymatic activity also decreased in the HHcy condition versus in WT mice (Fig. 1G). In order to know the effect of the HMD on the metabolic regulation of the Hcy pathway, two crucial metabolites, SAM and SAH were analysed by ELISA. The result revealed that the SAM/SAH ratio was significantly higher in the HHcy condition than in WT mice (Fig. 1 H, I and J). Concurrently, we tested both Hcy protein expression and plasma level of total homocysteine (tHcy) in the experimental groups. We found that the HMD exaggerated the level of both Hcy expression and plasma tHcy in the HHcy condition as compared to WT control (Fig. 1K, L, and M). Furthermore, H₂S levels were lower in the HHcy condition than in WT mice (Fig. 1N). Moreover, the data revealed that mouse plasma H₂S levels were indeed increased in the HHcy mice treated by intraperitoneal (i.p) injection of NaHS and were decreased in HHcy mice alone (Fig. S2A). Additionally, gene expression analysis and CBS enzyme activity assay also revealed that CBS gene expression and CBS enzyme activity were indeed upregulated in HHcy mice upon NaHS administration in comparison with HHcy alone (Fig. S2B, C). These data clarify the notion that HMD induced HHcy alters the normal homeostasis of H₂S production in Hcy metabolism.

Figure 1. Diet-induced Hyperhomocysteinemia (HHcy) alters normal homocysteine metabolism and hydrogen sulfide (H₂S) production.

(A and B) mRNA expression of mouse CBS and CSE in total BM cells. (C and D) mRNA expression of CBS and CSE in BMMSCs. (E and F) Protein western blot analysis of CBS and CSE in the experimental group. (G) CBS activity of the experimental groups. (H, I and J) Levels of SAM, SAH and the SAM/SAH ratio was estimated by ELISA. (K and L) Protein western blot analysis of Hcy expression. (M) Plasma-derived tHcy. (N) Plasma-derived H₂S after 6 weeks of supplemented HMD. Data are expressed as mean ± SEM. n = 7 mice per group. *p < 0.05 compared with the wild-type (WT) mice. See also Figure S1 and S2.

Administration of NaHS prevents HHcy-induced oxidative stress

To evaluate the osteoprotective effect of NaHS on HHcy induced oxidative stress, several parameters were assessed in relation to bone oxidative stress. As lipid peroxidation is an important indication of oxidative damage, we measured the MDA level in bone marrow (BM) plasma. The lipid peroxidation product, MDA, significantly increased in HHcy BM plasma compared to WT. This increase in MDA was rescued by administration of NaHS (Fig. 2A). On the contrary, protein western blot analysis shows that NOX-4, a marker of oxidative stress was significantly increased. However, NaHS reversed this effect (Fig. 2B, C). Similarly, anti-oxidant enzyme system glutathione peroxidase (GPx) was also markedly decreased in the HHcy condition. Addition of NaHS in the HHcy condition was sufficient to normalize the adverse HHcy effect (Fig. 2D). Since, reactive oxygen species (ROS) is an important mediator of oxidative stress; we therefore, sought to determine the production of total ROS in BMMSCs by flow cytometry analysis. The HHcy mice derived BMMSCs displayed a higher amount in total ROS as compared to WT (Fig. 2E, F). To further study ROS production under the HHcy condition, we tested the level of H₂O₂ production by ROS-Glo^™ H₂O₂ assay. The results showed a marked increase in H₂O₂ production in the HHcy condition with NaHS reversing these changes (Fig. 2G). This finding suggests that NaHS supplementation could improve the redox homeostasis in bone.

Figure 2. Effect of NaHS on HMD induced HHcy-mediated oxidative stress in mice.

(A) Lipid peroxidation level in the bone marrow plasma was assessed by MDA assay. (B) Representative Western blot analysis for NOX-4 protein and GAPDH control. (C) Densitometry analysis of NOX-4 protein expression as represented in the bar diagram (D) Cellular antioxidant enzyme GPx activity. (E, F) Flow cytometry analysis of Reactive Oxygen Species (ROS) using CM-H2DCFDA fluorescent dye in BMMSCs. (F) Densitometry analysis of ROS fluorescence represented in the bar diagram (G) Bar diagram represents quantification of H₂O₂ production in the experimental group. WT (label as WT), HMD-fed mice (label as HHcy), WT+NaHS (label as NaHS) and HMD+NaHS (label as HHcy+NaHS). Data represented as mean ± SEM from n =7 mice per group. *p < 0.05 compared with the wild-type (WT) mice, #p < 0.05 compared with the HHcy mice.

NaHS ameliorated Hcy-Induced RANKL synthesis via c-Jun/JNK signaling

RANK ligand (RANKL) and Osteoprotegerin (OPG), are proteins expressed by BM-derived osteoblasts. OPG a natural inhibitor of RANKL, interferes with RANKL-RANK association in the osteoclast, and thereby regulates osteoclast differentiation and bone resorption. To investigate the effects of Hcy on OPG and RANKL synthesis in in vitro conditions in BMMSCs collected from WT mice, we estimated these levels by ELISA in 24-hour conditioned medium taken from different dosage treatments of Hcy (0.25, 0.5, 1, 2, 3 Mm) (Fig. 3A). Our results suggested that a 3 mM concentration of Hcy, significantly reduced the OPG level and increased the RANKL level in BMMSCs culture supernatants. Time-dependent study of Hcy exposure to BMMSCs culture shows an increase in RANKL synthesis from 24–72 hours (Fig. 3B). This increase in RANKL was attenuated by NaHS (100 μM) supplementation to the BMMSCs culture. We further performed protein western blot analysis of RANKL and OPG in experimental mice. The results showed that RANKL protein expression was significantly increased in HHcy mice as compared to the WT group. However, NaHS supplementation was able to recover the RANKL level in the Hcy + NaHS mice (Fig. 3C, D). Similarly, the OPG level was significantly reduced in HHcy mice and administration of NaHS reversed this effect (Fig. 3C, D).

Figure 3. NaHS ameliorated HHcy-induced RANKL synthesis via c-Jun/JNK signaling.

(A) In Vitro BMMSCs culture supernatant derived OPG and secreted RANKL level was measured by ELISA, treated with increasing dosage concentrations of Hcy (0–3mM). The RANKL level was upregulated by increasing the concentration of Hcy. However, the OPG level was down-regulated by increasing the dosage concentration of Hcy. (B) Time-dependent study of Hcy exposure (3mM) to BMMSCs culture shows an increase in secreted RANKL (sRNAKL) level in culture supernatants. However, combination treatment of Hcy with NaHS (100 μM) reverses the Hcy effect, as measured by ELISA. (C) Representative Western blot analysis for RANKL, OPG proteins and GAPDH control. (D) Densitometry analysis of RANKL and OPG protein expression as represented in the bar diagram. (E) Bar diagram represents quantification of phosphorylation of JNK was determined by JNK (Thr183/Tyr185). (F) Bar diagram represents quantification of sRANKL levels in plasma (G) Bar diagram represents the quantification OPG level in plasma were estimated in by ELISA. Plasma isolated from WT (label as WT), HMD-fed mice (label as HHcy), WT+NaHS (label as NaHS), HMD+NaHS (label as HHcy+NaHS), HMD+SP600125 (label as HHcy+ SP600125 and HMD+ N-acetyl cysteine (label as HHcy+NAC). Data are expressed as mean ± SEM. n = 7 mice per group. *p < 0.05 compared with the wild-type (WT) mice, #p < 0.05 compared with the HHcy mice. See also Figure S3.

To determine the mechanism of HHcy induced RANKL synthesis via oxidative stress mediated c-Jun/JNK signaling, we tested the phosphorylation of JNK (T183/Y185) activity under treatment of JNK inhibitor SP600125 and antioxidant N-acetyl cysteine (NAC) in HHcy mice by ELISA method. The data show that HHcy triggered the phosphorylation of JNK (T183/Y185) as compared to the WT control. However, administration of NaHS, SP600125, and NAC reversed the HHcy induced effects (Fig. 3E). To further confirm the role of HHcy induced phosphorylation of JNK (T183/Y185), we applied a specific small interfering RNA (siRNA) against JNK and found that JNK-siRNA was able to substantially prevent increased phosphorylation of JNK (T183/Y185) in the HHcy condition (Fig. S3A, B, C). To further understand the molecular mechanism of RANKL/OPG synthesis through c-Jun/JNK activation, we tested the level of the RANKL (RANKL) and OPG in plasma by ELISA. We found that there were significant decreases in the sRANKL level in HHcy+SP600125 group as well as HHcy+NAC group as compared to mice with HHcy (Fig. 3F and also Fig. S3D). In contrast, OPG levels were markedly increased in the HHcy+SP600125 group as well as the HHcy+NAC group compared to HHcy (Fig. 3G and also Fig. S3E). Overall, administration of NaHS was able to reduce the HHcy mediated increased RANKL expression in HHcy mice. These results suggest that HMD induced HHcy mediated sRANKL upregulation by c-Jun/JNK signaling.

Effects of NaHS on HHcy induced JNK transcriptional activation and its binding to the DNMT1 Promoter

To understand the role of HMD induced HHcy on maintenance and de novo methylation, we performed both protein western blot and qRT-PCR to analyze the expression of the DNA methyl transferase enzymes (DNMT1, DNMT3A) and its mRNA levels respectively. DNMT-1 protein expressions were upregulated in the HHcy condition when compared to the control and (Fig. 4A, B). However, there was no changes has been detected in DNMT3A protein expression in all the condition. Likewise, mRNA level of DNMT-1 was significantly upregulated under HHcy condition as accessed by qRT-PCR analysis (Fig. 4C). DNMT3A levels also remained unchanged in all the experimental mice. However, NaHS treatment was able to attenuate the HHcy effect. This result indicates that maintenance of methylation through DNMT1 are induced during HHcy condition. We also tested the DNMT activity in BMMSCs cells nuclear extract and found a robust increased in DNMT activity in HHcy condition as compared to WT control. NaHS treatment restore the DNMT activity (Fig. 4D). This increased DNMT activity was further reduced in HHcy mice treated with JNK inhibitor SP600125 in BMMSCs cells nuclear extract. In addition, we also determined global DNA methylation from mouse BMMSCs derived genomic DNA by estimating the percent 5-mC levels. The results show that % 5-mC level was higher in HHcy DNA compared with WT control. Intently, Inhibition of c-Jun/JNK phosphorylation by JNK inhibitor SP600125 in HHcy mice markedly reduced global methylation status (% 5-mC). Interestingly, NaHS treatment ameliorated the HHcy induced methylation or hyper-methylation (Fig. 4E).

Figure 4. Effect of NaHS on HHcy induced DNMT1 expression via JNK mediated transcriptional gene regulation and its effects in the OPG/RANKL promoter methylation pattern.

Representative Western blot analysis for DNMT-1, DNMT3A proteins and GAPDH control. (B) Densitometry analysis of DNMT1 and DNMT3A protein expression as represented in the bar diagram. (C) qRT-PCR analysis of DNMT-1 and DNMT-3A mRNA expression levels as represented in a bar graph. (D) DNMT activity in BMMSCs nuclear extract samples expressed in OD per hour per milligram (mg) of protein represented in a bar graph. (E) Bar graph representing global DNA methylation status by %5-mC detected in different mouse BMMSCs genomic DNA samples. (F) PCR gel of DNMT ChIP assay. (G) Bar graph plot of ChIP results after JNK transcriptionally regulates DNMT1 expression by promoter binding in BMMSCs of experimental mice treated with NaHS and JNK inhibitor SP600125. (H) Representing PCR gel image of methylation (M) and un-methylated (U) shows that methylation changes at the promoter of candidate genes (OPG and RANKL) in experimental mice treated with NaHS and JNK inhibitor SP600125. (I) Bar diagram represents quantification of sRANKL levels in plasma after 5-Azacytidine treatment measured by ELISA. Data are expressed as mean ± SEM. n = 7 mice per group. *p < 0.05 compared with the wild-type (WT) mice, #p < 0.05 compared with the HHcy mice. See also Figure S3.

A role of c-Jun/JNK in the regulation of DNMT1 has been previously determined in some cancer model (45). To investigate the c-Jun/JNK mediated signaling that is involved in activation of DNMT1 gene at the transcriptional level in bone tissue during HHcy condition, we carried out ChIP assay to investigate whether JNK is able to activate the transcription from of the DNMT1 promoter in vivo. After cross-linking and immunoprecipitation, we found that JNK antibody readily precipitated chromatin containing the DNMT1 promoter in HHcy condition as compared to WT control and NaHS administration reversed this effect (Fig. 4F, G). In addition, treatment of JNK inhibitor SP600125 to HHcy condition resulted in decreased precipitation of the DNMT1 promoter, indicating that c-Jun/JNK binds to the DNMT1 promoter and regulates its expression in vivo.

Phospho-c-Jun/JNK epigenetically regulates the differential CpG methylation pattern in the OPG-RANKL gene promoter by regulating DNMT1 expression

The DNA methylation pattern of the gene regulates its expression during bone homeostasis. As we reported here that imbalance in the expression of OPG/RANKL in HHcy condition (Fig. 3), which may direct towards BMMSCs dysfunctions. We then tested potential involvement of phosphor-JNK in the establishment of global DNA methylation in bone tissue through DNMT1 expression during HHcy (Fig. 4). To test the hypothesis further, the increased DNMT1 activity may regulate methylation pattern in CpG islands of OPG and RANKL promoters under HHcy condition through JNK signaling. HHcy mice were treated with JNK inhibitor SP600125 and measured DNA methylation changes in the OPG and RANKL promoter by quantitative methylation specific PCR (qMSP-PCR). The result revealed that a significant increase in the DNA hyper-methylation of CpG regions of OPG gene promoter (regulator of osteoblast activity) of HHcy mice as compared to untreated WT control. Conversely, DNA methylation of CpG regions of OPG promoter was reduced in HHcy mice treated with JNK inhibitor SP600125 (Fig. 4H), and NaHS reversed this effect in HHcy + NaHS mice. In addition, we also checked the methylation pattern of CpG regions of RANKL gene promoter (regulator of osteoclast activity) under HHcy condition. We found that during HHcy, CpG regions of RANKL gene promoter was hypo-methylated as compared to WT control. However, this effect was rescued by supplementation of NaHS treatment (Fig. 4H). The details of CpG islands prediction in RANKL and OPG gene was performed by Methprimer analysis (Supplementary Fig. 4A, B). To dissect further the involvement of epigenetic mechanism, we tested the production of secreted RANKL with 5′-AZA (DNMT inhibitor) treatment in experimental mice. 5′-AZA and NaHS treatment significantly decrease the sRANKL production under HHcy condition as compared to WT (Fig. 4I). Together, these results demonstrate that RANKL produced by epigenetic DNA hypomethylation in HHcy mice. The increased synthesis of RANKL could facilitate the osteoclast maturation and differentiation by a paracrine mechanism. However, the reduced expression of OPG through epigenetic hyper-methylation further causes the BMMSCs derived osteoblast dysfunction.

Effect of NaHS on HHcy induced osteoclastogenesis ex vivo and in vivo

To determine whether the secretome of HHcy BMMSCs influenced osteoclastogenesis, we performed ex vivo osteoclastogenesis assays. It is well known that RANKL induces the differentiation of macrophage to osteoclast lineage and maturation [47]. Thus, BM cells from WT mice were cultured under osteoclast differentiation medium. BM cells were treated with conditioned medium (CM) derived from WT, HHcy, NaHS and HHcy + NaHS BMMSCs. The results showed that HHcy BMMSCs culture treated CM had more mature, tartrate-resistant acid phosphatase–positive (TRAP+) osteoclasts by Day 4 of differentiation, compared to cultures treated with CM from WT BMMSCs (Fig. 5, A and B). We also tested a series of osteoclast gene expressions from Nfatc1, Oc-Stamp, Oscar, Ctsk and Rank. We found that these genes are markedly upregulated in the CM of HHcy derived BMMSCs in comparison with WT. (Fig. 5C). To study the RANKL induced osteoclastogenesis, we treated a neutralizing antibody that binds to RANKL in the CM of RANKL associated groups. The results clearly suggested an increase in osteoclastogenesis from CM of HHcy derived BMMSCs as compared to CM from WT derived BMMSCs (Fig. 5D, E), in agreement with an increase in RANKL protein expression (Fig. 3C, F). However, CM treated with neutralizing RANKL antibody reduced HHcy induced osteoclastogenesis. Moreover, TRAP-5b enzyme activity was higher in HHcy osteoclast culture as compared to WT control, as determined by using ELISA. However, NaHS reversed this effect (Fig. 5F).

Figure 5. NaHS decreased HHcy induced osteoclastogenesis in vitro.

(A to F) BM cells were collected from WT mice and plated under 15% CM from WT and HHcy BMMSCs. (A) Day 5 TRAP-stained mature osteoclasts. (B) Quantitative analysis was showing the number of TRAP+ mature osteoclast (n = 7). (C) Total RNA was isolated from mature osteoclast cultures, and gene expression analysis for the indicated genes. (D) CM from HHcy-induced BMMSCs was pre-treated with a neutralizing antibody recognizing RANKL or an immunoglobulin G (IgG) control antibody (Ab). Day 5 mature osteoclasts were TRAP-stained and the number of osteoclasts was quantified (n = 7). (E) Bar diagram represents quantification of TRAP positive osteoclast (F) Bar diagram represents quantification of TRAP5b activity under HHcy. WT (label as WT), HMD-fed mice (label as HHcy), WT+NaHS (label as NaHS), HMD+NaHS (label as HHcy+NaHS). Data represented mean ± SEM from n =7 per group. *p < 0.05 compared with the wild-type (WT) mice and #p < 0.05 compared with the HHcy mice.

NaHS reverses the HHcy diminished osteoblast differentiation and mineralization in vitro

TRAP+ osteoclast assay surprisingly revealed a signi cant increase in osteoclastogenesis in vitro in HHcy condition. To address the role of HHcy in BMMSCs-derived osteoblasts, primary BM cells were cultured and differentiated in vitro under osteogenic induction medium for 14 days. MTT cell proliferation assay was con rmed the decreased proliferation of BMMSCs over time-dependent study (0–7 days) under the HHcy condition, as indicated by cell proliferation intensity and NaHS reverses the HHcy mediated effect. (Fig. 6A). We analyzed whether HHcy could suppress ALP activity in BMMSCs. Our data indicate that HHcy-BMMSCs exhibited a decrease in ALP activity and staining on day 6 under osteogenic medium. However, NaHS mitigates HHcy effect. Alizarin red staining was performed to investigate whether the HHcy resulted in a decreased mineralized nodule formation. HHcy reduced the mineralized nodule formation on Day 14 after confluence demonstrated positive ARS staining (Top panel) (Fig. 6C, D) compared with WT control. Similarly, Von kossa staining revealed a similar effect to alizarin red staining (Lower panel) (Fig. 6C). In parallel, we measured the calcium content of BMMSCs-osteoblast culture. The calcium levels were significantly reduced in HHcy BMMSCs culture in compared to WT (Fig. 6E). Collagen secretion was also markedly reduced in HHcy condition as compared to WT control (Fig. 6F). Western blot analyses revealed that protein expression levels of osteoblastogenesis-related proteins, Runx2, and osteocalcin were signi cantly reduced in HHcy condition as compared to WT control (Fig. 6G, H). Gene expression analysis by qRT-PCR of osteoblastogenesis-related genes, Runx2 and osteocalcin were also signi cantly reduced in HHcy condition as compared to WT control (I). However, the overall effects were improved by administration of NaHS. These data indicate that NaHS is a positive regulator of osteoblastogenesis in vivo by suppressing HHcy mediated detrimental effects.

Figure 6. NaHS increases osteogenesis in HHcy in vitro.

(A) In Vitro cell proliferation assay was measured by MTT method in BMMSCs culture in day dependent manner. Cell proliferation was low in HHcy culture in compared to WT. (B) Representative image and quantitative analysis for ALP staining of the BMMSCs culture after on day 6 under osteogenic medium. (C) Representative images of osteogenic bone mineralization assay (top panel with alizarin red staining and lower panel with Von Kossa staining) of BMMSCs after day 14. (D) Quantitative analysis for bone mineralization (Alizarin red assay) of the BMMSCs on day 14. (E) Bar diagram represents quantification of calcium deposition was measured in BMMSCs using calcium calorimetric assay kit. (F) Representative image and quantitative analysis for collagen secretion (Sirius red assay) of the BMMSCs after day 14. (G) Representative Western blot analysis for OCN, Runx2 proteins and GAPDH control. (H) Densitometry analysis of OCN and Runx2 protein expression as represented in the bar diagram. (I) mRNA expression of osteogenic marker genes (Runx2 and OCN). Data are expressed as mean ± SEM. n = 7 mice per group. *p < 0.05 compared with the wild-type (WT) mice, #p < 0.05 compared with the HHcy mice.

Pharmacological administration of H₂S (NaHS) improves bone formation in HHcy-induced bone loss

Confirming the results of the previous experiment, HHcy mice had lower plasma H₂S level and osteoblast differentiation, mineralization, and NaHS treatment normalized the plasma H₂S level and subsequently bone mineralization (Fig. 6). To investigate further, HHcy mice contributes to bone loss in in vivo model, 12-week-old female WT mice were treated with HMD fed diet (HHcy) or the NaHS (through i.p injections at 10 mg/kg/day) for 6 weeks (Fig. 7A). We observed no significantly gross developmental abnormalities in the whole skeleton, as observed by X-ray skeletal in vivo imaging (In-Vivo FX PRO; BRUKER Corporation) (Fig. 7B). Thinning of the femur and its length were affected in HHcy mice as compared to WT and HHcy + NaHS mice (Fig. 7C). There was a significant change in body weight (Fig. 7D) and high plasma TRAP5b activity indicated enhanced bone remodeling activity and higher susceptibility to bone loss in HHcy condition as compared to WT and HHcy+ NaHS mice (Fig. 7D, E). In parallel, we quantified various bone remodeling markers assessed by ELISA of plasma taken from the experimental mice. The plasma concentrations of CTx, a marker of bone resorption, were indeed increased in the HHcy mice in comparison to the WT mice and HHcy+ NaHS mice (Fig. 7F). The plasma concentrations of P1NP, a marker of bone formation, were reduced in the HHcy mice as compared to WT and HHcy+ NaHS mice (Fig. 7G). Moreover, NaHS administration to HHcy was able to rescue the reduced P1NP expression (Fig. 7G). Bone density was lower in the metaphysial area of the femur in the HHcy condition as compared to WT control and HHcy + NaHS (Fig. 7H).

Figure 7. Pharmacological administration of NaHS prevents HHcy-induced trabecular bone loss in vivo.

(A) H₂S donor NaHS was i.p. injected to HHcy mice every day for 6-weeks (10mg/kg/wt). After the last injection, the samples were harvested for additional experiments. (B) Representative X-ray images of the experimental mice. Arrows illustrate that NaHS increases bone density in femur (FROI, femur region of interest). (C) Bar diagram represents quantification of Femur length of the experimental mice. (D) Bar diagram represents quantification of Body weight was followed before (0-weeks) and after post feeding HMD diet (6-weeks) in the experimental group. HHcy mice show a significantly lower in body weight as compared with controls (n=7 per group). (E) Bar diagram represents quantification of Plasma TRAP5b activity, a specific marker of osteoclast activity in experimental mice. (F) Bar diagram represents quantification of Plasma CTX level, a marker of bone resorption. (G) Bar diagram represents quantification of Plasma P1NP level, a marker of bone formation. (H) Bar diagram represents quantification of BMD of the experimental mice. (I) Representative μCT cross-sectional images of distal femurs were demonstrating bone phenotypes. (J–M) Corresponding μCT measurements showed high trabecular bone mass phenotypes. Bone volume fraction (BV/TV) (%), trabecular number (Tb.N) (mm 1), trabecular thickness (Tb.Th.) (μm) and trabecular separation (Tb.Sp.) (μm) and (N) Femur trabecular bone volume, as shown by H&E staining in experimental mice.(O) Bar diagram represents quantification of H&E staining using Image J software. (P and Q) the representative bar graph shows the biomechanical quality of the femurs in mice. (P) Effect of NaHS on the ultimate load of the femur in three-point bending. (Q) The representative bar graph on the stiffness of femurs in three-point bending. Data are expressed as mean ± SEM. n = 7 mice per group. *p < 0.05 compared with the wild-type (WT) mice, #p < 0.05 compared with the HHcy mice.

3D reconstructions of μCT distal femur data revealed that bone volume fraction (BV/TV), trabecular number (Tb.N) and thickness (Tb.Th) were reduced in the trabecular bone of the distal femur of HHcy mice indicating osteoporosis (Fig. 7I–L). As expected with an osteoporotic phenotype, trabecular spacing (Tb.Sp) was also increased in HHcy condition (Fig. 7M). 2D histological evaluation of mouse distal femur also con rmed the osteoporosis with decreases in trabeculae and preservation of cortical bone in HHcy condition and NaHS recovers this effect (Fig. 7N, O). Moreover, HHcy mice showed a significant decrease in ultimate load and stiffness of the femur compared with WT control (Fig. 7P, Q). NaHS supplementation diminishes the HHcy induced detrimental effects in the bone. Collectively, these data indicates H₂S is required for normal bone homeostasis, and in its absence, mice develop osteoporosis (Fig. 7).

Discussion

In this study we report a molecular understanding of HHcy as a pathological biomarker for destructive bone disease, specifically in osteoporosis. Importantly, HHcy interferes with H₂S biosynthesis in BMMSCs in bone milieu by downregulating the key enzymes CBS and CSE in Hcy metabolism and thereby elevating the intracellular pool of Hcy (Fig. 1). The normal course of autoxidation of Hcy is exponentially increased, causing an imbalance in redox homeostasis by producing ROS (Fig. 2). Furthermore, the cellular oxidative mechanism works to activate RANKL synthesis in BMMSCs and causes osteoclast mediated bone loss via a paracrine manner. Pharmacological restoration of plasma H₂S prevents HHcy-induced osteoclastic bone loss by enhancing osteoblastic bone formation in vivo. Moreover, we have also demonstrated that administration of H₂S normalizes the HHcy induced downregulation of CBS and CSE key enzymes in BMMSCs population (Fig. 1).

The previous finding suggests that H₂S is a positive regulator in bone formation, which is also implicated in bone loss induced by ovariectomy in mice [35] or in H₂S-deficient CBS^+/− mice [28]. Consistant with earlier reports, we showed that diet induced HHcy regulates the BMMSCs level of CBS or CSE and decreases the plasma level of H₂S. Therefore, administration of H₂S to hyperhomocysteinemic mice, restores the plasma level of H₂S and maintains BMMSCs levels of CBS and CSE enzymes. This implies CBS and CSE levels to be essential in BMMSCs for maintenance of the physiological levels of H₂S, which mediates the redox homeostasis in bone, leading to recovery of pathological bone loss during HHcy.

RANKL and OPG are mostly expressed by osteoblasts, which regulate bone-remodeling events by coordinated regulation of osteoblastogenesis and osteoclastogenesis [48]. The results of the present study established, for the first time, that HMD induced HHcy exerts differential modulation of OPG and RANKL synthesis through phosphorylation c-Jun/JNK (T183/Y185) signaling. In this context, HHcy predominantly upregulates the osteoclastogenic factor; RANKL, in BMMSCs and in turn OPG production becomes inhibited, which is indispensable for bone formation (Fig. 3). However, inhibiting the JNK phosphorylation, by specific inhibitor SP600125 or by gene knockdown of JNK via siRNA against JNK in our mouse model, reversed the HHcy mediated effect on phosphorylation c-Jun/JNK (T183/Y185) signaling and OPG and RANKL production. In addition, we also found that administration of NaHS provided a protective effect from the HHcy prompted detrimental JNK signaling and RANKL expression induced-bone loss (Fig. 3). JNK is one of the three members from the mitogen-activated protein kinase (MAPK) family, and serves as a stress-activated protein kinase in response to ROS, UV radiation, cytokines, cytotoxic drugs, DNA-damaging agents and inflammatory cytokines [49, 50]. Activation occurs through a dual phosphorylation of threonine (Thr) and tyrosine (Tyr) residues, within a Thr-Pro-Tyr motif located in kinase subdomain VIII. The activated kinase subsequently phosphorylates c-Jun, a component of the AP-1 complex, and thereby activates cellular signaling. To confirm the involvement of JNK signaling in regulating HHcy enhanced oxidative stress mediated OPG and RANKL regulation, we administered cellular anti-oxidant, N-acetyl cysteine (NAC), to HHcy mice. The data confirmed that administration of NAC attenuated the HHcy mediated upregulation of JNK phosphorylation (T183/Y185), and subsequent decline in the RANKL expression as compared to HHcy mice alone. These data collectively link HHcy to OPG and RANKL regulation in the bone milieu via oxidant dependent JNK signaling (Fig. 3).

The increasing experimental evidence validating the notion that DNA methylation, especially occurring in CpG islands, plays an essential role in gene regulation and gene expression is undeniable [42, 43]. This was extensively reported in bone tissue and in human cancers [51, 52]. The recent report suggests that epigenetic mechanisms appear to be involved in the increased RANKL/OPG ratio of patients with hip fractures and hip osteoarthritis [53]. However, the potential role of epigenetic activation of DNA methylation in normal bone and in bone disease, in a murine model, is still unknown. A bioinformatics data search and analysis, revealed that at least a single CpG-rich region is present in both RANKL and OPG genes. Semi-quantitative qMSP confirmed the high degree of DNA methylation, associated with all the individual CpG sites that lead to gene expression suppression. Interestingly, our study clearly defined the differential DNA methylation pattern observed in the expression of RANKL and OPG in BMMSCs. Our data strongly signify that OPG production was significantly reduced through increased DNA hyper-methylation and epigenetic suppression of gene transcription in relation to exaggerated RANKL upregulation via epigenetic activation and DNA hypomethylation in the CpG regions of genes, in the HHcy condition. To study further evidence of DNA methylation, we tested both DNMT activity and global methylation. The protein western blot and DNMT activity assay by ELISA revealed that the overall enzyme activity was significantly upregulated in the HHcy condition. On the contrary, global methylation status (% of 5-mC level), of the genome was also higher, as confirmed by ELISA (Fig. 4). Furthermore, to acquire evidence explaining the role of DNA methylation during HHcy in the expression of RANKL and OPG, we administered DNA methylation inhibitor 5-azacytidine (5′-Aza) which specifically inhibits DNMTs in HHcy mice. The data implicated 5′-Aza as an effective repressor of gene methylation at the CpG sites and provided evidence that the 5′-Aza and NaHS treatment significantly decreased RANKL production in the HHcy condition as compared to WT (Fig. 4). Together, the results demonstrate that epigenetic DNA methylation was triggered during HHcy, leading to RANKL synthesis. The increased synthesis of RANKL facilitated osteoclast maturation via a paracrine dependent mechanism. A previous study showed that phospho-c-Jun, regulates DNMT1 expression and causes DNA hyper-methylation in low-grade gliomas [45]. Here, we showed that activation of Jun/JNK (T183/Y185) signaling is associated with increased methylation and consequent downregulation of OPG in BMMSCs. To further confirm the involvement of JNK signaling in regulating HHcy mediated DNMT1 promoter regulation, we completed a chromatin immunoprecipitation assay and showed that JNK physically binds to the DNMT1 promoter in the HHcy condition, as compared to WT. However, administration of NaHS reversed the HHcy mediated JNK signaling and its target DNMT1 promoter regulation (Fig. 4).

We have shown that activation of p-c-Jun/JNK was associated with increased DNA methylation by upregulating key DNMT enzymes (DNMT1 and DNMT3a) in the HHcy condition, which has been previously reported; partially, in human brain glioma [45, 54, 55]. Consistent with the earlier reported mechanism in the human brain glioma model, we showed that increased DNA methylation by p-c-Jun/JNK signaling was associated with increased RANKL hypomethylation (OPG downregulated), with activation of increased osteoclastogenesis in bone tissue. This mechanism was illustrated by DNMT expression, proving to be increased in the HHcy condition as compared to WT control.

The experimentation in the HHcy mice model, with the usage of cell-specific JNK inhibitor against the JNK activation, clearly suggested that the JNK signaling mechanism was integral to differential modulation of OPG/RANKL expression and secretion. It has been reported that RANKL induces osteoclastogenesis involving NF-Kb and JNK signaling [56, 57]. In this study, we also showed that BMMSCs derived CM containing RANKL secretion was poised to enhance osteoclastic activity via a paracrine mechanism. Furthermore, we showed that BMMSCs exaggerated RANKL secretion induced osteoclastogenesis by activating expression of key transcription factors; Nfatc1, Oc-Stamp, Oscar and its receptor RANK (Fig. 5). In this study, BMMSCs were used to examine the effect of NaHS on BMMSCs-derived osteogenesis in vitro. The administration of NaHS markedly increased ALP activity and bone mineralization, in vitro (Fig. 6). Both immunoblotting and qRT-PCR analysis, showed that NaHS is a potent inducer of osteogenic marker expression (Fig. 6). Our results also support the finding that NaHS supplementation normalized the level of BMD and improved trabecular microarchitecture (BV/TV, Tb.N, Tb.Th), in the HHcy condition as compared to WT control. In addition, bone mechanical properties (femur bending ultimate load and stiffness) also significantly improved under NaHS administration in the HHcy+NaHS mice group. In supplementary experiments, 2D histological analysis of bone by H&E staining, proved NaHS to efficiently recover the HHcy mediated increased bone marrow spacing and increased TRAP+ areas. This indicates that NaHS promoted osteogenesis and inhibited osteoclastogenesis. Furthermore, NaHS markedly increased cellular calcium levels in the HHcy+NaHS condition as compared to WT (Fig. 7).

In summary, HMD induced HHcy-mediated imbalance in redox homeostasis was a possible motive for the decline in OPG: RANKL expression. This study demonstrates that activated c-Jun/JNK signaling is mainly an indispensable mechanism for epigenetic DNA hyper-methylation of the OPG promoter, and in subsequent DNA hypomethylation of the RANKL promoter in BMMSCs in the HHcy condition. In this fashion, increased RANKL expression attributed to enhanced osteoclastogenesis in a paracrine dependent manner. Consequently, modulation of oxidative stress triggered in HHcy disturbed H₂S production. We established the novel gasotransmitter H₂S to be a potential therapeutic target to combat osteoclastic bone loss in the bone milieu. Our data repeatedly confirmed that NaHS exhibited potent abilities to treat the HMD induced HHcy mediated osteoporotic phenotype, through enhancing osteogenesis and inhibiting osteoclastogenesis (8).

Fig. 8.

Proposed mechanism for a role of H₂S on HHcy induced bone loss. First, HMD fed mice induce HHcy condition by altering normal Hcy metabolism and decreased H₂S production. Second, HMD enhanced HHcy condition further activates C-Jun/JNK-p signaling through oxidative stress. However, administration of NaHS (H₂S donor) or specific JNK inhibitor (SP600125) reverses this effects in HMD fed mice. Third, activated JNK-p further transcriptionally regulates DNMT1 expression through binding its promoter and, fourth, DNMT1 expression enhances OPG hyper-methylation, leads to BMMSCs-derived osteoblast dysfunction. Fifth, upregulation of RANKL during HHcy can accelerate osteoclastogenesis. Collectively, it causes bone loss during HHcy condition. Administration of DNMT inhibitor 5-Azacytidine reverses this changes.

Highlights.

• ✓ Diet-induced HHcy alters normal H₂S production.

• ✓ Administration of H₂S ameliorates Hcy-mediated c-Jun/JNK signaling.

• ✓ Administration of H₂S reverses HHcy diminished osteoblast differentiation.

• ✓ Pharmacological administration of H₂S improves bone formation.