RETRACTED ARTICLE: MIR221/MIR222-driven post-transcriptional regulation of P27KIP1 and P57KIP2 is crucial for high-glucose- and AGE-mediated vascular cell damage
- 1.4k Downloads
MicroRNAs (miRNAs) are a novel group of small non-coding RNAs that regulate gene expression at the post-transcriptional level and act on their target mRNAs in a tissue- and cell-type-specific manner. Herein, the relevance of MIR221/MIR222 in high-glucose- and AGE-mediated vascular damage was investigated.
Functional studies were performed using human mature endothelial cells and endothelial progenitor cells subjected to high glucose or AGE. Quantitative real-time amplification was performed to analyse MIR221/MIR222 expression in these experimental conditions. Luciferase assay was used to identify MIR221/MIR222 targets. Functional studies were performed in vitro and in vivo in mice using gain- and loss-of-function approaches.
Using an in vivo mouse model we demonstrated that exposure to AGE and high glucose impaired vessel formation. Moreover, in vitro functional studies revealed that both high glucose and AGE inhibit cell-cycle progression by modulating the expression of P27KIP1 (also known as CDKN1B) and P57KIP2 (also known as CDKN1C), which encode cyclin-dependent kinase inhibitor 1B (p27, Kip1) (P27KIP1) and cyclin-dependent kinase inhibitor 1C (p57, Kip2) (P57KIP2), respectively. Crucial to AGE- and high-glucose-mediated cell-cycle arrest was the downregulation of MIR221/MIR222 expression. Luciferase assay showed that MIR221 and MIR222 specifically bind to the P27KIP1 and P57KIP2 mRNA 3′-untranslated regions, implicating P27KIP1 and P57KIP2 as MIR221/MIR222 targets. These results were confirmed by gain-of-function experiments in vitro, and by injecting mice with endothelial cells overexpressing MIR221 and MIR222.
We provide evidence that high-glucose- and AGE-induced inhibition of vascular cell proliferation is controlled by MIR221/MIR222-driven post-transcriptional regulation of P27KIP1 and P57KIP2. These data add further insight to the possible contribution of miRNAs in vascular damage mediated by a high-glucose environment.
KeywordsAGE Cell cycle Endothelial cells High glucose miRNAs Vascular biology
Bovine calf serum
Basic fibroblast growth factor
Cyclin-dependent kinase inhibitor
Endothelial progenitor cells
Kinase insert domain receptor
Cyclin-dependent kinase inhibitor 1B (p27, Kip1)
Cyclin-dependent kinase inhibitor 1 C (p57, Kip2)
Gap 0/1/2 (cell cycle)
Mitosis (phase cell cycle)
Peripheral blood mononuclear cells
Smooth muscle cells
- S phase
DNA synthesis phase (cell cycle)
Vascular endothelial growth factor
Risk factors for coronary artery disease may modify an individual’s capacity for angiogenesis. Specifically, diabetes has been shown to be associated with a significant impairment in adaptive vascular growth of both capillary-like tube vessels and collateral vessels [1, 2]. Increasing evidence indicates that high glucose and AGEs are the initiating causes of vascular damage in diabetes [3, 4], acting on both resident endothelial cells (ECs) and endothelial progenitor cells (EPCs) [5, 6]. Although multiple growth factors have been shown to regulate vascular growth, little is known about the complex upstream regulation of gene expression and translation in these settings. MicroRNAs (miRNAs) are an emerging class of highly conserved non-coding small RNAs that regulate gene expression at the post-transcriptional level by inhibiting the translation of protein from mRNA or by promoting degradation of mRNA. More than 500 human miRNAs have been identified so far, and increasing evidence indicates that miRNAs have distinct expression profiles and play crucial roles in various physiological and pathological processes such as cardiogenesis, haematopoietic lineage differentiation and oncogenesis [7, 8, 9, 10]. Meanwhile, a few specific miRNAs that regulate endothelial cell functions and angiogenesis have been described [7, 11]. miRNAs that regulate angiogenesis include MIR17-5p (also known as MIR17), MIR-17-92 (also known as MIR17HG), MIR21, MIR27A, MIR27B, MIR126, MIR130A, MIR210, MIR221, MIR222, MIR378 and the LET-7 family, MIR27B and MIR130A have been identified as pro-angiogenic miRNAs [7, 11]. In contrast, MIR221 and MIR222 inhibit endothelial cell migration and proliferation by targeting the stem cell factor receptor C-KIT  and, as recently shown, by targeting STAT5A .
MIR221 and MIR222 were originally described as potent regulators of cell-cycle progression via direct targeting of cyclin-dependent kinase inhibitor 1B (p27, Kip1) (P27KIP1) and cyclin-dependent kinase inhibitor 1C (p57, Kip2) (P57KIP2) in various human malignancies . P27KIP1 and P57KIP2, together with cyclin-dependent kinase inhibitor 1A (p21, Cip1) (P21CIP1), are members of the CIP/KIP family of cyclin-dependent kinases inhibitors (CKIs) that share homology in their N-terminal regions and affect the complexes of cyclin-dependent kinases (CDKs) 2, 4 and 6 with cyclin A, D and E . The CIP/KIP family proteins block the progression through all stages of the gap 1 (G1)/DNA synthesis (S) phase of the cell cycle, thereby functioning as a ‘brake of cell cycle’ . Consistently, in vascular smooth muscle cells (SMCs) P27KIP1 and P57KIP2 act as cell-cycle regulatory proteins under the control of MIR221/MIR222 activity , indicating that MIR221 and MIR222, in addition to controlling tumour progression, might also regulate vascular cell biology.
Recently, an miRNA signature in insulin target tissues has been reported , and a plasma miRNA profile in type 2 diabetes revealed loss of endothelial MIR126 . MIR126 plays a pivotal role in maintaining endothelial cell homeostasis and vascular integrity by facilitating vascular endothelial growth factor (VEGF) signalling , thus suggesting that aberrant regulation of miRNAs might actually be crucial in dictating anti-angiogenic signals in the diabetic setting.
The present study aimed to investigate the contribution of MIR221/MIR222 to high-glucose- and AGE-driven vascular damage.
Reagents and antibodies, RNA isolation and quantitative real-time PCR for miRNAs or P27KIP1 (also known as CDKN1B) and P57KIP2 (also known as CDKN1C) expression, western blot analysis and immunofluorescence analysis are described in detail in the electronic supplementary material (ESM).
Isolation and culture of ECs
ECs were isolated from human umbilical vein within 4 h of delivery by trypsin treatment (0.1% [wt/vol.]), cultured in M199 with the addition of 20% (vol./vol.) bovine calf serum (BCS) and 5 ng/ml of basic fibroblast growth factor (bFGF) and used at early passage (II–III). Throughout the study ECs were cultured for 2 days in normal medium (5 mmol/l d-glucose) plus 10% (vol./vol.) BCS and bFGF (5 ng/ml) alone or in combination with 400 μg/ml AGE, 25 mmol/l d-glucose or 19 mmol/l d-mannitol (used as osmotic control). In selected experiments, ECs exposed to normal medium, AGE or high glucose were transfected for 48 h with pre-miRNA-negative control, pre-MIR221 or pre-MIR222 precursor oligonucleotides or, alternatively, with anti-miRNA-negative control, anti-MIR221 or anti-MIR222 antagonists (Applied Biosystems, Foster City, CA, USA), according to the manufacturer’s instructions.
Isolation, characterisation and culture of EPCs from peripheral blood mononuclear cells
The following method was used to isolate EPCs. Peripheral blood mononuclear cells (PBMNCs) were obtained by Ficoll Histopaque 1077 (Sigma-Aldrich, St Louis, MO, USA) and plated onto collagen-1-coated dishes for 21 days in EGM-2 medium (Cambrex, Walkersville, MD, USA), as described by Yoder et al. . FACS analysis was used to characterise EPC surface markers (anti-CD45, anti-CD14, anti-CD34, anti-CD31, anti-kinase insert domain receptor (KDR) and anti-CD146) at day 0 (2 days after isolation), when non-adherent cells were removed and at day 23, at the end of the experiments (21 days of EGM-2 culture plus 2 days with the stimuli: 400 μg/ml AGE, 25 mmol/l d-glucose glucose (‘high glucose’) or 19 mmol/l d-mannitol (‘high mannitol’)). The control conditions were 5 mmol/l d-glucose (‘normal’). Approval was obtained both from Servizio Immunoematologia e Medicina Trasfusionale and from the Institutional Review Board of S. Giovanni Battista Hospital, Turin, Italy. Informed consent was provided according to the Declaration of Helsinki. In selected experiments, EPCs exposed to normal medium, AGE or high glucose were transfected for 48 h with pre-miRNA-negative control, pre-MIR221 or pre-MIR222 precursor oligonucleotides or, alternatively, with anti-miRNA-negative control, anti-MIR221 or anti-MIR222 antagonists (Applied Biosystems), according to the manufacturer’s instructions.
To analyse cell-cycle progression, ECs and EPCs treated for 2 days as indicated were processed by FACS analysis, as previously described by Defilippi et al. . Briefly, after treatment, the cells were fixed with 70% (vol./vol.) ethanol and DNA was stained with propidium iodide (Sigma-Aldrich) and analysed with a flow cytometer (FACScan, Becton Dickinson, San Jose, CA, USA). The percentage of cells in each phase of the cell cycle was determined by ModFit LT software (Verity Software House, Topsham, ME, USA). The percentage of the cells in the DNA duplicating phase (S phase) was reported. Cell-cycle analysis by FACS was also performed on: (1) ECs treated with different AGE concentration (from 50 to 1,200 μg/ml); and (2) ECs and EPCs transfected for 48 h with pre-miRNA-negative control, pre-MIR221 or pre-MIR222 precursor oligonucleotides or, alternatively, with anti-miRNA-negative control, anti-MIR221 or anti-MIR222 antagonists in normal conditions or in the presence of AGE, high glucose or high mannitol.
In vitro endothelial cell migration assay
Analysis of chemotaxis of ECs was performed as previously described by Brizzi et al. and Dejana et al. [23, 24]. Briefly, assessment of EC migration was performed in Boyden’s chambers by counting the cells that passed across the filter (8 μm pore size) after addition in the lower compartment of the chamber of the vehicle alone (free medium with 0.25% [wt/vol.] BSA), high glucose (25 mmol/l), high mannitol (19 mmol/l), or AGE (400 μg/ml), in the presence of VEGF (20 ng/ml). Cell counting was performed by three different operators on 10 fields, ×20 magnification, of three individual experiments (n = 9).
Luciferase miRNA target reporter assay
The luciferase reporter assay was performed using a construct generated by subcloning the PCR products amplified from full-length 3′-untranslated region (UTR) of P27KIP1 and 3′-UTR of P57KIP2 DNA in the SacI restriction site of the luciferase reporter vector plasmid miR (pmiR) (Ambion, Applied Biosystems). The PCR products were obtained using the following primers: P27KIP1 sense 5′-AGAGCTCCAGATACATCACTGC-3′, antisense 5′-TGAGCTCTATACTTGGCTCAG-3′; P57KIP2 sense 5′-TTGAGCTCCCCTTCTTCTCGCTGTCCTCT-3′, antisense 5′-AAGAGCTCCTCTTTGGGCTCTAAATTGGC-3′.
The insert identities were verified by sequencing. The pmiR, pmiR-3′-UTR-P27KIP1 and pmiR-3′-UTR-P57KIP2 reporter vectors were transiently co-transfected in ECs and EPCs, cultured in normal medium alone or in combination with 400 μg/ml AGE or 25 mmol/l d-glucose or 19 mmol/l d-mannitol, at 30:1 molar ratio with the pRL vector, coding for the Renilla sp. luciferase, used as the internal control of the luciferase assay. Luciferase activities were analysed 48 h later by Dual-Luciferase Report Assay System (Promega, Madison, WI, USA), according to the vendor’s instructions, using a TD20/20 double injector luminometer (Turner Designs, Forlì, Italy). The results were expressed as relative luciferase activity (%), calculated by normalising the ratio of the firefly/Renilla sp. luminescences. Luciferase activities, using the pmiR reporter vectors, described above, were also evaluated in ECs or in EPCs, transfected 24 h previously with pre-miRNA negative control, pre-MIR221 or pre-MIR222 precursor oligonucleotides.
In vivo experiments
For the murine angiogenesis assay FVB mice (five mice, 8 weeks old, for each experimental group) (Charles River Laboratories International, Wilmington, MA, USA) were injected s. c. with Matrigel containing VEGF (50 ng/ml) [25, 26], high glucose (25 mmol/l), and AGE (400 μg/ml) alone or in combination as indicated. The negative control was NaCl solution (154 mmol/l). After 7 days, the FVB mice were killed and Matrigel plugs were processed for histological analysis with haematoxylin–eosin staining. In selected experiments for the angiogenesis assay, SCID mice (five mice, 7 weeks old, for each experimental group) (Charles River Laboratories) were injected s.c. with growth-factor-reduced Matrigel containing VEGF, AGE and 2 × 106 ECs, previously transfected with pre-miRNA negative control, pre-MIR221 or pre-MIR222 precursors and processed as described by Zeoli et al. .
Briefly, 4 days after injection Matrigel plugs were recovered and fixed in 10% (vol./vol.) buffered formalin and embedded in paraffin for histological and immunofluorescence analyses or digested for EC isolation. The vessel area and the total Matrigel area were planimetrically assessed from haematoxylin–eosin-stained sections as previously described by Zeoli et al. . Only the structures possessing a patent lumen and containing erythrocytes were considered vessels. Angiogenesis was expressed as the percentage ± SD of the vessel area relative to the total Matrigel area (% vessel area, ×10 magnification). Quantification of neo-formed vessels was also evaluated by CD31 staining of vascular ECs. Any stained EC or EC cluster, clearly separated from connective tissue elements, was considered as a single microvessel and counted according to Weidner et al. . Animal procedures conformed to the Guide for Care and Use of Laboratory Resources (National Institutes of Health publication no. 93–23, revised 1985) .
Isolation of ECs from Matrigel plugs
ECs were recovered from Matrigel plugs 4 days after injection into SCID mice . After digestion in Hank’s buffered salt solution containing 0.1% (vol./vol.) collagenase I for 30 min at 37°C, the cells were washed in medium plus 10% (vol./vol.) BCS and forced through a graded series of meshes to separate the cell component from Matrigel matrix. ECs were isolated via anti-human CD31 antibody coupled to magnetic beads, by magnetic cell sorting using the MACS system (Miltenyi Biotec, Auburn, CA, USA). Briefly, cells were labelled with the anti-human CD31 antibody for 20 min and then were washed twice and re-suspended in MACS buffer (PBS without Ca2+ and Mg2+, supplemented with 1% [vol./vol.] BSA and 5 mmol/l EDTA) at the concentration of 0.5 × 106 cells/80 μl. After washing, cells were separated on a magnetic stainless steel Wool column (Miltenyi Biotec) according to the manufacturer’s recommendation. The endothelial phenotype was verified by FACS analysis using an anti-human von Willebrand antibody (Sigma-Aldrich). The recovered cells were subjected to RNA isolation to detect MIR221 and MIR222 expression by quantitative real-time PCR or lysed for western blot analysis.
All in vitro and in vivo results are representative of at least three independent experiments, performed in triplicate. Densitometric analysis using a BioRad GS 250 molecular imager was used to calculate the differences in the fold induction of protein content, reported as ‘densitometric value’ and each western blot panel and relative densitometric histogram shown in the figures was representative of the results obtained in triplicate. The significance of differences between experimental and control values for both in vitro and in vivo experiments was calculated using analysis of variance with Newman–Keuls multi-comparison test and reported in detail in each figure legend (n = 9).
High glucose and AGE impair EC and EPC proliferation and neo-vessel formation in mice
Increasing evidence indicates that the vascular damage occurring in diabetes is the result of repeated acute changes in cellular glucose metabolism, or long-term accumulation of AGE [3, 4]. Thus, the effects of high glucose and AGE on EC functional activities were investigated. First, an AGE dose–response curve was constructed. The percentage of ECs in the different cell-cycle phases was evaluated and reported in ESM Fig. 1. As we failed to detect significant differences from 50 to 1,200 μg/ml of AGE, 400 μg/ml of AGE was used throughout the study.
Vascular remodelling does not exclusively rely on proliferation of resident ECs, but also involves EPCs . Thus, EPCs were isolated and characterised according to Yoder et al.  (ESM Figs 2–5 and ESM Table 1) and cultured in the presence of high glucose and AGE. Again, we found that cell-cycle progression was prevented when EPCs were exposed to high glucose and AGE (Fig. 1b), suggesting that exposure to AGE or high glucose might also hinder in vivo vessel formation. To investigate the in vivo effects of AGE and high glucose, a mouse model of angiogenesis was used. Data reported in Fig. 1d,e demonstrate that vessel formation in response to a well-known angiogenic stimulus, VEGF [25, 26], was prevented in the presence of AGE and high glucose.
MIR221 and MIR-222 targeting P27KIP1 and P57KIP2 are crucial for high-glucose- and AGE-mediated inhibition of cell-cycle progression in vascular cells
Gain- and loss-of function approaches identify MIR221 and MIR222 as crucial regulators of high-glucose- and AGE-induced inhibition of cell-cycle progression
High-glucose and AGE-mediated inhibition of vessel formation in vivo in mice is controlled by MIR221 and MIR222
The present study demonstrates: (1) high glucose and AGE induce inhibition of vascular cell progression in the cell cycle; (2) this effect is recapitulated in vivo in a mouse model of angiogenesis; and (3) under high glucose and AGE stress MIR221- and MIR222-driven post-transcriptional regulation of P27KIP1 and P57KIP2 controls cell-cycle progression of vascular cells both in vitro and in vivo in mice.
Endothelial injury is thought to represent a crucial step in the initiation and progression of vascular diseases in the diabetic setting. Acute changes in cellular glucose metabolism or accumulation of AGE, in addition to accelerating endothelial cell damage, also abate endothelial repair mechanisms. Vascular repair processes not only rely on resident endothelial cells but also on circulating EPCs . Compelling evidence indicates that changes in EPC number and functional activities are closely associated with cardiovascular risk factor profiles [31, 32], impacting on their delivery to sites of ischaemia, where angiogenesis might be required. Studies addressed to evaluate the functional role of EPCs in supporting neovascularisation have often provided controversial evidence. In particular, the use of different assays for EPC expansion and the reliance on cell-surface markers and morphology in the absence of functional assays have contributed to this confusion . Currently, EPC designation is based on EPC functional features and, in particular, on their in vivo vasculogenic capability [34, 35]. In the present study, by using EPCs previously shown to have vasculogenic capability when injected in collagen matrix in mice [36, 37], we were able to demonstrate that cell-cycle progression of EPCs was blocked under high-glucose and AGE stress. A similar effect was observed when ECs were exposed to high glucose and AGE, suggesting that high glucose concentrations or accumulation of AGE might exert a damaging effect in vivo. Indeed, AGE and high glucose were both able to prevent recruitment, proliferation and formation of vascular structures when injected in SCID mice together with a potent angiogenic stimulus such as VEGF [25, 26]. Progression into the cell cycle is a pre-requisite for cell proliferation. The cell cycle is controlled by a series of regulatory molecules known as cyclins, CDKs and CKIs. P27KIP1 and P57KIP2 belong to the CIP/KIP family of CKIs [15, 16] and act as a ‘brake of cell cycle’ as they exert a stringent control on all stages of the G1/S phase [15, 16]. The increased expression of P27KIP1 and P57KIP2 in response to high glucose and AGE stress supports the results of our functional studies.
Improvement in angiogenesis after critical ischaemia is of considerable interest as a therapeutic strategy. However, little is known about the complex regulation of gene expression during neovascularisation and vascular remodelling. miRNAs are a recently recognised class of highly conserved non-coding short RNA molecules that are considered fine-scale rheostats of protein-coding gene product abundance [7, 38]. The relative importance and mode of action of miRNAs in human complex diseases remain to be fully elucidated. Recently, miRNAs have been shown to be directly involved in cardiovascular diseases . Moreover, in diabetes, miRNAs have been implicated in the epigenetic regulation of key metabolic, inflammatory, and anti-angiogenic pathways . At this regard, a recent plasma miRNA signature in diabetic patients demonstrates loss of the vascular regulatory miRNA, MIR126 . Herein, we demonstrate that two additional angiogenic regulatory miRNAs, MIR221 and MIR222, are downregulated in response to high glucose and AGE stress. It has been proposed that MIR221 and MIR222 are oncogenic, based on their upregulation in tumour cells and on their suppressive effects on the production of the CKI protein P27KIP1 [40, 41, 42]. However, MIR221/MIR222 were subsequently recognised as being critical for the proliferation of SMC via their target gene products, P27KIP1 and P57KIP2 . We herein provide evidence that high-glucose- and AGE-mediated inhibition of cell-cycle progression as well as neo-vessel formation in mice are controlled by MIR221/MIR222-driven post-transcriptional regulation of P27KIP1 and P57KIP2. These results, together with data demonstrating that loss of pro-angiogenic MIR126 in endothelial apoptotic bodies correlates with plasma glucose concentrations , reveal additional molecular details to explain the defective adaptive vascular growth associated with diabetes.
One important feature of miRNA is tissue- and cell-specific expression pattern. However, it is becoming even more evident that the cell microenvironment dictates miRNA target specificity. This is particularly true in vascular cell biology, in which discrete targets, such as STAT5A  or, as reported herein, P27KIP1 and P57KIP2, are post-transcriptionally regulated by MIR221/MIR222 when vascular cells are exposed to an inflammatory microenvironment  or to high glucose and AGE stress, respectively.
Clinical studies have demonstrated that the level of the circulating AGE may be linked to various diabetes complications . However, until recently the sophisticated and expensive laboratory techniques required for measurement of specific AGEs have retarded any attempts at widespread use of such measurements in the clinic. Moreover, as AGEs are structurally heterogeneous, it remains unclear which circulating AGE should be measured. Similarly, AGE accumulates in tissues , and thus the level of AGE able to promote cell injuries remains to be established. For these reasons, to date, in vitro studies have been performed using concentrations ranging from 0.2 to 2 mg/ml [44, 45, 46]. The AGE concentration used throughout this study was chosen on the basis of a dose–response curve, which allowed us to define the boundary between harmless and damaging AGE concentrations in our experimental models. We are aware of the intrinsic limitations of in vitro studies; however, the validity of the results obtained with the selected AGE dosage is supported by the findings that inhibition of cell-cycle progression and downregulation of MIR221 and MIR222 expression also occurred when vascular cells were challenged with high glucose.
We are confident that our in vitro study presents a number of limitations mainly through the absence of a specific stromal microenvironment that, in vivo, may contribute to the response to AGE stress. In addition, ECs or EPCs used in vitro can be only considered surrogates of in vivo vascular cells located in a vessel microenvironment containing regulatory molecules that again can influence the response to AGE stress. Finally, in vivo studies in mice present, per se, a limitation as they may not reproduce the complexity of human model and can only provide speculative conclusions. Additional concerns may also derive from in vitro and in vivo mouse models in which natural molecules are artificially produced in abundance. Nonetheless, in vitro studies and mouse models are currently recognised as valuable tools as they may provide rational bases to set methods appropriate for human studies that, in this particular setting, may be able to assess the real impact of AGE in driving vascular damage.
Emerging evidence suggests that miRNAs play significant roles in insulin production, action and secretion and also in diverse aspects of glucose and lipid metabolism [18, 19]. Most importantly, microarray studies have highlighted an altered miRNA profile in insulin target tissues in diabetic models  and patients . The results reported herein provide evidence that deregulation of MIR221 and MIR222 expression, together with loss of MIR126 , might dictate and sustain high-glucose-driven anti-angiogenic signals. To boot, the finding that neo-vessel formation in response to high glucose and AGE stress is under the control of miRNA expression  identifies miRNAs as potential targets for pharmacological intervention to ameliorate vascular dysfunction in pathological settings, such as those associated with altered glucose metabolism.
This work was supported by grants of the Italian Association for Cancer Research, Ricerca Finalizzata Regione Piemonte and Ministero dell’Università e Ricerca Scientifica to M. F. Brizzi. The authors have no conflicting financial interests.
Duality of interest
The authors declare no duality of interest associated with this manuscript.
- 29.National Research Council (1996) Guide of care and use of laboratory animals. National Academy Press, Washington, NIH publication no. 93–23Google Scholar
- 33.Yoder MC, Ingram DA (2009) The definition of EPCs and other bone marrow cells contributing to neoangiogenesis and tumor growth: is there common ground for understanding the roles of numerous marrow-derived cells in the neoangiogenic process? Biochim Biophys Acta 1796:50–54PubMedPubMedCentralGoogle Scholar
- 45.Rodiño-Janeiro BK, González-Peteiro M, Ucieda-Somoza R, González-Juanatey JR, Alvarez E (2010) Glycated albumin, a precursor of advanced glycation end-products, up-regulates NADPH oxidase and enhances oxidative stress in human endothelial cells: molecular correlate of diabetic vasculopathy. Diab Metab Res Rev 26:550–558CrossRefGoogle Scholar