Introduction
The vertebrate skeleton is composed of two specialised tissues, cartilage and bone. Bone is essential for human beings as its main role is to provide structural support for the body, protect the organs, contribute to haematopoiesis and store minerals(Reference Pettifor, Prentice, Cleaton-Jones, Gibney, Macdonald and Roche1). Three distinctly different cell types can be found within bone: the osteoblasts responsible for the synthesis and mineralisation of bone, the osteoclasts implicated in bone tissue resorption and the osteocytes which are the most abundant cells present in the adult skeleton(Reference Sommerfeldt and Rubin2). The integrity of the skeleton requires a dynamic balance between bone formation and bone resorption (Fig. 1). If a dysfunction occurs in these two processes, pathologies, such as osteoporosis, can develop.
Fig. 1 Bone remodelling cycle (adapted from University of Michigan(122)). (1), Quiescence; (2), osteoclastic resorption; (3), osteoblast recruitment; (4), osteoblastic formation; (5), bone mineralisation.
Osteoporosis is the most common of all bone diseases in adults, especially in old age. This bone disorder is characterised by an increased risk of fracture due to a reduction in both bone quantity and quality. The concerns about the medical, social and financial impacts of this disease are increasing. Therefore it is important to develop strategies and recommendations not only for the treatment but also for the prevention of osteoporosis. Most of the current pharmacological approaches focus on inhibiting bone resorption in those who already have the disease or in those who are at risk of developing the disease(Reference Simon3, Reference Suzuki, Sekiguchi and Asano4). On the contrary, recent research has given insight into dietary components that may optimise bone mass and stimulate bone formation(Reference Mundy5). Concerning dietary matter, several nutritional factors are implicated in ensuring optimal bone health. Numerous studies have reported the essentiality of Ca and vitamin D for bone health(Reference New6–Reference Heaney8). However, in many countries in Europe and North America, where the dietary intake of Ca is adequate for most individuals compared with recommended daily allowances, very high rates of osteoporosis are nevertheless observed. These observations suggest that other nutrients and dietary factors may influence bone and mineral homeostasis and may be important for long-term bone health. Specifically, it is now well recognised that the human diet contains, in addition to essential macro- and micronutrients, a complex array of naturally occurring bioactive molecules, namely phytonutrients that may provide long-term health benefits. Indeed, these compounds have been linked to reductions in the risk of major chronic diseases including osteoporosis.
As previously described by Habauzit & Horcajada(Reference Habauzit and Horcajada9), phytonutrients and especially polyphenols can act on both osteoblasts and osteoclasts to modulate bone metabolism. The aim of the present paper is to review how they can modulate osteoblastic function responsible for bone formation and to report the cellular mechanisms involved.
Osteoblast metabolism
Bone formation is a tightly regulated process characterised by a sequence of events starting with the commitment of osteoprogenitor cells from mesenchymal cells, their differentiation into immature pre-osteoblasts and then into mature functional osteoblasts(Reference Marie10).
Characteristics of osteoblasts and their lineage
Morphologically, osteoblasts have a cuboidal shape and they form a dense monolayer of cells at the bone surface. Functionally, osteoblasts are the specialised cells within bone that produce extracellular matrix and also regulate mineralisation(Reference Sommerfeldt and Rubin2). More precisely, bone tissue consists of hydroxyapatite crystals and various extracellular matrix proteins (i.e. type I collagen, osteocalcin, osteonectin, osteopontin, bone sialoprotein and proteoglycans) secreted and deposited by mature osteoblasts(Reference Bringhurst, Demay, Krane, Kasper, Braunwald and Fauci11).
Osteoblasts are derived from precursors originating in the bone marrow. These precursors are multipotent mesenchymal stem cells that can differentiate into several kinds of tissue-specific cells such as osteoblasts, myoblasts, adipocytes and chondrocytes. Each differentiation pathway can be specifically regulated by transcription factors such as runt-related transcription factor-2 (Runx2)/Osterix, MyoD family, PPARγ and SOX5/6/7, respectively(Reference Katagiri and Takahashi12).
Regulation of osteoblastogenesis
In the process of osteoblastogenesis, different stages are involved including proliferation, extracellular matrix synthesis, maturation and mineralisation. Each stage is regulated by the coordinated expression of major transcription factors(Reference Marie10). The most important transcription factors controlling bone formation are Runx2, Osterix, β-catenin, activating transcription factor-4, activator protein-1 (AP-1) and CCAAT/enhancer binding proteins (C/EBP). Other transcription factors belonging to homeobox proteins (Msx and Dlx proteins) or to helix-loop-helix proteins (Id and Twist) may play a role in osteoblast development(Reference Marie10). Furthermore, the expression of all these transcription factors is known to be modulated by several hormones (parathyroid hormone, oestrogens, glucocorticoids, 1,25-dihydroxyvitamin D)(Reference Strewler13) or growth factors (bone morphogenetic protein (BMP), transforming growth factor-β, insulin-like growth factor-1, fibroblast growth factor-2)(Reference Baylink, Finkelman and Mohan14).
Osteoblast differentiation is stimulated by some hormones and local factors which act on signalling pathways in cells within the osteoprogenitor lineage(Reference Kalajzic, Staal and Yang15). Various pathways such as BMP, transforming growth factor-β, insulin-like growth factor, fibroblast growth factor, Hedgehog, WNT (wingless-type MMTV integration site family) and mitogen-activated protein kinase (MAPK) have been implicated(Reference Kalajzic, Staal and Yang15–Reference Huang, Yang and Shao19) (Fig. 2). Many markers are available and currently used to study the influence of several factors on osteoblast developmental stages or functions. Indeed, alkaline phosphatase (ALP) collagen type I, osteopontin and bone sialoprotein are early markers of the osteoblast phenotype while osteocalcin plays a role in mineralisation(Reference Stein and Lian20) (Fig. 3). Cytokines and/or growth factors produced by osteoblasts and other cells have an effect on differentiation and the maturation process through binding to the proteins of bone extracellular matrix. Mature osteoblasts surrounded by bone matrix become osteocytes. They probably function as mechanosensors which regulate the bone response to mechanical stimuli(Reference Ehrlich and Lanyon21). Activated osteocytes produce signalling molecules such as NO, which could stimulate the activity of osteoblasts and osteoclasts. Intercellular signalling is essential for bone adaptation(Reference Burger, Klein-Nulend and Smit22). Thus, bone cells cooperate together and regulate their own functions.
Fig. 2 Signal-transduction pathways implicated in regulation of osteoblast functions. BMP, bone morphogenetic protein; TGFβ, transforming growth factor-β; IGF-1, insulin-like growth factor-1; FGF, fibroblast growth factor; WNT, wingless-type MMTV integration site family; BMPR, bone morphogenetic protein receptor; TGFβR, transforming growth factor-β receptor; IGFR, insulin-like growth factor receptor; FGFR, fibroblast growth factor receptor; Lrp5/6, LDL-related protein 5/6; SMAD, mothers against decapentaplegic homologue; MAPK, mitogen-activated protein kinase; ERK, extracellular signal-regulated kinase; JNK, c-Jun N-terminal kinase; PKC, protein kinase C; Gli, glioma-associated oncogene; Runx2, runt-related transcription factor-2; AP-1, activator protein-1; ATF4, activating transcription factor-4; TF, transcription factors; Col1, type 1 collagen; ALP, alkaline phosphatase; OPN, osteopontin; BSP, bone sialoprotein; OCN, osteocalcin.
Fig. 3 Main factors produced by osteoblasts and used as markers of different developmental stages (adapted from Stein & Lian(Reference Stein and Lian20)). ECM, extracellular matrix; AP-1, activator protein-1; Col1, type 1 collagen; ALP, alkaline phosphatase; OPN, osteopontin; OCN, osteocalcin; BSP, bone sialoprotein.
Osteoblasts regulate osteoclastogenesis and osteoblasts play a direct and essential role in the regulation of osteoclast function within the bone microenvironment. Osteoclasts derive from haematopoietic progenitors (monocyte/macrophage lineage) in the bone marrow and they are formed by the fusion of precursor cells(Reference Ash, Loutit and Townsend23). They are responsible for bone resorption and therefore for normal skeletal development (growth and modelling). Macrophage colony-stimulated factor produced by osteoblasts/stromal cells is crucial for osteoclast formation and development(Reference Felix, Cecchini and Fleisch24–Reference Kodama, Nose and Niida26). It has been shown that the osteoprotegerin (OPG)/receptor activator of NF-κB (RANK) ligand (RANKL)/RANK system plays a role in osteoclastogenesis as the dominant and final mediator of this process.
There are many factors that can influence osteoblastogenesis and/or osteoclastogenesis. It is very important to maintain the balance between these cellular processes which correspond with bone formation and bone resorption. An imbalance between these processes could lead to a decrease in bone mineral density and thus to an increase in the risk of osteoporosis during ageing(Reference Sambrook and Cooper27).
Cellular models for bone formation
The molecular events that regulate osteoblastic differentiation and function are frequently studied using in vitro cellular models. In order to explore the impact of pharmacological agents or dietary molecules, various osteogenic cell-culture systems can be used: immortal cell lines (for example, murine osteoblastic cells (MC3T3-E1) or fetal human osteoblastic cell line (hFOB)) or genetically atypical cell lines derived from tumours (for example, sarcoma osteogenic cells (SaOS-2) or human osteosarcoma cell line (MG-63)). However, tumour origins and genetic alterations associated with the prolonged culture of cell lines can raise questions about the extent to which they match the normal in vivo physiology. That is why in vitro model systems using primary osteoblast precursor cells derived from either the developing calvaria of neonatal rodents or from the bone marrow (murine bone marrow mesenchymal stem cells, human bone marrow-derived mesenchymal stem cells) should be promoted.
Polyphenols: structure, metabolism and cellular bioactivities
Structure
Phytochemicals are a large group of plant-derived non-nutritive chemical compounds divided into several classes: phenolics (polyphenols), carotenoids, alkaloids, sterols, terpenes and fibre. Phytochemicals produced by plants are used to protect themselves but recent research demonstrates that certain molecules may protect humans against some pathologies such as cancer, cardiovascular diseases and osteoporosis(Reference Scalbert, Manach and Morand28).
Polyphenols can be divided into different groups depending on the number of phenol rings they contain and on the structural elements bound to these rings. Considering this, polyphenols have been classified as phenolic acids, flavonoids, stilbenes, tannins, coumarins and lignans. Moreover, among flavonoids, six subclasses exist and share a common structure of two aromatic rings (A and B) bound together by three carbon atoms that form an oxygenated heterocycle (ring C). These are flavones, flavonols, flavanones, isoflavones, flavanols (catechins and proanthocyanidins) and anthocyanidins(Reference Manach, Williamson and Morand29).
Metabolism of polyphenols
Knowledge on phenolic metabolites generated in an organism is useful to choose the appropriate molecule for in vitro studies. Biological activities of polyphenols are often examined in cell lines usually treated with aglycones or polyphenol-rich extracts (glycoside forms) at concentrations which elicit an appropriate response(Reference Kroon, Clifford and Crozier30). The compounds abundant in plants are not exactly the same as those found in plasma. Neither aglycones (except for green tea catechins) nor forms found in the diet reach the blood and tissue. The compounds, after consumption, are metabolised by the organism and/or microflora enzymes to conjugates of glucuronate or sulfate (with or without methylation across the catechol functional group) during absorption to the bloodstream and/or eventually eliminated(Reference Manach, Scalbert and Morand31). The circulating forms may possess different biological properties within cells and tissues compared with polyphenol aglycones. Moreover, the maximum plasma concentration of polyphenols from foods is in the range of 0·1–10 μm. As far as possible, conjugates of molecules should be used for in vitro studies in the physiological concentration(Reference Williamson, Barron and Shimoi32) even if they are poorly or not absorbed by cells(Reference Serra, Mendes and Bronze33). Unfortunately, many metabolites are not commercialised due to problems of isolation and synthesis and it is difficult to obtain them for in vitro experiments(Reference Williamson, Barron and Shimoi32). However, some cells (in organisms) could potentially deconjugate the circulating molecules through the activity of enzymes such as β-glucuronidase and be present in cells in the aglycone form(Reference O'Leary, Day and Needs34). Hence, this enzyme in human tissues could play a role in the turnover of flavonoids (conjugated/non-conjugated forms). It has also been shown that cells in culture are able to take up aglycone forms of flavonoids such as diosmetin, hesperetin and naringenin and conjugate them(Reference Serra, Mendes and Bronze33). Treatments of cells with conjugated forms are the most representative method to investigate the molecular mechanism of polyphenols. However, at the present time, there are no reported studies that have used conjugated forms in osteoblastic cells. In the present review, concerning nutrition we focused on studies of molecules at physiological doses (up to 10 μm) corresponding to the concentration of polyphenols found in plasma after polyphenol-containing food consumption.
Cellular bioactivities of polyphenols
The chemical structure of compounds is related to their biological activity and thus activation of different signalling pathways could be involved(Reference Khlebnikov, Schepetkin and Domina35) in their mode of action.
Flavonoids and more recently their metabolites (sulfates, glucuronides, O-methylated forms) have been shown to act on cells, independently of their antioxidant capacity(Reference Williamson, Barron and Shimoi32, Reference Loke, Proudfoot and McKinley36–Reference Suri, Taylor and Verity38). They may interact with specific proteins involved in intracellular signalling pathways(Reference Schroeter, Boyd and Spencer39). They can selectively act on protein kinase and lipid kinase signalling cascades such as MAPK, protein kinase C, protein kinase B (PKB/AKT) (serine/threonine kinase), tyrosine kinases and phosphoinositide 3-kinase (PI3K) (for a review, see Williams et al. (Reference Williams, Spencer and Rice-Evans40)). The regulation can occur at transcriptional and/or post-translational levels by the modulation of gene expression and/or phosphorylation of target proteins, respectively. The positive or negative effects of flavonoids depend on both cell type and metabolic disorder. For example, genistein (isoflavone) can inhibit protein tyrosine kinases and thus cause cell cycle arrest and apoptosis in leukaemic cells(Reference Spinozzi, Pagliacci and Migliorati41) whereas it can stimulate osteoblast differentiation(Reference Yamaguchi42, Reference Sugimoto and Yamaguchi43). The flavonoid-stimulated signalling pathways are more precisely described by Williams et al. (Reference Williams, Spencer and Rice-Evans40) and Orzechowski et al. (Reference Orzechowski, Ostaszewski and Jank44).
Flavonoids also have an impact at the transcriptional level. They may affect the expression of NF-κB and AP-1 composed of c-jun/c-fos complexes which are implicated in the MAPK pathway. Indeed, it has been shown that green tea polyphenols such as ( − )-epigallocatechin-3-gallate (EGCG) can modulate phosphorylation of extracellular signal-regulated kinase (ERK) 1/2, c-Jun N-terminal kinase (JNK) and p38 MAP kinase in normal human epidermal keratinocytes(Reference Katiyar, Afaq and Azizuddin45), which shows that this polyphenol can affect the MAPK signalling pathway(Reference Balasubramanian, Efimova and Eckert46). In the same way, resveratrol (trans-3,4′,5-trihydroxystilbene), a natural stilbene present abundantly in grape skins and red wine, has cardioprotective effects(Reference Bradamante, Barenghi and Villa47), cancer-preventive properties(Reference Aziz, Kumar and Ahmad48) and an impact on bone metabolism(Reference King, Bomser and Min49). It has been shown that resveratrol could act through ERK1/2 and p38 MAPK signalling(Reference Das, Tosaki and Bagchi50), similar to flavonoids.
Beside these indirect molecular effects, it is possible that the positive health effects of polyphenols occur through their direct action on response elements in the regulatory regions of target genes(Reference Orzechowski, Ostaszewski and Jank44). Some phenolic compounds can bind to specific cytoplasmic receptors which can play a role as DNA binding transcription factors to regulate target gene expression. This mechanism is well known for glucocorticoids(Reference Linder, Falkner and Srinivasan51) and oestrogens(Reference Klinge52) which bind to glucocorticoid and oestrogen receptors, respectively. These complexes enter the nucleus to stimulate the expression of glucocorticoid- or oestrogen-related genes. Some plant compounds, so-called phyto-oestrogens, can interact with oestrogen receptors. This is the case for soyabean isoflavones (mainly genistein, daidzein and glycitein) as well as flavonoids such as luteolin and naringenin that exhibit oestrogenic properties(Reference Kuiper, Lemmen and Carlsson53). These phytochemicals play a role as partial agonists and/or antagonists of oestrogen receptors and can modulate the oestrogen-dependent pathway(Reference Bennetau-Pelissero, Latonnelle and Sequeira54).
Mechanisms of polyphenols in osteoblastic cells
Cellular processes are complex and many components interact together. Cell signalling can be activated by interaction of compounds with a receptor on the cell surface and/or by interaction with extracellular proteins(Reference Williamson, Barron and Shimoi32). To understand cell signalling during development it is better to study whole signalling networks, not just their individual elements(Reference Niehrs and Meinhardt55). Moreover, gene expression depends on cellular development and time of molecule exposure. Unfortunately, most experiments reported to date were performed to study individual elements of signalling pathways at precise time-points.
Studies of modulation of cell functions by polyphenols have been performed in different in vitro models such as cancer cells(Reference Noe, Penuelas and Lamuela-Raventos56–Reference Yang, Lambert and Hou58), epithelial cells(Reference O'Prey, Brown and Fleming59, Reference Vittal, Selvanayagam and Sun60) or skeletal cells. However, little is known regarding osteoblastic cells. As previously reviewed by Habauzit & Horcajada(Reference Habauzit and Horcajada9), phenolic compounds can act on bone metabolism. Osteoblast proliferation, differentiation and mineralisation as well as osteoclast functions may be modulated by polyphenols. But how do cells read and translate the signal induced by polyphenols? The present review will now report what is known about molecular mechanisms activated when polyphenols (listed in Fig. 4) interact with osteoblasts, leading to a modulation of their function.
Fig. 4 Continued.
Effect of polyphenols on transcription factors
It is known that macro- and micronutrients may enter cells and then interact with transcription factors which are known to influence target gene expression(Reference Muller and Kersten61). Considering the field of bone metabolism and more specifically osteoblasts, it has been reported that some polyphenols such as EGCG, resveratrol and icariin can modulate some transcription factors which in turn can activate osteoblast-related genes (Table 1).
Table 1 Polyphenol regulation of transcription factors
Runx2, runt-related transcription factor-2; OB, osteoblasts; ↑ , increase; MBMSC, murine bone marrow mesenchymal stem cells; EGCG, ( − )-epigallocatechin-3-gallate; HBMSC, human bone marrow-derived mesenchymal stem cells; SaOS-2, sarcoma osteogenic cells; MC3T3-E1, murine osteoblastic cells; POB, primary osteoblasts; AP-1, activator protein-1; ↓ , decrease.
Runt-related transcription factor-2 and Osterix
There are many transcription factors implicated in the regulation of osteoblast functions(Reference Marie10). Among them, Runx2 and Osterix transcription factors are essential molecules to induce osteoblastic differentiation and can activate some genes implicated in this process(Reference Huang, Yang and Shao19, Reference Komori62, Reference Celil, Hollinger and Campbell63).
BMP signalling, well known to be implicated in osteoblast differentiation, may up-regulate expression of Runx2. This factor is known to activate the transcription of various osteoblast-related genes (leading to osteoblast differentiation)(Reference Yamaguchi, Komori and Suda16, Reference Ducy64). Another important transcription factor involved in modulating osteoblast function is Osterix, mediated by multiple signalling pathways(Reference Celil, Hollinger and Campbell63). Indeed, in an animal experiment, expression of osteoblast markers such as osteonectin, osteopontin, osteocalcin, and bone sialoprotein was decreased in Osterix-null mutant mice(Reference Nakashima, Zhou and Kunkel65) despite the fact that the Runx2 mRNA level remained unchanged compared with wild-type mice. In contrast, the expression of Osterix was diminished in Runx2/Cbfa1-null mice. These results suggested that Osterix can be downstream of Runx2 during osteoblast differentiation(Reference Nakashima, Zhou and Kunkel65). On the other hand, BMP-2-induced Osterix expression can be independent of Runx2. Indeed, Osterix expression was still induced by BMP-2 treatment in Runx2-null cells(Reference Lee, Kwon and Park66). Furthermore, it has been shown that during osteogenic lineage progression, in addition to the BMP/mothers against decapentaplegic homologue (SMAD) pathway, insulin-like growth factor-1 and MAPK signalling can mediate Osterix(Reference Celil, Hollinger and Campbell63).
EGCG has been demonstrated to have a stimulatory effect on osteogenesis of murine bone marrow mesenchymal stem cells. This green tea catechin after 48 h treatment, at concentrations of 1 and 10 μm, significantly up-regulated the expression of transcription factors Runx2 and Osterix as well as osteoblast markers ALP and osteocalcin in cells compared with non-treated control cells. Moreover the long-term treatment of EGCG can increase ALP activity and eventually stimulate mineralisation, in spite of its inhibitory effect on proliferation(Reference Chen, Ho and Chang67). The protein level of Runx2 was also studied at different time points in the osteoblast-like cell line SaOS-2. After 48 h treatment Runx2 expression was decreased by EGCG at 1–5 μm(Reference Vali, Rao and El-Sohemy68). The ALP activity and mineralisation in SaOS-2 human osteoblasts treated with EGCG were increased compared with the control cells(Reference Vali, Rao and El-Sohemy68). Independent experiments have shown stimulating activities of EGCG on cells; however, Chen et al. observed the up-regulation of Runx2 gene expression(Reference Chen, Ho and Chang67) and Vali et al. the decrease of Runx2 protein expression(Reference Vali, Rao and El-Sohemy68). It is common to observe that mRNA level does not correlate with protein level because expression of many genes can be regulated after transcription. To understand the molecular mechanisms it is necessary that transcript and protein levels are both assessed at the same time and in the same model.
Some isoflavones also have the capacity to increase ALP activity and mineralisation(Reference Yamaguchi42, Reference Sugimoto and Yamaguchi43, Reference De Wilde, Lieberherr and Colin69). It has been shown that incubation with 1 nm-daidzein for 48 h increased the nuclear content of Runx2/Cbfa1 protein in porcine osteoblasts(Reference De Wilde, Lieberherr and Colin69). This effect was blocked by the oestrogen receptor inhibitor ICI 182.780, which indicates that Runx2 activation is dependent on oestrogen-stimulated signalling.
Resveratrol (trans-3,4′,5-trihydroxystilbene), a natural stilbene, has an impact on bone metabolism(Reference King, Bomser and Min49) and, furthermore, trans-resveratrol has been shown to enhance the proliferation and differentiation of osteoblastic cells(Reference Mizutani, Ikeda and Kawai70). Dai et al. have demonstrated that this polyphenolic phyto-oestrogen at 1 μm concentration could increase Runx2 and Osterix mRNA levels in human bone marrow-derived mesenchymal stem cells as well as one of their target genes – osteocalcin(Reference Dai, Li and Quarles71). Addition of oestrogen receptor antagonist and ERK1/2 pathway inhibitors but not p38 MAPK pathway blockers attenuated resveratrol-induced gene expression(Reference Dai, Li and Quarles71).
Icariin, a flavonol glycoside isolated from herbs, was present in rat plasma after oral administration of Gushudan extract(Reference Liu, Liu and Lu72). This compound at different concentrations ranging from 0·1 nm to 10 μm induced osteogenic differentiation in MC3T3-E1 pre-osteoblasts and mouse primary osteoblasts(Reference Zhao, Ohba and Shinkai73). Moreover, the expression of Runx2 was up-regulated in a dose- and time-dependent manner.
These data support the involvement of both Runx2 and Osterix in the osteogenic effect of polyphenols, at a molecular level (Table 1).
Activator protein-1 and nuclear factor κB
AP-1 and NF-κB are also implicated in the regulation of genes involved in many important biological processes in osteoblasts as well as osteoclasts. It was claimed that AP-1 can be induced by both oxidant and antioxidant conditions and it is primarily transcriptionally regulated(Reference Pinkus, Weiner and Daniel74, Reference Fujioka, Niu and Schmidt75). Activating transcription factor, Jun and Fos are the part of the dimeric transcription factor complex AP-1 and are known activators of proliferation, differentiation and apoptosis-related genes(Reference Wagner76). Many bone-specific genes such as ALP, osteocalcin and collagen I have AP-1 sites in their promoters(Reference McCabe, Kockx and Lian77). Moreover, flavonoids are known to regulate AP-1 and NF-κB activity occurring as final stages in signal-transduction cascades(Reference Orzechowski, Ostaszewski and Jank44, Reference McCabe, Kockx and Lian77, Reference Frigo, Duong and Melnik78). NF-κB, a dimeric transcription factor composed of Rel (cRel), RelA (p65), RelB, NF-κB1 (p50) and NF-κB2 (p52) recognises a common DNA sequence (κB site). The activation of NF-κB in response to pro-oxidant conditions occurs by a post-translational mechanism, which involves dissociation of the inhibitory protein IκB and its phosphorylation(Reference Pinkus, Weiner and Daniel74). AP-1 and NF-κB transcription factors involved in MAPK signalling pathways are regulated by different mechanisms, but it appears that they could be activated simultaneously by the same agents. Thus these transcription factors work together and may modulate the activity of each other(Reference Fujioka, Niu and Schmidt75).
Pang et al. (Reference Pang, Ricupero and Huang79) have reported that kaempferol but not quercetin (both at 10 μm) blocked TNFα-induced NF-κB (p65) nuclear translocation in primary osteoblasts and inhibited the production of osteoclastogenic cytokines IL-6 and monocyte chemotactic protein-1. Recently, it has been demonstrated that quercetin (1–10 μm) could accelerate TNFα-mediated apoptosis in MC3T3-E1 cells via activation of caspases and the JNK pathway(Reference Son, Kook and Choi80). DNA binding activity of AP-1 and nuclear translocation of c-jun protein were increased after treating the MC3T3-E1 cells with TNFα and these effects were augmented in the presence of quercetin(Reference Son, Kook and Choi80). Quercetin at 10 μm alone did not have any effect on the nuclear induction of c-jun but was able to increase the level of Jun D translocation. However, this quercetin effect could not be found in primary osteoblasts(Reference Son, Kook and Choi80), this being consistent with the findings of Pang et al. (Reference Pang, Ricupero and Huang79).
Another phenolic compound, emodin (anthraquinone), present in the roots and bark of numerous plants of the genus Rhamus, stimulated differentiation and mineralisation and was able to activate NF-κB but not AP-1 in MC3T3-E1 cells(Reference Lee, Shin and Min81). Moreover, the assay to identify compounds activating BMP-2 promoter was performed after 2 d emodin treatment. Emodin at the range of 5–10 μm increased the activation of BMP-2 promoter containing NF-κB response element but not that without NF-κB(Reference Lee, Shin and Min81). It was suggested that NF-κB could be required for emodin anabolic activity.
The effect of the catechin EGCG on PGF2α-stimulated vascular endothelial growth factor synthesis via stress-activated protein kinase (SAPK)/JNK activation was studied in MC3T3-E1 cells(Reference Tokuda, Takai and Matsushima-Nishiwaki82). It has been shown that 60 min of pre-treatment with 10 μm-EGCG amplified PGF2α-induced phosphorylation of c-jun which is a part of the AP-1 complex and well known as a downstream effector of SAPK/JNK(Reference Tokuda, Takai and Matsushima-Nishiwaki82).
Regarding all the reported studies dealing with polyphenols at nutritional and physiological levels, there is good evidence that these compounds are able to interact with some well-known osteoblast-related transcription factors including Runx2, Osterix, AP-1 and NF-κB (Table 1) and possibly with all BMP-activated signalling pathways. Even though all the studies reported here provide some evidence of regulatory activity of specific transcription factors, it is difficult to make conclusions about precise molecular mechanisms of phenolics in osteoblastic cells. Moreover, if the present review focuses on osteoblasts, it is important to highlight the interface between polyphenols, bone formation and resorption. Indeed, it has been demonstrated that NF-κB and AP-1 are transcription factors also involved in osteoclast differentiation and their inhibition could play a role in the decrease of osteoclastogenesis.
A story beyond and around bone morphogenetic protein-stimulated pathways
Recent studies have shown that BMP induce osteoblast differentiation and play an important role in osteogenesis during development and in adult life(Reference Canalis, Economides and Gazzerro83). Furthermore, BMP stimulate the formation of mineralised bone-like nodules(Reference Yamashita, Ishii and Shimoda84). These proteins, by an autocrine action, may activate different signalling cascades. BMP regulatory molecules can transmit signals through SMAD-dependent and SMAD-independent pathways, including ERK, JNK, and p38 MAP kinase pathways(Reference Miyazono, Maeda and Imamura85). Members of the BMP family can bind to various serine/threonine kinase receptors I and II which are necessary for signal transduction(Reference Shi and Massague86). Moreover, BMP signalling can interact with other pathways such as transforming growth factor-β, Notch, Wnt, signal transducer and activator of transcription (STAT) and MAPK by cross-talk with them(Reference Miyazono, Maeda and Imamura85). Because BMP stimulate new bone formation(Reference Boden, Kang and Sandhu87), indicating a clear role in the regulation of osteoblast function, they represent molecular targets used to identify new agents to prevent osteoporosis(Reference Mundy5). BMP-2 is one of the major proteins implicated, as an important regulator, in bone metabolism.
Regulation of BMP signalling is important for cell functions and BMP activity can be regulated by many mechanisms (intra- and/or extracellular; for a review, see Canalis et al. (Reference Canalis, Economides and Gazzerro83)). One mechanism is regulation by extracellular antagonists such as noggin which can bind BMP and thus prevent stimulation of BMP receptors and cell signalling(Reference Zimmerman, De Jesús-Escobar and Harland88, Reference Groppe, Greenwald and Wiater89). Noggin selectively blocks BMP function; thus it can be used as a tool in cell biology to inhibit the BMP pathway(Reference Gazzerro, Gangji and Canalis90).
It has been shown that some phenolic compounds such as coumarin derivatives(Reference Kuo, Hsu and Chang91, Reference Tang, Yang and Chien92), piceatannol (stilbene)(Reference Chang, Hsu and Teng93), resveratrol (stilbene)(Reference Su, Yang and Zhao94) and myricetin (flavonoid)(Reference Hsu, Chang and Tsai95) increase BMP-2 production in osteoblastic cells (Table 2). Moreover, the ALP activity and osteocalcin production induced by these compounds was inhibited by noggin, suggesting that BMP-2 signalling is involved in osteoblast differentiation. Myricetin (3,3′,4′,4,5,5′,7-hexahydroxyflavone), a flavonoid present in grape juice and red wine, as well as piceatannol (3,3′,4,5′-tetrahydroxy-trans-stilbene) found in the skins of grapes, rhubarb and cane sugar exhibit potential healthy effects on organisms(Reference Wolter, Clausnitzer and Akoglu96, Reference Yanez, Vicente and Alcaraz97). More specifically, myricetin (1–20 μm)(Reference Hsu, Chang and Tsai95) and piceatannol (0·1–20 μm)(Reference Chang, Hsu and Teng93) dose dependently stimulated osteoblast differentiation at various stages and also mineralisation in both human osteoblastic cell line hFOB and human osteosarcoma cell line MG-63. It was demonstrated that myricetin induced phosphorylation of SMAD1/5/8 complex and p38 protein which can be abrogated by noggin pre-treatment. Addition of the specific inhibitor of p38 decreased ALP activity and osteocalcin secretion as well as phosphorylation of p38 in cells treated with myricetin. In this experiment, myricetin through BMP-2 activated not only the SMAD-dependent pathway but also the p38 kinase pathway that has been implicated in differentiation(Reference Hsu, Chang and Tsai95).
Table 2 Interaction of polyphenols with the bone morphogenetic protein (BMP) pathway
hFOB, fetal human osteoblastic cell line; ↑ , increase; MG-63, human osteosarcoma cell line; SMAD, mothers against decapentaplegic homologue; ERK, extracellular signal-regulated kinase; POB, primary osteoblasts; MC3T3-E1, murine osteoblastic cells.
Coumarin derivates such as imperatorin and bergapten increased BMP-2 gene expression in a time- and dose-dependent manner but not BMP-4 or BMP-7 in primary osteoblasts(Reference Tang, Yang and Chien92). On the contrary, icariin (a flavonol glycoside) induced the expression of BMP-4 in osteoblasts(Reference Zhao, Ohba and Shinkai73). Coumarin derivatives increased phosphorylation of SMAD1/5/8, p38 and ERK in a time-dependent manner. Pre-treatment with a p38 inhibitor and MAPK or ERK kinase (MEK) inhibitor markedly diminished coumarin-induced p38 and ERK1/2 phosphorylation, as well as BMP-2 gene expression(Reference Tang, Yang and Chien92). It has been observed that another coumarin derivate osthole (7-methoxy-8-isopentenoxycoumarin), present in Chinese medicine, used the same mechanisms to induce osteoblast differentiation in hFOB and MG-63 cells(Reference Kuo, Hsu and Chang91). Thus, BMP-2 protein appears to be required for coumarin-mediated maturation of osteoblasts.
Lee et al. have shown that emodin at the range of 2–5 μm accelerated osteoblast differentiation via stimulation of BMP-2 gene expression and the activation of PI3K-AKT/MAP kinases (especially JNK and p38). These effects were abolished by a PI3K inhibitor, suggesting that emodin-induced BMP-2 expression was dependent on PI3K-AKT/MAP kinase cascades(Reference Lee, Shin and Min81).
Results from these studies have shown that polyphenols were able to influence osteoblast differentiation and mineralisation through an increase in BMP-2 production, which in turn could activate phosphorylation of specific proteins (SMAD, p38, ERK1/2). Taken together, polyphenols can affect osteoblast function by the complex network of BMP-related signalling pathways (Table 2).
What about polyphenols and oestrogen receptor signalling?
Both oestrogen α and β receptors can be found in osteoblasts(Reference Kousteni, Bellido and Plotkin98–Reference Manolagas, Kousteni and Jilka100). Oestrogens can modulate gene expression through genomic and/or non-genomic mechanisms. Genomic transcriptional regulation is achieved through recruitment of oestrogen receptors to the promoter region of the target gene, either directly through interaction with DNA sequences (i.e. oestrogen receptor element) or through protein–protein interaction with other transcriptional factors such as AP-1 and NF-κB(Reference Ascenzi, Bocedi and Maria101, Reference Nilsson, Makela and Treuter102). Non-genomic action (called ‘rapid’) can also initiate gene transcription but through oestrogen activation of signal-transduction pathways which are insensitive to inhibitors of transcription and translation(Reference Losel, Falkenstein and Feuring103). However, it is possible that different oestrogen receptor ligands induce conformational changes in oestrogen receptors, blocking the activation of signalling cascades(Reference Watson, Bulayeva and Wozniak104). The anti-apoptotic effects of oestrogens and oestrogenic compounds on osteoblastic cells are mediated by oestrogen receptor-mediated ERK pathway activation, which is distinct from genomic actions of oestrogen receptors(Reference Prouillet, Maziere and Maziere105–Reference Kono, Oshima and Hoshi109). The oestrogen receptor–oestrogen complex interacts with Src tyrosine kinase and results in the rapid activation of ERK. ERK in turn is responsible for downstream activation of Elk-1, cAMP responsive element binding protein (CREB) and CCAAT/enhancer binding protein-β, transcription factors involved in anti-apoptotic effects of sex steroids on osteoblasts. Also inactivation by phosphorylation of a pro-apoptotic protein Bad takes place. On the other hand, JNK activity and downstream transcription driven by transcription factor AP-1 are down-regulated via a non-genomic oestrogen receptor-mediated pathway(Reference Kousteni, Han and Chen99, Reference Driggers and Segars110). In general, oestrogen acts on osteoblasts by modulating the activity of cytosolic kinases and subsequent post-transcriptional modification of transcription factors, which results in up- or down-regulation of gene expression(Reference Kousteni, Bellido and Plotkin98).
Isoflavones such as daidzein and genistein are classified as phyto-oestrogens and are thought to act via the oestrogen receptors to improve bone health. Plant-derived flavonoids and phyto-oestrogens can change oestrogen receptor-α and oestrogen receptor-β activities in different ways and act as oestrogen agonists and/or antagonists depending on tissue type and molecule concentration(Reference Kuiper, Lemmen and Carlsson53). Some nutritional flavonoids such as quercetin and naringenin acting as oestrogen-mimetics can activate p38/MAPK in the presence of oestrogen receptor-β(Reference Totta, Acconcia and Leone111). These molecules have an antioxidant function in plants but they could act on oestrogen receptors as agonists and/or antagonists in the organism(Reference Cassidy, Dalais, Gibney, Macdonald and Roche112). It has been shown that isoflavones have anabolic effects on bone metabolism by stimulating osteoblasts and decreasing osteoclast functions(Reference Yamaguchi42). The oestrogenic properties of isoflavones were investigated by Chen et al. (Reference Chen, Garner and Anderson113) in two hFOB osteoblastic cell lines expressing different levels of oestrogen receptors. The expression and the production of osteoclastogenesis-regulatory cytokines such as IL-6 and OPG were dependent on the level of oestrogen receptors. The human fetal osteoblastic cell lines hFOB1·19 (400 oestrogen receptors/nucleus) and hFOB/ER9 (8000 oestrogen receptors/nucleus) were treated with 10 nm-17β-oestradiol, 0·1–10 nm-genistein or daidzein and/or 10 μm-ICI (inhibitor of oestrogen receptor) for 48 h. IL-6 protein production was decreased by about 30–40 % in hFOB1·19 cells and 40–60 % in hFOB/ER9 cells by isoflavone treatments in comparison with the non-treated control cells. The effect of genistein and daidzein was slightly dose dependent and was greater in cells that were most abundant in oestrogen receptors (hFOB/ER9)(Reference Chen, Garner and Anderson113). Moreover, it has been shown that daidzein could enhance the cell content of oestrogen receptor-β, which seems to be involved in the daidzein action in osteoblasts(Reference De Wilde, Lieberherr and Colin69).
Concerning other polyphenols, ellagitannins and ellagic acid, mainly found in red fruit, such as raspberries, strawberries, blackcurrants and pomegranate(Reference Clifford and Scalbert114), are also oestrogenic but have not been studied as much. It has been shown that inhibitors of oestrogen receptors decreased the activity of ALP and mineralisation induced by ellagic acid in osteoblastic cells(Reference Papoutsi, Kassi and Tsiapara115). Indeed, ellagic acid may act as a natural selective oestrogen receptor modulator(Reference Papoutsi, Kassi and Tsiapara115) and may act through oestrogen receptors, but the molecular mechanisms are not well known and they need to be further investigated.
Resveratrol enhanced proliferation and osteoblastic differentiation in a dose-dependent manner in human bone marrow-derived mesenchymal stem cells(Reference Dai, Li and Quarles71). This stimulating effect could be induced by an oestrogen receptor-dependent mechanism coupled to ERK1/2 activation. Moreover, ICI completely abolished the resveratrol-induced phosphorylation of ERK1/2 and p38, suggesting that oestrogen receptors were implicated in resveratrol MAPK activation(Reference Dai, Li and Quarles71).
As reported above, some phenolic compounds can act on bone remodelling through the oestrogen receptors as well as via non-oestrogenic mechanisms, for example, resveratrol. This increases the difficulty associated with elucidating the molecular mechanisms of action of a given compound in a specific cell or tissue.
Polyphenols and osteoprotegerin/receptor activator of nuclear factor-κB ligand
Osteoblasts are implicated in the regulation of osteoclastogenesis through the OPG/RANKL/RANK regulatory system. OPG is a member of the TNF receptor family. This protein was characterised as an essential osteoclast differentiation factor expressed by osteoblastic/stromal cells and is involved in cell development(Reference Simonet, Lacey and Dunstan116, Reference Rodan and Martin117). Osteoblasts produce OPG, which acts as a decoy receptor for RANKL and thereby neutralises its function in osteoclastogenesis. Indeed, RANKL activates osteoclastogenesis by binding to RANK while OPG may inhibit the activation of osteoclasts and promote osteoclast apoptosis. Bone homeostasis seems to depend on the local RANKL:OPG ratio(Reference Lacey, Timms and Tan118, Reference Yasuda, Shima and Nakagawa119) and a mechanism involving the RANKL/OPG system is plausible because RANK is present in pre-osteoclastic cells(Reference Hsu, Lacey and Dunstan120).
Within the polyphenols there is most evidence to show that isoflavones are able to influence OPG/RANKL/RANK regulatory machinery implicated in the osteoblast and osteoclast relationship. Daidzein and 17β-oestradiol at low concentration (1 nm) stimulated osteoblast differentiation and were able to increase OPG secretion into the medium. Daidzein and genistein at physiological concentrations (0·1 and 10 nm) up-regulated OPG gene expression, but RANKL expression was not detectable in human osteoblastic cell lines(Reference Chen, Garner and Anderson113). Moreover, daidzein enhanced RANKL protein secretion and content in the membrane in porcine osteoblasts(Reference De Wilde, Lieberherr and Colin69). This stimulatory effect was reversed by ICI, suggesting that the RANKL:OPG ratio is dependent on the oestrogen receptor pathway(Reference De Wilde, Lieberherr and Colin69, Reference Chen, Garner and Anderson113). It is possible that OPG and RANKL secretion induced by daidzein leads to an increase in the binding capacity of OPG to RANKL. In consequence, the amount of RANKL available to induce osteoclast formation and differentiation is reduced(Reference De Wilde, Lieberherr and Colin69).
However, Pang et al. demonstrated, in mouse primary calvaria osteoblasts, that kaempferol and quercetin dose dependently (5–20 μm) inhibited RANKL-induced expression of osteoclastic differentiation markers, which led to a decrease in RANKL-induced formation of multinucleated osteoclasts(Reference Pang, Ricupero and Huang79). The relationship between osteoblasts, osteoclasts and phytochemicals remains to be further investigated.
Effects of polyphenol extracts on osteoblasts
The effects of polyphenols on osteoblasts were mostly studied using pure molecules, unfortunately most of the time in aglycone forms. However, individuals consume not isolated molecules but fruit and vegetables, which are rich in many different polyphenols. It could be thus interesting to assess effects of phenolic mixtures on osteoblast function. It is important to take into consideration that polyphenols are metabolised before reaching the target cells.
The only available data with respect to osteoblasts are from Bu et al. who studied the influence of dried plum polyphenol extract at concentrations of 2·5, 5 and 10 μg/ml on MC3T3-E1 cells under normal and inflammatory conditions(Reference Bu, Hunt and Smith121). This extract, non-toxic for osteoblasts, was able to stimulate ALP activity and mineralise nodule formation. The plum extract was not able to alter the expression of the transcription factors Runx2 and Osterix nor genes implicated in osteoclast regulation such as RANKL and OPG. However, this extract could restore Runx2 and Osterix expression in cells that had been pre-treated with TNFα (inflammatory condition) and as a result exhibited reduced Runx2 and Osterix expression. The extract also suppressed the TNFα-induced up-regulation of RANKL expression(Reference Bu, Hunt and Smith121), but OPG expression was not affected by any treatments. These results suggest that consumption of dried plums could modulate both bone formation and resorption but the molecular mechanisms are poorly elucidated.
Conclusions
Regarding osteoblasts, polyphenols can act on these cells at different stages including proliferation, differentiation and mineralisation(Reference Habauzit and Horcajada9). Physiological effects in response to polyphenols have been observed but the molecular mechanisms schematically presented in Fig. 5 are not fully understood. Many signalling pathways, often talking to each other, need to be activated to stimulate cell functions, and it is expected that many years of scientific work will be required to elucidate the exact molecular mechanisms by which polyphenols affect osteoblast regulation. Moreover, it is important to keep in mind that polyphenols may act through not only an action on osteoblasts but also on osteoclasts. These combined effects are helpful to maintain a balance in bone remodelling, a key endpoint in the management of osteoporosis.
Fig. 5 Putative mechanisms of action of polyphenols on osteoblast signalling. Runx2, runt-related transcription factor-2; AP-1, activator protein-1; BMP, bone morphogenetic protein; MAPK, mitogen-activated protein kinase; RANKL, receptor activator of NF-κB ligand; OPG, osteoprotegerin; SMAD, mothers against decapentaplegic homologue; ERK, extracellular signal-regulated kinase. Polyphenols: (1), daidzein, ( − )-epigallocatechin-3-gallate, emodin, icariin, kaempferol, resveratrol, quercetin; (2), coumarins, emodin, icariin, myricetin, piceatannol, resveratrol; (3), coumarins, myricetin; (4), daidzein, genistein, kaempferol, quercetin.
Acknowledgements
The authors have no conflicts of interest to declare.
