In general, GFs influence MSCs acting both in a paracrine and autocrine manner; in fact, MSCs express on their surface GF receptors.

Proteins of the TGFβ/BMP superfamily are the most abundant GFs in BM-ME and originate mainly from bone matrix degradation and activated T cells (Tang et al., 2009; Croes et al., 2016; Li et al., 2016). The TGFβ family comprises three members: TGFβ1, TGFβ2, and TGFβ3. TGFβ1, released from the bone matrix by the osteoclast resorptive activity, has been demonstrated in vitro and in vivo to induce MSCs migration to the remodeling sites, thus coupling bone formation and resorption. The mechanism through which this function is exerted, whether through the canonical signaling pathway, involving SMADs, or the non-canonical one, involving AKT, ERK1/2, FAK, and p38, is debated (Tang et al., 2009; Dubon et al., 2017). Moreover, in vitro TGFβ1 drives MSCs fate toward osteoblasts generation and inhibits adipogenic differentiation; accordingly, TGFβ1 induces the switching from adipogenesis to osteogenesis when added to an adipogenic medium, acting mainly on the SMAD/C/EBPs/PPARγ signaling (Choy and Derynck, 2003; van Zoelen et al., 2016). In other reports, during in vitro MSC expansion, TGFβ1 reduces the number of osteoprogenitor cells and limits their expansion inducing a rapid terminal differentiation, suggesting that TGFβ1 effects on MSCs might depend on the commitment state of the cells (Walsh et al., 2003; Claros et al., 2014). Furthermore, TGFβ1 is a key molecule in chondrogenesis by stabilizing SOX9 via the canonical SMAD or the non-canonical p38 pathways (Coricor and Serra, 2016; Dexheimer et al., 2016). A similar potent effect on chondrogenic differentiation has been demonstrated for TGFβ2 and TGFβ3 (Vinatier et al., 2009).

Furthermore, TGFβ1 induces ADAM12 expression in MSCs/pericytes triggering myofibroblast transdifferentiation and contributing to fibrosis (Cipriani et al., 2016).

The BMP family comprises at least 15 members, which usually exert synergistic effects with TGFβs and activate SMAD transcription factors and expression of genes such as Runx2, Pparγ, or Sox9. This ultimately results in the promotion of MSCs differentiation toward the adipogenic (e.g., BMP2, BMP4, and BMP7), osteogenic (e.g., BMP2, BMP6, and BMP9), or chondrogenic (e.g., BMP2 and BMP7) lineage, depending on the microenvironmental concentrations (Kang et al., 2009; Chen et al., 2012).

IGF1, a polypeptide with a high-binding affinity to IGF1R and insulin-like features (Wang et al., 2013), is one of the most abundant GFs deposited in the bone matrix. In vitro and in vivo in mouse and rat models, IGF1 released from bone matrix degradation induces osteoblast differentiation via the mTOR pathway and enhances osteoblasts function (Xian et al., 2012; Crane and Cao, 2014). Regarding chondrogenesis, IGF1 in combination with TGFβs enhances in vitro chondrocytes proliferation and collagen II production (Fukumoto et al., 2003; Indrawattana et al., 2004), while for the adipogenic fate commitment, IGF1 activates IGF1R-dependent AKT/PKB signaling, increasing Pparγ expression and lipid accumulation (Scavo et al., 2004).

VEGF is a key soluble molecule involved in endothelial cell proliferation, migration, and tissue vascularization (Ferrara et al., 2003). MSCs/osteoblasts themselves express VEGF and its receptors, so this GF may exert both a paracrine and an autocrine regulation (Niida et al., 1999; Nakagawa et al., 2000; Kaigler et al., 2003; Yang et al., 2008; Liu and Olsen, 2014; Marsano et al., 2016). VEGF is essential for coupling angiogenesis to bone formation during skeletal development, by promoting chondrocytes survival in hypoxic regions of cartilaginous templates, vascularization of developing bones, and proliferation and differentiation of osteoblasts (Maes et al., 2004; Zelzer et al., 2004). In post-natal bone homeostasis, VEGF favors MSCs osteoblastogenesis at the expense of adipogenesis, through intracrine regulation of Runx2 and Pparγ (Liu et al., 2012), while intracellular blockade of VEGF signaling in MSCs activates TGFβ signaling, thus inducing spontaneous in vivo chondrogenesis and formation of a hypoxic microenvironment and a stable hyaline cartilage (Marsano et al., 2016).

In FGF family, many members positively regulate MSCs proliferation and osteogenic differentiation, by interacting with FGFR2 and activating PLCγ/PKCα, MAPK/ERK 1/2, and PI3K/AKT pathways (Marie, 2012). FGFs are also crucial for the regulation of MSCs chondrogenic differentiation through FGF/FGFR3, as demonstrated by their involvement in the pathogenesis of different forms of chondrodysplasia (Ornitz and Legeai-Mallet, 2017). However, also for these GFs, results in literature are discordant: recent data indicate that FGF1 and FGF2 maintain MSCs in an uncommitted state, preventing their differentiation (Le Blanc et al., 2015; Simann et al., 2017). Finally, the hormone-like FGF23, produced by bone cells and by cells of different tissues, favors osteogenic differentiation at the expenses of adipogenesis by binding to its receptor Klotho, which mediates the activation of MAPKs signaling (Li et al., 2013c).

The skeleton is widely recognized as both an endocrine organ and a target for other endocrine tissues (Fukumoto and Martin, 2009).

The prototypical example of MSCs-regulating hormone is estrogens, the main molecules involved in post-menopausal osteoporosis. Estrogens bind their α and/or β receptors and induce MSCs proliferation and osteogenic and chondrogenic differentiation (Rodriguez et al., 2008), through the activation of BMPs/WNT/β-catenin and p38 MAPKs/NF-κB signaling pathways (Gopalakrishnan et al., 2006; Li et al., 2013b; Kim et al., 2015; Cong et al., 2016). Moreover, estrogens induce early osteoblast differentiation and inhibit adipogenesis in mice (Okazaki et al., 2002). Furthermore, they reduce LPL levels, impair adipocyte progression into hypertrophic state, and influence body adipose tissue depots distribution and glucose metabolism (Post et al., 2008). Overall, this evidence points to a role of estrogens in regulating both bone and glucose homeostasis.

PTH is one of the principal modulator of calcium homeostasis through cAMP/PKA/PLCγ signaling. It also displays both catabolic and anabolic functions in bone remodeling: the former is exerted by inducing RANKL production, which fosters osteoclasts’ generation and activity, the latter by affecting MSCs fate (Hock and Gera, 1992; Qin et al., 2004). Indeed, PTH induces in vitro osteogenic differentiation via LRP6-dependent BMP/SMAD signaling (Polo and Di Fiore, 2006; Jilka, 2007; Qiu et al., 2010; Yu et al., 2012). Accordingly, in vivo deletion of PTHR in murine MSCs reduces bone formation and increases bone resorption and marrow adiposity, while intermittent PTH administration to control mice reduces marrow adipogenesis (Fan et al., 2017). On the contrary, PTHrP affects MSCs differentiation, preventing chondrocyte hypertrophy and blocking osteogenesis through regulation of Sox9 and Runx2 gene expressions (Provot et al., 2006; Zhang et al., 2009; Fischer et al., 2014).

GH regulates linear growth during development (Gomes et al., 2013; Ma et al., 2016) and is involved in BM adiposity maintenance. It enhances adipocytes and osteoblast precursor pool size, while it induces MSCs osteogenesis and inhibits BM fat accumulation (Menagh et al., 2010). Its regulator, GHRH, has receptors (GHRHR) also on MSCs (Gomes et al., 2013; Ma et al., 2016) and through their binding promotes MSCs proliferation and survival via the cAMP/PKA/PLCγ signaling, activates MAPK signals, and induces osteogenic differentiation (Jaiswal et al., 2000; Xia et al., 2016).

Finally, the adipose tissue-derived hormone leptin contributes to guide MSCs commitment but contradictory results are reported (Ducy et al., 2000; Kontogianni et al., 2004; La Cava and Matarese, 2004). MSCs highly express leptin receptor (Zhou et al., 2014), and in vitro, leptin enhances osteogenic differentiation and reduces the adipogenic one (Thomas, 2004). On the contrary, leptin in vivo regulates MSCs, increasing marrow adipogenesis and reducing osteogenesis in response to diet and adiposity, through the JAK2/STAT3 pathway (Yue et al., 2016).

Among the cell populations present in BM-ME, immune cells participate in directing MSCs fate by secreting a variety of cytokines, with anabolic or anti-anabolic effects depending on the inflammatory state of the bone tissue.

For example, T cells activate bone formation by producing Wnt ligands that initiate Wnt signaling and osteoblastogenesis (Ouji et al., 2006; Terauchi et al., 2009). They also produce CD40L that binds CD40 on MSCs inducing their proliferation and survival (Ahuja et al., 2003; Gao et al., 2008; Li et al., 2013a).

Conflicting results are reported regarding the effects of proinflammatory cytokine on MSCs differentiation. In a proinflammatory environment, the interleukins IL-1β, IL-6, and IL-23 (mainly derived from Th17 cells) have been reported to increase the differentiation performance of human MSC toward the osteogenic and adipogenic lineages (Pourgholaminejad et al., 2016). Accordingly, TNFα, IL-1β, and IL-6 enhance osteoblast differentiation by triggering NF-κB signaling or modulating BMP2 pathway (Nakase et al., 1997; Hess et al., 2009; Huh and Lee, 2013; Croes et al., 2015).

On the contrary, IL-1 and TNFα inhibit MSCs osteogenesis and adipocyte generation, acting not only through the canonical NF-κB signaling (Lacey et al., 2009; Sullivan et al., 2014) but also via other mechanisms such as IL-1R1/MyD88 signal transduction, as described for murine MSCs obtained from different genetic backgrounds (Martino et al., 2016). Similarly, IL-6 can impair MSCs ability to generate adipocytes and chondrocytes and keep them in an undifferentiated state by activating ERK1/2 (Pricola et al., 2009).

Recently, great attention has been paid to the skeletal muscle-derived myokine Irisin, which is able to directly target the bone tissue, thus regulating its physiology. Specifically, Irisin can induce MSCs osteoblast differentiation through p38/ERK MAPK signaling pathways, leading to the upregulation of osteogenic marker genes, such as Atf4, Runx2, Osx, Lrp5, β-catenin, Alp, and Col1a1 (Colaianni et al., 2015; Qiao et al., 2016).

Multiple myeloma is a common hematological malignancy mainly characterized by osteolytic lesions due to increased osteoclast number and activity and strongly decreased bone formation (Kassen et al., 2014). MSCs and osteoblasts support MM cells survival, proliferation, and progression (Azab et al., 2009; Reagan et al., 2014; Roccaro et al., 2014; Fairfield et al., 2016), while osteogenic differentiation is reduced in MM patients, which might be a putative strategy of MM cells to preserve cells (e.g., MSCs) necessary for their support (Corre et al., 2007; Reagan et al., 2014). Cell-to-cell contact and production of soluble factors are likely involved in these mechanisms. For example, MM cells inhibit Runx2 and inactivate the non-canonical Wnt5a/Ror2 pathway; a putative role of IL-7 produced by MM cells can be hypothesized (Giuliani et al., 2005; D’Souza et al., 2011; Bolzoni et al., 2013). Moreover, MM cells secrete Wnt inhibitory factors, i.e., Dkk1 and sclerostin; TGFβ, which impairs osteoblast differentiation (Lee et al., 2003; Tian et al., 2003; Colucci et al., 2011); and also factors inducing MSCs growth that, in turn, produce osteoclast-activating factors (i.e., IL-6, MCSF, TNFα, and RANKL) leading to osteolysis (David Roodman and Silbermann, 2015).

Breast cancer cells preferably metastasize to bone inducing purely osteolytic lesions, via the production of osteoclast-activating factors (mainly RANKL and MCSF). Furthermore, osteolytic lesions are production sites of several soluble factors derived from osteoclasts’ resorption of the bone matrix, such as TGFβ (Kang et al., 2003). These molecules can inhibit osteoblast development and functions and are able to induce BC cell proliferation and progression that, in turn, sustain the secretion of osteoblast inhibitory factors (Mundy, 2002).

Bone metastases in PC tend to be osteosclerotic, rather than osteolytic. PC cells produce soluble factors, e.g., BMPs, TGFβ, IGF1, FGFs, and VEGF, which increase MSCs osteogenic differentiation, osteoblast development, and bone deposition, leading to elevated mineral apposition, even though the newly formed bone is immature and of poor quality (Guise et al., 2006; David Roodman and Silbermann, 2015). Of note, the osteoclast-inducing hormone PTHrP, above reported as osteogenic inhibitor, is highly produced by PC cells and in this context enhances osteoblast progenitors’ proliferation and early osteogenesis (Liao et al., 2008).

Bone marrow failure syndromes are hematological disorders characterized by impaired hematopoiesis comprising different phenotypes, i.e., myelodysplastic syndromes (MDS), aplastic anemia (AA), and chronic idiopathic neutropenia (CIN). MSCs play an important role in maintaining and restoring hematopoiesis, thanks to the secretion of regulatory factors for HSC functionality (Kastrinaki et al., 2013). Scanty data are available the other way round. For example in AA, MSCs osteogenic capacity is inhibited in favor of adipogenesis (Papadaki et al., 2001; Shipounova et al., 2009; Xu et al., 2009). Recently, oncostatin M, a member of the IL-6 family, has been reported to stimulate HSC expansion, inhibiting adipogenic differentiation and enhancing osteogenic differentiation of MSCs. Upon administration in mice bearing BM injury, it decreases marrow adipogenesis and restores HSC number (Sato et al., 2014).