Several lines of evidence support that T1D alters MSC lineage selection. Most notably, mouse models of T1D display an increase in bone marrow adipocyte number and a decrease in the number of surface osteoblasts [14, 16, 18, 19, 45, 60, 87]. In vitro studies further indicate that factors present in T1D patients and mice (i.e., high glucose, pro-inflammatory cytokines) are capable of promoting MSC adipogenesis at the cost of osteogenesis [45, 49, 50, 60, 83, 88–90]. In addition, a recent study in rats found that the osteoprogenitor pool in T1D rat bone marrow was significantly depleted [88], further suggesting decreased lineage selection towards osteogenesis and/or a reduction in overall MSC number. Interestingly, the T1D adiposity is bone marrow specific, as T1D mice have depleted subcutaneous and visceral fat stores and exhibit weight loss [48, 83, 87]. This is consistent with the weight loss seen in T1D patients [87, 91]. Both T1D male and female mice display increased bone marrow adiposity in tibia, femur, and calvaria, and correspondingly display increased levels of adipocyte markers such as PPARγ2 and adipocyte protein 2 [aP2, also known as fatty acid binding protein 4 (FABP4)] in bone [45, 62]. T1D-induced marrow adiposity could be either an active or passive shunt of MSCs towards adipogenesis and away from osteogenesis [17–19, 92, 93] (see section on “Potential Mechanisms Contributing to Altered Bone Remodeling”).

Several in vivo studies have examined the role of T1D-induced marrow adiposity in mediating bone loss in T1D mouse models. Pharmacologically, an antagonist of PPARγ (bisphenol-A-diglycidyl ether) was given to T1D mice to inhibit adipogenesis. This approach was successful in preventing T1D-induced marrow adiposity, but T1D bone loss still occurred [23–25, 79]. Genetic manipulations, such as the knockout of C/EBP-β (a transcription factor required early in adipogenesis) have not been effective in preventing T1D bone loss [94, 95]. The lack of an association between bone marrow adiposity and bone loss was also observed in T1D patients [96], though marrow adiposity did correlate with serum lipid levels [96] suggesting a link to hyperlipidemia. Larger clinical studies are needed to determine significant relationships. Consistent with the lack of a link between marrow fat and T1D bone loss, mouse vertebral bone which has reduced density in T1D does not display an increase in marrow adiposity [45]. Taken together, these studies indicate that T1D is associated with changes in MSC lineage selection but T1D bone loss is not the direct result of increased marrow adiposity. Additional pathways regulating osteoblast activity must be involved.

Once osteoblast lineage is selected, pre-osteoblasts progressively mature through stages of matrix production and mineralization. Identifying a specific change in maturation is somewhat difficult since altered lineage selection and increased death can also reduce markers of maturation in T1D bone. It is known that bone defect healing is decreased in T1D rats [97]. Similarly, T1D decreases bone formation on titanium and hydroxyapatite implants [80, 98, 99]. Though a reduction in pre-osteoblasts would cause similar outcomes, cell culture studies provide evidence to suggest that T1D reduces osteoblast maturation. Pre-osteoblast cell lines (i.e., MC3T3-E1 cells) cultured under conditions associated with T1D (high glucose or pro-inflammatory cytokines such as TNFα) display reduced levels of maturation markers [100–104]. In many of these conditions there is also evidence of enhanced expression of adipocyte markers (PPARγ and aP2) [45, 60, 62] suggesting a lineage or functional shift to an adipocyte.

Increased osteoblast apoptosis can also contribute to reduced T1D osteoblast activity. T1D associated factors, such as reactive oxygen species (ROS), advanced glycation end-products (AGEs) and pro-inflammatory cytokines, are linked to increased osteoblast death [33, 48, 105–107]. In T1D mouse models, the number of osteoblasts stained by terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL), a method of detecting cellular apoptosis, was increased twofold compared to levels in control mouse femurs and tibias [52, 108]. The increase in apoptotic osteoblasts was confirmed in spontaneously diabetic Ins2 ^+/− mice where the increase was over threefold when compared to control [52]. Similarly, mRNA levels of Bax (a pro-apoptotic protein) were increased in streptozotocin (STZ)-induced diabetic mice and Ins2 ^+/− mice [52]. When osteoblasts (primary cells and cell lines) were cocultured with control versus T1D mouse bone marrow, the T1D marrow caused a twofold increase in osteoblast caspase-3 activity, a marker of cell apoptosis [52]. These findings support a role for the diabetic bone marrow microenvironment in inducing osteoblast death [52].

Osteoclasts are multinucleated cells responsible for the resorption phase of the bone remodeling process. These cells are derived from HSCs, which reside in the bone marrow. Differentiation of HSCs into active osteoclasts is a multistep process (Fig. 8.2) and is regulated by three main cytokines: macrophage colony stimulating factor (M-CSF), RANKL and osteoprotegerin (OPG). These factors are derived from osteoblasts as well as other cells of the immune system including T-cells and B-cells [131]. While RANKL binds to RANK on osteoclasts to activate resorption, OPG (a secreted as well as membrane-bound protein produced by osteoblasts) acts as a decoy for RANKL, preventing it from binding to the transmembrane receptor RANK [137–139]. Both M-CSF and RANKL can increase nuclear factor-kappa (NF-kB) activity to stimulate osteoclast maturation [132–136]. RANKL does this by first inducing TNF receptor-associated factor (TRAF) proteins to start a signaling pathway that ends in the activation of NF-kB. NF-kB is typically bound in the cytoplasm by the inhibitor of kB (IkB). Binding of RANKL induces the phosphorylation of IkB to allow NF-kB to enter the nucleus to stimulate genes required for osteoclast function. Mutations in this signaling pathway lead to defective osteoclast formation and function [140–143]. Similarly, mutations in the gene encoding M-CSF causes osteopetrosis due to the inability of osteoclasts to fully mature.

Other factors involved in osteoclast differentiation include PU.1, which is a transcription factor that is required for the normal formation of osteoclasts. PU.1 regulates the receptor for macrophage colony stimulating actor (M-CSF) and integrin proteins needed for osteoclast maturation and binding of the osteoclast to the bone surface [125–127]. Mice lacking PU.1 die immediately after being born and have symptoms of osteopetrosis, an increase in bone density [128]. PU.1 interacts with other transcription factors (MITF, micro-ophthalmia-associated transcription factor) to regulate genes important for osteoclast function such as tartrate-resistant acid phosphatase (TRAP) and carbonic anhydrase 2 (CA-II) [129, 130].

The majority of rodent studies examining T1D show no change or decreases in markers of osteoclastic resorption [33, 48, 62, 85, 144–147]. However, there is some controversy since a few studies have found increased osteoclast parameters in animals [147]. Clinical studies support no change in T1D osteoclast activity as determined by levels of serum markers of resorption such as deoxypyridinoline and c-terminal telopeptide of type 1 collagen (CTX) [53, 69, 86, 148]. In mice, the typical STZ dosing regimen used to induce T1D does not cause a change in osteoclast parameters (consistent with spontaneous mouse models [60]), however, increasing the dose causes an increase in osteoclast activity possibly due to more rapid onset of T1D and/or inflammation (as seen in the liver with high STZ dose) [67]. As noted previously, T1D increases bone marrow adiposity [60, 62] and thus MSC lineage selection. Not only does this change the overall physical morphology of the bone marrow compartment, but it also reduces the number of MSC maturing to the osteoblasts that are needed for HSC maintenance and maturation to osteoclasts (via M-CSF and RANKL) [149]; this can contribute to reduced osteoclastogenesis in the presence of inflammation. Other contributors to reduced resorption in T1D include hyperglycemia which can directly decrease RANKL-induced osteoclastogenesis [150], a reduced number of osteoblasts which can reduce RANKL levels in bone, and elevated levels of anti-inflammatory cytokines.

T2D effects on patient bone densities are variable with reports indicating lower, the same, or higher bone density than nondiabetic individuals. In T2D animal models, such as the leptin receptor knock out db/db mouse, mice display suppression of bone remodeling, with both bone formation and resorption decreased [151]. Yet, osteoclast activity in T2D male patients has been reported to increase [152, 153]. Serum concentrations of TRAP and urine markers for bone resorption, such as CTX, deoxypyridinoline, and N-telopeptide of type I collagen (NTX), are significantly elevated in T2D patients [154–156]. However, T2D patients can also exhibit increased levels of OPG which can reduce the increased resorption caused by T2D [157].

Immune cells are capable of regulating osteoblast and osteoclast activity and vice versa. The role of immune cell populations and their effect on bone marrow hematopoiesis has been demonstrated in a variety of pathology conditions associated with bone loss. For example, activated T-cells are involved in regulating bone homeostasis in estrogen deficiency, an effect that was mediated by IFNγ [158]. In ovariectomized (ovx) mice, T-cells express high levels of TNF-α (enough to induce osteoclastogenesis) [159]. T-cell deficiency prevents ovx induced bone loss, a benefit lost when mice are given T-cells from WT mice but not T-cells from TNFα-deficient mice [160]. Similarly, blockade of TNF-α and IL-1 by Anakinra or Etanercept decreases bone resorption in postmenopausal women [161]. This suggests that an increase in these cytokines can be one of the causes of bone loss. However, TNF-α, IFNγ, and IL-1 are not the only cytokines associated with bone loss. Another cytokine IL-7 promotes osteoclastogenesis by upregulating T- and B-cell-derived RANKL [162]. IL-23 also stimulates the differentiation of human osteoclasts from peripheral blood mononuclear cells (PBMC) in the absence of osteoblasts or exogenous RANKL and in vivo blockade of IL-23 activity prevents inflammation and bone destruction in collagen-induced arthritis in rats [163]. These findings provide a new insight into the interaction between immune system, bone marrow and bone loss.

Another intracellular system that is associated with changes in the bone marrow, immune cells, and bone density is Wnt signaling. Wnt signaling regulates the reciprocal relationship between adipocytes and osteoblast. Part of the mechanism whereby Wnt10b inhibits adipogenesis and stimulates osteoblastogenesis is by suppression of PPAR-γ. In addition, it has been found that expression of Wnt10b in marrow increases trabecular bone mass and strength [37]. Wnt ligands are also required for hematopoietic progenitor cells (HSPC) expansion and survival. Experiments in ovx mice has shown that T-cell-expressed CD40 ligand (CD40L) enhance T-cell production of Wnt10b which results in activation of stromal cell (SCs) and HSPCs, along with improvement of cytokine production by SCs [164]. In summary, these results recognized the impact of cellular interaction between immune cells, HSPCs and the effect in bone homeostasis.

T1D is an autoimmune disease characterized by destruction of pancreatic β-cells producing insulin. White blood cells, like T-cells and macrophages, and the interaction and secretion of cytokines, nitric oxide, and free radicals are involved in this autoimmune pathology. Cytokines, such as IFNγ, IL-1β, and TNFα, can act together to induce beta cell death and can also affect the expression of genes that are protective or harmful for beta cell survival such as signal transducers and activators of transcription 1 (STAT-1) and interferon regulatory factor 1 (IRF-1) [169]. One study examining T-helper cell cytokine profiles indicated higher intracellular TNFα in CD8⁺T cells at the time of T1D diagnosis and higher intracellular TNFα in CD4⁺T lymphocytes in patients at 3 months post-diagnosis, compared with controls [170]. This suggests that there are distinct roles for cytokine release during different stages of T1D. Diabetic mouse models also display elevated TNFα levels in serum and increased TNFα mRNA level/expression in bone [52, 72]. In the STZ-induced diabetic mouse model IL-1α, IL-6, IFNγ, and TNFα expression are increased in bone with the onset of T1D and during the onset of altered expression of bone phenotype markers [72] (Fig. 8.3). Furthermore, it has been shown that T1D reduces the number of MSCs found at sites of fracture repair and that suppression of TNFα with Pegsunercept treatment restores the level of MSCs to control levels. It was further determined that TNFα is capable of reducing MSC proliferation and increasing MSC apoptosis [171].

Emerging evidence indicates that T1D directly compromises the function of the bone marrow [57]. Increased local cytokine levels in the bone marrow of mice under hyperglycemic conditions can disrupt hematopoiesis by shifting monocytes towards a pro-inflammatory phenotype [167, 172]. This change generated more monocytes and less endothelial progenitor cells (EPCs) in T1D, which contribute to the development of microvascular complications in this condition [166]. This finding was also evident in individuals with non-proliferative diabetic retinopathy [173]. In another experiment using Yorkshire pigs, after STZ-induction of diabetes a significant impairment of bone marrow and circulating EPCs as well as endothelial cells function were observed [174]. Chronic diabetes is also associated with dysfunction in the number and repopulation activity of HSPC, which is likely mediated by alterations in cytokines involved in stem cell renewal and maintenance [165]. Longstanding STZ-induced and spontaneous T1D mice further display impairment of colony-forming capability of the bone marrow and an increase in the number of Ly6C^hi (pro-inflammatory) monocytes in the peripheral blood [166].

The bone marrow is also the source of cells that contribute to diabetes-associated inflammation. Bone marrow transplantation experiments done in both mice and rats, demonstrate that most of the extrapancreatic proinsulin producing cells (found in liver, adipose tissue, spleen, and thymus) originated from the bone marrow [168]. These cells were also demonstrated to produce the pro-inflammatory cytokine TNF-α which could contribute to damage of the target organs [168]. This suggests that the damage seen in organs such as bone during diabetes may be due to the irregularity in bone marrow stem cells.

T2D is not an autoimmune disease and is not always associated with bone loss [175]. However, there are changes in hematopoetic profiles. For example, T2D patients have 33 % less circulating progenitor cells (CPCs) and 40 % less circulating EPCs compared with healthy subjects [176]. Peripheral vascular complications of T2D mellitus are associated with an extensively low number of EPCs [176] which could result from alterations in bone marrow function. Altered Wnt signaling is also associated with T2D. A single nucleotide polymorphism locus in the Wnt5b gene conferred susceptibility to T2D by modifying adipocyte function [177]. Wnt signaling is also altered in immune cells. A link between cellular glucose sensing and the Wnt/b-catenin pathway was recently reported in two macrophage cell lines (J774.2 and RAW264.7 cells) [178]. T2D is also accompanied by an increased inflammatory state that can disrupt osteoclast function and survival [67]. However, a small study in T2D patients examining changes in bone turnover markers found that treatment with nonsteroidal anti-inflammatory drugs causes an increase in osteocalcin levels in men (not in women) and increases serum C-terminal telopeptide (s-CTx) [179]. This small study suggests that T2D inflammation decreases bone turnover.

Hyperglycemia has many negative effects on bone and bone formation. High glucose can result in the production of ROS [180, 181], induction of cellular osmotic responses [182, 183], and increased nonenzymatic glycosylation of proteins and DNA [144, 184–188] as well as suppress calciotropic hormones and growth factors [62, 144, 180–185, 187–189], all of which can lead to decreased osteoblast activity. In vitro studies exposing MC3T3-E1 cells (osteoblastic cell line) to high glucose conditions (30 mM) showed a significant decrease in markers of osteoblast maturation, such as alkaline phosphatase and osteocalcin [90, 190]. Even when MC3T3-E1 cells are exposed to hyperglycemic conditions for short time periods, 48 h, osteocalcin mRNA levels are significantly decreased compared to cells cultured under normal (5 mM) glucose levels [90]. MC3T3-E1 cells cultured in high glucose also show decreased mineralization [190, 191]. These in vitro results suggest hyperglycemia decreases osteoblast maturation.

Hyperglycemia also promotes MSC adipogenesis over osteogenesis [45, 60, 88–90]. Specifically, culturing primary rat osteoblasts under high glucose conditions (25.5 and 35.5 mM) increases levels of adipogenic markers (PPARy and aP2) and decreases levels of osteogenic markers (osteocalcin) [192]. A similar lineage switch was seen in human osteoblastic MG-63 cells cultured under high glucose conditions. The decrease in osteogenesis appears to be dependent on cAMP/protein kinase A (PKA)/extracellular signal-regulated kinase (ERK) signaling [193] and the PI3K/Akt signaling [192]. Hyperglycemia can also reduce osteoblast growth by 40 % [192] and induce osteoblast apoptosis. Both UMR and MC3T3-E1 osteoblasts undergo increased apoptosis when exposed for 24 h to AGE-BSA compared to unmodified BSA [194]. Thus, hyperglycemia can affect osteoblast activity through alterations in MSC lineage selection, osteoblast maturation, and osteoblast death.

Hyperglycemia decreases osteoclastogenesis in vitro. Using RAW264.7 cells and bone marrow macrophages, it was found that high levels of D(+)glucose inhibited osteoclast formation, ROS production (needed for bone resorption), caspase-3 activity (necessary for RANKL-induced differentiation) and migration to bone resorption pits [150]. In addition to this, osteoclasts derived from bone marrow cells extracted from mice which were exposed to high glucose had decreased mRNA expression of RANK and cathepsin K, decreased TRAP activity and decreased resorption activity [195]. Together, these studies indicate that a high glucose environment decreases osteoclast differentiation and activity.

Hyperglycemia also promotes a nonenzymatic reaction between glucose and proteins that results in AGEs. AGEs build up over time in diabetic tissues [196, 197] because they are irreversible modifications [198]. For example, T2D elderly patients display increased urine pentosidine (an AGE) which was associated with increased clinical fracture incidence and increased vertebral fracture prevalence in those with diabetes but not in those without diabetes [199]. Pentosidine, pyrraline and N(carboxymethyl)lysine (CML) are some of the more commonly characterized AGE products whose levels increase in tissues with collagen matrices, particularly pentosidine, under conditions of diabetes [200–202] and aging [199, 203, 204]. AGE modifications can change the structural and functional properties of proteins and correspondingly their levels correlate with diabetes complication severity, including glycosylated hemoglobin levels and bone loss [188, 201, 202, 205–209]. Type 1 collagen, the most abundant constituent of the bone matrix, is a target for AGE [210, 211], which can accumulate and alter the mechanical properties of bone, decreasing its toughness and could therefore contribute to skeletal fragility [212–214]. AGE increases collagen crosslinking, which can decrease bone strength and ductility, and increase brittleness [209, 214–218]. An increase in AGEs have been linked to a reduction in serum osteocalcin levels and are positively associated with levels of pentosidine which impairs the biomechanical properties of bone [216, 219–221]. AGEs can modify bone cell behavior by interacting with specific cellular receptors, such as RAGE (receptor for AGEs). When AGEs bind to RAGE on osteoclasts, NF-kB activity and ROS increase [222, 223]. RAGE is also expressed in osteoblasts and AGE treatment of osteoblasts causes decreased osteoblast maturation [184, 222]. In vitro and in vivo studies showed that AGEs increase calvarial periosteal cell apoptosis and induced apoptosis in primary cultures of human or neonatal rat osteoblastic cells. This effect was mediated through RAGE and increased p38 and c-Jun N-terminal kinase (JNK) [224]. The presence of AGE can also induce osteoblast apoptosis through activation of caspase-3 and caspase-8 [105, 108, 197]. Consistent with these findings, deficiency of RAGE results in an increased BMD compared with control mice [225]. Interestingly, in vitro studies using AGE-modified bone slices and osteoclastogenesis assays have shown that resorption was markedly inhibited. This was confirmed by a marked decrease in the release of type I collagen fragments generated by the collagenolytic enzymes secreted by osteoclasts [226]. This decrease in bone resorption is thought to be due not only to the alteration of the structural integrity of the bone matrix by AGEs, but also their effect on osteoclastic differentiation [226]. AGEs also increase inflammation by monocytes and macrophages due to upregulation of TNFα and IL-1β [227].

Hyperglycemia can also increase oxidative stress. Oxidative stress is triggered through two ways: the mitochondria becoming overloaded with glucose or through AGEs and Polyol signaling [88]. Oxidative DNA damage markers such as 8-hydroxydeoxyguanosine are increased in hyperglycemic diabetic mice but decreased when the mice are treated with insulin and become euglycemic [120, 228]. This oxidative stress marker is also elevated in T2D rats and could contribute to decreased osteoblast maturation based on the in vitro findings of oxidative stress negatively affecting osteoblast maturation [120, 122, 229, 230]. In addition, hyperglycemia-induced osteoblast apoptosis has been linked with an increase in oxidative stress [51, 231]. ER-stress activates the unfolded protein response mechanism leading to increased C/EBP homologous protein (CHOP) signaling and cellular apoptosis [51]. T1D rats showed an increase in CHOP positive osteoblasts in the femurs which also showed decreased bone mineral density [51].