A novel Keap1 inhibitor iKeap1 activates Nrf2 signaling and ameliorates hydrogen peroxide-induced oxidative injury and apoptosis in osteoblasts

An ultra-large structure-based virtual screening has discovered iKeap1 as a direct Keap1 inhibitor that can efficiently activate Nrf2 signaling. We here tested its potential effect against hydrogen peroxide (H2O2)-induced oxidative injury in osteoblasts. In primary murine and human osteoblasts, iKeap1 robustly activated Nrf2 signaling at micromole concentrations. iKeap1 disrupted Keap1-Nrf2 association, causing Nrf2 protein stabilization, cytosol accumulation and nuclear translocation in murine and human osteoblasts. The anti-oxidant response elements (ARE) activity and transcription of Nrf2-ARE-dependent genes (including HO1, NQO1 and GCLC) were increased as well. Significantly, iKeap1 pretreatment largely ameliorated H2O2-induced reactive oxygen species production, lipid peroxidation and DNA damage as well as cell apoptosis and programmed necrosis in osteoblasts. Moreover, dexamethasone- and nicotine-induced oxidative injury and apoptosis were alleviated by iKeap1. Importantly, Nrf2 shRNA or CRISPR/Cas9-induced Nrf2 knockout completely abolished iKeap1-induced osteoblast cytoprotection against H2O2. Conversely, CRISPR/Cas9-induced Keap1 knockout induced Nrf2 cascade activation and mimicked iKeap1-induced cytoprotective actions in murine osteoblasts. iKeap1 was ineffective against H2O2 in the Keap1-knockout murine osteoblasts. Collectively, iKeap1 activated Nrf2 signaling cascade to inhibit H2O2-induced oxidative injury and death of osteoblasts.

Thus, Nrf2 signaling activation, using genetic methods or pharmacological strategies, can efficiently protect osteoblasts/ osteoblastic cells from H 2 O 2 -induced oxidative injury [4,7,9,10,23]. A recent study by Gorgulla et al. has carried out an ultralarge structure-based virtual screening on computer clusters and identified a novel Keap1 inhibitor, iKeap1. It can bind to Keap1 with high affinity and block Keap1-Nrf2 association at submicromolar concentrations [24]. In the present study, we found that this novel Keap1 inhibitor activated Nrf2 signaling to inhibit H 2 O 2 -induced oxidative injury and death of osteoblasts.

Culture of primary murine and human osteoblasts
The trabecular bone fragments of healthy donors (undergoing preimplant bony reconstruction of the mandible) were minced into small pieces, washed with cold PBS, and then digested with 2 mg/mL collagenase type II (300 U/mg; Sigma) for 2 h. As reported [25], the primary human osteoblasts were then placed in culture flasks and cultured in DMEM nutrient mixture F-12 (DMEM/F12) supplemented with 10% Fetal Clone I (Hyclone; Thermo Fisher Scientific) and antibiotics, and incubated in a humidified air with 5% CO 2 at 37°C. The medium was changed twice a week until cells reached confluence. Written informed consent was obtained from each donor. The culture of primary murine osteoblasts was described in our previous study [4]. Primary osteoblasts were utilized at passage 3-10. The primary osteoblasts were subjected to mycoplasma and microbial contamination examination. STR profiling, population doubling time, and morphology were checked to confirm their genotypes. The protocols of using primary osteoblasts were approved by Ethics Board of Shanghai Ruijin Hospital, in according with the Declaration of Helsinki.

Quantitative real-time PCR (qRT-PCR)
Murine or human osteoblasts were seeded into six-well plates at 1.2 × 10 5 cells per well and were subjected to applied treatments. As reported previously [4], total cellular RNA was extracted using TRIzol reagents and was quantified. A SYBR Green PCR kit (Applied Biosystems, Suzhou, China) was employed to perform the qRT-PCR assays under an ABI Prism-7900H Fast Real-Time PCR system [26]. The melting curve analysis was always performed and a 2 −ΔΔCt method was utilized for data quantification. Glyceraldehyde-3-phosphatedehydrogenase(GAPDH) was always tested as the internal control and the reference gene. Primers utilized in this study were provided by Dr. Jiang at Nanjing Medical University [26,27].

ARE reporter activity
Osteoblasts were seeded into six-well plates at 1.2 × 10 5 cells per well and transduced with an ARE-inducible firefly luciferase vector (from Dr. Jiang [26] at Nanjing Medical University). After applied treatments, the relative ARE firefly luciferase activity was tested via quantification of the luminescence, and results were always normalized to control.

Western blotting
Murine or human osteoblasts were seeded into six-well plates at 1.2 × 10 5 cells per well and were subjected to treatments. Cell lysates were achieved by incubating cells with the cell lysis buffer (Beyotime Biotechnology, Wuxi, China). The nuclei isolation kit (from Sigma, Shanghai, China) was utilized to separate nuclear fraction lysates [28]. Western blotting procedures were described previously [26]. The quantification of the indicated protein bands was through an ImageJ software (from NIH). The same set of lysates were run in parallel gels when necessary.

Co-immunoprecipitation (Co-IP)
As reported previously [4], following the applied treatments total cell lysates (800 μg proteins per treatment) were pre-cleared and then incubated with an anti-Keap1 antibody (Santa Cruz Biotech, Shanghai, China) overnight. Proteins that were immunoprecipitated with Keap1 were captured by protein IgA/G beads, and were subsequently tested by western blotting. The detailed protocols of isolating mitochondrial fraction lysates and mito-IP were described before [29].

Cell viability
Murine or human osteoblasts were seeded into 96-well plates (at 3.5 × 10 3 cells per well) and were subjected to applied treatments. A CCK-8 assay kit was utilized to test cell viability according to the attached protocol. CCK-8 optical density (OD) was tested at 490 nm in each well.

NQO1 activity assay
The detailed protocols of analyzing NQO1 activity in osteoblasts were described elsewhere [21]. In brief, the inducer potency was quantified by the use of the NQO1 bioassay. Murine or human osteoblasts were seeded into 96-well plates (at 3.5 × 10 3 cells per well) and were subjected to applied treatments. The NQO1 enzyme activity was quantified in cell lysates using menadione as the substrate. Its value was always normalized to that in control osteoblasts.

Mitochondrial depolarization
JC-1 dye (Sigma) will accumulate in mitochondria in cells with mitochondrial depolarization to form monomers and will emit green fluorescence [32]. Murine or human osteoblasts were seeded into six-well plates at 0.8 × 10 5 cells per well and were subjected to applied treatments. Cells were incubated with JC-1 dye [33]. The JC-1 green fluorescence intensity (at 490 nm) was recorded and the representative JC-1 images were presented.

ROS detection
The detailed protocols were described in our previous study [4]. Murine or human osteoblasts were seeded into six-well plates at 0.8 × 10 5 cells per well and were subjected to applied treatments. Osteoblasts were then stained with CellROX (5 μM, Invitrogen-Thermo Fisher) for 30 min at room temperature. A fluorescent spectrophotometer was employed to examine CellROX fluorescence intensity. The representative CellROX images were presented as well.
was employed to quantitatively measure cellular lipid peroxidation levels using the described protocols [34,35]. Its level was always normalized to that of control.
Single strand DNA (ssDNA) ELISA. Murine or human osteoblasts were seeded into 96-well plates (at 3.0 × 10 3 cells per well) and were subjected to applied treatments. ssDNA contents were tested through an ApoS-trandTM ELISA kit (BIOMOL International, Plymouth Meeting, PA). The ssDNA ELISA absorbance was tested at 450 nm in each well.

Nrf2 short hairpin RNA (shRNA)
As reported early [4], the primary murine osteoblasts were seeded into sixwell plates at 0.8 × 10 5 cells per well in polybrene (2.5 μg/mL)-containing complete medium and were treated with the Nrf2 shRNA lentiviral particles (sc-37030V, Santa Cruz Biotech, Santa Cruz, CA). Afterward, osteoblasts were returned back to the complete medium, and puromycin (2.5 μg/mL) was added to select stable osteoblasts (for four passages). Nrf2 knockdown was verified by western blotting and qRT-PCR assays.

Statistical analysis
Quantitative data, all with normal distribution, were shown as mean ± standard deviation (SD). Statistical analyses between multiple groups were examined using ANOVA plus a Scheffe's f-test (SPSS 23.0, SPSS Co. Chicago, IL). To examine significance between two treatment groups, a two-tailed unpaired T test (Excel 2007) was employed. Values of P < 0.05 were considered as statistically significant.

RESULTS
iKeap1 activates Nrf2 cascade in murine and human osteoblasts First, we examined whether iKeap1 could activate Nrf2 signaling in osteoblasts. The primary murine osteoblasts (see our previous study [4]) were treated with iKeap1 at gradually increased concentrations, from 0.1-10 μM, and cultured for 6 h. Cellular ARE activity was tested and results showed that iKeap1 dosedependently increased ARE activity in murine osteoblasts (Fig. 1A). ARE activity increase was significant after 2.5-10 μM of iKeap1 treatment (P < 0.05 vs. vehicle control, Fig. 1A). In addition, iKeap1 , cells were further cultured for indicated time points, the relative ARE activity, NQO1 activity, and cell viability (CCK-8 OD) were tested (A); Keap1-Nrf2 association was tested by the coimmunoprecipitation (Co-IP) assays (B, H); Expression of listed proteins in cytosol fraction lysates and nuclear fraction lysates were examined by western blotting assays (C, E, G, I), with relative expression of listed mRNAs tested by qRT-PCR assays (F, J). Expressions of the listed proteins were quantified and normalized to the loading control. Quantified values were mean ± standard deviation (SD, n = 5). "C" stands for the untreated control cells. "Veh" stands for the vehicle control (0.1% DMSO). * P < 0.05 vs. "Veh" cells. Experiments were repeated five times, with similar results obtained.
iKeap1 ameliorates H 2 O 2 -induced ROS production and oxidative injury in murine and human osteoblasts We have previously shown that activation of Nrf2 cascade can protect osteoblasts from H 2 O 2 [4], we therefore analyzed whether iKeap1 could attenuate oxidative injury in osteoblasts. The primary murine osteoblasts were treated with H 2 O 2 (400 μM, 6 h) and the cellular ROS contents (CellROX intensity) were significantly increased ( Fig. 2A, B). Pretreatment with iKeap1 potently inhibited H 2 O 2 -induced ROS production in murine osteoblasts ( Fig. 2A, B). Quantitative analyses showed that 10 μM of iKeap1 was more potent than 2.5 μM in suppressing H 2 O 2 -induced CellROX intensity increase (Fig. 2B). Single treatment of iKeap1 failed to alter cellular ROS contents in murine osteoblasts ( Fig. 2A, B). To further support the anti-oxidant activity by the Keap1 inhibitor, we found that H 2 O 2 -induced lipid peroxidation was significantly inhibited after iKeap1 (2.5/10 μM) pretreatment (Fig. 2C). Lipid peroxidation in murine osteoblasts was quantified via TBAR activity assays (Fig. 2C).
We next tested whether iKeap1 could protect osteoblasts from these two stimuli. The CellROX fluorescence images, Fig. 4A, indicated that ROS intensity was significantly increased in primary murine osteoblasts after DEX (2 μM) and nicotine (1 μM) stimulation, which was largely attenuated with iKeap1 (10 μM) pretreatment (Fig. 4A, B). DEX-and nicotine-induced viability (CCK-8 OD) reduction (Fig. 4C), apoptosis activation (TUNEL-positive nuclei ratio increase, Fig. 4D), and cell necrosis (medium LDH Fig. 2 iKeap1 ameliorates H 2 O 2 -induced ROS production and oxidative injury in murine and human osteoblasts. The primary murine osteoblasts (A-F) or the primary human osteoblasts (G and H) were pretreated (for 2 h) with iKeap1 (2.5/10 μM), followed with or without H 2 O 2 (400 μM) stimulation, and cells were cultured for applied time periods; ROS contents (CellROX intensity assay, A, B, G), lipid peroxidation (TBAR activity, C), mitochondrial depolarization (JC-1 staining assays, D, E, and H) and DNA damage [single strand DNA (ssDNA) ELISA OD, F] were tested by the mentioned assays, and results were quantified and normalized. Quantified values were mean ± standard deviation (SD, n = 5). "C" stands for the untreated control cells. *P < 0.05 vs. "C" cells. # P < 0.05 vs. cells with H 2 O 2 stimulation but "Veh" pretreatment. Experiments were repeated five times, with similar results obtained. Scale bar = 100 μm (A and D). Caspase-3 activity was tested (B and I); Expression of apoptosis-associated proteins was tested by western blotting assays (C); Cell apoptosis was examined by nuclear TUNEL staining assays (D and J) and Annexin V FACS (E and K) assays, and results were quantified. Mitochondrial CyPD-ANT1-p53 association and their expressions were shown (F and L), and cell necrosis examined by quantifying medium LDH release (G and M). Expressions of the listed proteins were quantified and normalized to the loading control. Quantified values were mean ± standard deviation (SD, n = 5). "C" stands for the untreated control cells. *P < 0.05 vs. "C" cells. # P < 0.05 vs. cells with H 2 O 2 stimulation but "Veh" pretreatment. Experiments were repeated five times, with similar results obtained.

Nrf2 activation is absolutely required for iKeap1-induced osteoblast cytoprotection against H 2 O 2
To test whether Nrf2 cascade activation is required for iKeap1induced osteoblast cytoprotection against H 2 O 2 , we utilized genetic strategies to silence Nrf2. As reported early [4] primary murine osteoblasts were transfected with Nrf2 shRNA lentiviral particles and stable osteoblasts were established via selection by puromycin. These cells were named as "sh-Nrf2" osteoblasts. Alternatively, a CRISPR/Cas9-Nrf2-KO-GFP-puro construct (see our previous study [4]) was transduced to murine osteoblasts, and single stable "ko-Nrf2" osteoblasts were established by FACS-mediated sorting and Nrf2-KO screening. As compared to the control osteoblasts with scramble control shRNA plus the CRISPR/Cas9 empty vector ("shC+Cas9-C"), Nrf2 mRNA levels were almost depleted in sh-Nrf2 osteoblasts and ko-Nrf2 osteoblasts (Fig. 5A), where Keap1 mRNA levels were unchanged (Fig. 5A).

DISCUSSION
Studies have shown that Nrf2 activation using various pharmacological agents could protect osteoblasts/osteoblastic cells from H 2 O 2 -induced oxidative injury and cell death [7,9,10]. However, most of these agents are not direct Nrf2 activators and are often utilized at relatively higher concentrations [4,10,[52][53][54]. A very recent study has identified iKeap1 as a novel and highly efficient Nrf2 activator [24]. iKeap1 is engaged in the Nrf2-binding pocket of Keap1 and locates at the entrance to the tunnel formed by the β-barrel [24]. Nuclear magnetic resonance analyses confirmed a direct binding between iKeap1 and Keap1 [24]. Surface plasmon resonance (SPR) studies demonstrate that iKeap1 binds to Keap1 at a binding affinity of 114 nM [24]. A fluorescence polarization assay results found that iKeap1 can displace Nrf2 peptide at the IC 50 of 258 nM [24]. Its potential effect on H 2 O 2 -induced osteoblast injury was studied here.
In human and murine osteoblasts, iKeap1 activated Nrf2 cascade signaling, causing Keap1-Nrf2 disassociation, Nrf2 protein stabilization, cytosol accumulation, and nuclear translocation. In addition, the novel Keap1 inhibitor increased ARE activity and expression of Nrf2-dependent genes (HQ1, NQO1, and GCLC) in murine and human osteoblasts. Functional studies showed that iKeap1 largely attenuated H 2 O 2 -induced ROS accumulation, lipid peroxidation, mitochondrial depolarization, and DNA breaks in murine and human osteoblasts. Furthermore, H 2 O 2 -induced osteoblast apoptosis was significantly ameliorated after iKeap1 pretreatment.
Besides apoptosis, H 2 O 2 could promote p53 translocation to mitochondria to form the CyPD-ANT1-p53 complex, leading to mPTP opening and cell necrosis [4,41,42]. Inhibition of this process can exert significant osteoblastic cytoprotection [4,41,42]. Here we found that H 2 O 2 -induced CyPD-ANT1-p53 association and necrosis were largely inhibited by iKeap1. These results further explained the superior osteoblast cytoprotective activity by the novel Keap1 inhibitor.
Existing studies have reported that nicotine exposure could inhibit proliferation, differentiation, alkaline phosphatase activity, oxidative metabolism, and collagen synthesis as well as calcium absorption and mineralized nodule formation in osteoblastic cells and osteoblasts [49][50][51]. Furthermore, nicotine is shown to induce human osteoblast apoptosis, which could also be linked to cigarette smoke-induced osteoporosis and dental implant failure [49][50][51]. Marinucci et al. found that nicotine-induced osteoblast cell apoptosis was driven by H 2 O 2 -induced glyoxalase 1 inhibition and MG-H1 accumulation [49]. Liang et al. reported that nicotine altered genes and signaling pathways associated with bone formation in the rat osteoblasts, and induced osteoblast cell apoptosis [50]. Ma et al. reported that nicotine decreased expression of osteogenic and angiogenic genes, including TGFβ, BMP-2, PDGF-AA, and VEGF, in primary rat osteoblasts [51]. Here we found that nicotine-induced oxidative injury and apoptosis in murine and human osteoblasts were potently alleviated after iKeap1 pretreatment. Therefore, this novel Keap1 inhibitor might have important translational value for the treatment of nicotineassociated osteoblast injury.

CONCLUSION
Taken together, iKeap1 activated Nrf2 signaling cascade to inhibit H 2 O 2 -induced oxidative injury and osteoblast death. This novel Nrf2 activator could be a novel strategy to protect osteoblasts from various oxidative stimuli.