Bone Marrow Stem Cell Abnormality and Diabetic Complications



Long-lasting diabetes impairs the function of multiple organs, which consists of degeneration of various tissues with increasing apoptosis of target cells. Recently, we found that hyperglycemia induced the appearance of abnormal cells in the bone marrow and cell fusion between bone marrow-derived cells and hepatocytes, peripheral neural cells, or renal tubular cells occurred in diabetic animals. Fused cells in these organs expressed TNFα, and accelerated apoptosis, suggesting that these events might be a cause of diabetic complications. In this review, we propose a new concept that bone marrow stem cell abnormality causes diabetic complications, and this concept might provide new strategies for treatment of diabetes-associated tissue damage. Anat Rec, 2012. © 2012 Wiley Periodicals, Inc.


Diabetic complications comprise the dysfunction of various organs including kidney, liver, neurons, retina, blood vessels, bone, heart, gut, hematopoietic, and immune cells. These multiorgan abnormalities induced by diabetes have been thought as a result of glucotoxicity, proinflammatory cytokine production, or superoxide generation in affected organs.

Recently, we found that abnormal proinsulin and TNFα producing cells appeared in the bone marrow in streptozotocin (STZ)-induced diabetic, high fat diet (HFD)-induced diabetic and db/db mice, and these cells underwent fusion with peripheral neurons (Terashima et al., 2005), hepatocytes (Fujimiya et al., 2007), and renal tubular epithelial cells. Fused cells produced tumor necrosis factor α (TNFα) and accelerated apoptosis and then caused tissue degeneration.

These findings suggest that diabetic tissue degenerations seem due to the abnormality in the bone marrow stem cells (Fig. 1). If so, the treatment on the bone marrow stem cells may successfully reverse diabetic complications.


Insulin-producing cells normally occur only in the pancreas and thymus. Surprisingly, we found widespread insulin mRNA and protein expression in different diabetic mouse and rat models, including STZ-treated mice and rats, ob/ob mice, and mice with fed high fat diets (HFDs) (Kojima et al., 2004). We detected in diabetic mice proinsulin-positive cells in the liver, adipose tissue, spleen, bone marrow, and thymus. By in situ hybridization, diabetic, but not nondiabetic, mouse liver exhibited insulin transcript-positive cells, indicating that proinsulin was synthesized by these cells. In transgenic mice that express green fluorescence protein (GFP) driven by the mouse insulin promoter (MIP-GFP mice), STZ-induced diabetes led to the appearance of GFP-positive cells in liver, adipose tissue, and bone marrow; the fluorescence signals showed complete concordance with the presence of immunoreactive proinsulin. The above observation indicates that extrapancreatic, extrathymic insulin gene expression occurs in multiple tissues of the different diabetic models. The STZ diabetes model is one of insulin deficiency, whereas the diabetes in ob/ob and HFD-induced obese mice is associated with hyperinsulinemia and insulin resistance. The plasma triglyceride and cholesterol levels were different in these diabetes models. Therefore, hyperglycemia appears to be the common denominator in these models. We examined the hypothesis that hyperglycemia induces extrapancreatic insulin gene expression by administration of i.p. injection of 25% glucose in nondiabetic mice. This led to the appearance of proinsulin-positive cells in multiple organs within merely 3 days. To examine the origin of these proinsulin-positive cells, we performed bone marrow transplantation (BMT) from mice that constitutively express beta-galactosidase to C57BL/6 mice and induced diabetes by STZ treatment. The results showed that most of the extrapancreatic proinsulin-producing cells originated from the bone marrow.

Although the appearance of proinsulin-producing cells in multiple organs was detected in diabetes, these cells do not possess therapeutic potential to lower the glucose level but rather play harmful effects on target organs. Because these proinsulin-positive cells originated from bone marrow produce proinflammatory cytokine TNFα and, therefore, contribute to damage target organs (Chan et al., 2006; Kojima et al., 2006).


Diabetic neuropathy is a major complication of diabetes and is a leading cause of nontraumatic limb amputations. Although many biochemical perturbation have been implicated in the pathogenesis of diabetic neuropathy, including oxidative stress, hypoxia and ischemia, increased advanced glycation end products, activation of protein kinase, and growth factor deficiency, the pathogenesis of neuronal degeneration is not fully understood.

We identified proinsulin-positive cells in the dorsal rot ganglia (DRG) and sciatic nerves of STZ-diabetic rats but not in the nondiabetic controls. Approximately 10% of all DRG neurons expressed proinsulin in diabetic rats, and treatment with insulin markedly reduced the proportion of proinsulin-positive cells as well as preventing the prolongation in motor nerve conduction velocity (Terashima et al., 2005). These proinsulin-positive cells were also positive for TNFα and CD45 by immunohistochemistry, and reverse transcription-PCR (RT-PCR) analysis on isolated cells confirm the mRNA expression of these markers. To determine whether the proinsulin-positive cells came from the bone marrow, we transplanted bone marrow from nondiabetic mice that constitutively expressed beta-galactosidase (Rosa mice) to nondiabetic C57BL/6 mice and induced diabetes by STZ. We found proinsulin-positive cells in DRG and sciatic nerve of diabetic mice displayed overlapping proinsulin and beta-galactosidase staining, indicating that they originated from the bone marrow. DRG neurons that had undergone fusion with bone marrow derived (BMD) cells would be expected to contain more than one set of chromosomes. We further quantified the nuclear ploidy of DRG neurons isolated from diabetic and nondiabetic rats. In nondiabetic rats, 99.1% of the DRG neurons were diploid (2n), tetraploid (4n) cells comprised only 0.9%, and no cells were higher than tetraploid. In contrast, in diabetic rats, 86.8% were diploid, 12.5% were tetraploid, and 0.9% contained 6n or 8n. The presence of polyploidy was strongly associated with proinsulin expression.

Diabetic neuropathy in rodents has been shown to be associated with perturbed calcium homeostasis. For example, DRG neurons from STZ-diabetic rats display increased resting intracellular calcium concentration and a prolonged phase of recovery from KCl-induced depolarization. To explore the pathophysiological basis of the abnormal neuronal function in diabetes, we isolated DRG neurons and measured the kinetics of changes in [Ca2+]i produced by KCl-induced depolarization. The proinsulin-positive neurons of diabetic rats displayed an almost two-fold higher resting [Ca2+]i and prolonged recovery time for depolarization-induced [Ca2+]. In addition to functional defects such as perturbed calcium homeostasis, nerve cells affected by diabetic neuropathy also exhibit enhanced apoptosis. We found that proinsulin-expressing DRG neurons and sciatic nerve display readily detectable levels of caspase-3 mRNA expression, whereas caspase-3 mRNA is undetectable in non-proinsulin expressing cells from either diabetic or nondiabetic mice.

These findings show that a subpopulation of BMD cells marked by proinsulin expression migrate and fuse with neurons in the sciatic nerves and DRG, resulting in neuronal dysfunction and accelerated apoptosis (Terashima et al., 2005).


Hepatocyte degeneration is a common complication of diabetes, which includes a well-known pathology of nonalcoholic steatohepatitis caused by type 2 diabetes, and is a leading cause of liver cirrhosis.

We found that in different types of diabetic mouse models, including STZ-induced, ob/ob, and HFD-induced diabetic mice, produce proinsulin-positive cells in the liver and these cells also produce TNFα simultaneously (Fujimiya et al., 2007). We performed BMT from GFP mice (transgenic mice that constitutively express GFP) to Rosa mice and induced diabetes in the recipients with STZ injection. GFP protein of donor origin and beta-gal of recipient origin were overlapped in the cytoplasm of hepatocytes in diabetic animals, and in such GFP- and beta-gal-positive cells were also positive for proinsulin and TNFα. To confirm the cell fusion between hepatocytes and BMD cells, we further transplanted bone marrow cells from male MIP-GFP donor mice (proinsulin-positive cells are marked by GFP expression) to female Rosa recipients and induced diabetes. In the hepatocytes, GFP and beta-gal were coexpressed in the cytoplasm, and these cells possessed Y chromosome in the cytoplasm. These results suggest cell fusion occurred between hepatocytes and BMD cells in diabetes and fused cells produce proinsulin and TNFα. Such abnormal gene expression in hepatocytes contributes to pathophysiology of diabetic liver injury.

Previous reports document that BMD cells can fuse with hepatocytes in vivo, and such fusion events can contribute to liver regeneration. However, in normal conditions, hepatocytes and BMD cell fusion is very rare at ranges <0.05% of hepatocytes. We found that fusion occurs between hepatocytes with a frequency that is at least >100- to 1,000-fold higher than that in nondiabetic controls (Fujimiya et al., 2007). Therefore, diabetes produces major perturbation in the behavior of BMD cells that may have significant pathological implications.


Diabetic nephropathy (DN) is a major cause of end-stage renal disease and an increase in the incidence of DN has become a worldwide problem. It is established that tubulointerstitial injury, which consists of tubular cell damage with infiltration of inflammatory cells to the interstitium, is a primary feature of DN and leading albuminemia.

We performed BMT from GFP mice to C57BL/6 mice and induced diabetes by STZ injection and feeding a HFD. We found that a large amount of BMD cells accumulated in the renal outer cortex in both types of diabetic models. Accompanied by an increasing number of BMD cells, renal proximal tubules, which are a cause of albuminemia when damaged, showed a decrease in number. Most interestingly, BMD cells as well as renal proximal tubules expressed both proinsulin and TNFα (Yamashita et al., in press).

To confirm the hypothesis that BMD cells fuse with renal tubules and lead to renal dysfunction, we performed sex-mismatched BMT and FISH for the Y chromosome. Results showed that 15.4% of renal tubular epithelial cells were fused with BMD cells in STZ-diabetic mice, 8.6% of renal tubular epithelial cells were fused with BMD cells in HFD-diabetic mice, while 1.5% of renal tubular epithelial cells were fused with BMD cells in nondiabetic mice. Overlap staining showed that Y chromosome containing cells were positive for TNFα and caspase-3. These results suggest that diabetes activates fusion of BMD cells with renal tubular epithelial cells, causing chromosomal instability, producing proinflammatory cytokine and subsequently tubular cell degeneration. Therefore, not only migration of cytokine-producing BMD cells in the tubular interstitium but also active fusion with renal tubular cells may play a crucial role in DN (Yamashita et al., 2011).


It has been widely known that diabetes causes abnormality in the bone marrow stem cells. Mesenchymal stem cells (MSCs) in the bone marrow are characterized as differentiating into bone, cartilage, and adipose tissues. In diabetic patients, it has been observed that the number of MSCs committed to the adipocyte lineage increases while the number of those committed to the osteoblastic lineage decreased (Rosen et al., 2009). Such unbalanced proadipocytic and antiosteoblastic MSC allocation could result from increased activity of Peroxisome Proliferator-Activated Receptor γ2 (PPARγ2) or the decreased expression of Transforming growth factor-β (TGF-β)/Bone Morphogenic Protein (BMP), Wnt/βcatenin, and Insulin-like growth factor-1 (IGF-1) signaling pathway induced by diabetes. Not only the differentiation abnormality but also the limited proliferation capacity in the MSC has been shown in the diabetic patients (Phadnis et al., 2009). It is well known that MSCs in the bone marrow play important roles for regeneration of various tissues. For example, autologous bone marrow-derived MSCs transplantation can effectively improve cardiac function after myocardiac infarction. However STZ-induced diabetic rats-derived bone marrow MSCs have impaired ability in proliferation, paracrine release of vascular endothelial growth factor (VEGF) and IGF-1, anti-apoptosis and myogenic differentiation in transplanted tissues (Jin et al., 2010).

Diabetes impairs the function of hematopoietic stem cells (HSCs) in the bone marrow, which is associated with endothelial progenitor cell dysfunction and reduced neovascularization in response to tissue ischemia (Orlandi et al., 2010). Long-term diabetes in mice induced a reduction in Lin, Sca-1+, c-kit+ hematopoietic progenitor cells, which include endothelial progenitor cells. In human type 2 diabetes, the reduction of bone marrow-derived circulating progenitor cells has been proposed as a mechanism of cardiovascular complication (Fadini et al., 2010). In fact, peripheral blood CD34+ cells were significantly reduced and this was associated with the reduction of circulating endothelial progenitor cells (Fadini et al., 2010). Osteoclasts are derived from HSCs and delayed bone formation in diabetes has been shown as a result of malfunction of bone marrow-derived osteoclasts (Kasahara et al., 2010). The size of osteoclasts is significantly smaller in STZ-diabetic mice and the expression of dendritic cell- specific transmembrane protein (DC-STAMP) is lower compared to nondiabetic controls (Kasahara et al., 2010). A delayed would closure was found in diabetic mice and this was associated with diminished circulating Lin-, Sca-1+, c-kit+ hematopoietic progenitor cells (Tepper et al., 2010). Such impaired hematopoietic progenitor cell mobilization is caused by diminished expression of stromal cell-derived factor-1 (SDF-1) in the bone marrow stem cells (Tepper et al., 2010). For the maintenance of stem cell function in the bone marrow, bone marrow niche which includes osteoblastic niche, vascular niche or other reticular cells provide the essential role to keep microenvironment for stem cell function. Therefore, diabetes might cause the disturbed stem cell niches in the bone marrow, thereby, impairing progenitor cell-dependent tissue repair (Fig. 2).

Figure 1.

Bone marrow stem cell abnormality causes diabetic complications.

Figure 2.

Functions of MSCs and HSCs in the bone marrow.


A series of our study demonstrate that hyperglycemia induces abnormality in the bone marrow stem cells and abnormal BMD cells fuse with hepatocytes, peripheral neurons, and renal tubular epithelial cells that leads degeneration of target organs (Fig. 1). Further studies are necessary to elucidate the fundamental mechanism to cause abnormality in MSCs and/or HSCs in the bone marrow, and mechanism to raise fusion potential induced by hyperglycemia. It is highly possible not only bone marrow cells but also peripheral cells in various organs may respond to hyperglycemia and chemokine/cytokine expression from these cells may accelerate homing of circulating BMD cells to make efficient cell fusion. In addition, the clinical relevance of diabetes-induced bone marrow stem cell abnormality may provide the therapeutic strategies for diabetic complications.