Quercetin as an Agent for Protecting the Bone: A Review of the Current Evidence

Quercetin is a flavonoid abundantly found in fruits and vegetables. It possesses a wide spectrum of biological activities, thus suggesting a role in disease prevention and health promotion. The present review aimed to uncover the bone-sparing effects of quercetin and its mechanism of action. Animal studies have found that the action of quercetin on bone is largely protective, with a small number of studies reporting negative outcomes. Quercetin was shown to inhibit RANKL-mediated osteoclastogenesis, osteoblast apoptosis, oxidative stress and inflammatory response while promoting osteogenesis, angiogenesis, antioxidant expression, adipocyte apoptosis and osteoclast apoptosis. The possible underlying mechanisms involved are regulation of Wnt, NF-κB, Nrf2, SMAD-dependent, and intrinsic and extrinsic apoptotic pathways. On the other hand, quercetin was shown to exert complex and competing actions on the MAPK signalling pathway to orchestrate bone metabolism, resulting in both stimulatory and inhibitory effects on bone in parallel. The overall interaction is believed to result in a positive effect on bone. Considering the important contributions of quercetin in regulating bone homeostasis, it may be considered an economical and promising agent for improving bone health. The documented preclinical findings await further validation from human clinical trials.


Introduction
The skeletal system is metabolically active and capable of continuous self-renewal in response to biomechanical use, microdamage and fracture [1,2]. The homeostasis of the skeletal system is maintained via concerted communication between bone-building osteoblasts, bone-degrading osteoclasts and mechanosensory osteocytes through a process called bone remodelling. Perturbation of the equilibrium between bone resorption and formation (in the condition of excessive removal of mineralised bone by osteoclasts without the corresponding amount of newly-formed bone by osteoblasts) results in bone loss and delayed bone regeneration [3]. The long-term progression of bone-related disorders generates significant personal impacts and a growing socioeconomic burden due to huge medical costs, sustained disability and reduced quality of life.
The conventional treatment options for osteopenia and osteoporosis consist of antiresorptive agents (such as bisphosphonates, hormone-replacement therapy, selective oestrogen-receptor modulators and anti-RANKL antibodies) and/or anabolic agents (such as intermittent low doses of teriparatide and antisclerostin antibodies) [4,5]. However, these treatment strategies are not free from adverse effects that limit their use. For instance, the use of bisphosphonates and anti-RANKL antibodies is closely associated with gastrointestinal tract symptoms [6]. Hormone-replacement therapy and selective oestrogen-receptor modulators can cause increased risk for venous thromboembolism and cardiovascular events [6,7]. Treatments administered as injections may cause injection-site reactions [5,8].
A prolonged treatment period and high administration dose of teriparatide has been reported to increase the incidence of bone neoplasms [9]. The concerns linked to pharmacological sclerostin inhibition (e.g., romosozumab) include the probability of bone overgrowth and skeletal deformities [5]. Thus, natural compounds with potent bone-conserving properties and fewer side effects could present alternative strategies to overcome the shortcomings of existing therapies.
In this review, the authors provide a comprehensive summary of literature related to the effects of quercetin on bone in animal models. In particular, the underlying cellular and molecular mechanisms of quercetin that can modulate skeletal homeostasis and protect the bones are discussed.

In Vivo Evidence on the Effects of Quercetin on Bone
The bone-conserving effects of quercetin have been extensively established using animal models of osteopenia induced by sex-hormone deficiency, streptozotocin (STZ), glucocorticoids, bile-duct ligation, retinoic acid and zinc oxide nanoparticles. Animals with osteolysis, bone defects and healthy animals were also used to examine the protective effects of quercetin on bone (Table 1). Table 1. The effects of quercetin on bone in in vivo studies. Symbol ↑ indicates an increase or upregulation, ↓ indicates a decrease or downregulation and ↔ indicates no change.  Female rats subjected to bilateral ovariectomy are the most commonly used osteoporotic animal model [54]. Yuan et al. (2018) reported that eight weeks of oral quercetin (50 mg/kg/day) administration to ovariectomised rats increased the bone mineral density (BMD), trabecular number (Tb.N) and trabecular thickness (Tb.Th) of rat femoral metaphysis. Three-point flexural testing indicated increases in elastic and maximum radial degrees as well as elastic and maximum loads of rat femur [29]. Another group of researchers demonstrated that oral supplementation of quercetin (50 mg/kg/day) for 30 days reduced acid phosphatase (ACP) and alkaline phosphatase (ALP) activities, but increased serum calcium and phosphorus levels [30]. In another study, quercetin (2.5%) was added into the diets of female C57BL/6J mice for four weeks after undergoing bilateral ovariectomy. The findings obtained from this study were increases in total lumbar BMD, total femur BMD, cortical bone area (Cr.Ar), cortical thickness (Cr.Th), section modulus, bone volume/total volume (BV/TV), Tb.N, Tb.Th, osteoid volume, osteoid surface and decrease in trabecular separation (Tb.Sp) [31], which indicated that quercetin improved bone density.
Siddiqui and colleagues assessed the bone-sparing effects of quercetin and its structurally related compound, quercetin-6-C-β-d-glucopyranoside, in ovariectomised animals. In the first study, adult female rats were given oral 5 mg/kg/day quercetin or quercetin-6-C-β-d-glucopyranoside for 12 weeks after bilateral ovariectomy [32]. Quercetin-6-C-β-d-glucopyranoside raised BMD level at femur epiphysis, proximal tibia, lumbar vertebra 4 (L4), femur diaphysis and tibia-fibula separating point (TFSP). All trabecular microarchitectural parameters of femur epiphysis and proximal tibia were improved after treatment with quercetin-6-C-β-d-glucopyranoside, characterised by higher BV/TV, Tb.Th and Tb.N and lower Tb.Sp. Cortical parameters of femur and tibia were also improved, as shown by increased Cr.Ar, Cr.Th, periosteal area (T.Ar), periosteal perimeter (T.Pm), bone perimeter (B.Pm) and endosteal perimeter (E.Pm). Bone biomechanical strength was increased while bone turnover markers (osteocalcin and C-terminal telopeptide of type 1 collagen (CTX)) were reduced in ovariectomised rats supplemented with quercetin-6-C-β-d-glucopyranoside. In contrast, treatment with quercetin resulted in partial maintenance in all these measured parameters [32]. Quercetin and its derivatives differ in the attachment of sugar moieties that affect the pharmacokinetic profile [55]. Hence, a possible explanation for the greater bone-protecting efficacy of quercetin-6-C-β-d-glucopyranoside may be that it has better bioavailability and bioactivity as compared to free quercetin. In another study, the same group of researchers administered 5 or 10 mg/kg/day quercetin-6-C-β-d-glucopyranoside to ovariectomised rats for 12 weeks [33]. An improvement of bone microstructure at the proximal tibia, diaphysis and epiphysis of the femur was detected. The observation was evidenced by increased BV/TV, Tb.N, Tb.Th, connectivity density (Conn.D), mineral apposition rate (MAR) and bone formation rate (BFR), but decreased Tb.Sp, trabecular pattern factor (Tb.Pf) and structure model index (SMI) [33].
Bone-sparing effects were also observed in the ovariectomised animal model when quercetin was incorporated into nanoparticles or hydroxyapatite bioceramic microspheres. The application of nanoparticles and microspheres has received great attention for their efficient enhancement of bioavailability and bioactivity by prolonging half-life, managing drug particle size and controlling drug release [56,57]. Female albino rats underwent bilateral ovariectomy and were orally treated with 10 or 50 mg/kg quercetin-loaded phytosome nanoparticles for 30 days. The treatment significantly decreased bone turnover markers and increased mineral levels in the sera of the animals [30]. In another study, hydroxyapatite bioceramic microspheres loaded with quercetin were fabricated. It was implanted into a bone defect created at the distal region of the femur diaphysis in ovariectomised rats. It was noted that the BMD and Tb.Th of the bone defects were improved after eight weeks of implantation [34].
On the other hand, the addition of phytochemical blends into animals' basal diet failed to exert anti-osteoporotic effects in ovariectomy-induced bone loss in aged retired-breeder Fischer 344 rats. Ambati and coresearchers used two types of phytochemical blends containing (a) 1000 mg/kg diet quercetin, 2400 IU/kg diet vitamin D 3 , 500 mg/kg diet genistein and 200 mg/kg diet resveratrol, or (b) 2000 mg/kg diet quercetin, 2400 IU/kg diet vitamin D 3 , 1000 mg/kg diet genistein and 400 mg/kg diet resveratrol. No change was observed in the BMD values or bone microstructure in the femurs of the treated rats after four weeks of feeding with diets containing either intervention [35]. The discrepancy in the outcomes obtained in this study compared to other studies might be because the bone-protecting action of quercetin was inhibited by other compounds.
Quercitrin, a glycoside formed from the flavonoid quercetin and the deoxy sugar rhamnose, has also been investigated for its bone-protecting actions by researchers. Xing et al. (2017) performed bilateral removal of ovaries in female Sprague-Dawley rats and assigned the animals to oral quercitrin (50, 100 or 200 mg/kg/day) treatments for 60 days. The results showed that quercitrin elevated BMD at the distal femur, and increased maximum energy absorption as well as maximum fracture load and stiffness at the femoral neck. Serum analyses of minerals, osteogenic markers and bone turnover markers revealed higher calcium, phosphorus, osterix, runt-related transcription factor 2 (Runx-2), bone formation markers (ALP and N-terminal propeptide of type 1 procollagen (P1NP)) and lower bone resorption markers (CTX and tartrate-resistant acid phosphatase (TRAP)) in the quercitrin-treated ovariectomised rats [36]. Fayed et al. (2019) further delineated the anti-osteoporotic activities of isoquercitrin in ovariectomised rats. The highest dose of isoquercitrin applied in this study (60 mg/kg/day) significantly increased lumbar compression strength. Bone turnover markers were reduced in the isoquercitrin-treated ovariectomised rats [37]. Isoquercitrin was used as previous studies indicated better bioavailability and higher antioxidative activity than quercetin and quercitrin [58,59].
Diabetes mellitus is closely associated with osteopenia, an increased risk of bone fractures and a delay in fracture healing. Therefore, the role of quercetin in diabetic osteopenia was studied by two groups of investigators using an STZ-induced diabetic rat model. In an earlier study, diabetes was induced using a single intraperitoneal (i.p.) injection of STZ at the dose of 50 mg/kg. After eight weeks of diabetes induction, the animals were injected (i.p.) with 15 mg/kg/day quercetin for four weeks. It was noted that quercetin treatment increased BV/TV, Tb.N, Tb.Th, maximum load at femoral mid-diaphysis and neck, and calcium and magnesium concentrations in the diabetic animals. Quercetin treatment also decreased blood glucose level and increased plasma insulin in the STZ-induced diabetic rats [38]. In another experiment, male Sprague-Dawley rats were injected with STZ (100 mg/kg, i.p.) for two consecutive days to induce diabetes, followed by quercetin (30 or 50 mg/kg) treatment after four weeks for a duration of eight weeks. The untreated diabetic rats exhibited higher serum glucose levels and lower bone remodelling compared to the normal controls, as reflected by serum bone markers and histomorphometric evaluation, which subsequently translated to deterioration of bone microstructural and mechanical properties. All these degenerative changes were reversed by the treatment of quercetin [39].
The use of glucocorticoids is one of the most common causes of secondary osteoporosis [60]. Two studies created an osteoporotic model by subcutaneously (s.c.) injecting glucocorticoids in rats. In the first study, methylprednisolone sodium succinate (40 mg/kg) was used as an agent to induce bone loss in female Sprague-Dawley rats. Quercetin at the dose of 150 mg/kg improved bone histomorphometric indices (Tb.Th, Cr.Th osteoblast number and moment of inertia were increased), bone strength and osteocalcin level [40]. More recently, researchers noted that the levels of calcium and phosphorus in serum, as well as weight, density and tensile strength of the femur were significantly raised after topical treatment with quercetin-loaded transfersome film (10 mg/kg) on the dorsal surface for a duration of 15 days. Bone turnover markers were reduced. Histomicrographic examination showed fewer disruptive and lytic changes in the femurs of the treatment group as compared to those without treatment [41].
Besides the more common osteoporotic conditions described above, quercetin has also been studied in osteoporosis related to other stressors. Osteoporosis and increased fracture risk are common complications in patients with primary biliary cholangitis/cirrhosis. Using a rat model of osteoporosis induced by complete bile-duct ligation, the animals receiving quercetin (150 µmol/kg/day) for four weeks showed reversal of the deteriorating effects of cirrhosis on bone strength and histomorphometric indices of trabecular and cortical bones [42]. The use of high-dose vitamin A (retinoic acid) has been reported to cause reductions in BMD and structural bone parameters [61]. Administration of quercetin (100 mg/kg/day) orally via a gastric tube for 14 days increased BMD, ash, calcium and phosphorus content in the femur. Femur weight and femur length were also elevated in the quercetin-treated group [43]. Exposure to zinc oxide nanoparticles caused perturbation of bone turnover, reflected by alteration of bone formation and bone resorption markers. The bone ALP level was higher and serum CTX level was lower in zinc-oxide-nanoparticle-intoxicated rats following treatment with intragastric quercetin (200 mg/kg) for three weeks [44]. Osteolysis is defined by pathological destruction or disappearance of bone tissues. Zhang et al. (2017) determined the protective effects of quercetin against titanium-particle-induced calvarial osteolysis utilising male BALB/c mice. Quercetin at 50 or 100 mg/kg per day inhibited titanium-particle-induced osteolysis by increasing bone area and decreasing osteoclast formation in vivo [45]. In another study, female C57BL/6 mice with titanium-particle-induced osteolysis and treated with quercetin (2 or 5 mg/kg/day) for 14 days had higher BV/TV but lower total porosity, erosion area and osteoclast number [46].
A bone defect is a lack of bone continuity usually caused by trauma, tumour or infection. Quercetin provided beneficial effects on bone defects in rats and rabbits. Song and colleagues created a circular calvarial defect with 4 mm diameter in female Sprague-Dawley rats. The defect was implanted with a quercetin/silk fibroin/hydroxyapatite scaffold with bone-marrow-derived mesenchymal stem cells (MSCs). Six weeks postimplantation, microcomputed tomography results revealed higher values of BMD, BV, BV/TV, BS, Tb.N and Tb.Th. Histological observation also demonstrated that the defect site implanted with bone-marrow-derived MSCs seeded on the quercetin/silk fibroin/hydroxyapatite scaffold had thicker bone matrix, increased formation of new collagenous tissue and tissue ingrowth [47]. A subsequent study explored the effects of quercetin/duck's foot collagen/hydroxyapatite sponge on calvarial bone defects. Similar outcomes were found, with increased BMD, BV and new bone formation 8 weeks after surgery [48]. In another study, parietal bone defects in New Zealand rabbits were grafted with quercetin solution mixed with collagen matrix for 14 days and bone histological assessment was performed to quantify new bone formation. More new bone was present in quercetin-grafted animals [49,50]. Córdoba et al. (2018) drilled holes and placed four quercetin-nanocoated implants in both tibias of model rabbits. The quercetin implants significantly decreased the expression of osteoclast-related genes (cathepsin K (CTSK), vacuolar type proton ATPase (H + ATPase) and matrix metalloproteinase-9 (MMP-9)). The level of receptor activator of nuclear factor-kappa B ligand (RANKL) was also significantly reduced in rabbits implanted with quercetin-nanocoated implants [51].
Heterogenous findings on bone parameters were obtained in healthy rats upon treatment with quercetin, whereby three studies reported positive outcomes while one study showed a negative impact on bone. In growing rats, higher MAR and BFR were observed at the periosteal and endosteal sites of femur mid-diaphysis in the animals treated with quercetin-6-C-β-d-glucopyranoside. Dual-energy absorptiometry and histomorphometry analyses reported higher femoral BMD, tibial BMD, Cr.Ar, Cr.Th, T.Ar, T.Pm, B.Pm and E.Pm at mid-diaphysis and TFSP after quercetin-6-C-β-d-glucopyranoside supplementation [33]. In senescence-accelerated OXYS rats, dihydroquercetin (5.06 mg/kg/day) was given in combination with glucosamine alendronate (1.26 mg/kg/day) for two months. The BMD values (total, lumbar and humerus), force applied at failure point and strength of femur were increased, whereas the level of CTX was decreased in the treated ageing rat model [52]. A study by Babosova et al. (2016) also yielded results showing an increase in Cr.Th in healthy 5 month old female rabbits treated intramuscularly (i.m.) with quercetin (10 or 100 µg/kg) three times a week [53]. In contrast, Oršolić et al. (2014) found that quercetin (100 mg/kg/day) given orally for 14 days did not change the weight, length, or ash, calcium and phosphorus content in normal female fertile Y59 rats [43].
Taken together, most of these in vivo studies have shown that quercetin enhances bone quality by improving BMD, trabecular and cortical bone microstructure, bone strength and bone histomorphometric parameters in animals with osteopenia, osteolysis and bone defects. In general, quercetin was administered to animals via oral gavage, injection (i.p. or i.m.) or implantation. Two studies reported paradoxical findings, probably due to the following reasons: Firstly, the beneficial effects of quercetin on preventing bone loss may be inhibited by other phytochemicals, which is worth further validation. Secondly, quercetin may not augment bone health significantly in healthy animals. In view of the potential skeletal benefits of quercetin in vivo, we looked further into the effects of quercetin on bone cells and the molecular mechanisms involved in orchestrating bone metabolism.

The Effects of Quercetin on Osteoblastogenesis
In in vitro studies, quercetin has been incubated with MSCs, murine or human osteoblast cells to investigate its effects on cell viability, proliferation, mineralisation and expression of osteogenic genes ( Table 2). The proliferation and differentiation of osteoblasts are mainly regulated by a number of transcription factors (osterix, Runx-2, core binding factor alpha 1 (Cbfα1) and type 1 collagen (COL1)), growth factors (bone morphogenetic protein-2 (BMP-2), transforming growth factor-beta 1 (TGF-β1) and vascular endothelial growth factor (VEGF)) and bone markers (ALP, bone sialoprotein (BSP), osteopontin and osteocalcin). Hence, these are the parameters that are commonly measured to determine the osteogenic ability of quercetin.
Extensive research has shown favourable properties of quercetin and its related compounds on bone; however, there have been a few studies reporting negative effects. Using rat calvarial osteoblast-like cells exposed to quercetin (1-10 µM), one study reported significant reductions in proliferation, differentiation and mineralisation (as shown by decreased cell number, ALP, osteocalcin, calcium deposition and mineralised nodules) [92]. A study done by Son et al. (2006) reported that quercetin treatment itself did not reduce cell viability and did not exert a substantial cytotoxic effect on MC3T3-E1 cells. The presence of TNF-α decreased cell viability and increased cytotoxicity, but treatment with quercetin caused further acceleration of TNF-α-induced decrease in cell viability and TNF-α-induced cytotoxicity [90]. A recent study by Casado-Díaz et al. (2016) also pinpointed that quercetin inhibited cell proliferation, ALP activity and mineralisation, as well as downregulating markers of osteogenesis (COL1 and osteocalcin) in MSCs induced to differentiate into osteoblasts [93].
In short, the majority of studies have reported osteoblastogenesis-activating effects of quercetin, with a few exceptions. These discrepancies may be related to the cell types and doses of quercetin used. It is important to look further into the mechanisms of action that underlie the osteogenic potential of quercetin.

The Effects of Quercetin on Osteoclastogenesis
Numerous studies have found that quercetin inhibits the formation of osteoclast-like cells (evidenced by TRAP-positive multinucleated cells), bone resorption pit and F-actin ring formation in RAW264.7 cells, human peripheral-blood mononuclear cells (PBMCs) or bone-marrow macrophages treated with macrophage colony-stimulating factor (M-CSF) and/or RANKL [31,32,45,46,68,74,[94][95][96]. In another study using highly purified rabbit osteoclasts, treatment with 50 µM quercetin caused a reduction in the total area of resorption pit [97]. The measurement of hydroxylysylpyridinoline, a collagen cross-link molecule released during bone resorption, was used by researchers to confirm the bone resorption activity in cultures [97]. Quercetin was found to decrease hydroxylysylpyridinoline content in human PBMCs treated with M-CSF and RANKL, as well as in highly purified rabbit osteoclasts [96,97]. Quercetin at a dose of 15, 25 or 50 µM also blocked LPS-induced osteoclast formation and viability in RAW264.7 cells [98]. Mouse bone-marrow cells cultured in the presence of parathyroid hormone (PTH), a bone-resorbing factor, caused an increase in osteoclast-like cell formation. This increase was suppressed in the presence of quercetin (0.01-1 µM) [71]. Similarly, other structurally similar derivatives of quercetin, such as quercetin 3-O-β-D-glucuronide, quercetin-6-C-β-d-glucopyranoside and quercitrin, decreased the number of TRAP-positive cells and bone resorption pit area in RANKL-induced osteoclast differentiation in RAW264.7 cells and bone-marrow macrophages [31,32,83]. Another cell culture study evaluated the effects of calcium-deficient hydroxyapatite loaded with quercetin on cell proliferation in RAW264.7 cells exposed to RANKL. Their findings indicated a dramatic reduction in cell number and osteoclast proliferation [68]. Two recent studies demonstrated a reduction in cell viability and proliferation of osteoclasts when co-culturing osteoblast-osteoclast or triculturing osteoblast-osteoclast-endothelial cells on hydroxyapatite loaded with quercetin [84,85]. In summary, the current literature demonstrated that quercetin consistently has inhibitory effects on osteoclast formation, proliferation and maturation.

Regulation of the Receptor Activator of Nuclear Factor-Kappa B (RANK)/RANKL/Osteoprotegerin System
Osteoclastogenesis is a multistep process that requires the commitment of osteoclast progenitors and two haematopoietic factors, M-CSF and RANKL. Osteoclastogenesis is an osteoblast-and osteocyte-mediated process wherein these bone cells are the main source of osteoprotegerin and RANKL governing the RANK/RANKL/osteoprotegerin system. RANKL has a pivotal role in activating RANK expressed on osteoclasts and its precursors, subsequently supporting pre-osteoclast differentiation and osteoclast maturation and survival [102]. Osteoprotegerin, a decoy receptor for RANKL, plays an anti-osteoclastogenic role by blocking the RANK-RANKL interaction. The completion of osteoclast proliferation, differentiation and maturation mediated by RANK-RANKL recognition requires sequential recruitment of tumour necrosis factor receptor-associated factor 6 (TRAF6) and activation of downstream signalling molecules (such as mitogen-activated protein kinase (MAPK), nuclear factor-kappa B (NF-κB) and activator protein (AP-1)), followed by the transcriptional amplification of various genes encoding for cellular proto-oncogene (c-Fos), nuclear factor of activated T-cells cytoplasmic 1 (NFATc1), TRAP, CTSK, calcitonin receptor (CalcR), alpha v integrin, MMP-9 and H + ATPase [103]. Dendritic-cell-specific transmembrane protein (Dc-Stamp) is also a key regulator for osteoclast differentiation, wherein its mechanism of action is involved in mediating cell-cell fusion and formation of fully functional osteoclasts [104].

Regulation of MAPK Signalling
The MAPKs are a set of serine/threonine kinases that function as key signal transducers in converting a wide range of extracellular stimuli to cellular responses. In general, activation of the MAPK cascade occurs via the consecutive phosphorylation of three protein kinases. Upon the arrival of a stimulus, mitogen-activated protein kinase kinase kinase (MAP3K) is activated to phosphorylate mitogen-activated protein kinase kinase (MAP2K) and, in turn, phosphorylate MAPK [105]. The three best-studied subgroups of MAPK signalling pathways include extracellular-signal-regulated kinase (ERK), p38 MAPK and c-hJun N-terminal kinase (JNK) signalling pathways, which have been recognised to be critical in modulating the differentiation and activation of both osteoblasts and osteoclasts. ERK1/2 is activated by growth factors, hormones and pro-inflammatory cytokines, whereas p38 MAPK and JNK are activated by pro-inflammatory mediators and cellular and environmental stresses.
Previous studies have confirmed the direct relationship of ERK and p38 MAPK activation with osteogenesis [106][107][108]. Deletion of Mek1 and Mek2 (the upstream kinases of ERK signalling) in mice was correlated with lower bone mass and decreased expression of osteoblast-specific genes (Runx-2,

Regulation of MAPK Signalling
The MAPKs are a set of serine/threonine kinases that function as key signal transducers in converting a wide range of extracellular stimuli to cellular responses. In general, activation of the MAPK cascade occurs via the consecutive phosphorylation of three protein kinases. Upon the arrival of a stimulus, mitogen-activated protein kinase kinase kinase (MAP3K) is activated to phosphorylate mitogen-activated protein kinase kinase (MAP2K) and, in turn, phosphorylate MAPK [105]. The three best-studied subgroups of MAPK signalling pathways include extracellular-signal-regulated kinase (ERK), p38 MAPK and c-hJun N-terminal kinase (JNK) signalling pathways, which have been recognised to be critical in modulating the differentiation and activation of both osteoblasts and osteoclasts. ERK1/2 is activated by growth factors, hormones and pro-inflammatory cytokines, whereas p38 MAPK and JNK are activated by pro-inflammatory mediators and cellular and environmental stresses.
Previous studies have confirmed the direct relationship of ERK and p38 MAPK activation with osteogenesis [106][107][108]. Deletion of Mek1 and Mek2 (the upstream kinases of ERK signalling) in mice was correlated with lower bone mass and decreased expression of osteoblast-specific genes (Runx-2, ATF4 and β-catenin) [107]. Furthermore, oestrogen-receptor signalling can activate ERK signalling in many cells, including osteoblasts [109], and p38-mediated phosphorylation promotes the expression of osteoblast master regulators like Runx-2, Dlx5, osterix and ATF4 [108]. In contrast, the role of JNK in osteogenesis regulation remains inconsistent; its activation may positively or negatively regulate osteogenesis. The increase of JNK phosphorylation enhanced ALP activity and mineralisation and increased cellular responsiveness to BMP-2 [110,111]. On the other hand, another study hinted at a reverse association between JNK and osteoblastic differentiation. A loss of function in JNK enhanced ALP expression and mineralisation, while the gain of function in JNK reduced BMP-2-mediated osteoblastic differentiation [112]. Activation of JNK and p38 MAPK also induced apoptosis in many cell types, including osteoblasts and osteoclasts [113,114].
Consistently, increased phosphorylation and activation of ERK1/2 was observed in rat bone-marrow-derived MSCs, human adipose-tissue-derived stromal cells, human osteoblast-like MG-63 cells and murine osteoblastic MC3T3-E1 cells incubated with quercetin [34,62,63,67,69,75,76,86]. The data presented by Prouillet et al. (2004) showed that the activation of ERK by quercetin was most likely downstream of the oestrogen-receptor activation [69]. Quercetin exerted dual actions on p38 MAPK to promote osteogenic proliferation and differentiation in MSCs upon its activation, as well as suppressing apoptosis in MC3T3-E1 cells following its inhibition [34,62,63,75]. For the JNK pathway, some studies have reported that quercetin might activate, inhibit or exert negligible effects on osteoblasts. Although the exact mechanism of action of quercetin on JNK signalling in osteoblasts is inconclusive, its net effect has been shown to favour osteogenesis [34,62,63,76]. On the other hand, the activation of JNK pathway by quercetin was associated with acceleration of apoptosis in both osteoblast-and osteoclast-like cells. Quercetin triggered increases in the phosphorylated form of JNK and caspase activation in MC3T3-E1 cells treated with TNF-α, indicating TNF-α-mediated apoptosis via activation of JNK pathway [91]. A study done by Guo et al. (2012) pointed out that quercetin increased protein levels of p-p38 MAPK and p-JNK, thereby increasing Bcl-2-associated X protein (Bax) and decreasing B-cell lymphoma 2 (Bcl-2) in RAW264.7 cells treated with LPS. These findings suggested the augmentation of LPS-induced osteoclast apoptosis upon activation of p38 MAPK and JNK apoptotic signalling pathways [98].
The regulation of osteoblastogenesis and osteoclastogenesis governed by MAPK signalling pathways is highly complex (Figure 2). The current state of knowledge shows that even though quercetin has biphasic effects on activation or inhibition of MAPKs (particularly p38 MAPK and JNK), the net outcomes obtained are the enhancement of osteoblast function and osteoclast apoptosis, as well as the suppression of osteoblast apoptosis. ATF4 and β-catenin) [107]. Furthermore, oestrogen-receptor signalling can activate ERK signalling in many cells, including osteoblasts [109], and p38-mediated phosphorylation promotes the expression of osteoblast master regulators like Runx-2, Dlx5, osterix and ATF4 [108]. In contrast, the role of JNK in osteogenesis regulation remains inconsistent; its activation may positively or negatively regulate osteogenesis. The increase of JNK phosphorylation enhanced ALP activity and mineralisation and increased cellular responsiveness to BMP-2 [110,111]. On the other hand, another study hinted at a reverse association between JNK and osteoblastic differentiation. A loss of function in JNK enhanced ALP expression and mineralisation, while the gain of function in JNK reduced BMP-2-mediated osteoblastic differentiation [112]. Activation of JNK and p38 MAPK also induced apoptosis in many cell types, including osteoblasts and osteoclasts [113,114]. Consistently, increased phosphorylation and activation of ERK1/2 was observed in rat bonemarrow-derived MSCs, human adipose-tissue-derived stromal cells, human osteoblast-like MG-63 cells and murine osteoblastic MC3T3-E1 cells incubated with quercetin [34,62,63,67,69,75,76,86]. The data presented by Prouillet et al. (2004) showed that the activation of ERK by quercetin was most likely downstream of the oestrogen-receptor activation [69]. Quercetin exerted dual actions on p38 MAPK to promote osteogenic proliferation and differentiation in MSCs upon its activation, as well as suppressing apoptosis in MC3T3-E1 cells following its inhibition [34,62,63,75]. For the JNK pathway, some studies have reported that quercetin might activate, inhibit or exert negligible effects on osteoblasts. Although the exact mechanism of action of quercetin on JNK signalling in osteoblasts is inconclusive, its net effect has been shown to favour osteogenesis [34,62,63,76]. On the other hand, the activation of JNK pathway by quercetin was associated with acceleration of apoptosis in both osteoblast-and osteoclast-like cells. Quercetin triggered increases in the phosphorylated form of JNK and caspase activation in MC3T3-E1 cells treated with TNF-α, indicating TNF-α-mediated apoptosis via activation of JNK pathway [91]. A study done by Guo et al. (2012) pointed out that quercetin increased protein levels of p-p38 MAPK and p-JNK, thereby increasing Bcl-2-associated X protein (Bax) and decreasing B-cell lymphoma 2 (Bcl-2) in RAW264.7 cells treated with LPS. These findings suggested the augmentation of LPS-induced osteoclast apoptosis upon activation of p38 MAPK and JNK apoptotic signalling pathways [98].
The regulation of osteoblastogenesis and osteoclastogenesis governed by MAPK signalling pathways is highly complex (Figure 2). The current state of knowledge shows that even though quercetin has biphasic effects on activation or inhibition of MAPKs (particularly p38 MAPK and JNK), the net outcomes obtained are the enhancement of osteoblast function and osteoclast apoptosis, as well as the suppression of osteoblast apoptosis.

Regulation of Apoptosis
Apoptosis is a process of programmed cell death mediated by two main apoptotic pathways, i.e., the intrinsic (mitochondria-dependent) pathway and extrinsic (death-receptor-mediated) pathway. The mitochondrial pathway is driven by the presence of intracellular cues such as cellular damage and oxidative stress. These stimuli downregulate anti-apoptotic (Bcl-2 and B-cell lymphoma-extra large (Bcl-XL)) and upregulate pro-apoptotic (Bax and Bcl-2 homologous antagonist/killer (Bak)) molecules, leading to the mitochondrial release of cytochrome c. The binding of cytochrome c with apoptotic protease activating factor-1 (Apaf-1), deoxyadenosine triphosphate (dATP) and procaspase-9 forms the apoptosome, catalysing the conversion of procaspase-9 to caspase-9. Eventually, the downstream effector caspases (caspase-3 and caspase-7) are activated, acting as a molecular switch towards cell apoptosis [115]. In the extrinsic pathway, two examples of death ligands are Fas ligand (FasL) and tumour necrosis factor-related apoptosis-inducing ligand (TRAIL), recognised by their receptors, the Fas and TRAIL receptors. Their interaction results in the formation of the death-inducing signalling complex (DISC), consisting of Fas-associated death domain (FADD) and caspase-8, to activate downstream effector caspase-3, which further induces typical apoptosis features including cell fragmentation and cell death [116].
Quercetin has been demonstrated to inhibit cell growth via promoted cell apoptosis in mouse embryo 3T3-L1 pre-adipocytes [117]. Increases in lactate dehydrogenase (LDH) activity (a marker of cell-membrane damage), number of apoptotic cells and collapse of mitochondrial membrane potential have been evidenced. Quercetin also modulates apoptosis-related activity and proteins by downregulating poly(ADP-ribose) polymerase (PARP) and Bcl-2 and upregulating caspase-3 activity, Bax and Bak [117]. On the other hand, quercetin protected murine osteoblastic MC3T3-E1 cells against LPS-induced apoptosis. Pretreatment with quercetin restored the downregulated Bcl-2 and Bcl-XL expression as well as the upregulated caspase-3, Bax and cytochrome c expression caused by LPS [75]. Zhang et al. (2017) reported that quercetin had anti-apoptotic effects in vitro and in vivo, shown by inhibition of titanium-particle-induced endoplasmic-reticulum-stress-related cell apoptosis. The levels of protein kinase RNA-like endoplasmic reticulum kinase (PERK), inositol-requiring enzyme 1 (IRE1), glucose-regulated protein (GRP78), CCAAT/enhancer-binding protein homologous protein (CHOP), caspase-12 and caspase-3 were decreased, whereas Bcl-2 level was increased in the titanium-particle-treated RAW264.7 cells and titanium-particle-implanted calvarium. [45]. Seeding the co-culture of osteoblasts and osteoclasts onto hydroxyapatite loaded with quercetin resulted in the lowering of LDH activity in osteoblasts and upregulation of caspase-3 in osteoclasts [84]. Quercetin also increased the number of apoptotic osteoclasts in highly purified rabbit osteoclast cultures in a dose-dependent manner [97].
In short, quercetin plays a role in inhibiting apoptosis in osteoblasts but promoting apoptosis in adipocytes and osteoclasts (Figure 3). Osteoblasts and adipocytes share a common mesenchymal progenitor. The increase in osteoblast formation decreases the MSC reserve for adipocyte differentiation.

Antioxidative Effects
The physiological intracellular redox balance is delicately regulated, with the production of reactive oxygen species (ROS) or/and reactive nitrogen species (RNS) neutralised by an interacting network of antioxidants in the body. Mitochondria are the primary site of production for ROS. The synthesis of superoxide anions (O2• − ) during oxidative phosphorylation is converted to hydrogen peroxide (H2O2) by superoxide dismutase (SOD). The less-reactive H2O2 can be further detoxified by glutathione peroxidase (GPx) or/and catalase (CAT) to form water. If antioxidants do not inactivate H2O2, it can be subjected to photoexcition to generate hydroxyl radicals (OH•) [118,119]. On the other hand, the synthesis of RNS is initiated in the production of nitric oxide (NO•) by nitric oxide synthase (NOS). The reaction between NO• and O2• − forms a stronger oxidant, the peroxynitrite anion (ONOO − ). Subsequently, ONOO − reacts with other molecules to form other RNS such as nitrogen dioxide (NO2•) and dinitrogen trioxide (N2O3) [120]. The supraphysiological levels of ROS or/and RNS and depletion of antioxidant capacity result in oxidative and nitrosative stress that severely damages lipids, proteins and deoxyribonucleic acid (DNA). Due to the short life of ROS, oxidative stress is usually measured by tracing the level of modified oxidative products such as lipid peroxides (malondialdehyde (MDA)), oxidised proteins (nitrotyrosine and protein carbonyl), oxidised nucleic acid bases (8-hydroxy-2′-deoxyguanosine (8-OHdG)) and enzymatic markers (myeloperoxidase) [121].
The major signalling pathway that responds to oxidative and nitrosative stress is the nuclear factor erythroid 2-related factor 2 (Nrf2) pathway. The key signalling molecules, Nrf2 binds to Kelchlike ECH-associated protein 1 (Keap1) in the cytoplasm to form an inactive complex under basal conditions, thus suppressing its transcriptional activity. Under oxidative conditions, Nrf2 dissociates from Keap1, allowing its translocation into the nucleus, followed by heterodimerisation with small musculoaponeurotic fibrosarcoma oncogene homolog (Maf) proteins and association with antioxidant-responsive element (ARE). Upon binding, the expression of downstream Nrf2 target cytoprotective genes and antioxidative enzymes (including heme-oxygenase 1 (HO-1), nicotinamide

Antioxidative Effects
The physiological intracellular redox balance is delicately regulated, with the production of reactive oxygen species (ROS) or/and reactive nitrogen species (RNS) neutralised by an interacting network of antioxidants in the body. Mitochondria are the primary site of production for ROS. The synthesis of superoxide anions (O 2 • − ) during oxidative phosphorylation is converted to hydrogen peroxide (H 2 O 2 ) by superoxide dismutase (SOD). The less-reactive H 2 O 2 can be further detoxified by glutathione peroxidase (GPx) or/and catalase (CAT) to form water. If antioxidants do not inactivate H 2 O 2 , it can be subjected to photoexcition to generate hydroxyl radicals (OH•) [118,119]. On the other hand, the synthesis of RNS is initiated in the production of nitric oxide (NO•) by nitric oxide synthase (NOS). The reaction between NO• and O 2 • − forms a stronger oxidant, the peroxynitrite anion (ONOO − ). Subsequently, ONOO − reacts with other molecules to form other RNS such as nitrogen dioxide (NO 2 •) and dinitrogen trioxide (N 2 O 3 ) [120]. The supraphysiological levels of ROS or/and RNS and depletion of antioxidant capacity result in oxidative and nitrosative stress that severely damages lipids, proteins and deoxyribonucleic acid (DNA). Due to the short life of ROS, oxidative stress is usually measured by tracing the level of modified oxidative products such as lipid peroxides (malondialdehyde (MDA)), oxidised proteins (nitrotyrosine and protein carbonyl), oxidised nucleic acid bases (8-hydroxy-2 -deoxyguanosine (8-OHdG)) and enzymatic markers (myeloperoxidase) [121].
The major signalling pathway that responds to oxidative and nitrosative stress is the nuclear factor erythroid 2-related factor 2 (Nrf2) pathway. The key signalling molecules, Nrf2 binds to Kelch-like ECH-associated protein 1 (Keap1) in the cytoplasm to form an inactive complex under basal conditions, thus suppressing its transcriptional activity. Under oxidative conditions, Nrf2 dissociates from Keap1, allowing its translocation into the nucleus, followed by heterodimerisation with small musculoaponeurotic fibrosarcoma oncogene homolog (Maf) proteins and association with antioxidant-responsive element (ARE). Upon binding, the expression of downstream Nrf2 target cytoprotective genes and antioxidative enzymes (including heme-oxygenase 1 (HO-1), nicotinamide adenine dinucleotide phosphate quinone dehydrogenase 1 (NQO1), γ-glutamyl-cysteine ligase catalytic subunit (GCLC), glutathione-S-transferase (GST), SOD, CAT and reduced glutathione (GSH)) are driven to circumvent oxidative and nitrosative insult [122,123]. Apart from that, there are molecular interactions between Nrf2 and other signalling proteins like ERK and NF-κB. The activation of the ERK signal transduction pathway regulates Nrf2-dependent transcription by promoting the release of Nrf2 from Keap1 [124]. In contrast, Nrf2 signalling has an opposing effect on NF-κB activity, whereby the activation of Nrf2 pathway inhibits NF-κB by upregulating antioxidant defence and HO-1 expression [125].
In experimental in vitro studies, researchers have found that the antioxidative properties of quercetin and its related glycoside contribute to their osteoblastogenic and anti-osteoclastogenic properties. Primary human osteoblasts exposed to cigarette-smoke medium rapidly produced ROS, leading to a decrease in cell viability. Incubation with quercetin reversed these effects by increasing the expression of antioxidative enzymes such as HO-1 and SOD [78]. Another study showed that treatment of co-culture of osteoblasts and osteoclasts with H 2 O 2 induced the generation of ROS. The elevation in ROS level was prevented when the co-culture was seeded on hydroxyapatite loaded with quercetin in the presence of H 2 O 2 [84]. Quercitrin also provided protective effects against H 2 O 2 -induced oxidative stress in osteoblastic MC3T3-E1 cells, whereby the levels of ROS-damaged biological molecules such as MDA, protein carbonyl and nitrotyrosine were reduced [82]. Messer and collaborators reported that osteoblasts derived from fetal rat calvaria treated with quercetin aglycone showed marked upregulation of HO-1, CAT, GCLC and peroxiredoxin-5 (Prdx5) [79]. In the subsequent year, the same group of investigators cultured osteoblasts from fetal rat calvaria in the presence of an oxidative stressor (H 2 O 2 ) and treated them with quercetin aglycone. Sustained upregulation of HO-1 and GCLC was observed in osteoblasts without treatment, but it was blocked by the addition of quercetin aglycone [80]. At this juncture, it might be postulated that these increases in HO-1 and GCLC expression might be due to the need to counteract the overwhelming oxidative response induced by H 2 O 2 . The decreases in antioxidant gene expression upon quercetin aglycone incubation indicate the role of quercetin in maintaining a balanced profile between oxidative stress and antioxidant capacity.
The findings of mechanistic studies conducted by researchers remain inconsistent. Quercetin protected the primary human osteoblasts from toxic effects of oxidative stress via the Nrf2 and ERK signalling pathways in one study. Increased phosphorylation of Nrf2 and ERK1/2 in osteoblasts became evident upon stimulation with quercetin [78]. However, a study by Messer et al. (2015) showed no alteration in Nrf2 accumulation in the nucleus and cytoplasm of fetal rat calvarial osteoblasts, even though there was upregulation of antioxidant genes. The authors postulated that the role of the Nrf2 signalling pathway in controlling the transcription of antioxidant genes might be complex, whereby multiple post-translational modifications of Nrf2 and Kaep1 might be involved. In addition, they also reported that the upregulation of antioxidant genes was associated with suppressed phosphorylation of ERK1/2, which was in direct contrast with the findings from a study by Braun et al. (2011). The authors suggested that the effects of quercetin might be different depending on the cell and culture conditions. For NF-κB activity, there was a downregulation in the total level of NF-κB p65 in osteoblasts after treatment with quercetin aglycone [79].
Quercetin also attenuated oxidative and nitrosative stress in osteoclasts. As reported by Wattel et al. (2003), the intracellular ROS production in highly purified rabbit osteoclasts was reduced after treatment with quercetin [97]. Later on, Tang et al. (2019) indicated that both quercetin and quercitrin reduced the LPS-stimulated elevation in NO• and ROS production in RAW264.7 [99]. In the following year, quercetin was shown to inhibit the expression of inducible nitric oxide synthase (iNOS) in RAW264.7 cells treated with M-CSF and RANKL [46].
Several in vivo studies have reported these oxidative-stress-prevention effects alongside the bone-protecting actions of quercetin. Oral administration of free quercetin and quercetin-loaded phytosome nanoparticles reduced MDA level and increased GSH content in ovariectomised rats [30] and in retinoic-acid-induced bone loss model [43]. The concentration of urinary 8-OHdG was lowered, whereas the serum total antioxidant capacity, SOD, GPx, CAT and GST were increased in STZ-induced diabetic rats orally treated with quercetin for 8 weeks [39]. Along with the elevation of bone formation markers and reduction in bone resorption markers, Abdelkarem et al. (2016) also found a decrease in NO• and DNA damage in rats exposed to zinc oxide nanoparticles and treated with quercetin [44]. Nonetheless, no effect was observed in oxidative-stress biomarkers and antioxidant activities when quercetin was orally supplemented to healthy rats [43].
In brief, the evidence shows that quercetin exerts antioxidative properties favouring an increase in osteogenic activities and decrease in osteoclastogenic activities (Figure 4). However, further experiments are recommended to elucidate the exact role of the Nrf2 and ERK signalling pathways in the quercetin-induced upregulation of the antioxidative response in osteoblasts. Meanwhile, investigation into the underlying molecular mechanism governing the antioxidative action of quercetin in osteoclasts is also required. STZ-induced diabetic rats orally treated with quercetin for 8 weeks [39]. Along with the elevation of bone formation markers and reduction in bone resorption markers, Abdelkarem et al. (2016) also found a decrease in NO• and DNA damage in rats exposed to zinc oxide nanoparticles and treated with quercetin [44]. Nonetheless, no effect was observed in oxidative-stress biomarkers and antioxidant activities when quercetin was orally supplemented to healthy rats [43].
In brief, the evidence shows that quercetin exerts antioxidative properties favouring an increase in osteogenic activities and decrease in osteoclastogenic activities (Figure 4). However, further experiments are recommended to elucidate the exact role of the Nrf2 and ERK signalling pathways in the quercetin-induced upregulation of the antioxidative response in osteoblasts. Meanwhile, investigation into the underlying molecular mechanism governing the antioxidative action of quercetin in osteoclasts is also required.

Anti-Inflammatory Effects
Chronic inflammation is associated with systemic bone loss. The unresolved inflammatory milieu is often characterised by the release of pro-inflammatory factors such as TNF-α, interleukin-1 beta (IL-1β), interleukin-6 (IL-6) and C-reactive protein (CRP), which is not counterbalanced by the production of anti-inflammatory factors such as interleukin-10 (IL-10) and arginase-1 (Arg-1). These inflammatory cytokines represent cell signalling molecules involved in inflammation, and their perturbation implicates abnormal bone metabolism favouring the activation of bone degradation and inhibition of bone formation. The secretion of pro-inflammatory cytokines under inflammatory conditions inhibits osteoblast gene products, partly through the activation of suppressor of mothers against decapentaplegic ubiquitylation regulatory factor 1 (SMURF1), SMURF2 and NF-κB as well as via the suppression of MAPK [126]. Pro-inflammatory mediators also promote Dickkopf-related

Anti-Inflammatory Effects
Chronic inflammation is associated with systemic bone loss. The unresolved inflammatory milieu is often characterised by the release of pro-inflammatory factors such as TNF-α, interleukin-1 beta (IL-1β), interleukin-6 (IL-6) and C-reactive protein (CRP), which is not counterbalanced by the production of anti-inflammatory factors such as interleukin-10 (IL-10) and arginase-1 (Arg-1). These inflammatory cytokines represent cell signalling molecules involved in inflammation, and their perturbation implicates abnormal bone metabolism favouring the activation of bone degradation and inhibition of bone formation. The secretion of pro-inflammatory cytokines under inflammatory conditions inhibits osteoblast gene products, partly through the activation of suppressor of mothers against decapentaplegic ubiquitylation regulatory factor 1 (SMURF1), SMURF2 and NF-κB as well as via the suppression of MAPK [126]. Pro-inflammatory mediators also promote Dickkopf-related protein 1 (DKK1) and sclerostin (SOST), inhibiting the Wingless (Wnt)/beta (β)-catenin signalling cascade [126]. Additionally, pro-inflammatory cytokines interact with their respective receptors to stimulate the production of M-CSF and RANKL by osteoblasts, essential for osteoclast proliferation, differentiation and maturation.
In recent years, researchers have investigated whether the bone-sparing actions of quercetin are mediated through its anti-inflammatory properties. An osteoblast and osteoclast co-culture treated with H 2 O 2 showed an increased level of TNF-α, which was found to be restored to normal level when seeded on hydroxyapatite loaded with quercetin [84]. Another experiment done by the same group of investigators found that quercetin-functionalised hydroxyapatite inhibited IL-6 production on a triculture model consisting of three types of cells, namely human osteoblast-like MG63 cells, osteoclast precursors 2T-110 and human umbilical-vein endothelial cells (HUVECs) [85]. The concentrations of pro-and anti-inflammatory cytokines were also determined in vitro using RAW264.7 cells. Quercetin reduced the levels of IL-1β, TNF-α and IL-6 while increasing those of IL-10 and Arg-1 in RAW264.7 cells exposed to M-CSF, RANKL or LPS [45,46,99]. In vivo, a reduction in TNF-α, IL-6 and CRP levels was observed following quercetin or quercetin-loaded phytosome nanoparticle interventions in ovariectomised [30] and zinc-oxide-nanoparticle-treated rats [44].
Indeed, the process of bone remodelling depends on the tight coupling between pro-and anti-inflammatory mediators. Quercetin helps to resolve overwhelming inflammatory responses by inhibiting pro-inflammatory cytokines and stimulating anti-inflammatory cytokines ( Figure 5). The molecular machinery that underlies the anti-inflammatory action of quercetin remains a research gap in the field. cascade [126]. Additionally, pro-inflammatory cytokines interact with their respective receptors to stimulate the production of M-CSF and RANKL by osteoblasts, essential for osteoclast proliferation, differentiation and maturation. In recent years, researchers have investigated whether the bone-sparing actions of quercetin are mediated through its anti-inflammatory properties. An osteoblast and osteoclast co-culture treated with H2O2 showed an increased level of TNF-α, which was found to be restored to normal level when seeded on hydroxyapatite loaded with quercetin [84]. Another experiment done by the same group of investigators found that quercetin-functionalised hydroxyapatite inhibited IL-6 production on a triculture model consisting of three types of cells, namely human osteoblast-like MG63 cells, osteoclast precursors 2T-110 and human umbilical-vein endothelial cells (HUVECs) [85]. The concentrations of pro-and anti-inflammatory cytokines were also determined in vitro using RAW264.7 cells. Quercetin reduced the levels of IL-1β, TNF-α and IL-6 while increasing those of IL-10 and Arg-1 in RAW264.7 cells exposed to M-CSF, RANKL or LPS [45,46,99]. In vivo, a reduction in TNF-α, IL-6 and CRP levels was observed following quercetin or quercetin-loaded phytosome nanoparticle interventions in ovariectomised [30] and zinc-oxide-nanoparticle-treated rats [44].

Canonical Wnt/β-Catenin Signalling
The canonical Wnt/β-catenin signalling pathway has a central regulatory role in bone metabolism. In unstimulated conditions, β-catenin is sequestered into a destruction complex consisting of axis inhibition protein 2 (Axin), casein kinase 1 alpha (CK1α), adenomatosis polyposis coli (APC) and glycogen synthase kinase-3 beta (GSK3β). Consequently, β-catenin is phosphorylated, ubiquitinylated and degraded by the proteasome. In the presence of extracellular Wnt protein binding to the Frizzled receptor and low-density-lipoprotein receptor-related protein 5/6 (LRP5/6) co-receptors, the ubiquitination and degradation of β-catenin mediated by GSK3β are inhibited, thus facilitating the accumulation and translocation of β-catenin into the nucleus [127]. The activation of Wnt signalling leads to expression of the Wnt-targeted gene Runx-2, which is essential for osteoblast differentiation [128]. Apart from its well-established role in regulating inflammatory cytokines, the NF-κB signalling pathway has been suggested to interact with the canonical Wnt/β-catenin signalling pathway. A study by Le Henaff et al. (2015) reported that the overactivation of NF-κB transcriptional activity promoted β-catenin phosphorylation and reduced β-catenin target gene expression in mutant osteoblasts. The defective osteoblast differentiation and function were rescued by pharmacological inhibition of NF-κB and activation of Wnt/β-catenin signalling [129].
Quercetin rescued LPS-induced impairment of osteogenesis in murine osteoblastic MC3T3-E1 cells by enhancing the protein levels of Wnt3 and β-catenin and decreasing the protein level of GSK3β [75]. Yuan et al. (2018) also found that quercetin protected rat bone-marrow-derived MSCs against TNF-α-induced inhibition of osteoblast differentiation. The activation of NF-κB and degradation of β-catenin were noted in cells treated with TNF-α. These results were reversed after treatment with TNF-α and quercetin [29]. In a preclinical animal experimentation setting, ovariectomy reduced the expression of β-catenin and treatment with isoquercitrin reversed the condition [37]. Taken together, these studies indicate that quercetin potentially orchestrates osteoblast differentiation and function through activation of the canonical Wnt/β-catenin signal transduction pathway ( Figure 6). Further investigation into the role of quercetin on Wnt antagonists such as DKK-1 and SOST is required to provide a better understanding of this signalling pathway.

Canonical Wnt/β-Catenin Signalling
The canonical Wnt/β-catenin signalling pathway has a central regulatory role in bone metabolism. In unstimulated conditions, β-catenin is sequestered into a destruction complex consisting of axis inhibition protein 2 (Axin), casein kinase 1 alpha (CK1α), adenomatosis polyposis coli (APC) and glycogen synthase kinase-3 beta (GSK3β). Consequently, β-catenin is phosphorylated, ubiquitinylated and degraded by the proteasome. In the presence of extracellular Wnt protein binding to the Frizzled receptor and low-density-lipoprotein receptor-related protein 5/6 (LRP5/6) co-receptors, the ubiquitination and degradation of β-catenin mediated by GSK3β are inhibited, thus facilitating the accumulation and translocation of β-catenin into the nucleus [127]. The activation of Wnt signalling leads to expression of the Wnt-targeted gene Runx-2, which is essential for osteoblast differentiation [128]. Apart from its well-established role in regulating inflammatory cytokines, the NF-κB signalling pathway has been suggested to interact with the canonical Wnt/β-catenin signalling pathway. A study by Le Henaff et al. (2015) reported that the overactivation of NF-κB transcriptional activity promoted β-catenin phosphorylation and reduced β-catenin target gene expression in mutant osteoblasts. The defective osteoblast differentiation and function were rescued by pharmacological inhibition of NF-κB and activation of Wnt/β-catenin signalling [129].
Quercetin rescued LPS-induced impairment of osteogenesis in murine osteoblastic MC3T3-E1 cells by enhancing the protein levels of Wnt3 and β-catenin and decreasing the protein level of GSK3β [75]. Yuan et al. (2018) also found that quercetin protected rat bone-marrow-derived MSCs against TNF-α-induced inhibition of osteoblast differentiation. The activation of NF-κB and degradation of β-catenin were noted in cells treated with TNF-α. These results were reversed after treatment with TNF-α and quercetin [29]. In a preclinical animal experimentation setting, ovariectomy reduced the expression of β-catenin and treatment with isoquercitrin reversed the condition [37]. Taken together, these studies indicate that quercetin potentially orchestrates osteoblast differentiation and function through activation of the canonical Wnt/β-catenin signal transduction pathway ( Figure 6). Further investigation into the role of quercetin on Wnt antagonists such as DKK-1 and SOST is required to provide a better understanding of this signalling pathway.

BMP and TGF-β Signalling
BMPs are multifunctional growth factors that belong to the TGF-β superfamily. They bind to a tetrameric receptor complex and transduce intracellular signals via SMAD proteins to express osteoblastogenic genes directly or through Runx-2, suggesting their fundamental roles in skeletal development and bone repair [126]. Another mechanism involved in BMP-2 and TGF-β-mediated osteogenesis is the noncanonical-SMAD-independent pathway (i.e., the MAPK signalling), which phosphorylates Runx-2 and promotes its transcriptional activity [130]. Furthermore, BMP-2 signalling might work dependently with oestrogen-receptor signalling. Oestrogen-stimulated BMP-2 transcription was inhibited by selective oestrogen-receptor-modulator treatment in mouse MSCs [131]. Apart from the well-documented anabolic osteoinductive effects of BMP and TGF-β, they directly support osteoclast formation through the modulation of RANKL expression [132]. Phosphorylated SMAD proteins were detected on bone-marrow-derived osteoclast precursors during RANKL-stimulated differentiation, which was an indicator of BMP activation. The presence of noggin, a BMP antagonist, inhibited RANKL-mediated osteoclast differentiation [133].
The levels of BMP-2 and SMAD proteins and TGF-β1 were evidently increased in rat and mouse bone-marrow-derived MSCs treated with quercetin [34,62,64]. The cross-talk between the BMP-2 and oestrogen-receptor signalling pathways was confirmed by Pang et al. (2018). Their study found that the upregulation of BMP-2 and its downstream targets, as well as other osteogenic genes, by quercetin and oestrogen, was inhibited after the addition of ICI182780 (an oestrogen-receptor antagonist) [64]. In contrast, Yamaguchi and Weitzmann reported that quercetin failed to alleviate the suppressive effects of TNF-α on BMP-2-and TGF-β-induced SMAD activation in MC3T3-E1 cells. However, the osteogenic effects of quercetin were still considered to be due to its anti-NF-κB activity resulting in increased mineralisation in the cultures [74]. Forte and co-researchers reported a slight decrease in TGF-β1 level in the presence of quercetin in an in vitro osteoblast-osteoclast-endothelial cell triculture model. It was noteworthy that the reduction of TGF-β1 level exerted an effect on both osteoblast and osteoclast performance [85].
BMP and TGF-β signalling have been demonstrated to positively correlate with osteoblastogenesis and osteoclastogenesis. Quercetin exerts complex competing effects on BMP-and TGF-β-mediated signalling in bone cells (Figure 7). Further studies are recommended to confirm the effects of quercetin on these signalling pathways in osteoclasts.

Regulation of Angiogenesis
Angiogenesis (defined as the formation of new blood vessels) and osteogenesis are the foundational steps of bone regeneration and bone repair. In a bone defect, angiogenesis precedes osteogenesis to provide key elements (such as oxygen, nutrients, minerals, growth factors and bone precursor cells) required for osteogenic processes at the site of injury [134]. Vascular endothelial growth factor (VEGF) is the master player in coupling angiogenesis and osteogenesis, capable of promoting endothelial cells' migration and proliferation, as well as stimulating the secretion of osteogenic factors [135]. As an angiogenic inducer, angiogenin-1 (ANG-1) also enhances osteoblast differentiation, bone matrix deposition and mineralisation [136]. Basic fibroblast growth factor (bFGF) promotes vascular tissue regeneration, osseous formation and bone remodelling after the implantation of tissue-engineered bone [137]. TGF-β activates VEGF signalling, which is essential for angiogenesis, as well as phosphorylation of SMAD and protein kinase B (Akt), which are required for osteoinduction [138]. The interaction of BMP-2 and TGF-β with their respective receptors results in the upregulation of osteogenic genes and osteoclastogenic genes via activation of SMAD proteins and/or inhibition of MAPK. Quercetin may cause positive (indicated by the green arrows) or negative effects (indicated by the red arrows) on the levels of BMP-2, TGF-β and SMAD protein in bone cells. Abbreviations: AP-1 = activator protein-1; BMP-2 = bone morphogenetic protein-2; BMPR = bone morphogenetic protein-2 receptor; ERK = extracellular signal-regulated kinase; JNK = c-Jun N-terminal kinase; MAPK = mitogen-activated protein kinase; Runx-2 = runt-related transcription factor 2; SMAD = Suppressor of mothers against decapentaplegic; TGF-β1 = transforming growth factor-beta 1; TGFβR = transforming growth factor-beta receptor

Regulation of Angiogenesis
Angiogenesis (defined as the formation of new blood vessels) and osteogenesis are the foundational steps of bone regeneration and bone repair. In a bone defect, angiogenesis precedes osteogenesis to provide key elements (such as oxygen, nutrients, minerals, growth factors and bone precursor cells) required for osteogenic processes at the site of injury [134]. Vascular endothelial growth factor (VEGF) is the master player in coupling angiogenesis and osteogenesis, capable of promoting endothelial cells' migration and proliferation, as well as stimulating the secretion of osteogenic factors [135]. As an angiogenic inducer, angiogenin-1 (ANG-1) also enhances osteoblast differentiation, bone matrix deposition and mineralisation [136]. Basic fibroblast growth factor (bFGF) promotes vascular tissue regeneration, osseous formation and bone remodelling after the implantation of tissue-engineered bone [137]. TGF-β activates VEGF signalling, which is essential for angiogenesis, as well as phosphorylation of SMAD and protein kinase B (Akt), which are required for osteoinduction [138].
Using MSCs derived from rat bone marrow, Zhou and colleagues conducted two in vitro studies to investigate the effects of quercetin on angiogenic factors. The expressions of VEGF, ANG-1, bFGF and TGF-β were higher in MSCs obtained from bone marrow of normal and ovariectomised rats [34,63]. Mechanistically, Akt was significantly phosphorylated in rat MSCs under quercetin treatment The interaction of BMP-2 and TGF-β with their respective receptors results in the upregulation of osteogenic genes and osteoclastogenic genes via activation of SMAD proteins and/or inhibition of MAPK. Quercetin may cause positive (indicated by the green arrows) or negative effects (indicated by the red arrows) on the levels of BMP-2, TGF-β and SMAD protein in bone cells. Abbreviations: AP-1 = activator protein-1; BMP-2 = bone morphogenetic protein-2; BMPR = bone morphogenetic protein-2 receptor; ERK = extracellular signal-regulated kinase; JNK = c-Jun N-terminal kinase; MAPK = mitogen-activated protein kinase; Runx-2 = runt-related transcription factor 2; SMAD = Suppressor of mothers against decapentaplegic; TGF-β1 = transforming growth factor-beta 1; TGFβR = transforming growth factor-beta receptor Using MSCs derived from rat bone marrow, Zhou and colleagues conducted two in vitro studies to investigate the effects of quercetin on angiogenic factors. The expressions of VEGF, ANG-1, bFGF and TGF-β were higher in MSCs obtained from bone marrow of normal and ovariectomised rats [34,63]. Mechanistically, Akt was significantly phosphorylated in rat MSCs under quercetin treatment [34]. A quercetin-copper (II) complex was also found to promote angiogenesis, evidenced by increased blood vessel size, length and junctions, in a standard in vivo angiogenesis assay [87]. In animals, quercetin-loaded hydroxyapatite bioceramic microspheres promoted blood-vessel formation after eight weeks in female Sprague-Dawley rats following surgeries for bilateral ovariectomy and femoral bone defect creation [34]. Fayed et al. (2019) reported higher VEGF expression in isoquercitrin-treated ovariectomised rats [37]. Notably, quercetin potentially promotes angiogenesis in vitro and in vivo, leading to the enhancement of bone regeneration (Figure 8).
increased blood vessel size, length and junctions, in a standard in vivo angiogenesis assay [87]. In animals, quercetin-loaded hydroxyapatite bioceramic microspheres promoted blood-vessel formation after eight weeks in female Sprague-Dawley rats following surgeries for bilateral ovariectomy and femoral bone defect creation [34]. Fayed et al. (2019) reported higher VEGF expression in isoquercitrin-treated ovariectomised rats [37]. Notably, quercetin potentially promotes angiogenesis in vitro and in vivo, leading to the enhancement of bone regeneration (Figure 8).

Perspectives
The pharmacokinetics of quercetin is an important consideration in understanding its pharmacological effects in vivo and its establishment as an agent for improving bone health. Dietary fat affects the rate and extent of quercetin absorption. A randomised cross-over study found that ingestion of quercetin aglycone together with a standardised formulated high-fat breakfast increased the maximum concentration of total plasma quercetin as compared to those who ate a low-fat breakfast. Hence, the consumption of adequate dietary fat assists the micellisation efficiency of quercetin aglycone in the intestine for absorption [139]. Quercetin has a low oral bioavailability [140]. Generally, ingested flavonoids diffuse passively from the intestinal lumen into enterocytes in unmetabolised form. They can be either metabolised in enterocytes or rapidly transported to the liver for metabolism via phase I and II metabolism. The produced quercetin metabolites are then transported into the systemic circulation for distribution to body tissues [55]. In terms of excretion, the oral clearance of quercetin is rapid, with quertecin having a short half-life in the blood [141]. Various strategies have been employed to overcome these drawbacks and improve the biological performance of quercetin. The use of nanoparticles, microspheres [72], transfersomes [41], hydroxyapatite [68,84,85], transition metals [89] or polyethylene glycolylated graphene oxide [88] functionalised with quercetin helps to promote sustained drug-delivery performance to achieve longterm effects while enhancing rapid osteointegration, bone regeneration and biomechanical properties.
It is noteworthy that higher doses of quercetin exerted either suppressive or lessened activity on the expression of osteoblast-specific genes, osteoblast growth and mineralisation in some of the in Figure 8. The effects of quercetin on the regulation of angiogenesis. Quercetin increases angiogenic factors (TGF-β, VEGF, ANG-1 and bFGF) to facilitate endothelial-cell proliferation and migration as well as vascular tissue regeneration. Additionally, these angiogenic factors play an important role in promoting osteogenesis. Abbreviations: Akt = protein kinase B; Ang-1 = angiogenin-1; bFGF = basic fibroblast growth factor; SMAD = Suppressor of mothers against decapentaplegic; TGF-β = transforming growth factor-beta; VEGF = vascular endothelial growth factor.

Perspectives
The pharmacokinetics of quercetin is an important consideration in understanding its pharmacological effects in vivo and its establishment as an agent for improving bone health. Dietary fat affects the rate and extent of quercetin absorption. A randomised cross-over study found that ingestion of quercetin aglycone together with a standardised formulated high-fat breakfast increased the maximum concentration of total plasma quercetin as compared to those who ate a low-fat breakfast. Hence, the consumption of adequate dietary fat assists the micellisation efficiency of quercetin aglycone in the intestine for absorption [139]. Quercetin has a low oral bioavailability [140]. Generally, ingested flavonoids diffuse passively from the intestinal lumen into enterocytes in unmetabolised form. They can be either metabolised in enterocytes or rapidly transported to the liver for metabolism via phase I and II metabolism. The produced quercetin metabolites are then transported into the systemic circulation for distribution to body tissues [55]. In terms of excretion, the oral clearance of quercetin is rapid, with quertecin having a short half-life in the blood [141]. Various strategies have been employed to overcome these drawbacks and improve the biological performance of quercetin. The use of nanoparticles, microspheres [72], transfersomes [41], hydroxyapatite [68,84,85], transition metals [89] or polyethylene glycolylated graphene oxide [88] functionalised with quercetin helps to promote sustained drug-delivery performance to achieve long-term effects while enhancing rapid osteointegration, bone regeneration and biomechanical properties.
It is noteworthy that higher doses of quercetin exerted either suppressive or lessened activity on the expression of osteoblast-specific genes, osteoblast growth and mineralisation in some of the in vitro studies included [34,47,48,[63][64][65]74,76,90,92,93]. The bioactivity of quercetin might depend on the type and position of the sugar attachments of the flavonoid [55]. In other words, a high concentration of quercetin does not always imply its efficacy and effectiveness. Meanwhile, dose-dependent inhibition of osteoclast proliferation and dose-dependent enhancement of osteoclast apoptosis were consistently observed in quercetin-treated cells. In this context, the selection of the optimum dose and type of quercetin to exert bone-protecting effects in humans requires careful consideration and further validation from clinical trials.
Quercetin has been tested in humans for its anti-hypertensive effects [142][143][144][145], lipid-lowering activities [143,145,146] and reduction of upper respiratory tract infections [147]. The dosages used in these clinical trials ranged from 150-1000 mg/day. The safety profile of quercetin has also been evaluated. No notable adverse event was observed in individuals taking up to 5000 mg quercetin daily, characterised by no significant changes in full blood count, metabolic and cholesterol panels, blood coagulation, serum electrolytes, and liver and kidney function [143,148]. In animals, the no-observed-adverse-effect level (NOAEL) for quercetin was found to be 300 mg/kg/day in hamsters and 416 mg/kg/day in rats, which was equivalent to~2400 mg/day and~4000 mg/kg for a 60 kg adult human, respectively [149]. Since quercetin has a high NOAEL, it has been presumed that there are no theoretically expected side effects at high doses of quercetin, which requires further validation. However, mild stomach discomfort may develop if quercetin is taken before a meal [148]. Based on its actions on bone, an interventional double-blind placebo randomised controlled study investigated the effects of quercetin supplementation on bone mineralisation biomarkers in patients with type 2 diabetes mellitus. In this study, patients receiving quercetin oral supplement at 500 mg/day for three months had higher levels of serum osteocalcin, calcium and vitamin D compared to their pretreatment values [28]. However, the direct effects of quercetin on bone in osteoporotic patients are still unconfirmed.

Conclusions
In conclusion, quercetin may represent a useful prophylactic or therapeutic option in the management of skeletal diseases, with the evidence showing that the effects of quercetin on the skeleton are primarily protective. Different types of quercetin formulation have been used, including its free form, derivatives or combined with other compounds. Among these formulations, quercetin derivatives seem to have more potent bone-protective effects than free quercetin and the combination of other phytochemicals may inhibit the anti-osteoporotic effects of quercetin. This review has expanded our understanding of the bone-protecting effects of quercetin, particularly in terms of the regulation of bone mass via control of bone cell differentiation and function. Even though documented works exist to both support and refute the bone-conserving possibility of quercetin, the net effect of quercetin on bone likely depends on the multiple targets influenced by quercetin that modulate the bone microenvironment. Several gold-standard protocol recommendations for future studies are suggested. In in vivo experiments, quercetin at doses lower than the NOAEL can be used to study its beneficial effects on the skeleton. Different routes of administration can be tested for different medical conditions, for instance oral gavage for anti-osteoporotic effects and local injection for anti-fracture effects. Aged animals instead of sexually mature animals might be a more appropriate model. The bone-sparing effects of quercetin must be compared with standard anti-osteoporotic treatments. For cellular studies, three-dimensional co-cultures of osteoblast and osteoclast can be adopted to mimic the in vivo skeletal microenvironment. Secondly, osteoblast cultures should not be subjected to differentiation media for proliferation assays and osteoblast culture should be incubated in differentiation media for differentiation assays. Thirdly, more research in osteocytes is needed, but this is limited to the number of commercial cell lines available. Fourthly, the use of PBMCs from osteoporotic patients is recommended to showcase the effects on osteoclasts more accurately. Human studies investigating the clinical utility of quercetin and its structurally relevant compounds on bone health are limited, thus presenting a subject for further investigation.