Diabetes Causes Bone Marrow Autonomic Neuropathy and Impairs Stem Cell Mobilization via Dysregulated p66Shc and Sirt1

Diabetes reduces the availability of bone marrow (BM)-derived circulating angiocompetent CD34⁺ cells, especially in the presence of chronic vascular complications (1,2). This is believed to represent a risk factor for adverse outcomes, as a low CD34⁺ cell count is associated with incident cardiovascular events (3). Recently, the BM has emerged as a previously neglected target of hyperglycemic damage in experimental and human diabetes, with features of microangiopathy and niche dysfunction (4–7). As a result of BM remodeling, mobilization of stem and progenitor cells is impaired in diabetic animals and humans (7–10). Since mobilization of BM-derived stem/progenitor cells contributes to angiogenesis and endothelial repair (11,12), such stem cell “mobilopathy” (13) is expected to promote cardiovascular disease in diabetes.

The sympathetic nervous system (SNS) is prominently involved in BM niche function (14) and stem cell trafficking is regulated by catecholaminergic neurotransmitters (15,16). Recently, it has also been demonstrated that BM denervation strongly contributes to chemotherapy-induced myelotoxicity (17). Diabetic patients often experience autonomic neuropathy, which is characterized by pauperization of SNS fibers in multiple organs (18–20). Thus, we herein hypothesize that diabetic autonomic neuropathy (DAN) can affect BM function and stem cell mobilization. Starting from the observation that patients with DAN have a marked pauperization of peripheral blood (PB) CD34⁺ cells, we used several experimental models and conditions to demonstrate that autonomic neuropathy in the BM is causally linked to impaired stem cell mobilization in diabetes, which is associated with defective reperfusion after ischemia. Our data show that denervation in the diabetic BM is mediated by the 66-kDa protein from the src homology and collagen homology domain (p66Shc) and that impaired mobilization relies on sirtuin 1 (Sirt1) dysregulation, both of which are therefore potential targets to restore BM cell–mediated vascular repair.

The protocol was approved by the Ethics Committee of the University Hospital of Padova, and experiments were conducted in accordance with the Declaration of Helsinki, as revised in 2000. Patients were recruited at the diabetes outpatient clinic of the University Hospital of Padova. All consecutive type 1 diabetes (T1D) or type 2 diabetes (T2D) patients aged 20–90 years and undergoing autonomic neuropathy tests were eligible, provided they were clinically stable and free from exclusion criteria (recent acute disease or infection, trauma, or surgery; uncontrolled hypertension or hyperglycemia, prevalent hypoglycemia; immunological disorders, organ transplantation, or immunosuppression; pregnancy and lactation). The following data were recorded: age, sex, type of diabetes and disease duration, HbA_1c level, BMI, waist circumference, smoking habit, systolic and diastolic blood pressure, history of hypertension and use of antihypertensive medications, resting heart rate, lipid profile, albumin/creatinine excretion ratio, diabetes complications, and medications. Diabetic nephropathy was defined as a urinary albumin/creatinine ratio >30 mg/g or an estimated glomerular filtration rate <60 mL/min/1.73 m². Coronary artery disease was defined as a history of myocardial infarction or angina, confirmed by coronary artery angiography or a myocardial stress test. Peripheral arterial disease was defined as a history of claudication or rest pain with evidence of stenosis of leg arteries upon invasive or noninvasive examination. Cerebrovascular disease was defined as a history of stroke or evidence of >30% carotid stenosis upon ultrasound examination. Diagnosis of DAN was performed according to standard interpretation of R-R interval variations during a change from lying to standing, deep breathing, and Valsalva maneuver, as well as according to the degree of orthostatic hypotension. Age-specific thresholds were used to define pathological vs. normal DAN tests.

All procedures were approved by the local ethics committee and from the Italian Ministry of Health. Experiments were conducted according to the National Institutes of Health Principles of Laboratory Animal Care. Vav1-Sirt1^−/−(generated by crossing Vav1^cre/+ mice with Sirt1^stop-flox mice), Vav1-Sirt1^TG (generated by crossing Vav1^cre/+Sirt1^ex4.Δflox/+ with Sirt1^{ex4.flox/flox}), and Vav1-YPF mice were bred at the external mouse facilities of the University of Frankfurt (mfd Diagnostics GmbH). Sell^−/− mice were bred at the mouse facilities of the Institut für Labortierkunde, University of Zurich (Zurich, Switzerland). Sell^−/− mice were generated by Richard Hynes, a Howard Hughes Medical Institute Investigator at the Massachusetts Institute of Technology (Cambridge, MA). p66Shc^−/− and ob/ob p66Shc^−/− animals were bred at the animal facility of the University of Padova. The ob/ob mice were from The Jackson Laboratory (Bar Harbor, ME). Age-matched wild-type animals were obtained from the in-house colony of the animal facility of the Venetian Institute of Molecular Medicine (Padova, Italy). All animals were on a C57BL/6 background. Diabetes was induced with a single intraperitoneal injection of 150 mg/kg streptozotocin (STZ) (Sigma-Aldrich, St. Louis, MO) in citrate buffer 50 mmol/L, pH 4.5. Blood glucose was measured with Glucocard G-meter (Menarini, Florence, Italy); animals with blood glucose levels ≥300 mg/dL in at least two measurements within the first week were classified as diabetic and were housed for 4 weeks with feeding and drinking ad libitum before performing experiments. In a separate set, animals were killed at week 1, 2, and 3 after STZ treatment for a time-course analysis (n = 4–5/group). For SNS disruption, animals received three intraperitoneal injections of 100 mg/kg 6-hydroxydopamine (6-OHDA; Tocris Bioscience, Bristol, UK), which destroys sympathetic neurons through oxidative stress and other mechanisms. Animals were housed for 4 weeks before a second round of injections with two doses of 6-OHDA (100 mg/kg on day 0, 250 mg/kg on day 2), and experiments were started on day 5.

Animals were sedated with 10 mg/kg zolazepam/thylamine (Zoletil; Virbac, Nice, France) and 7 mg/kg xylazine (Xilor; Bio98 Srl, Milan, Italy). The femoral artery and the vein were surgically dissected from the femoral nerve, then cauterized with low-temperature cautery, and excised between inguinal ligament and hackle. Fifteen days after surgery, we measured hind-limb microvascular perfusion with the Periscan-Pim II Laser Doppler System (Perimed AB, Järfälla, Sweden). Each measure was repeated five times.

Progenitor cell levels were quantified in PB after peripheral ischemia or after 4 days of subcutaneous injection of 200 μg/kg/day human recombinant granulocyte colony-stimulating factor (G-CSF) (Filgrastim; Roche, Basel, Switzerland). Based on a preliminary analysis of the kinetics of endothelial progenitor cells (EPCs) and Lin⁻cKit⁺Sca1⁺ (LKS) cells after ischemia, EPC levels peaked at day 3 after ischemia, and LKS cell levels peaked at day 14. Therefore, these time points were used for all subsequent analyses. AMD3100 (Tocris Bioscience) was administered intraperitoneally at 6 mg/kg, and mobilization was assessed after 4 h. Desipramine treatment (10 mg/kg i.p.; Tocris Bioscience) was started 4 days before G-CSF treatment and continued for the 4 days of G-CSF treatment. PB was collected at baseline and after mobilization. One hundred fifty microliters of PB was stained with a rat anti-mouse allophycocyanin lineage cocktail (BD, Franklin Lakes, NJ), phycoerythrin rat anti-mouse Sca-1 (Ly6A/E; BD) and fluorescein isothiocyanate rat anti-mouse cKit (BD) to quantify LKS cells or with Alexa Fluor 647 rat anti-mouse CD34 (BD) and Alexa Fluor 488 anti-mouse Flk-1 (BioLegend, San Diego, CA) to quantify endothelial-committed progenitors. CD62L (l-selectin) expression on BM LKS cells was performed by staining 10⁶ nonfractioned BM cells with LKS antibodies and PerCp Cy5.5 anti-Mouse CD62L (BD). A total of 250,000 events were acquired for each analysis, and the level of progenitor cells was expressed as the number of positive events per 1,000,000 total events. Data were acquired using a FACSCalibur instrument (BD Biosciences, San Jose, CA) and analyzed with FlowJo X (Tree Star Inc., Ashland, OR). For quantification of circulating stem cells in humans, we stained 150 μL of freshly collected PB with a phycoerythrin-conjugated anti-human CD34 monoclonal antibody (BD) and gated CD34⁺ cells in the mononuclear cell fraction. CD34⁺ cell count was expressed either as the number of events per 10⁶ total events or cells per milliliter by multiplying the fractional CD34⁺ cell count by white blood cells obtained from the automated hemocytometer.

Adductor femoris longus muscles were frozen in isopentane cooled in liquid nitrogen immediately after dissection. Muscles were then stored at −80°C for further analysis. Muscle sections were incubated with a DyLight594-conjugated Isolectin B4 (Bandeiraea simplicifolia, 1:100; Vector Laboratories) and costained with Oregon Green–conjugated wheat-germ agglutinin (1:400; Invitrogen) to allow identification of capillaries and muscle fibers. Nuclei were counterstained with Hoechst 33258 (Sigma-Aldrich). Femurs were fixed for 24 h in 4% paraformaldehyde at 4°C, washed with PBS, and then decalcified by incubation with 5% formic acid for 2 days at 4°C. Demineralized femurs were then covered with OCT embedding medium and frozen in liquid nitrogen. Ten-micrometer-thick longitudinal femur or transversal muscle sections were obtained with a Leica CM1950 cryostat. Femur sections were fixed in 4% paraformaldehyde and incubated with tyrosine hydroxylase antibody overnight at 4°C (rabbit polyclonal antibody, 1:100, AB152; Millipore). A mock incubation, without primary antibody, was performed in parallel. Images were acquired with a Leica DM5000B microscope equipped with a Leica DFC300 FX charge-coupled device camera, and were processed with ImageJ software (National Institutes of Health). Ten random ×20 magnification pictures were obtained from each muscle sample.

Human PB mononuclear cells (PBMCs) were obtained from healthy donors and separated by gradient centrifugation with Histopaque-1077 (Sigma-Aldrich). A total of 10⁷ cells/well were plated in a six-well culture plate (BD) with RPMI 1640 medium supplemented with 10% FBS and 1% penicillin/streptomycin/glutamine. Isoproteronol (Sigma-Aldrich) was added for 48 h. Then cells were collected and lysed for gene expression analysis. To test chemical toxicity, BM cells from control mice were incubated for 1 h with STZ and 6-OHDA, followed by 24 h of recovery at 37°C 5% CO₂ with RPMI 1640 supplemented with 10% FBS and 1% penicillin/streptomycin/glutamine, at concentrations that are supposed to be achieved in vivo in mice. Live early and late apoptotic and necrotic cells (total BM cells or gated CD34⁺ cells) were quantified by flow cytometry staining with propidium iodide and Annexin-V (BD). Hematopoietic colonies were grown from unfractioned BM cells and quantified using the MethoCult system (StemCells Inc., Vancouver, BC, Canada).

After mice were killed, bones were harvested, cleaned of debris, and flushed with ice-cold PBS; cells were collected through a 40-μm sterile filter, total RNA was extracted using the RNeasy kit (Qiagen), and RNA was quantified using a Nanodrop Spectrometer (Thermo Fisher Scientific). Four hundred nanograms of RNA were reverse-transcribed to cDNA using the First-Strand cDNA Synthesis Kit (Invitrogen). Duplicates of sample cDNA were then amplified on the 7900HT Fast Real-Time PCR System (Applied Biosystems) using the Fast SYBR Green RT-PCR kit (Applied Biosystems) in 96-well plates (Applied Biosystems). Expression data of selected genes were normalized against housekeeping genes and were analyzed using the 2^(-ΔΔct) method. Primer sequences are reported in Supplementary Table 3.

Data are expressed as the mean ± SE or percentage. Comparison between two or more groups was performed using Student t test or ANOVA (for normal variables), and with Mann-Whitney U test and Kruskal-Wallis test (for non-normal variables). For human studies, a multivariable analysis was run to detect the independent effect of DAN: the CD34⁺ cell count was the dependent variable, and clinical characteristics that were different at univariate analysis in Table 1 were explanatory independent variables. Statistical significance was accepted at P < 0.05.

First, we found that the coexistence of multiple pathologic cardiovascular autonomic tests used to diagnose DAN was associated with a progressive reduction of CD34⁺ cells in PB of diabetic patients (Fig. 1A and B). Independent of confounders, patients with DAN had a 40% reduction in PB CD34⁺ cells compared with DAN-free patients (Fig. 1C and D, Table 1, and multivariate analysis in Supplementary Table 1). A breakdown analysis by DAN test showed trend reductions of PB CD34⁺ cells in the presence of the alteration of each of the autonomic function tests (Fig. 1E and F). These data suggest that autonomic neuropathy in diabetic patients reduces the number of circulating CD34⁺ cells, which have been previously shown to exert proangiogenic activity in vivo (21), that their levels are informative of the BM status (8) and predict future cardiovascular events (22).

Next, we report that experimental T1D induced by STZ caused a marked reduction in tyrosine-hydroxilase⁺ (Tyr-OH⁺) SNS fibers in the BM. This was equal to ∼80% of Tyr-OH staining reduction obtained by chemical sympathectomy with 6-OHDA. STZ is considered to exert little or no myelosuppression because it enters the cells via glucose transporters (23). However, to rule out that BM sympathectomy was attributable to STZ toxicity and not to diabetes, we analyzed BM Tyr-OH staining in ob/ob mice, a model of T2D, and found a similar pauperization of SNS fibers (Fig. 2).

To test the consequence of sympathectomy on BM function in diabetes, we analyzed LKS cells and CD34⁺Flk1⁺ cells (EPCs) in T1D mice and 6-OHDA–treated mice before and after stimulation with G-CSF or hind-limb ischemia. Consistent with previous findings (8), diabetic animals were unable to mobilize LKS cells and EPCs after G-CSF or ischemia. Mice treated with 6-OHDA also showed impaired mobilization of LKS cells and EPCs in response to both stimuli (Fig. 3A and B). While the SNS dependency of the mobilizing effect of G-CSF is well-established (14), the role of BM SNS in ischemia-induced mobilization was hitherto unknown. The BM content of EPCs, but not of LKS cells, was reduced in diabetic animals, while steady-state circulating levels of EPCs and LKS cells were unaffected by diabetes or sympathectomy (Supplementary Fig. 1). In vitro, STZ and 6-OHDA induced minor changes in the survival of total BM cells and CD34⁺ BM cells, and in the generation of hematopoietic colonies (Supplementary Fig. 2). These data rule out that chemical toxicity or absolute pauperization of intramarrow stem cells account for the diabetes- and neuropathy-associated mobilopathy.

Because sympathectomy caused similar mobilization impairment as that in diabetes, we reasoned that the diabetic stem cell mobilopathy could be attributed to autonomic denervation. In support of this, we show that diabetic mice were able to mobilize LKS cells in response to AMD3100/Plerixafor (Fig. 3C), which is a SNS-independent mobilizer (24). In addition, the norepinephrine (NE) reuptake inhibitor desipramine, which potentiates residual SNS outflow in autonomic ganglia (25), partially restored G-CSF–induced LKS cell mobilization in diabetic mice (Fig. 3D).

As a functional readout of impaired mobilization after ischemia, we found that, although the microvasculature was mildly affected, hind paw perfusion was significantly impaired in diabetic and sympathectomized mice (Fig. 3E). This is in compliance with the known reperfusion defect of diabetes (26,27) and the recent finding that catecholamines regulate postischemic tissue remodeling (28). Collectively, these data indicate that diabetes and autonomic neuropathy hamper BM stem cell mobilization and postischemic responses.

Previous studies in mice show that p66Shc mediates hyperglycemic damage and diabetes complications (29). Therefore, we tested whether p66Shc is involved in BM denervation induced by diabetes. First, we report that gene expression of p66Shc was increased in PBMCs of diabetic patients, which was worsened by the presence of DAN (Fig. 4A and Supplementary Table 2). In addition, BM cells from T1D (induced by STZ), T2D (by crossing with ob/ob mice), and sympathectomized (6-OHDA) mice showed increased p66Shc gene expression compared with controls (Fig. 4B). The functional consequence of this association was tested using p66Shc^−/− mice, which, once made diabetic by STZ injection (T1D) or by crossing with ob/ob mice (T2D), were protected from BM sympathectomy, as evidenced by the preserved number of Tyr-OH⁺ fibers (Fig. 4C and D). As a result, T1D (induced by STZ) and T2D (by crossing with ob/ob mice) p66Shc^−/− mice were able to mobilize LKS cells and EPCs after G-CSF administration, despite the fact that they were as hyperglycemic as wild-type diabetic mice (Fig. 4E). To confirm that restored mobilization in p66Shc^−/− diabetic mice was due to preserved BM innervation, we show that STZ diabetic p66Shc^−/− mice treated with 6-OHDA, which effectively denervates the BM despite p66Shc knockout (KO) (Fig. 4D), were unable to mobilize LKS cells after G-CSF stimulation (Fig. 4F).

Because Sirt1 has been implicated in EPC survival and proangiogenic activity (30), we analyzed its contribution in the mobilopathy model. We first found that Sirt1 gene expression was reduced in PBMCs of diabetic patients with DAN compared with those without (Fig. 5A and Supplementary Table 2). Consistently, Sirt1 gene expression was reduced in BM cells of T1D, T2D, and sympathectomized mice compared with controls (Fig. 5B). The link between neuropathy and BM Sirt1 downregulation is strengthened by the evidence that preservation of BM innervation by p66Shc deletion prevented BM Sirt1 mRNA reduction induced by T1D (by STZ administration) and T2D (by crossing with ob/ob mice) (Fig. 5B). A cAMP-dependent activation of Sirt1 has been recently described (31), and we show here that treatment of human PBMCs with the nonselective β-adrenergic agonist isoproterenol stimulated Sirt1 mRNA expression in vitro (Supplementary Fig. 3), providing a link between sympathectomy and reduced BM Sirt1 expression. Consistently, we also found that potentiation of NE signaling with desipramine restored Sirt1 gene expression in the BM of T1D (by STZ administration) mice (Fig. 5C).

To understand the functional consequences of Sirt1 downregulation in the BM, we generated hematopoietic-restricted Sirt1^−/− mice (Vav1-Sirt1^−/−), lacking Sirt1 in cells that express the Vav1 guanine nucleotide exchange factor, which is specific for the hematopoietic system (Supplementary Fig. 4) (32). Despite the fact that inconsistent data have been reported on hematopoietic function in ubiquitous and hematopoietic Sirt1^−/− mice (33,34), Vav1-Sirt1^−/− mice, when subjected to G-CSF stimulation and hind-limb ischemia, recapitulated the poor mobilizer phenotype observed in diabetic and sympathectomized mice (Fig. 5D), supporting the idea that Sirt1 downregulation in the BM impairs mobilization. To confirm this hypothesis, we generated Vav1-Sirt1 transgenic (TG) mice (Vav1-Sirt1^TG), which showed a 12-fold increased expression of Sirt1 in BM LKS cells (Supplementary Fig. 4). Once made diabetic with STZ or sympathectomized with 6-OHDA, Vav1-Sirt1^TG mice were able to mobilize LKS cells and EPCs after G-CSF and hind-limb ischemia (Fig. 5E and F), and showed improved postischemic paw perfusion (Fig. 5G and H). To test the role of Vav1-expressing cells in the ischemic vasculature, we analyzed Vav1-YFP mice, which express the yellow fluorescent protein (YFP) only in hematopoietic (Vav1⁺) cells. Based on the integration of YFP⁺ cells into the muscle vascular endothelium of ischemic compared with nonischemic hind limbs, we show that Vav1⁺ cells can contribute to postischemic angiogenesis (Supplementary Fig. 5), explaining why modulation of Sirt1 in Vav1⁺ cells can translate into improved recovery after ischemia. These data indicate that reduced Sirt1 expression in the BM is responsible for the impaired stem cell mobilization in diabetic and sympathectomized mice and that this can impact on vascular recovery.

Data obtained in Vav1-Sirt1^KO/TG mice indicate that mobilization is affected by hematopoietic cell-intrinsic mechanisms in diabetic and sympathectomized mice. In addition, the mechanism of impaired stem cell mobilization associated with DAN are likely to be very downstream in the cascade of events that make cells leaving the niche, as is common to G-CSF and ischemia, which act on different niche components (35). Therefore, we analyzed the gene expression of a series of typical niche adhesion molecules (CD11a, CD11c, CD49d, CD49e, CD62L, and intracellular adhesion molecule 1) in BM cells, and found that they were globally upregulated in diabetic and sympathectomized mice compared with controls (Supplementary Fig. 6). The link between diabetic BM neuropathy and changes in gene expression is supported by the finding that upregulation of adhesion molecule genes occurred after the development of neuropathy in T1D mice (Supplementary Fig. 7). We focused on CD62L/l-selectin because overexpression of this molecule was previously noted by others in the diabetic BM (7). Upregulation of CD62L was confirmed by flow cytometry on LKS cells of T1D (by STZ administration), T2D (by crossing with ob/ob mice), sympathectomized, and Vav1-Sirt1^−/− mice, whereas it was prevented in p66Shc^−/− LKS cells of diabetic and sympathectomized mice (Fig. 6A and B). The link between adrenergic signaling, Sirt1 and CD62L expression is supported by the decrease in CD62L mRNA after treatment of PBMCs with isoproterenol (Supplementary Fig. 1). l-Selectin (CD62L) is prominently involved in the trafficking of blood cells to lymphoid tissues (36), and excess l-selectin expression may render the cells more adherent to the BM stroma, preventing mobilization. The relevance of l-selectin overexpression in this setting is confirmed by the observation that genetic l-selectin deletion (Sell^−/−) partially restored mobilization in diabetic and sympathectomized mice (Fig. 6C).

This study shows for the first time that autonomic neuropathy in the BM impairs stem cell mobilization in diabetes with dysregulation of the life span–determinant genes p66Shc and Sirt1. Previous data on BM SNS in diabetes models were inconsistent, reporting either reduced (37) or increased (7) fibers. Our data indicate that T1D (by STZ administration) and T2D (by crossing with ob/ob mice) in mice are characterized by a profound depletion of BM SNS fibers. We also show that BM denervation in models of diabetes occurs via p66Shc. p66Shc is induced by hyperglycemia (38) and mediates the development of several complications of experimental diabetes, including cardiomyopathy (39), nephropathy (40), and delayed wound healing (41). In addition, p66Shc contributes to the hyperglycemic damage of BM-derived EPCs (42). Although the link between DAN and mobilopathy cannot be firmly established, the preservation of BM innervation and the rescue of mobilization in p66Shc^−/− diabetic mice suggest that neuropathy impairs mobilization in models of diabetes. Other eventual effects of p66Shc deletion seem important in this model, because chemical denervation abolished the mobilization rescue in diabetic p66Shc^−/− mice. The mobilizing activity of G-CSF is dependent on the SNS (14), and we show that hind-limb ischemia also requires a functional SNS system to elicit stem/progenitor mobilization, because it is defective in sympathectomized mice. On the other hand, AMD3100 directly disrupts the CXCR4/CXCL12 interaction to promote mobilization independently of the SNS (24). The effectiveness of AMD3100 in inducing mobilization in diabetic mice and the inefficacy of both SNS-dependent stimuli (G-CSF and ischemia) are in line with the hypothesis that neuropathy impairs mobilization in diabetes. In addition, potentiation of residual SNS output by treatment with the NE reuptake inhibitor desipramine partially restored G-CSF–induced mobilization in diabetic mice. It has been demonstrated that G-CSF stimulates the SNS by reducing NE reuptake, while desipramine can rescue G-CSF–triggered mobilization in mice that mobilize poorly (25). These data suggest that a defective adrenergic signaling is responsible for impaired mobilization in diabetes and suggests a potential treatment strategy. We have previously shown that T1D and T2D patients have impaired G-CSF–induced mobilization, irrespective of the neuropathy (9). However, SNS fiber loss can be present in up to 50% of patients without clinical/instrumental evidence of DAN (43). Thus, investigation into the history and features of diabetic BM dysautonomy is an interesting matter for future research.

Sirt1 has been implicated in EPC survival and proangiogenic activity (30), and its expression is reduced in EPCs of T2D patients (44) and is lowered by high glucose levels in THP-1 cells (45). Diabetic as well as chemically induced neuropathy reduced Sirt1 expression in the BM. The mechanism whereby the SNS regulates Sirt1 expression remains unclear, but we show that activation of the β-adrenergic receptor increases Sirt1 mRNA in blood cells in vitro. The link between neuropathy and Sirt1 repression is strengthened in vivo by the observation that potentiating NE signaling with desipramine rapidly restored Sirt1 expression in the BM of T1D mice. In addition, the induction of diabetes in p66Shc^−/− was not followed by Sirt1 mRNA reduction in the BM, which can be attributed to the preserved BM innervation, although we cannot definitely rule out that metabolic dysregulations independent of neuropathy can affect Sirt1 expression. Importantly, we show that Sirt1 downregulation in BM cells is causally linked to impaired mobilization. In fact, hematopoietic-specific Sirt1 KO recapitulated the poor mobilizer phenotype, while hematopoietic Sirt1 overexpression rescued mobilization in diabetic and sympathectomized mice. As hematopoietic cell–intrinsic modulation of Sirt1 was sufficient to influence mobilization, we focused on a series of adhesion molecules typically implicated in the trafficking of stem cells in the BM. The generalized induction of adhesion molecule genes in diabetic and sympathectomized mice, by rendering the cells more adherent to the stroma, can thus prevent mobilization. Excess CD62L (l-selectin) expression on BM LKS cells of diabetic mice was previously noted (7), and we now report that this is in fact implicated in the diabetes- and neuropathy-induced mobilopathy, because l-selectin deletion was sufficient to rescue mobilization in both models. Other adhesion molecules likely provide redundant signals, such that moderate increases in their expression may additively promote stem cell retention.

Together, our data suggest that BM autonomic neuropathy induced by diabetes via p66Shc plays a central role in the impaired BM stem cell mobilization, involving dysregulation of hematopoietic Sirt1 and cell-intrinsic mechanisms including adhesion molecule overexpression. As multiple redundant mechanisms regulate the complex stem cell niche, it is likely that SNS-independent pathways contribute to diabetes-induced impairment in stem cell mobilization. Importantly, both in mice and in humans, diabetes induces extensive microvascular remodeling of the BM (6). We now expand this notion by showing that neuropathy is an important component of diabetic niche remodeling, with negative functional implications for mobilization.

Vascular stem/progenitor cells mobilized from the BM by G-CSF have been shown to stimulate endothelial healing and angiogenesis (46,47). In addition, ischemia-induced mobilization is considered a natural response to tissue damage that promotes angiogenesis and reperfusion (11). Therefore, impaired mobilization of cells with vasculotropic activity is expected to promote cardiovascular disease. We show that the poor mobilizer phenotype of diabetic and sympathectomized mice is associated with defective recovery after ischemia. While several BM-independent mechanisms can contribute to this phenomenon, it is of interest that cell-intrinsic Sirt1 overexpression rescues postischemic mobilization and reperfusion at the same time. The relevance of Sirt1 modulation in Vav1-expressing cells is supported by the contribution played by Vav1⁺ cells to the microvasculature of ischemic muscles. Therefore, our data suggest that the rescue of BM function in diabetes can promote recovery from ischemia.

DAN is a disabling condition with reduced life expectancy and increased cardiovascular risk (18). Our data suggest a novel possible mechanism, whereby DAN worsens vascular health. Future investigation into the features of BM neuropathy in humans and its relationships with cardiovascular DAN will shed light on this novel diabetes complication. In addition, diabetes is considered an accelerated aging disease (48), and dysregulation of longevity-determinant hubs, such as Sirt1 and p66Shc, emerges as a critical factor in inducing aging features, including autonomic neuropathy, impaired stem cell mobilization, and defective recovery from ischemia. Therefore, countering aging-associated pathways has the potential to tackle vascular complications of diabetes by restoring BM function and BM-derived stem/progenitor cells.

Funding. This study was supported by LOEWE Center for Cell and Gene Therapy and SBF 834 B8/Z1 grants to M.T.; and a European Foundation for the Study of Diabetes/Lilly Grant and Fellowship Program grant and an Italian Ministry of Health grant (GR-2010-2301676) to G.P.F.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. M.A. designed and performed in vivo and ex vivo experiments and contributed to manuscript writing. N.P., S.C., and G.C. performed molecular biology analyses. M.T. and R.M. generated and analyzed transgenic Vav1 mice. L.M., M.G., and P.P. generated and analyzed p66Shc KO mice. S.V.d.K. contributed to patient recruitment and characterization. A.A. supervised the project, provided financial support, and wrote the manuscript. G.P.F. designed the project, recruited and characterized patients, analyzed data, wrote the manuscript, and provided financial support. G.P.F. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.