Hyperglycemia Inhibits Retinoic Acid–Induced Activation of Rac1, Prevents Differentiation of Cortical Neurons, and Causes Oxidative Stress in a Rat Model of Diabetic Pregnancy

This study was conducted using protocols approved by the institutional animal care and use committee. Adult, age-matched female SD rats were housed in rooms controlled for temperature (23 ± 1°C) and light/dark cycle (12 h light/12 h dark) and provided with standard rat diet, and water was available ad libitum. In vitro studies were performed using cortical neurons isolated from 16-day-old rat embryos (E16). In vivo studies were performed using cortices or cortical neurons isolated from E16 embryos of diabetic and nondiabetic female SD rats.

The baseline blood glucose levels of the animals were measured after a 12-h fast. Diabetes was induced by a single intraperitoneal injection of freshly prepared streptozotocin (60 mg/kg body wt; Sigma, St. Louis, MO) dissolved in sterile saline (0.85% NaCl). Control rats received an equal volume of the vehicle. Nonfasting blood glucose levels were quantified after 1 week by using a commercially available glucometer (Elite; Bayer, Newbury, U.K.). Streptozotocin-injected rats having initial blood glucose levels of <300 mg/dl were considered to be nondiabetic. Nonfasting blood glucose levels were monitored every 3rd day throughout the course of the study and were tested again just before anesthetizing the animals for embryo extraction.

Diabetic and nondiabetic rats were put into the home cages of individually housed, proven, breeder male SD rats just before the end of a light cycle. The following morning, each female was examined for the presence of an ejaculatory plug in the vagina. The day the plug was first observed was defined as the first day of gestation, or embryonic day 1 (E1) (25).

E16 timed-pregnant SD rats were anesthetized with isoflurane (2.5% for 10 min), and embryos were removed for terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) and caspase-3 activity assessment. Whole brains were dissected out from the embryos and fixed in 10% formaldehyde for 72 h. For studying oxidative stress parameters, the cortex was separated, immediately frozen in liquid nitrogen, and stored at −80°C until use.

The brains were removed from the embryos, and cortical tissue was trypsinized (0.06% in Dulbecco’s modified Eagle’s medium; Sigma-Aldrich) for 10 min at 37°C followed by treatment with trypsin inhibitor and DNAase I solution (40 mg/ml soybean trypsin inhibitor and 15 mg/ml DNAase I) for an additional 5 min. Neurons were dissociated by trituration, and Dulbecco’s modified Eagle’s medium supplemented with 4% rat serum and 1% fetal bovine serum was added to the single-cell suspension. The suspension was plated on tissue culture plates (60 mm) coated with poly-l-lysine (40 μg/ml) and laminin (1 μg/ml) at a density of 2 × 10⁶ cells per culture plate. After 12–14 h of incubation at 37°C (5% CO₂), 10 μmol/l cytosine-d-arabinofuranoside was added to prevent growth of non-neuronal cells. This protocol produced >95% pure cortical neuronal cultures based on manual counting of phase-bright cells (presumed neurons) and firmly adherent, phase-dark cells (presumed non-neuronal cells).

Cortical neuronal cultures were subjected to the hyperglycemic insult 2 h after plating by exposing them to a high-glucose solution (final concentration 60 mmol/l; 35 mmol/l d-glucose added to the 25 mmol/l d-glucose already present in the medium). The control set of cortical neurons was exposed to the 25 mmol/l d-glucose present in the medium. For osmotic control, separate cultures were exposed to medium containing 35 mmol/l l-glucose plus the 25 mmol/l d-glucose present in the standard culture medium. Cortical neurons were also exposed for 48 h to retinoic acid (300 nmol/l) throughout the experiment with or without 35 mmol/l glucose.

To assess the effects of hyperglycemia on the expression of neuronal markers (microtubule-associated protein [MAP-2], total tau, and growth-associated protein [GAP-43]), SOD-2, and activated caspase-9, cortical neurons were isolated and lysed as previously reported (23). Cell lysates (30 μg protein for neuronal markers and 100 μg protein for SOD-2 and caspase-9) were separated by SDS-PAGE, transferred onto nitrocellulose membranes, and probed with appropriate antibodies. The membranes were reprobed with anti-actin antibody to assess loading differences among samples. To estimate oxidative stress, cortical tissue or cultured cortical neurons were thoroughly washed with PBS (20 mmol/l, pH 7.0), and homogenates were prepared in phosphate buffer (pH 7.0) by sonication (duty cycle 50; time 10 s) followed by centrifugation at 12,000g for 30 min. The supernatant was used for estimating SOD activity, GSH, and total thiol levels. Lipid peroxidation (LPO) levels were estimated directly in the homogenates after sonication.

Activation of Rac1 was studied by pull down of the GTP-bound form of Rac1 using glutathione S-transferase (GST)-fused PAK binding domain (GST-PBD), as described previously (23). Levels of total Rac1 in the cell lysates were determined by Western blotting.

Cortical neurons were infected with adenovirus-expressing green fluorescent protein (GFP)-Rac1 N17 at 100 multiplicities of infection for 48 h. Cells infected with adenovirus-expressing GFP were used as controls.

Hyperglycemia-induced apoptotic stress was studied in whole-brain sections using TUNEL and cleaved caspase-3 staining. Whole brains from E16 embryos fixed in formaldehyde were embedded in paraffin, and 4-μm-thick sections were cut in the horizontal plane. Serial sections were mounted onto glass slides in triplicate, heat dried, deparaffinized in xylene, and dehydrated/rehydrated in a series of graded ethanols. The first section was stained with thionin for detection of Nissl bodies present in the cytoplasm of neurons. The second section was stained to detect the presence of apoptotic nuclei using a TUNEL labeling kit (Promega, Madison, WI) according to the manufacturer’s protocol. The third section was immunostained to detect activated caspase-3. Sections were incubated with a rabbit monoclonal anti–cleaved caspase-3 antibody (1:200; Cell Signaling Technology, Danvers, MA) overnight at 4°C, followed by fluorescein isothiocyanate–conjugated goat anti-rabbit IgG (1:400; Santa Cruz Biotechnology, Santa Cruz, CA) for 30 min at room temperature.

SOD activity in tissue homogenates was assessed by measuring the delay in SOD-inhibited auto-oxidation of epinephrine (26). The reaction was monitored at 12-s intervals for 1 min at 480 nm. Reaction mixtures lacking enzyme were run simultaneously as controls. SOD activity was expressed as the amount of the SOD required to inhibit 50% auto-oxidation of epinephrine.

Formation of thiobarbituric acid reactive substances as a product of LPO was estimated in tissue homogenates (27). Absorbance was measured at 532 nm, and a molecular extinction coefficient of 1.56 × 10⁵ mol/l per centimeter was used to calculate nanomoles malondialdehyde (MDA) formed per milligram protein in the lysate.

GSH levels in cell and tissue homogenates were estimated using 5-5′-dithiobis (2-nitrobenzoic acid) (DTNB) reagent (28). Absorbance of the product was read at 412 nm. The amount of GSH in the sample was calculated in micromoles per liter from a standard curve obtained using known quantities of GSH (10–50 μg).

Total thiol content was estimated in cell and tissue homogenates using DTNB reagent (29). Absorbance of the supernatant was measured at 412 nm. Reduced GSH was used as a standard. A molar extinction coefficient of 13,600 mol/l per centimeter was used for calculations, and total thiol present in the sample was expressed in millimoles per milligram protein in the lysate.

Statistical significance of the data was determined by one-way ANOVA, Tukey-Kramer multiple comparisons tests, and paired t tests. The results were considered significant at P < 0.05.

During pregnancies, the average nonfasting blood glucose level of diabetic rats was 362.5 mg/dl (358–377) and that of nondiabetic rats was 84.5 mg/dl (79.5–98). The rate of successful pregnancies in diabetic rats was 66.67% (20 of 30) and in nondiabetic rats was 80% (20 of 25). The average number of embryos per pregnancy was 12 for both diabetic and nondiabetic rats.

To determine whether retinoic acid plays a role in differentiation of cortical neurons, we treated cultured cortical neurons with retinoic acid for 48 h. Compared with vehicle-treated neurons, retinoic acid–treated neurons displayed significantly increased neurite formation and neuronal marker expression (Fig. 1A–C). To determine whether hyperglycemia affects neuronal differentiation, we exposed cortical neurons to high glucose concentrations (60 mmol/l) in the presence of retinoic acid. High glucose significantly inhibited retinoic acid–induced neurite outgrowth (Fig. 2A and B). To confirm this inhibitory effect in vivo, cortical neurons isolated from embryos of nondiabetic and diabetic rats were exposed to retinoic acid. With cultures derived from embryos of diabetic rats, we observed a significant decrease in cells with neurites. These neurons also formed fewer neurites when treated with retinoic acid, suggesting that diabetes might impair retinoic acid signaling during differentiation of cortical neurons (Fig. 2C and D).

We treated cortical neurons with retinoic acid, retinoic acid plus glucose (60 mmol/l), or glucose alone and monitored Rac1 activation by GST-PBD assay. Retinoic acid promoted Rac1 activation, whereas glucose inhibited retinoic acid–induced Rac1 activation (Fig. 3A and B). We used l-glucose (35 mmol/l) along with d-glucose (25 mmol/l) as an osmotic control. To assess the effects of osmolarity on retinoic acid–induced Rac1 activation, we treated cortical neurons with retinoic acid, retinoic acid plus glucose (25 mmol/l d-glucose and 35 mmol/l l-glucose), or glucose (25 mmol/l d-glucose and 35 mmol/l l-glucose) alone and monitored Rac1 activation. Retinoic acid once again promoted Rac1 activation, but neurons exposed to retinoic acid plus l-glucose/d-glucose showed no significant inhibition of retinoic acid–induced Rac1 activation (Fig. 3C and D), indicating that l-glucose/d-glucose failed to inhibit retinoic acid–induced neurite outgrowth, and osmolarity had no effect on neurite outgrowth in our model system.

To verify that retinoic acid–induced neurite outgrowth is mediated by Rac1, we infected neurons with an adenovirus expressing a dominant-negative form of Rac1 (GFP-Rac-N17); the neurons were then treated with retinoic acid. Overexpression of Rac1-N17 blocked the induction of neurite outgrowth, suggesting that Rac1 is involved in regulating neurite outgrowth (Fig. 4A–C).

Type 1 diabetes is known to induce apoptosis in hippocampal neurons (30). Because high glucose concentrations inhibited retinoic acid–induced neurite outgrowth, we hypothesized that hyperglycemia may also promote oxidative stress in cortical neurons that may ultimately induce neuronal apoptotic stress. To address this issue, we evaluated and compared hyperglycemia-induced changes in GSH, LPO, and total thiol levels. Compared with cortex from control rats, E16 cortex from diabetic rats displayed significantly decreased GSH levels (Fig. 5A) but significantly increased lipid peroxide and total thiol levels (Fig. 5B and C).

Next, we evaluated SOD-1 and SOD-2 expression and the degree of caspase-9 cleavage in E16 cortex from control and diabetic rats. E16 cortex of diabetic rats displayed significantly decreased expression of the enzymatic antioxidant SOD-2 and significantly increased caspase-9 cleavage (Fig. 6). We did not detect a change, however, in SOD-1 expression (data not shown).

After demonstrating that hyperglycemia induces oxidative stress, we studied its effects on neuronal differentiation by assessing the expression of neuronal markers. Cortical neurons isolated from E16 embryos of diabetic rats showed significantly decreased expression of neuronal markers (MAP-2, total tau, and GAP-43) (Fig. 7A and B). However, neuronal marker expression in cortical neurons exposed to hyperglycemic conditions in vitro remained unaltered (data not shown).

To determine whether diabetes affects activation of Rac1, we performed GST-PBD assays on cortical neurons isolated from embryos of nondiabetic and diabetic rats. Cortical neurons isolated from diabetic rats displayed significantly decreased Rac1 activation (Fig. 7C and D).

Retinoic acid is known to reduce stauro-sporine-induced apoptosis and oxidative stress in hippocampal neurons by preserving SOD protein levels (31). To study the role of retinoic acid in glucose-induced oxidative stress, we treated cultured cortical neurons with retinoic acid, retinoic acid plus glucose (60 mmol/l), or glucose alone for 48 h. Cell homogenates were analyzed for SOD activity and GSH, LPO, and total thiol levels. High glucose treatment significantly decreased SOD activity and GSH levels but increased LPO and total thiol levels (Table 1). Retinoic acid treatment, however, inhibited the high glucose–induced decrease in SOD activity and GSH levels and increase in LPO and total thiol levels. These findings demonstrated that retinoic acid functions to protect neurons from oxidative stress.

Our findings showing that hyperglycemic conditions subject the developing cortex to significantly high levels of oxidative stress prompted us to determine whether pathological conditions inherent to diabetes also induce apoptotic stress in the developing brain. Hyperglycemia-induced apoptotic stress in whole-brain sections from E16 embryos of nondiabetic and diabetic rats was assessed with TUNEL and cleaved caspase-3 staining. Specific brain regions were identified by Nissl-thionin staining (Fig. 8A). The neocortex of embryos from diabetic rats displayed significantly denser basal TUNEL staining compared with that of embryos from nondiabetic, control rats (Fig. 8B).

Because thionin selectively stains Nissl bodies that are present only in the cytoplasm of neuronal cells, dense Nissl staining (purple-blue) within neocortex confirms the presence of large numbers of neuronal cells (Fig. 8C). Hence, the majority of TUNEL-positive cells that we observed within the neocortex are most likely to be neuronal cells. Apoptotic stress was further evaluated on the basis of caspase-3 activation in the neocortex. Staining for cleaved caspase-3 revealed significantly more immunopositive cells in the neocortex of embryos from diabetic rats than in the neocortex of embryos from nondiabetic rats (Fig. 8D).

Rho GTPases (RhoA, Rac1, and Cdc42), small–molecular weight G-proteins of the Ras super family, have been implicated in cellular processes such as cytoskeletal rearrangement and cell cycle control (32,33). Active (GTP-bound) and inactive (GDP-bound) Rho conformations function as regulators in a myriad of signal transduction events. Studies conducted in primary neuronal cultures suggest that Rho GTPases play a central role in growing axons and neuronal processes. In general, Rac1 and Cdc42 enhance axonal extension, whereas RhoA enhances process retraction (34,35). One common denominator found in neurite outgrowth inhibition, extension, and retraction is that all three display actin rearrangement within growth cones (36). Using the neuroblastoma cell line SH-SY5Y in our previous studies, we have shown that Rac1 mediates retinoic acid–induced neurite formation and neuronal marker expression (23).

In the present study, we selected cortical neurons as a model system for clarifying the role of retinoic acid in neuronal differentiation and for understanding the molecular events underlying hyperglycemia-induced complications in the CNS. Exposing cortical neurons to 300 nmol/l retinoic acid significantly induced neurite outgrowth and increased expression of MAP-2, total tau, and GAP-43, all of which are involved in microtubule polymerization and in stabilization of growing neurites (37). These observations were indicative of retinoic acid–induced differentiation of cortical neurons and were well within the range of effects observed with physiological concentrations (10⁻¹⁰ to 10⁻⁶ mol/l) of retinoic acid (38,39).

Using primary cortical neuronal cultures, we demonstrated that exposure to high glucose in vitro and in vivo prevents neurite formation and MAP-2, total tau, and GAP-43 expression. We also showed that Rac1 mediates retinoic acid–induced neurite outgrowth in cortical neurons and that exposure to high glucose levels in vitro and in vivo inhibits the activation of Rac1. Rac1 inhibition was not due to elevated osmotic pressure because Rac1 activation in retinoic acid–treated and in retinoic acid plus l-glucose/d-glucose–treated neurons was comparable. These findings not only demonstrate that diabetes impairs retinoic acid signaling in developing cortical neurons but also establish a molecular link between neuronal differentiation and hyperglycemia in diabetes.

Free radicals have been implicated in a number of human diseases, including myocardial infarction, kidney damage, diabetes, atherosclerosis, and cancer (40). They have also been implicated in the underlying processes of aging (41). Cells have developed a comprehensive array of antioxidant defenses (enzymatic and nonenzymatic) to prevent free radical formation or to limit the damaging effects of free radicals. Antioxidant concentrations are interesting parameters because they reflect susceptibility to oxidative stress.

There is considerable evidence that hyperglycemia-induced oxidative stress is one of the main causes of complications resulting from diabetes. In mammalian cells, reactive oxygen species (ROS) are generated during normal aerobic metabolism. Increased ROS levels, however, cause oxidative stress that leads to neuronal apoptosis in diseases like amyotrophic lateral sclerosis, Alzheimer’s disease, and Parkinson’s disease (41–43). In addition, a genetic predisposition to embryonic dysmorphogenesis has been proposed to be linked to impaired expression of SOD in response to increased ROS levels (44). SOD is a potent antioxidant enzyme that scavenges ROS. Recently, it was demonstrated that C-peptide has a preventive effect in neuronal hippocampal apoptosis in type 1 diabetes, although C-peptide has no effect on oxidative stress. These studies suggest that in addition to oxidative stress, other factors are also involved (45,46). We observed decreased GSH levels and increased LPO and total thiol levels in cortex of E16 embryos from diabetic rats. SOD-2 protein levels in cortex were also significantly reduced. Taken together, these findings indicate that diabetes-related oxidative stress may reduce SOD activity.

To further determine whether diabetes-related oxidative stress results from hyperglycemia, we exposed cortical neurons to high glucose solutions and assessed SOD activity and GSH, LPO, and total thiol levels. Exposure to high glucose induced oxidative stress, as indicated by the reduced SOD activity and GSH levels, and enhanced LPO and total thiol levels in cortical neuronal cultures treated with high glucose medium (Table 1). Retinoic acid reduced this high glucose–induced oxidative stress, indicating that retinoic acid may function as an antioxidant.

Retinoic acid is primarily known to be apoptotic; only a few reports have demonstrated retinoic acid to be anti-apoptotic (31,47). The reduced SOD activity that we observed in high glucose–treated cortical neurons may be related to the high levels of superoxide anion produced during oxidative stress, which undergoes dismutation to form elevated levels of hydrogen peroxide. Decreased SOD activity can be due to the inhibition of SOD by hydrogen peroxide (48). Interestingly, we observed that retinoic acid inhibits the high glucose–induced decrease in SOD activity, which is consistent with the protective role retinoic acid in neonatal rat hippocampal neurons undergoing staurosporine-induced oxidative stress and apoptosis in which SOD prevents oxidative stress–related decreases in SOD-1 and SOD-2 (31). These studies proposed that retinoic acid promotes antioxidant signaling mechanisms that protect neurons from oxidative stress.

GSH is a scavenger of hydroxyl radicals and singlet oxygen and functions as a substrate for GSH peroxidase, a hydrogen peroxide–quenching enzyme. Exposing tissues to a large flux of hydrogen peroxide and hydroxyl radicals might cause an imbalance in the GSH/GSSG GSH–to–GSSG (reduced-to-oxidized GSH) ratio. GSSG subsequently accumulates and contributes to the inhibition of protein synthesis and inactivation of various enzymes, which would normally lower GSH levels in cortical neurons subjected to high glucose conditions. High MDA levels in cortical neurons cultured under high glucose conditions clearly reflect enhanced peroxide formation due to free radical–mediated destruction of lipids. Because lipids represent the major constituents of cell membranes, a physiological and biochemical disturbance in activities related to the cell membrane is expected.

In our cortical neuronal model, exposure to high glucose increased thiol levels in the cortical neurons. Free sulfhydryl groups in proteins play the role of highly reactive functional groups in biological systems and participate in several reactions such as alkylation, arylation, oxidation, and thiol-disulfide exchange. The modification of protein thiol groups may result in severe functional damage, including loss of enzyme activity in biological systems (49).

Our data also indicate that hyperglycemia induces neuronal apoptotic stress in addition to oxidative stress in the cortex of E16 embryos from diabetic rats, as demonstrated by positive TUNEL and increased expression of caspase-9 and downstream cleaved caspase-3. The neocortex represents a differentiating field of the developing brain and therefore has a large number of neuronal cells (as shown by Nissl staining). The presence of apoptotic stress in the cells within the neocortex of E16 embryos from diabetic rats indicates that there might be a significant progressive neuronal loss in this region of the developing cortex. This decrease in neurons may contribute to various dysfunctions typical of diabetes-related neuronal complications.

In the diabetic group, the presence of positive TUNEL staining, cleaved caspase-3, and high oxidative stress is indicative of apoptotic activity and reflects failure of the cellular antioxidant defense system, ultimately leading to cell death. The cognitive impairment associated with type 1 diabetes has been linked to apoptosis-induced neuronal loss in the hippocampus (30). Moreover, diabetes not only induces apoptosis but also enhances apoptosis induced by other pathological conditions, such as cerebral ischemia (50). To our knowledge, our findings are the first of their kind to suggest that hyperglycemic conditions during development produce abnormal apoptotic stress in the neocortex.

The present study shows that subjecting the developing cerebral cortex to hyperglycemic conditions leads to the disruption of various signal transduction components, ultimately resulting in neuronal damage. In addition to the known function of retinoic acid in neuronal differentiation and brain patterning, our studies show that retinoic acid might also strengthen endogenous antioxidant defense systems. Hence, retinoic acid can be used to design preventive and therapeutic strategies against CNS complications associated with diabetes.

This work was supported in part by Veteran Affairs Grant VISN-17 and by Scott and White Resident Training Grant 8271.

We thank Drs. Eric Rachut and Douglas Toler (Division of Pathology, Central Texas Veterans Health Care System, Temple, TX) for their help in the histological studies.