Inflammation, glucose, and vascular cell damage: the role of the pentose phosphate pathway

The Ethics Committees for Clinical Research from Hospital Universitario de Getafe (reference numbers 11-20 and 14-117) and Hospital Universitario La Paz (reference numbers PI-1111 and PI-1878) approved the protocol in human cell cultures. The studies on animals were performed according to national and European guidelines (2010/63/EU), approved by the ethics committee of Universidad Autónoma de Madrid (CEI 27-670), and developed in registered animal facilities (ES-28079-000097).

Human aortic smooth muscle cells (HASMC) were obtained by enzymatic dissociation from the aortae of five organ donors, who had no evidences for cardiometabolic diseases or risk factors, and according to Spanish legal regulations as previously described [6]. The cells were routinely cultured in Dulbecco’s Modified Eagle Medium (DMEM, Biological Industries, Beit-Hamek, Israel) supplemented with 10 % fetal calf serum (FCS, Biological Industries), 100 µg/ml streptomycin (Reig-Jofré, Barcelona, Spain), 100 U/ml penicillin (ERN SA, Barcelona, Spain) and 2.5 µg/ml amphotericin B (Sigma Chemical Co., St. Louis, MO, USA). At confluence, the cells were passaged using a 0.02 % EDTA-0.05 % trypsin solution (Biological Industries). The cells were counted using a hemocytometer. Cell characterization was carried out based on both cell morphology and indirect immunofluorescence staining of α-smooth muscle actin. For experiments, confluent cells were serum-deprived for 24 h prior to the addition of the different test compounds in serum-free DMEM supplemented with 0.1 % bovine serum albumin (BSA, Sigma). Confluent cultures between passages 3 and 12 were used.

The cells were plated onto 48-well plates at a density of 20,000 cells per well. After serum-deprivation the cells were incubated for 24 h in DMEM with 0.1 % BSA and different initial concentrations of glucose (5.5, 11, or 22 mmol/L); in some experiments, the cultures were also treated with 1–10 ng/mL interleukin-1β (IL1β; Peprotech, London, UK) or 0.5 mmo/L sodium azide (Sigma). Glucose consumption and lactate release over 8 or 24 h of incubation were determined by measuring the respective levels in the culture medium at time 0 and after 8 or 24 h using commercial kits (Glucose HK Assay kit, Sigma; Lactate Reagent kit, Trinity Biotech, Wicklow, Ireland). Glucose consumption was calculated as the difference between initial and final glucose levels, while lactate production was calculated as the difference between final and initial lactate levels. Results were normalized by cell number. Both glucose consumption and lactate production were expressed as pmol per cell over 8 or 24 h.

Cytochalasin B-sensitive ³H-deoxyglucose uptake was used as an indicator of the cellular GLUT1 transporter capacity. Briefly, cells were incubated for 18 h in 24-well plates containing DMEM with 0.1 % BSA and 5.5 or 22 mmol/L glucose; in some experiments, the cells were also treated with 1–10 ng/mL IL1β 1 ng/mL anakinra (Biovitrum, SOBI, Stockholm, Sweden), or 0.5 mmol/L sodium azide. The 18 h optimal period for such treatments was established in preliminary experiments (data not shown). After washing in a Krebs–Ringer solution free of glucose (KRS; composition in mmol/L, NaCl 136, KCl 4.7, MgSO₄ 1.25, CaCl₂ 1.25, pH 7.4 HEPES 10), the cultures were pre-incubated for 15 min in KRS without d-glucose, in the presence or absence of the GLUT1 transporter blocker cytochalasin B (Sigma; 20 µmol/L). The cultures were then incubated for 5 min in KRS without cold d-glucose containing 1 µCi/mL ³H-deoxyglucose (activity 8.00 Ci mmol; Amersham, Buckinghamshire, UK), in the absence or presence of cytochalasin B. The optimal 5 min time for such incubation period was established in preliminary experiments (data not shown). After washing with KRS, the protein content was determined by the bicinchoninic acid (BCA) assay and the radioactivity incorporated to the cells was measured in a liquid scintillation counter (Tri-Carb 2800 TR, Perkin Elmer, Waltham, MA, USA). Results were expressed in dpm/µg protein.

A pCB6 plasmid containing the human GLUT1 cDNA sequence was a kind gift from Prof. David James (Garvan Institute of Medical Research, Darlinghurst Sydney NSW, Australia). The GLUT1 cDNA was subcloned into a lentiviral vector (pCDH-CMV-MCS-EF1-Puro, System Biosciences SBI) using the XbaI restriction site. All plasmids were confirmed by PCR screening and DNA sequencing. Viral particles were generated according to the manufacturer’s protocol (pPACKH1 Lentivector Packaging Kit, SBI, Mountain View, CA, USA). The cells were infected with viral particles (GLUT1+) in the presence of 4 µg/mL polybrene (Sigma) and selected in 1 µg/mL puromycin (Sigma) for 3 days. Matched cells subjected to infection with empty vector lentiviral plasmids (EV) were used as controls.

For GLUT1 mRNA quantification, total RNA was extracted from vascular cells after 18 h of treatment using a commercial kit (RNeasy, Qiagen; Crawley, UK) and quantified by absorbance at 260 nm. Aliquots of 1 µg RNA were reverse-transcribed using Moloney murine leukemia virus-reverse transcriptase (Sigma). The resulting cDNA was amplified using a human GLUT1 gene-specific relative RT-PCR kit (Ambion, Austin, TX, USA) containing an 18S internal standard, according to the manufacturer’s instructions. The reaction was conducted in a Peltier PTC100 thermocycler (M&J Research, Waltham, MA, USA) with an initial denaturation step at 95 °C for 3 min, followed by 23 cycles each consisting of incubation at 94 °C for 30 s, 59 °C for 30 s, and 72 °C for 30 s. Aliquots of the resulting PCR were loaded on 2 % agarose gels containing ethidium bromide, the resulting bands were visualized under ultraviolet light and quantified using NIH Image free software.

As previously described [6, 7], cells grown on glass coverslips were exposed for 18 h to the different treatments, fixed with acetone at −20 °C and blocked in a PBS solution containing 1 % BSA and 0.1 % Triton X-100. The cells were then incubated with a primary polyclonal antibody against GLUT1 (dilution 1/50; Neomarkers, Fremont, CA, USA), followed by incubation with an appropriate Alexa Fluor 488-secondary antibody (dilution 1/200; Molecular Probes-Invitrogen Corporation). Nuclei were counterstained with 4′,6-diamino-2-phenylindole (DAPI). Cells were observed with an Eclipse TE300 epifluorescence microscope (Nikon, Tokyo, Japan) and a spectral confocal microscope (Leica TCSSP5-AOBS, Leica microsystems, Heidelberg, GMBH, Germany). Confocal images were analyzed with LAS AF software, version 1.5.1 Build 869 (Leica).

After washing with PBS, the cells were subjected to extraction in lysis buffer containing 10 mmol/L Tris pH 7.4, 1 % sodium dodecyl sulfate (SDS), 10 mmol/L sodium orthovanadate, 2 mmol/L phenylmethylsulfonyl fluoride and 12.5 µg/mL aprotinin (all from Sigma). Proteins (20 µg per lane) were equally loaded and separated on 8 % SDS-PAGE and transferred onto a nitrocellulose membrane (BioRad Laboratories, Madrid, Spain). After blocking overnight, the membrane was incubated for 1 h at room temperature with polyclonal antibodies against iNOS (dilution 1/10,000; Transduction Laboratories, Lexington, KY, USA) or glucose-6-phosphate dehydrogenase (G6PD, dilution 1/5000, Sigma), followed by incubation for 1 h with a horseradish peroxidase-conjugated secondary antibody (dilution 1/10,000; Chemicon, Temecula, CA, USA), as described previously [6]. Immunoreactive bands were detected using an enhanced chemiluminescence detection kit (GE Healthcare, Uppsala, Sweden) and quantified by densitometry using NIH Image software. The membranes were stripped and reprobed with an anti α-tubulin primary antibody (dilution 1/10,000; Sigma) to ensure equal loading.

After 18 h of treatment, nuclear extracts from cell cultures were prepared as previously described [6, 7]. For binding reactions, nuclear extracts (5 μg) were incubated on ice for 15 min in a reaction buffer (pH 7.0 40 mmol/L HEPES, 140 mmol/L NaCl, 5 mmol/L dithiothreitol, 10 μg/mL BSA, 0.01 % Nonidet P-40, 4 % Ficoll and 0.05 μg/mL poly [dI-dC]poly [dI-dC]). A commercial oligonucleotide (Promega, Madison, WI, USA) encoding the NF-κB consensus sequence (5′-AGTTGAGGGGACTTTCCCAGGC-3′) and 5′ end-labelled with ©-³²P (~50,000 cpm) was then added, and the reaction mix incubated for 20 min at room temperature. DNA–protein complexes were resolved on 4 % nondenaturing polyacrylamide gels in 0.5 × TBE (45 mmol/L Tris–borate, 1 mmol/L EDTA, pH 8.0) at 4 °C. Gels were dried and exposed for autoradiography at −80 °C.

Carbon flux from glucose through the PPP was measured in cells treated for 120 min, as previously described [8], with minor modifications. Briefly, suspensions of known amounts of cells (3–5 × 10⁵ cells) obtained by smooth detaching from the cultures were incubated in sealed vials containing a central well with KOH (4 mol), which was used for ¹⁴CO₂ trapping, in the presence of 0.5 μCi of D-[1-¹⁴C]glucose (activity 56 mCi/mmol/L; Perkin Elmer) or D-[6-¹⁴C]glucose (activity 60 mCI/mmol/L; Perkin Elmer) in a Krebs–Ringer phosphate buffer (5.7 mmol/L Na₂HPO₄, 145 mmol/L NaCl, 4.86 mmol/L KCl, 1.22 mmol/L MgSO₄, 0.54 mmol/L CaCl₂; pH 7.4) containing 5.5 or 22 mmol/L glucose at 37 °C. In order to ensure an adequate O₂ supply for oxidative metabolism by the cells throughout the 120 min incubation period, the gas phase in the vials containing the cells was supplied with extra O₂ before the vials were sealed. The PPP flux was measured by assessing the difference between ¹⁴CO₂ production from [1-¹⁴C]glucose—which decarboxylates via the 6-phosphogluconate dehydrogenase-catalyzed reaction and the tricarboxylic acid cycle—and that of [6-¹⁴C]glucose—which decarboxylates only via the tricarboxylic acid cycle [8, 9].

NADPH oxidase activity was measured by lucigenin-derived chemiluminescence, as described previously [10]. After the various treatments described, the cells were washed in ice-cold PBS, scraped and centrifuged at 13,000 rpm for 1 min at 4 °C. The resulting cell pellet was homogenized in lysis buffer (pH 7.0) containing 50 mmol/L KH₂PO₄, 1 mmol/L EGTA and 150 mmol/L sucrose for 5 min at 4º C. The activity of NADPH oxidase was measured by lucigenin-derived chemiluminescence. For oxidase assay, cells samples (5 µg protein) were incubated in PBS containing 5 µmol/L lucigenin and 100 µmol/L NADPH. Luminiscence was measured every 30 s for 5 min in a tube luminometer (Optocomp, MGM Instruments, Hamden, CT, USA). The enzymatic activity was obtained as relative light units (RLU)/µg protein min⁻¹ determined by the BCA method.

After 24 h of serum deprivation, cells were incubated for a further 18 h period, with the different treatments in serum-free medium. After the treatment described, the cells were washed in ice-cold PBS, scraped and centrifuged at 1200 rpm for 8 min at 4 °C. Total glutathione and GSH were determined in the resulting pellet using a commercial kit (Glutathione Colorimetric Detection kit, Arbor Assays, Ann Arbor, MI, USA).

G6PD silencing was achieved by a procedure previously described [11]. For silencing, the On-TARGETplus SMARTpool for human G6PD siRNA (Dharmacon, Uppsala, Sweden) was used, according to manufacturer’s instructions. The On-Targetplus Non-targeting siRNA #1 (Dharmacon) was used as a negative control. Afterwards, cells were serum-deprived for 24 h and then submitted to the appropriate treatments for Western blot or NADPH oxidase activity determination, as described above.

Isolated mesenteric microvessels from 16-week male old Sprague–Dawley (SD) rats (300–350 g) were studied ex vivo. The animals were briefly exposed to a chamber filled with carbon dioxide until they fell unconscious and then immediately killed by cervical dislocation. The mesentery was removed, and placed in a Petri dish containing Krebs-Henseleit solution (KHS) at 4 °C. The third branch mesenteric arteries were dissected (mean internal diameter ranged between 200 and 400 µm; non-significant differences were observed among the different groups of rats). The arteries were dissected and cleaned free of fat and connective tissue under a light microscope and mounted as ring preparations on a small vessel myograph [12] to measure isometric tension. Arteries were bathed in KHS at 37 °C continuously bubbled with a 95 % O₂-5 % CO₂ mixture, which yields a pH of 7.4 and their passive tension and internal circumference were determined. The arteries were subjected to optimal tension (90 % of the tension equivalent to a intramural pressure of 100 mm Hg. After 30 min of equilibration, the vessels were exposed to 125 mmol/L KCL (KKHS, equimolar substitution of KCl for NaCl in KHS) for 2 min in order to check their functional integrity. Segments failing to achieve a maximum active tension equivalent to a pressure of 100 mmHg were rejected [12].

The bath was then washed three times with KHS and a further period of 180 min washout period was allowed with KHS containing 5.5 or 22 mmol/L glucose before the arteries were contracted with the concentration of NA (1 µmol/L) required to produce approximately 80 % of the maximum response to KKHS. Relaxations to acetylcholine (ACh) were subsequently assessed by adding cumulative concentrations of the drug at 2 min intervals (final bath concentrations 0.1 nmol/L–10 µmol/L). Some microvessels were pre-incubated with 2.5 ng/mL IL1β for 120 min before and during the administration of NA and ACh. Other vascular segments were treated with 2 mmol/L 6-aminonicotinamide or 10 µmol/L apocynin for 30 min in advance and during the administration of IL1β, NA, and ACh. The composition of KHS (mmol/L) was NaCl 115, CaCl₂ 2.5, KCl 4.6, KH₂PO₄ 1.2, MgSO₄.7H₂O 1.2, NaHCO₃ 25, glucose 5.5 or 22, and Na₂EDTA 0.03.

Culture plasticware was obtained from TPP (Tragadingen, Switzerland). All the reagents and drugs were purchased from Sigma unless otherwise stated.

At a physiological concentration of 5.5 mmol/L glucose, the cells consumed 2.1 ± 0.3 and 5.3 ± 0.3 pmol per cell during 8 and 24 h, respectively, exhibiting a predominantly glycolytic metabolism, as shown by the lactate: glucose ratio (Table 1a). Increasing the extracellular glucose concentration to 11 or 22 mmol/L affected neither the glucose consumption nor the lactate: glucose ratio (Additional file 1: Figure S1 in the online-only data supplement and Table 1a). Furthermore, it did not increase the total glucose transport capacity (Fig. 1a) nor the expression of GLUT1, the predominant glucose transporter described in vascular cells [13] (Fig. 1b and Additional file 1: Figure S2).

When cells were activated with IL1β (1–10 ng/mL), however, glucose consumption was enhanced in a concentration-dependent manner (Fig. 1c). Moreover, at each concentration of IL1β, the consumption of glucose was proportional to its extracellular concentration (Fig. 1c), although the lactate: glucose ratio was not significantly modified (Table 1a). The increased consumption of glucose caused by IL1β was related to increased GLUT1 mRNA (by 70.4 ± 16.3 %; P < 0.05; n = 3) and a higher expression of GLUT1 (Fig. 1b and Additional file 1: Figure S2) that enhanced glucose transport capacity (Fig. 1d). The IL-1 receptor antagonist anakinra (1 µg/mL) abolished the ability of IL1β to enhance the transport capacity of glucose (³H-deoxyglucose uptake was 171.6 ± 19.3, 493.3 ± 65.0, and 177.8 ± 4.0 dpm/µg protein in control, IL1β and IL1β plus anakinra, respectively, incubated in 5.5 mmol/L glucose; P < 0.05, n = 4). Enhancing glucose concentration, however, did not affect the effects of IL1β on GLUT1 expression or glucose transport capacity (Fig. 1a and 1b, and Additional file 1: Figure S2).

We confirmed our previous findings that activation of cells with IL1β resulted in cell inflammation [6], determined as the activation of NFκB and the expression of iNOS. This effect was exaggerated in the presence of high glucose (Fig. 2a, b), through a mechanism independent of hyperosmolarity [6]. Because of the above, we decided to investigate whether the simple increase in the entry of glucose was sufficient to cause inflammation.

In order to achieve this, we first over-expressed GLUT1 transporters by infecting the cells with a lentiviral vector containing human GLUT1 cDNA (GLUT1+). Over-expression of GLUT1 was verified by quantification of GLUT1 mRNA levels (Additional file 1: Figure S3A). While the glucose transport capacity was increased by approximately 50 % in GLUT1+ cells but not in control cells transduced with an empty vector (EV) (Additional file 1: Figure S3B), glucose consumption was not significantly enhanced when the extracellular glucose concentration was increased to 22 mmol/L (Table 1b). Moreover, lactate production and the lactate: glucose ratio in GLUT1+ and EV cells was similar to those in control cells (Table 1b).

In a different set of experiments, we increased glucose entry in the cells by inhibiting mitochondrial respiration using sodium azide (0.5 mM). This treatment did not affect cell survival after 24 h (data not shown), however it increased glucose transport (³H-deoxyglucose uptake of 171.6 ± 19.3 and 345.3 ± 46.7 dpm/µg protein in control and plus sodium azide, respectively; P < 0.05, n = 4). This was associated with an increase in surface GLUT1 transporters (Additional file 1: Figure S4), without a modification in GLUT1 mRNA levels (98.6 ± 17.0 % vs non-treated cells, n = 3). Sodium azide treatment produced a marked increase in glucose utilization, without changing the lactate: glucose ratio significantly (Table 1a).

Neither the over-expression of GLUT1 nor the glucose overload after sodium azide treatment were sufficient to induce inflammation (Fig. 2a, b).

We next monitored the intracellular utilization of glucose by measuring the generation of ¹⁴CO₂ from glucose labelled in different carbon atoms. The production of ¹⁴CO₂ from [6-¹⁴C]glucose, which decarboxylates only via the tricarboxylic acid cycle, was not modified by incubation in a high concentration of glucose (Fig. 3a). There was, however, a significant increase following treatment with IL1β, which was similar at both 5.5 and 22 mmol/L glucose (Fig. 3a). The activity of the pentose phosphate pathway (PPP) was not significantly affected by increasing the concentration of glucose in the medium or by incubating cells in 5.5 mmol/L glucose with IL1β (10 ng/ml); however, a significant enhancement was observed when the cells were activated by IL1β in a medium containing 22 mmol/L glucose (Fig. 3b). The protein levels of glucose-6-phosphate dehydrogenase (G6PD), as the step-limiting enzyme of the PPP, were slightly increased after IL1β treatment in 5.5 mmol/L glucose, but markedly enhanced in 22 mmol/L glucose (Fig. 3c).

NADPH generated by the PPP can be used either to regenerate reduced glutathione (GSH) through the action of glutathione reductase or as a substrate to fuel NADPH oxidase. We found that incubation in 5.5 mmol/L glucose with IL1β increased both total glutathione and GSH; however, this effect was abolished in cells incubated in 22 mmol/L glucose (Fig. 3d). In the absence of the cytokine, simply enhancing the extracellular concentration of glucose failed to modify the intracellular glutathione profile (Fig. 3d).

Measurements of the NADPH oxidase showed that IL1β elicited a concentration-dependent activation of this enzyme, which was exaggerated in the presence of high glucose (Fig. 4a). NADPH oxidase activity was not modified by just enhancing glucose transport capacity or following mitochondrial inhibition (Fig. 4b). By using the antioxidant and NADPH oxidase inhibitor apocynin as well as anakinra, we observed that the activation of NADPH oxidase via IL-1 receptors was necessary to induce inflammation by IL1β and its enhancement by high glucose (Figs. 4c, d).

To study the role of the PPP on the activation of NADPH oxidase, we used the G6PD pharmacological inhibitor 6-aminonicotinamide (6-ANAM). In cells activated with IL1β this compound prevented the activation of NADPH oxidase and the resulting inflammation at 5.5 and 22 mmol/L glucose (Figs. 5a, b, c). To obtain a more selective blockade of G6PD activity, cells were treated with G6PD siRNA (Fig. 6a). There was a significant increase (P < 0.005) in the basal NADPH oxidase activity and iNOS expression produced both by scrambled or G6PD siRNA, suggesting the experimental procedure may originate some unspecific inflammation. Interestingly, however, G6PD silencing specifically blocked the over-activation of NADPH oxidase and the enhanced iNOS expression induced by high glucose (Figs. 6b, c, d).

We first confirmed our previous results [12] demonstrating that treating isolated segments of rat mesenteric microvessels with 2.5 ng/ml produced an impairment of the endothelium-dependent relaxations evoked by ACh through a mechanism dependent on NADPH oxidase (Fig. 7a; Additional file 1: Figure S5). Moreover, we observed that the endothelial dysfunction induced by the cytokine was further enhanced when the vascular segments were pre-incubated in a medium with 22 mmol/L glucose (Fig. 7a). The role of the PPP was studied by treating the mesenteric microvessels with 6-ANAM; this drug did not affect the endothelial dysfunction induced by IL1β in normal glucose but prevented the potentiation observed in high glucose conditions (Figs. 7b, c).