Prolonged Cadmium Exposure and Osteoclastogenesis: A Mechanistic Mouse and in Vitro Study

Abstract

Background: Cadmium (Cd) is a highly toxic and widespread environmental oxidative stressor that causes a myriad of health problems, including osteoporosis and bone damage. Although nuclear factor erythroid 2-related factor 2 (NRF2) and its Cap 'n' Collar and basic region Leucine Zipper (CNC-bZIP) family member nuclear factor erythroid 2-related factor 1 (NRF1) coordinate various stress responses by regulating the transcription of a variety of antioxidant and cytoprotective genes, they play distinct roles in bone metabolism and remodeling. However, the precise roles of both transcription factors in bone loss induced by prolonged Cd exposure remain unclear.

Objectives: We aimed to understand the molecular mechanisms underlying Cd-induced bone loss, focusing mainly on the roles of NRF2 and NRF1 in osteoclastogenesis provoked by Cd.

Methods: Male wild-type (WT), global Nrf2-knockout ($Nrf{2}^{-/-}$) and myeloid-specific Nrf2 knockout [Nrf2(M)-KO] mice were administered Cd (50 or $100\text{ppm}$) via drinking water for 8 or 16 wk, followed by micro-computed tomography, histological analyses, and plasma biochemical testing. Osteoclastogenesis was evaluated using bone marrow-derived osteoclast progenitor cells (BM-OPCs) and RAW 264.7 cells in the presence of Cd (10 or $20\text{nM}$) with a combination of genetic and chemical modulations targeting NRF2 and NRF1.

Results: Compared with relevant control mice, global $Nrf{2}^{-/-}$ or Nrf2(M)-KO mice showed exacerbated bone loss and augmented osteoclast activity following exposure to $100\text{ppm}$ Cd in drinking water for up to 16 wk. In vitro osteoclastogenic analyses suggested that Nrf2-deficient BM-OPCs and RAW 264.7 cells responded more robustly to low levels of Cd (up to $20\text{nM}$) with regard to osteoclast differentiation compared with WT cells. Further mechanistic studies supported a compensatory up-regulation of long isoform of NRF1 (L-NRF1) and subsequent induction of nuclear factor of activated T cells, cytoplasmic, calcineurin dependent 1 (NFATc1) as the key molecular events in the Nrf2 deficiency-worsened and Cd-provoked osteoclastogenesis. L-Nrf1 silenced (via lentiviral means) Nrf2-knockdown (KD) RAW cells exposed to Cd showed dramatically different NFATc1 and subsequent osteoclastogenesis outcomes compared with the cells of Nrf2-KD alone exposed to Cd, suggesting a mitigating effect of the Nrf1 silencing. In addition, suppression of reactive oxygen species by exogenous antioxidants $N$-acetyl-l-cysteine ($2\text{mM}$) and mitoquinone mesylate (MitoQ; $0.2\mu M$) mitigated the L-NRF1-associated effects on NFATc1-driven osteoclastogenesis outcomes in Cd-exposed Nrf2-KD cells.

Conclusions: This in vivo and in vitro study supported the authors' hypothesis that Cd exposure caused bone loss, in which NRF2 and L-NRF1 responded to Cd and osteoclastogenic stimuli in a cooperative, but contradictive, manner to coordinate Nfatc1 expression, osteoclastogenesis and thus bone homeostasis. Our study suggests a novel strategy targeting NRF2 and L-NRF1 to prevent and treat the bone toxicity of Cd. https://doi.org/10.1289/EHP13849.

Figure 1.

The effects of cadmium (Cd) exposure on bone metabolic homeostasis in global Nrf2 knockout ($Nrf{\mathit{2}}^{-/-}$) and WT littermate ($Nrf{\mathit{2}}^{+/+}$) control mice. (A) Schematic illustration of the experimental design. Male mice (14 wk old) were exposed to cadmium chloride (${\text{CdCl}}_2$; $100\mathrm{mg}$ Cd/L) in drinking water for 16 wk. (B) Representative images of the left femur of mice examined by micro-CT. Scale bar: $0.5\mathrm{mm}$. (C) Bone density estimations, including bone volume per tissue volume (BV/TV), trabecular separation (Tb.Sp), trabecular thickness (Tb.Th), trabecular number (Tb.N), total area (Tt.Ar), cortical area (Ct.Ar), structure thickness (St.Th) and cortical area per total area (Ct.Ar/Tt.Ar), were calculated using the Skyscan CT-Analyser software (version 1.1.7; Skyscan CTAn) based on micro-CT measurements. $n=6–9$. Values are expressed as $\text{mean}\pm \text{SD}$. Two-way ANOVA with Bonferroni multiple comparison posttest was used. *$p<0.05$ vs. $Nrf{\mathit{2}}^{+/+}$ with the same treatment; ^#$p<0.05$ vs. control (Cont) of the same genotype. (D) Results of the three-point bending test performed on the left femurs. The maximum load, breaking load, energy absorption, and stiffness were calculated using Bluehill Elements. $n=4–5$. Values are expressed as $\text{mean}\pm \text{SD}$. Two-way ANOVA with Bonferroni multiple comparison posttest was used for the maximum load and breaking load; Kruskal–Wallis test with Dunn’s multiple comparisons posttest was used for the energy absorption and stiffness. *$p<0.05$ vs. $Nrf{\mathit{2}}^{+/+}$ with the same treatment. ^#$p<0.05$ vs. Cont of the same genotype. (E) Plasma levels of TRAP5b and CTX-I. $n=5–6$. Values are expressed as $\text{mean}\pm \text{SD}$. Two-way ANOVA with Bonferroni multiple comparison posttest was used. *$p<0.05$ vs. $Nrf{\mathit{2}}^{+/+}$ with the same treatment; ^#$p<0.05$ vs. Cont of the same genotype. (F) Representative histological images of TRAP staining of the right femur of mice. Black arrow, osteoclasts. Scale bar: $1\mathrm{mm}$. (G) Osteoclast number (OcN) and osteoclast surface (OcS) in the femur sections with TRAP staining normalized to bone perimeter (Bpm) and bone surface (BS), respectively. $n=4–6$. Values are expressed as $\text{mean}\pm \text{SD}$. Two-way ANOVA with Bonferroni multiple comparison posttest was used. *$p<0.05$ vs. $Nrf{\mathit{2}}^{+/+}$ with the same treatment; ^#$p<0.05$ vs. Cont of the same genotype. Summary data are provided in Tables S6–S21. Note: ANOVA, analysis of variance; CTX-I, C-terminal telopeptides of type I; micro-CT, micro-computed tomography; Nrf2, nuclear factor erythroid 2-related factor 2; SD, standard deviation; Trap, tartrate-resistant acid phosphatase; W, weeks; WT, wild type.

Figure 2.

The impacts of cadmium (Cd) exposure on bone metabolic homeostasis in myeloid-specific Nrf2 knockout [Nrf2(M)-KO] and their littermate control (Cont) mice. (A) Schematic illustration of the experimental design. Male mice (14 wk old) were exposed to cadmium chloride (${\text{CdCl}}_2$; $100\mathrm{mg}$ Cd/L) in drinking water for 8 or 16 wk. (B) Representative images of the left femur of mice examined by micro-CT. Scale bar: $0.5\mathrm{mm}$. (C) Bone density estimations, including bone volume per tissue volume (BV/TV), trabecular separation (Tb.Sp), trabecular thickness (Tb.Th), trabecular number (Tb.N), total area (Tt.Ar), cortical area (Ct.Ar), structure thickness (St.Th), and cortical area per total area (Ct.Ar/Tt.Ar), were calculated using the Skyscan CT-Analyser software (version 1.1.7; Skyscan CTAn) based on micro-CT measurements. $n=5–9$. Values are expressed as $\text{mean}\pm \text{SD}$. Two-way ANOVA with Bonferroni multiple comparison posttest was used for BV/TV, Tb.Sp, Tb.Th, and Tb.N; Kruskal–Wallis test with Dunn’s multiple comparisons posttest was used for Tt.Ar, Ct.Ar, St.Th, and Ct.Ar/Tt.Ar. *$p<0.05$ vs. KI with the same treatment; ^#$p<0.05$ vs. Cont of the same genotype. (D) Representative histological images of TRAP staining of the right femur of mice. Black arrow, osteoclasts. Scale bar: $1\mathrm{mm}$. (E) Osteoclast number (OcN) and osteoclast surface (OcS) in the femur sections with TRAP staining normalized to bone perimeter (Bpm) and bone surface (BS), respectively. $n=4–5$. Values are expressed as $\text{mean}\pm \text{SD}$. Two-way ANOVA with Bonferroni multiple comparison posttest was used. *$p<0.05$ vs. KI with the same treatment; ^#$p<0.05$ vs. Cont of the same genotype. (F–H) Plasma levels of TRAP5b and CTX-I (F), P1NP and OCN (G), and RANKL and OPG (H). $n=5–7$. Values are expressed as $\text{mean}\pm \text{SD}$. Two-way ANOVA with Bonferroni multiple comparison posttest was used for OcN/Bpm, OcS/BS, TRAP5b, CTX-I, P1NP, and OPG; Kruskal–Wallis test with Dunn’s multiple comparisons posttest was used for OCN and RANKL. *$p<0.05$ vs. KI with the same treatment; ^#$p<0.05$ vs. Cont of the same genotype. Summary data are provided in Tables S22–S37. Note: ANOVA, analysis of variance; CTX-I, C-terminal telopeptides of type I; KI, $Nfe\mathit{2}l{\mathit{2}}^{\text{LoxP}/\text{LoxP}}$; KO, Nrf2(M)-KO (knockout); micro-CT, micro-computed tomography; Nrf2, nuclear factor erythroid 2-related factor 2; OCN, osteocalcin; OPG, osteoprotegerin; P1NP, procollagen type I N-terminal propeptide; RANKL, receptor activator of nuclear factor kappa-B ligand; SD, standard deviation; Trap, tartrate-resistant acid phosphatase; W, week.

Figure 3.

The effects of low-level cadmium (Cd) exposure on osteoclastogenesis in BM-OPCs and RAW 264.7 cells with genetic manipulation of Nrf2 expression. (A,G,M) Representative images of BM-OPCs (A) and RAW 264.7 cells with Nrf2 silencing (G) and Nrf2 overexpression (M) treated with RANKL ($50\text{\hspace{0.17em}ng}/\mathrm{mL}$) and M-CSF ($30\text{\hspace{0.17em}ng}/\mathrm{mL}$) for 7 d in the presence of vehicle (Veh), 10 or $20\text{\hspace{0.17em}nM}$ Cd, followed by TRAP staining. BM-OPCs were isolated from $Nrf{\mathit{2}}^{+/+}$ and $Nrf{\mathit{2}}^{-/-}$ mice and cultured as detailed in the “Methods” section. Scale bars: $200\mathrm{\mu }\mathrm{m}$. (B,H,N) TRAP-positive multinucleated cells containing three or more nuclei counted in the images. $n=3–6$. Values are expressed as $\text{mean}\pm \text{SD}$. Two-way ANOVA with Bonferroni multiple comparison posttest was used. *$p<0.05$ vs. $Nrf{\mathit{2}}^{+/+}$, Scr or Cont with the same treatments; ^#$p<0.05$ vs. Veh of the same cytotype. (C–F,I–L,O–R) mRNA expression of Cathepsin K, $H^{+}\text{-}atpase$, and Atp6v0d2, as well as protein expression of NFATc1 in the cells following the osteoclastogenic treatments. $n=3–6$. Values are expressed as $\text{mean}\pm \text{SD}$. Two-way ANOVA with Bonferroni multiple comparison posttest was used. *$p<0.05$ vs. $Nrf{\mathit{2}}^{+/+}$, Scr or Cont with the same treatments; ^#$p<0.05$ vs. Veh of the same cytotype. Summary data are provided in Tables S38–S49. Quantifications of (F), (L), and (R) are found in Figure S7. Note: ANOVA, analysis of variance; Atp6v0d2, ATPase H positive Transporting V 0 Subunit D 2; BM-OPCs, bone marrow-derived osteoclast progenitor cells; Cont, control; KD, Nrf2-knockdown (Nrf2-KD); M-CSF, macrophage colony-stimulating factor; MNC, multinucleated cells; NFATc1, nuclear factor of activated T cells, cytoplasmic, calcineurin dependent 1; Nrf2, nuclear factor erythroid 2-related factor 2; OE, overexpression; RANKL, receptor activator of nuclear factor kappa-B ligand; RAW 264.7, mouse leukemic monocyte/macrophage cells; Scr, Scramble; SD, standard deviation; Trap, tartrate-resistant acid phosphatase.

Figure 4.

Genetic manipulation of L-Nrf1 expression in Nrf2-deficient RAW 264.7 cells and osteoclast differentiation following cadmium exposure. The (A) mRNA (A) and (B) protein levels of L-NRF1 in RAW cells with lentiviral Nrf2 silencing treated with RANKL ($50\text{\hspace{0.17em}ng}/\mathrm{mL}$) and M-CSF ($30\text{\hspace{0.17em}ng}/\mathrm{mL}$) for indicated days in the presence of $20\text{\hspace{0.17em}nM}$ Cd. $n=3–9$. Values are expressed as $\text{mean}\pm \text{SD}$. Two-way ANOVA with Bonferroni multiple comparison posttest was used. *$p<0.05$ vs. Scr with the same treatment; ^#$p<0.05$ vs. the same genotype of D0. (C,D,I,J) Representative images (C) and (I) and quantifications (D) and (J) of TRAP staining of RAW cells with Nrf2-KD and L-Nrf1 knockdown or overexpression treated with RANKL ($50\text{\hspace{0.17em}ng}/\mathrm{mL}$) and M-CSF ($30\text{\hspace{0.17em}ng}/\mathrm{mL}$) for 5 d in the presence (i.e., Cd) or absence (i.e., Veh) of $20\text{\hspace{0.17em}nM}$ Cd. For quantification, TRAP-positive multinucleated cells containing three or more nuclei on the images were counted. $n=3–4$. Values are expressed as $\text{mean}\pm \text{SD}$. Two-way ANOVA with Bonferroni multiple comparison posttest was used. *$p<0.05$ vs. the same genotype with Veh; ^#$p<0.05$ vs. Scr with the same treatment; ^&$p<0.05$ vs. KD with the same treatment. (E–H, K–N) The mRNA expression of Cathepsin K, $H^{+}\text{-}atpase$, and Atp6v0d2 (E–G) and (K-M), as well as the protein levels of NFATc1 (H) and (N) were measured by RT-qPCR and immunoblotting, respectively, in the cells treated as detailed in (C), (D), (I), and (J). $n=3–4$. Values are expressed as $\text{mean}\pm \text{SD}$. Two-way ANOVA with Bonferroni multiple comparison posttest was used. *$p<0.05$ vs. the same genotype with Veh; ^#$p<0.05$ vs. Scr with the same treatment; ^&$p<0.05$ vs. KD with the same treatment. Summary data are provided in Tables S50–S58. Quantifications of (B), (H), and (N) are found in Figure S10. Note: ANOVA, analysis of variance; Atp6v0d2, ATPase H positive Transporting V 0 Subunit D 2; D0–D5, the day following osteoclastogenic treatments; DKD, Nrf2 and L-Nrf1 double knockdown; KD, Nrf2-knockdown (Nrf2-KD); KD+OE, Nrf2-KD and L-NRF1-741 overexpression; L-NRF1, long isoform nuclear factor-erythroid 2-related factor 1; M-CSF, macrophage colony-stimulating factor; MNC, multinucleated cells; NFATc1, nuclear factor of activated T cells, cytoplasmic, calcineurin dependent 1; Nrf2, nuclear factor erythroid 2-related factor 2; RANKL, receptor activator of nuclear factor kappa-B ligand; RAW 264.7, mouse leukemic monocyte/macrophage cells; RT-qPCR, reverse transcription quantitative polymerase chain reaction; Scr, Scramble; SD, standard deviation; Trap, tartrate-resistant acid phosphatase; Veh, vehicle.

Figure 5.

Effects of $N$-acetyl-l-cysteine (NAC) and mitoquinone mesylate (MitoQ) treatment on the protein levels of L-NRF1 and NFATc1 and osteoclast differentiation in Nrf2-KD and Scr RAW 264.7 cells. The cells were treated with RANKL ($50\text{\hspace{0.17em}ng}/\mathrm{mL}$) and M-CSF ($30\text{\hspace{0.17em}ng}/\mathrm{mL}$) for 5 d in the presence of $20\text{\hspace{0.17em}nM}$ cadmium (Cd) with NAC ($2\text{\hspace{0.17em}mM}$) or MitoQ ($0.2\mathrm{\mu }\mathrm{M}$). (A,G) Representative images of immunoblotting of L-NRF1 (upper panels) and NFATc1 (lower panels). (B,C,H,I) Representative images of TRAP staining (B) and (H) and the quantification of TRAP-positive multinucleated cells containing three or more nuclei (C) and (I). $n=3–5$. Values are expressed as $\text{mean}\pm \text{SD}$. Two-way ANOVA with Bonferroni multiple comparison posttest was used. *$p<0.05$ vs. Scr with the same treatment; ^#$p<0.05$ vs. the same genotype with Veh; ^&$p<0.05$ vs. the same genotype with Cd alone. (D–F,J–L) mRNA expression of Cathepsin K, $H^{+}\text{-}atpase$, and Atp6v0d2. $n=3–4$. Values are expressed as $\text{mean}\pm \text{SD}$. Two-way ANOVA with Bonferroni multiple comparison posttest was used, *$p<0.05$ vs. Scr with the same treatment; ^#$p<0.05$ vs. the same genotype with Veh; ^&$p<0.05$ vs. the same genotype with Cd alone. Summary data are provided in Tables S59–S66. Quantification of (A) and (G) is found in Figure S16. Note: ANOVA, analysis of variance; Atp6v0d2, ATPase H positive Transporting V 0 Subunit D 2; Cd, Cd alone; Cd+MitoQ, Cd and MitoQ co-treatment; Cd+NAC, Cd and NAC co-treatment; KD, Nrf2-knockdown (Nrf2-KD); L-NRF1, long isoform nuclear factor-erythroid 2-related factor 1; M-CSF, macrophage colony-stimulating factor; MitoQ, MitoQ alone; NAC, NAC alone; NFATc1, nuclear factor of activated T cells, cytoplasmic, calcineurin dependent 1; RANKL, receptor activator of nuclear factor kappa-B ligand; Scr, Scramble; SD, standard deviation; Trap, tartrate-resistant acid phosphatase; Veh, vehicle.

Figure 6.

A schematic depiction of the potential molecular mechanisms underlying the cadmium (Cd) exposure-augmented osteoclastogenesis. NFATc1, which is a direct downstream transcriptional target of L-NRF1 and is also influenced by intracellular ROS levels, is a key driver controlling the osteoclastogenesis signaling cascade. NRF2, as a master transcription factor regulating many antioxidant genes, is involved in the regulation of osteoclast differentiation via the NRF2–antioxidants–ROS negative feedback loop. Cd exposure may stimulate intracellular ROS production via a complex mechanism and thus aggravate osteoclast differentiation, which can be mitigated by the NRF2-mediated antioxidant response. As a result, deficiency of Nrf2 augments Cd exposure-stimulated osteoclast differentiation in a ROS-dependent manner, in which ROS may up-regulate NFATc1 via L-NRF1-dependent and independent mechanisms. Note: Atp6v0d2, ATPase H positive Transporting V 0 Subunit D 2; KD, Nrf2-knockdown (Nrf2-KD); KO, myeloid-specific Nrf2 knockout [Nrf2(M)-KO]; L-NRF1, long isoform nuclear factor-erythroid 2-related factor 1; NFATc1, nuclear factor of activated T cells, cytoplasmic, calcineurin dependent 1; Nrf2, nuclear factor erythroid 2-related factor 2; RANK, receptor activator of nuclear factor kappa-B; ROS, reactive oxygen species; TRAF6, tumor necrosis factor receptor–associated factor 6; WT, wild type.