Diabetes Impairs Stem Cell and Proangiogenic Cell Mobilization in Humans

The study was approved by the local ethics committee and is registered in ClinicalTrials.gov (NCT01102699). This was a prospective, parallel group study of direct BM stimulation with hrG-CSF in subjects with and without DM. The primary end point was change in circulating CD34⁺ cells from baseline. Secondary end points were changes in other progenitor cell phenotypes, proangiogenic capacity of peripheral blood mononuclear cells (PBMCs), white blood cells, and safety. The study was not designed and powered to detect baseline differences in progenitor cell levels. DM patients were recruited at the outpatient clinic of the University Hospital of Padova, and healthy control subjects were volunteers from the local community. Both type 1 DM and type 2 DM patients were eligible because preclinical studies have shown similar BM alterations and progenitor cell reductions in both types of DM (4,5). Exclusion criteria were as follows: age <25 or >65 years; any acute disease or infection; recent trauma, surgery, or cardiovascular event; chronic immune or infectious diseases; current or past hematological disorders or malignancy; leukocytosis, leukopenia, or thrombocytopenia; organ transplantation or immune suppression; advanced diabetic retinopathy; altered liver function; severe renal failure (estimated glomerular filtration rate <30 mL/min/m²); anomalies in lymphocytes subpopulations; allergy to Filgrastim; bronchial asthma or other chronic lung disorders; and impossibility to provide informed consent. For each patient, we collected anthropometric measures, data on concomitant risk factors, HbA_1c, eventual DM complications, and therapy.

After providing informed consent, patients were subjected to baseline examination and blood samples, including determination of the complete leukocyte counts and lymphocyte subpopulations, liver enzymes, renal function, plasma protein electrophoresis, erythrocyte sedimentation rate, C-reactive protein, prothrombin time, uric acid, and standard urine examination. After verification of inclusion and exclusion criteria, at 8:30 a.m. after an overnight fast, eligible patients were subjected to a baseline peripheral blood sampling for circulating progenitor cell quantification and collection of PBMCs. Immediately after, they were injected subcutaneously with 5 μg/kg Filgrastim (Granulokine; Amgen). Twenty-four hours later, another peripheral blood sample was obtained to evaluate the effects of Filgrastim. Study subjects were invited to register and report any eventual side effect occurred after Filgrastim injection. Dosage of the drug was chosen as the minimum effective dose based on available pharmacodynamic data on Filgrastim, showing that 5 μg/kg is sufficient to increase absolute count of circulating CD34⁺ cells in healthy control subjects (13).

Circulating progenitor cells were quantified using flow cytometry as previously described in detail (14). In brief, after erythrocyte lysis, 150 μL peripheral blood were stained with 10 μL fluorescein isothiocyanate-conjugated anti-human CD34 mAb (Becton Dickinson), 10 μL phycoethrin-conjugated anti-human KDR mAb (R&D Systems), and 10 μL allophycocyanin-conjugated anti-CD133 mAb (Miltenyi Biotech). The frequency of peripheral blood cells positive for these reagents was determined by a two-dimensional side-scatter fluorescence dot plot analysis after appropriate gating. We gated CD34⁺ or CD133⁺ peripheral blood cells in the mononuclear cell fraction and then examined the resulting population for the dual expression of KDR. At the intersection of the CD34 and CD133 gates, we identified CD34⁺CD133⁺ cells, which were examined for KDR expression. In all patients, we also quantified the expression of CD26/DPP-4 on CD34⁺ cells using a phycoethrin-labeled anti-CD26 mAb (Becton Dickinson). In separate analyses, CD45 costaining was performed and showed that >90% of CD34⁺ cells are CD45dim. For fluorescence-activated cell sorter analysis, 5 × 10⁵ cells were acquired and scored using a FACSCalibur (BD). Data were processed using the Macintosh CELLQuest software program (BD). The same trained operators, blind to the clinical status of the patients, performed the tests throughout the study. Absolute progenitor cell counts per unit of blood were derived by multiplying fractional data per white blood cell count. We previously have shown that reproducibility of CD34⁺ cell quantification with this method is high (intraclass correlation coefficient 0.94; 95% CI 0.88–0.96; coefficient of variation [CV] 6.3%) (14).

To gather information on the presence of functional circulating proangiogenic cells and how they are modulated by hrG-CSF in subjects with and without DM, we used the in vivo Matrigel plug angiogenesis assay with patients’ PBMCs. Data suggest that diverse monocyte subsets, including monocytic EPCs and Tie2-expressing monocytes, have proangiogenic capacity (15,16). In brief, PBMCs were isolated with Histopaque (Sigma-Aldrich). Cell count and viability were assayed with an automated BioRad TC20 cell counter, and then, 3 × 10⁶ PBMCs were resuspended in 500 μL phenol-free Matrigel (catalog number 356237; BD) and implanted subcutaneously into the dorsum of immunodeficient RAG-2/γ(c) double knockout mice (in-house colony). The experiment was performed with pre-hrG-CSF and post-hrG-CSF PBMCs of five non–DM control subjects, five type 1 DM patients, and five type 2 DM patients. To minimize variability, the same mouse received pre-GCSF and post-GCSF PBMCs of the same subject. Plugs were explanted 10 days later for macroscopic inspection, histology (hematoxylin and eosin staining), and determination of the hemoglobin-to-protein content ratio (Drabkin solution and Bradford reagent, respectively; Sigma-Aldrich), which is a surrogate of perfusion. Hemoglobin-to-protein ratio was adjusted for the change in monocyte count after hrG-CSF to gather information on the proangiogenic capacity of circulating PBMCs at each time point.

Data are expressed as mean ± SE. Normal distribution of the variables of interest was verified with the Kolmogorov-Smirnov test. Comparisons between the diabetic group and nondiabetic group were performed using Student t test for normal variables, Mann-Whitney U test for nonnormal variables, and the χ² test for categorical variables. Because CD34⁺ cell count is a normally distributed variable, the change in CD34⁺ cell count from baseline to 24 h after Filgrastim was assessed using paired Student t test. We then calculated the mean ± SE change of CD34⁺ cells in the diabetic and nondiabetic groups, which were compared using unpaired Student t test. Linear associations were assayed using the Pearson r correlation coefficient. To control for variables that were different between the two groups at P ≤ 0.10 and that may bias results, a multiple linear regression analysis was performed with change in progenitor cell levels as the dependent variable. Nonnormal dependent variables in secondary end point analyses were log transformed. Statistical significance was accepted at P < 0.05; SPSS version 17.0 was used.

A total of 24 DM patients (10 type 1 DM and 14 type 2 DM patients) and 14 control subjects without DM have been enrolled and treated. DM patients had a higher prevalence of hypertension and tended to be older than control subjects. Type 1 DM patients had significantly longer disease duration and lower prevalence of cardiovascular disease compared with type 2 DM patients (Table 1).

In subjects without DM, absolute CD34⁺ cell level significantly increased 2.2-fold after hrG-CSF, whereas CD34⁺ cells completely failed to mobilize in DM patients (mean ± SE change vs. baseline; non–DM: 3,475 ± 800 cells/mL; DM: 52 ± 332 cells/mL; P = 5 × 10⁻⁵) (Fig. 1A). Results were similar when type 1 or type 2 DM patients, considered separately, were compared with age-matched non–DM control subjects (Fig. 1B and C). Clinical characteristics of the subgroups are shown in Table 1.

After adjusting for potential confounders that were different between the two groups at P ≤ 0.10 (age, hypertension, and cardiovascular disease), DM remained significantly associated with reduced CD34⁺ cell increase after hrG-CSF (P = 0.002).

In response to hrG-CSF, DM was associated with impaired mobilization of all the other progenitor cell phenotypes, such as CD133⁺, CD34⁺CD133⁺, CD34⁺KDR⁺, CD133⁺KDR⁺, and CD34⁺CD133⁺KDR⁺ cells (Fig. 2). On correction for potential confounders, DM remained significantly associated with defective mobilization of CD133⁺ (P = 0.015), CD34⁺CD133⁺ (P = 0.011), and CD133⁺KDR⁺ (P = 0.013) cells, whereas the association between defective CD34⁺KDR⁺ mobilization and DM was marginally significant (P = 0.056) and blunted by age (P = 0.024).

Percentage expression of CD26/DPP-4 on CD34⁺ cells significantly increased after hrG-CSF in non–DM control subjects (δ = +14.1 ± 3.9%), consistent with previous findings in vitro (17). In DM, CD26/DPP-4 expression was elevated at baseline and tended to decline after hrG-CSF treatment (δ = −8.8 ± 5.5%; P = 0.013 vs. non–DM) (Fig. 3).

In the DM groups, progenitor cell mobilization was not significantly correlated to HbA_1c, disease duration, pattern of complications, or treatment regimen.

After treatment with hrG-CSF, white blood cell counts, neutrophil counts, and monocyte counts significantly increased in both groups, and there were no differences between patients with DM and control subjects without DMs (Table 2 and Supplementary Fig. 1). This suggests that DM affects immature, but not mature, cell mobilization.

Mononuclear cells collected from patients before and after hr-GSCF administration were embedded into Matrigel plugs and implanted in immunodeficient mice to assess the presence of proangiogenic cells. Baseline PBMC from non–DM subjects showed higher neovascularization capacity compared with DM patients, which was statistically significant versus type 1 DM. Plugs with non–DM PBMCs before hrG-CSF (baseline) showed vascular invasion at gross inspection and the presence of vascular structures containing erythrocytes at histology. Plugs implanted with type 1 DM PBMCs at baseline showed almost no vascularization, lower hemoglobin content, and no evidence of perfused vascular structures. Plugs containing baseline type 2 DM PBMCs showed nonsignificantly lower vascularization capacity compared with non–DM at baseline. After hrG-CSF, the proangiogenic capacity of patients’ cells significantly increased in non–DM, but not in DM groups, and post-hrG-CSF neovascularization capacity was lower in type 1 DM (P = 0.058) and type 2 DM (P = 0.045) versus non–DM (Fig. 4). Matrigel plug neovascularization capacity was significantly correlated with circulating CD34⁺ (r = 0.47; P = 0.003), CD34⁺KDR⁺ (r = 0.38; P = 0.024), CD133⁺KDR⁺ (r = 0.46; P = 0.005), and CD34⁺CD133⁺KDR⁺ (r = 0.41; P = 0.014) cell levels (Supplementary Fig. 2). These data suggest that hrG-CSF mobilizes functional proangiogenic cells in non–DM subjects but not in DM patients.

Treatment with hrG-CSF was safe and uneventful. Five subjects in the non–DM group (35.7%) and five patients in the DM group (20.8%; P = 0.53 DM vs. non–DM) reported mild back pain 12 to 18 h after Filgrastim injection, which resolved at 24 h and required analgesic therapy with acetaminophen in three non–DM cases.

We show that DM is associated with impaired stem and progenitor cell mobilization after direct BM stimulation, independently of potential confounders. Remarkably, this was true for both hematopoietic stem cells and EPCs despite baseline cell levels not being reduced in this DM cohort. Therefore, the present prospective trial substantiates the existence of a BM defect in human DM and suggests that the BM mobilization failure precedes reduction of circulating progenitors. Both type 1 DM and type 2 DM patients showed almost complete unresponsiveness to stem/progenitor cell mobilization, suggesting that this complication is independent of DM etiology; however, the role of autoimmunity in determining BM response may be worthy of investigation.

In mice, long-term DM causes BM microangiopathy and altered oxygen gradients (4) that, in addition to reduced expression of prosurvival genes (6), lead to a pauperization of the stem cell pool. Microvascular BM alterations in experimental DM include capillary rarefaction, increased permeability, endothelial cell apoptosis, and dysfunction (4), features that resemble diabetic microangiopathy of other organs, such as the kidney and retina. These histopathological aspects suggest that the BM is a hitherto unrecognized site of DM complication and is likely responsible for the mobilization failure. Data on the amount of BM stem cells in DM are discordant, with some studies showing normal (3,18) or even increased (7) primitive Sca-1⁺c-kit⁺Lin⁻ hematopoietic progenitors. Therefore, the low CD34⁺ cell count in BM aspirates from type 2 DM patients that we have shown previously (9) might reflect true stem cell deficiency or reduced accessibility of the niches to aspiration, attributable to the sticky property of the diabetic niche, which is more prone to stem cell retention than mobilization (7). It is noteworthy that DM did not impair mobilization of mature leukocytes, which are more loosely retained by the BM stroma than stem cells.

Normally, G-CSF stimulates expression and activity of CD26/DPP-4 and other proteases, with subsequent degradation of the chemokine and retention signal SDF-1α (17,19). Thus, stem/progenitor cells migrate to the peripheral circulation following SDF-1α gradients. Herein, we suggest a possible mechanism of stem cell unresponsiveness to G-CSF in DM by showing a maladaptive CD26/DPP-4 response. Systemic CD26/DPP-4 activity is increased in DM (11), and our new data indicate that G-CSF fails to upregulate CD26/DPP-4 on BM-derived cells in DM, likely preventing modification of the SDF-1α gradient. Studies show that DM mice mobilize stem cells after treatment with the SDF-1α receptor CXCR4 antagonists AMD3100 (7,18,20) and NIBR1816 (21). Therefore, the vascular niche containing stem cells that can be readily mobilized on disruption of the SDF-1α retention signal seems to be preserved, but responsiveness to CXCR4 antagonists should be confirmed in DM patients.

The existence of a BM mobilopathy in DM might be ascribed to both structural alterations affecting the stem cell niche (microangiopathy) and functional defects preventing the cells from being mobilized (e.g., the CD26/SDF-1α/CXCR4 axis). In addition, DM alters the activity of endothelial nitric oxide synthase (22), which is pivotal for EPC mobilization (23), and although G-CSF activity is mainly endothelial nitric oxide synthase–independent (24), this may be another mechanism accounting for depressed mobilization in DM. Among strategies to reverse BM dysfunction, our data showing no correlation between HbA_1c and mobilization suggest that glucose control might not be effective, whereas experimental data indicate that boosting the antioxidative defense is a suitable strategy to prevent BM alterations (4). Although CD26/DPP-4 inhibition increases EPCs (25), whether this represents a therapeutic target to restore BM responsiveness in DM needs to be ascertained.

We also found that DM impairs neovascularization by mononuclear cells mobilized by hrG-CSF, as shown by the in vivo Matrigel plug assay. It is noteworthy that type 1 DM patients already had defective neovascularization capacity at baseline. This is possibly related to the longer disease duration in type 1 DM compared with type 2 DM patients (Table 1). Both type 1 DM and type 2 DM patients were unable to increase significantly their PBMCs proangiogenic capacity after hr-GCSF, again suggesting that mobilization failure precedes reduction of proangiogenic cells. Although we did not identify the subpopulation of PBMCs involved (15,16), change in neovascularization capacity was correlated with stem/progenitor mobilization. These important results indicate that the stem cell mobilization failure can be pathophysiologically linked to impaired tissue repair and development of cardiovascular DM complications, which are characterized by defective angiogenesis (26,27).

Our findings have clinical implications. Reduced progenitor cell levels are powerful predictors of future cardiovascular events (28), and replenishment of progenitor cells may lower cardiovascular risk. Although the pathogenesis of diabetic vascular complications is complex and possibly diversified in type 1 DM and type 2 DM, it is remarkable that alterations in BM-derived cells have been consistently reported for both type 1 DM (29) and type 2 DM patients (30). Understanding the causes of progenitor cell reduction and the role of BM can identify novel strategies to reverse this defect and prevent vascular disease. Moreover, the immunomodulatory activity of mobilized BM-derived progenitors may be important in type 1 DM (21). In addition, clinicians should be aware of the likelihood of mobilization failure in DM patients undergoing stem cell collection for BM autotransplantation or for angiogenic cell therapy (7,8).

The study has limitations. First, replication in other, possibly larger, cohorts is needed and comparison between recent-onset and long-term type 1 DM patients would allow a description of the natural history of BM dysfunction. Second, whereas the imbalances in clinical characteristics between groups were adjusted by multivariate analyses, residual confounding may be present. Finally, whereas a single-dose hrG-CSF was used in this study, DM patients might respond to a full 5-day course of hrG-CSF, but it would not be ethical to perform maximal BM stimulation for research purposes only. Several DM patients have been treated with high-dose hrG-CSF in cell therapy protocols (31), but a formal comparison of the mobilization effect between DM and non–DM patients never has been performed. Nonetheless, our data represent a proof-of-concept of diabetic BM dysfunction in humans. As the BM emerges as a novel target organ in DM, intensive investigation to reverse this complication becomes compelling.

This study was supported by grants from the University of Padova (60A07-8452) and the European Foundation for the Study of Diabetes (EFSD)/Lilly Fellowship and EFSD/Sanofi Clinical Investigator Fellowship to G.P.F.

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

G.P.F. designed the study and wrote the manuscript. G.P.F., M.A., S.V.d.K., E.B., R.C., M.M., and N.P. researched data. C.A. designed the study and contributed to discussion. A.A. reviewed and edited the manuscript, contributed to discussion, and supervised the project. 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.