Thioredoxin-1 Overexpression in the Ventromedial Nucleus of the Hypothalamus Preserves the Counterregulatory Response to Hypoglycemia During Type 1 Diabetes in Male Rats

Hypoglycemia is a severe side effect of intensive insulin therapy (1). Recurrent hypoglycemia (RH) impairs the brain’s ability to sense subsequent hypoglycemia and generate the autonomic and neuroendocrine counterregulatory responses (CRRs) that restore euglycemia (1). This condition, known as hypoglycemia-associated autonomic failure (HAAF), is the major limiting factor in treating diabetes with insulin.

The ventromedial hypothalamus (VMH) plays a key role in hypoglycemia detection and CRR initiation (2,3). VMH glucose-sensing neurons are likely candidates for detecting hypoglycemia. The glucose sensitivity of VMH glucose-inhibited (GI) neurons, in particular, parallels CRR magnitude (4); for example, RH blunts activation of VMH GI neurons by decreased glucose (5–7). Manipulations that increase activation of VMH GI neurons in vitro also enhance the CRR (6–9). Nitric oxide (NO) production by neuronal NO synthase and activation of the NO receptor soluble guanylyl cyclase (sGC) are critical for activation of VMH GI neurons in low glucose and for the CRR (7,10). Elevated reactive oxygen species (ROS) level causes sGC s-nitrosation, a posttranslational modification that desensitizes sGC to NO (11). Hypoglycemia increases VMH ROS (6). Thus, ROS production during hypoglycemia may impair NO signaling in VMH GI neurons and blunt their activation by decreased glucose. In support of this hypothesis, we found that enhancing the glutathione antioxidant defense system with N-acetylcysteine (NAC) in nondiabetic rats reduced VMH s-nitrosation, restored normal glucose sensing by VMH GI neurons after RH, and completely prevented HAAF (6).

Although these results in nondiabetic rats are promising, for clinical relevance, the effect of NAC must persist in the presence of diabetic hyperglycemia, which exacerbates brain oxidative stress by triggering mitochondrial ROS production (12). VMH GI neurons from rats with diabetes induced by streptozotocin (STZ) show a severely blunted activation in low glucose. High glucose incubation also prevents subsequent activation of VMH GI neurons by low glucose in vitro (13). These data are consistent with the impaired CRR in diabetes (14). Therefore, diabetic hyperglycemia per se may also impair VMH GI glucose sensing and the CRR by increasing oxidative stress. However, NAC treatment did not restore activation of VMH GI neurons in low glucose after RH in STZ rats (Supplementary Figs. 1 and 2). This suggests that the combined oxidative stress resulting from RH during diabetic hyperglycemia exceeded the capacity of the glutathione system or that recruitment of a different antioxidant defense system is necessary.

The thioredoxin (Trx) system, composed of NADPH, Trx, and Trx reductase (TrxR), facilitates enzymatic protein reduction and denitrosation (15). Trx is the primary regulator of cellular nitrosative status (16). Hyperglycemia reduces Trx activity in the aorta of diabetic rats (17) and blunts exercise-induced upregulation of Trx activity in rat brain (18). Conversely, Trx overexpression in pancreatic β-cells slowed the progression of STZ-induced type 1 diabetes (19) and prevented the formation of type 2 diabetes in mice (20). Moreover, overexpression of Trx-1, the cytosolic isoform of Trx, attenuated oxidative stress–induced osteopenia in STZ mice (21). Therefore, this study tests the hypothesis that VMH Trx-1 overexpression preserves glucose sensing and the CRR in rats with STZ-induced type 1 diabetes.

All procedures were approved by the Rutgers University Institutional Animal Care and Use Committee. Male Sprague-Dawley rats (3 or 5 weeks old) were purchased from Charles River Laboratories. Animals were housed and maintained on a 12-h light-dark cycle at 22–23°C with ad libitum access to food and water.

Purified and packaged adeno-associated virus (AAV)/DJ9-hybrid cytomegalovirus/chicken β-actin promoter (CAG)-green fluorescent protein (GFP)-2A-Trx-1 and AAV/DJ9-CAG-GFP were purchased from Vector BioLabs. Three-week-old rats were stereotaxically injected bilaterally in the VMH over 10 min with 1 μL PBS containing 2 × 10⁹ genomic particles of AAV/DJ9 per side expressing Trx-1 + EGFP or EGFP. The coordinates were 2.3 mm caudal to the bregma, 0.58 mm lateral to the midline, and 8.80 mm down from the dural surface.

Two weeks after the AAV injections, rats received one intraperitoneal injection of STZ (65 mg/kg in 0.1 mol/L citrate buffer; pH 4.5). Controls received vehicle. Rats with glycemia below 150 mg/dL 24 h after the STZ injection received a second dose of STZ. Glycemia exceeding 300 mg/dL by the third day after the STZ injection was considered type 1 diabetes. This was defined as day 1 of diabetes. For the following 2 weeks, diabetic rats received one daily injection of long-acting insulin, ProZinc insulin (PZI, Boehringer Ingelheim), to prevent illness caused by severe hyperglycemia. The glycemic response to PZI was monitored for 3 h postinjection. Daily doses were adjusted to each individual’s response. Rats that experienced inadvertent hypoglycemia were excluded from the study. Nondiabetic rats received saline.

Two hours after PZI or saline treatment, rats were fasted for one hour and intraperitoneally injected with 6, 5, and 4 units/kg regular human insulin (Humulin R, Lilly) respectively, for the final 3 days of the study. Less insulin was administered on subsequent days to decrease the chance of severe hypoglycemia. Controls received saline. Tail blood was taken to measure glycemia every 30 min to verify 2 h of hypoglycemia (∼30 to 60 mg/dL).

One week after diabetes induction, rats were anesthetized with ketamine/xylazine (80/10 mg/kg i.p.) and surgically implanted with vascular catheters in the right jugular vein. For RH, the animals were allowed 3 or 4 days to recover from surgery before the first episode of hypoglycemia. The CRR was examined 5–7 days after cannulation of the jugular vein or the day after RH. Thus, the range for the study was from day 13 to 16 of diabetes or the corresponding days for nondiabetic controls. On the day of the CRR evaluation, rats were fasted for 1 h starting 2 h after the PZI or saline treatment, and then Humulin R (1.3–1.5 units/kg) was administered via jugular catheter. Blood glucose was monitored every 15 min from 0 to 120 min after the insulin infusion. Blood samples withdrawn from the jugular catheter were collected at 0, 15, 30, 60, and 90 min in chilled tubes containing EGTA (1.6 mg/mL, Sigma-Aldrich) and aprotinin (250 kallikrein inhibitor units, Sigma-Aldrich) for measurement of plasma glucagon (6). For catecholamines, blood was collected in chilled tubes containing reduced glutathione (1.2 mg/mL, Sigma-Aldrich) and EDTA (1.8 mg/mL, Sigma-Aldrich) at 0, 60, and 120 min (6). Plasma samples were shipped to the Yale Diabetes Center for analysis. Erythrocytes were resuspended in isovolumetric sterile artificial plasma and infused back via the jugular catheter to prevent volume depletion and anemia (22). The animals were sacrificed at the end of study, and the brain was harvested. Slices containing the VMH were made to verify the position of the stereotaxic injections.

Five-week-old rats were anesthetized with ketamine/xylazine (80/10 mg/kg, i.p.) and then underwent catheter placement in the jugular vein and stereotaxic surgery to implant cannula bilaterally in the VMH according to the following coordinates: 2.4 mm caudal to the bregma, 0.60 mm lateral to the midline, and 9.0 mm down from the dural surface. The CRR was examined 5–7 days later. On the day of the experiment, 1 µL of 1 μmol/L auranofin in 0.1% DMSO in artificial cerebrospinal fluid or same amount of vehicle was infused into the VMH bilaterally via the brain cannula over 15 min. Humulin R (3 units/kg) was subcutaneously administered 1 h after the auranofin infusion. Blood glucose was monitored every 15 min from 0 to 120 min after the insulin injection.

VMH neurons from tissue punches that excluded the arcuate nucleus were dissociated and placed in 2.5 mmol/L glucose Neurobasal-A Media (Life Technologies) with B27, as previously described (10), and used within 6 h. Neurons were visualized on an Olympus BX81 WI microscope with a 10× objective equipped with a red filter for visualization of fluorescence imaging plate reader membrane potential dye (FLIPR-MPD; excitation, 548 nm; emission, 610–675 nm). Perfusion of VMH neurons in extracellular solution (in mmol/L: 25 HEPES, 121 NaCl, 4.7 KCl, 1.2 MgSO₄, 5 NaHCO₃, 2 CaCl₂, 0.23 KH₂PO₄, 0.97 K₂HPO₄; pH 7.4; ∼300 μOsm) containing 0.5% MPD (FLIPR MPD Blue, Molecular Devices) at 34°C began 30 min before and continued throughout the duration of all experiments. Glucose concentrations are indicated in the figures. Fluorescence images were recorded to evaluate changes in membrane potential as glucose level changes. Increases in fluorescence indicate depolarization. Cells that have a reversible increase in fluorescence ≥12% compared with that observed in 2.5 mmol/L glucose are considered to be GI neurons. This threshold was defined as exceeding twice the SD of fluorescence variation observed in response to a solution change in which the glucose concentration was held constant (i.e., noise). The amplitude of fluorescence intensity change was not correlated with the change in the glucose concentration. However, the number of neurons that depolarized as glucose decreased (GI neurons) was correlated with the glucose concentration, with an IC₅₀ of ∼0.7 mmol/L, as we have shown previously (23). Hence, data are expressed as the percentage of GI neurons detected per dish.

Total RNA was extracted from ∼2–4 mg VMH punches using TRIzol reagent (Invitrogen) and then transcribed to cDNA using SuperScript III Reverse Transcriptase (Invitrogen) according to the manufacturer’s instructions. cDNA quantization was performed using the TaqMan assay by the LightCycler 480 system (Roche) for 45 PCR cycles. Results were calculated from the standard curves relative to cyclophilin A mRNA levels in the same samples. Tumor necrosis factor-α mRNA was measured and found to be below detection, indicating a minimal inflammation response caused by the AAV injections.

VMH punches were sonicated in Tris-EDTA buffer (pH 7.5). Trx activity was measured using the Trx activity fluorescent assay kit (Cayman Chemical) following the manufacturer’s manual and normalized to the amount of protein.

All data are presented as mean ± SEM. Statistical analysis was performed using GraphPad Prism 5.0 software by two-way ANOVA, followed by the Bonferroni post hoc test, one-way ANOVA, followed by the Tukey post hoc test, or the unpaired Student t test, as indicated in the figures. P < 0.05 indicates statistical significance.

Bilateral VMH infusion of the TrxR inhibitor auranofin 1 h before insulin-induced hypoglycemia administration compared with vehicle‐treated controls (Fig. 1A) showed the absolute value for the area under the curve (AUC) for glucose after the insulin injection was increased 22% (P < 0.05) by VMH auranofin (−6,264 ± 523 mg/dL × min, n = 11), compared with the vehicle-treated controls (−4,907 ± 492 mg/dL × min, n = 11). This indicates a slower recovery to euglycemia after insulin-induced hypoglycemia. Incubation of isolated VMH neurons in 1 µmol/L auranofin for 1 h decreased the percentage of detectable VMH GI neurons observed in response to both 0.1 and 0.5 mmol/L glucose by 50% (P < 0.05) and 43% (P < 0.05), respectively (Fig. 1B and C). Together, these data support the hypothesis that the Trx antioxidant defense plays a role in both the CRR and glucose sensing by VMH GI neurons.

Trx activity was significantly lower in STZ-induced diabetic rats compared with nondiabetic controls (Fig. 2A). Therefore, we determined whether VMH Trx-1 overexpression preserved hypoglycemia counterregulation and glucose sensing by VMH GI neurons in these animals. To do this, we injected AAV/DJ9-CAG-EGFP-2A-Trx-1 (Trx-1 and EGFP expression) or AAV/DJ9-CAG-EGFP (EGFP expression alone; control) into the VMH bilaterally. Figure 2B shows that EGFP expression was largely constrained to the VMH. Trx-1 AAV injection increased VMH Trx-1 mRNA expression by fivefold and doubled Trx activity 2 weeks after the injection (Fig. 2C and D). Trx-1 overexpression was maintained in control and STZ rats for the duration of the study (Fig. 2E).

Interestingly, the proportion of rats with bilateral VMH Trx-1 AAV injections that failed to develop diabetes in response to a single STZ injection (24 of 90) was significantly greater than that of rats that received the control EGFP AAV injection (2 of 86; P < 0.0001 two-sample z test (Table 1). Moreover, glycemia was lower on the day after injection in the Trx-1–overexpressing rats that developed diabetes after the first STZ injection compared with the analogous controls (Fig. 3A). Although nearly all of the rats in both groups developed diabetes after a second STZ injection, the proportion of rats that did not develop diabetes was still larger for those with VMH Trx-1 overexpression (P < 0.05) (Table 1). Glycemia in both diabetic groups reached the same level 3 days after the last STZ injection (Fig. 3A). Furthermore, Trx-1 AAV–injected rats had lower blood glucose levels during 2 weeks of diabetic hyperglycemia than did rats expressing the control AAV (Fig. 3B). This was associated with a decreased amount of PZI required to lower morning glycemia from >300 to 150–200 mg/dL (Fig. 3C and D). Although STZ rats weighed less than the nondiabetic rats, VMH Trx-1 overexpression had no effect on body weight (Supplementary Fig. 3).

In response to insulin-induced hypoglycemia, STZ rats with the control EGFP AAV injection did not recover to euglycemia during the 2 h monitoring period, and the absolute value of their glucose AUC was significantly greater than nondiabetic rats (Fig. 4A). In contrast, glucose levels and the glucose AUC in diabetic rats with VMH Trx-1 overexpression were not significantly different from the nondiabetic animals. A similar effect was observed for the glucagon response (Fig. 4B). Interestingly, epinephrine levels at 120 min were higher in both groups of diabetic rats compared with the nondiabetic rats (Fig. 4C). Two-way ANOVA indicated that the absolute value of the AUC for epinephrine was greater in diabetic than in nondiabetic rats; however, there was no interaction between Trx-1 overexpression and diabetes, and post hoc analysis showed no differences between individual groups. The norepinephrine level at each time point was comparable among all the groups, whereas the AUC for norepinephrine was similar to that of epinephrine (Fig. 4D).

Neurons that were transfected with the Trx-1 EGFP or control EGFP AAV were analyzed separately from neurons in the same animals that were not transfected (e.g., did not show EGFP fluorescence). Compared with VMH neurons isolated from nondiabetic controls, the percentage of control EGFP+ VMH GI neurons that were detected in response to 0.1 and 0.5 mmol/L glucose was reduced by 84% and 85%, respectively, in rats with STZ-induced diabetes. Trx-1 overexpression completely normalized activation of VMH GI neurons in both 0.1 and 0.5 mmol/L (Fig. 5A and B, left panels). Activation by low glucose of nontransfected GI neurons from control EGFP AAV-STZ rats but not Trx-1 AAV-STZ rats was also significantly lower than in nondiabetic rats (Fig. 5A and B, right panels).

We next examined the CRR in STZ rats that experienced one daily 2-h episode of hypoglycemia for the final 3 days of the study. After RH, VMH Trx-1 overexpression did not restore the glycemic recovery after insulin-induced hypoglycemia in diabetic rats. However, the AUC for glycemia in the Trx-1–overexpressing diabetic rats showed a slight improvement, but this was not significantly different from the nondiabetic controls or the EGFP-expressing diabetic controls (P = 0.052 [Trx-1 vs. EGFP] two-way ANOVA). In contrast to our previous results with NAC, VMH Trx-1 overexpression did not enhance glycemic recovery in nondiabetic animals after RH (Fig. 6A). VMH Trx-1 overexpression also failed to increase glucagon secretion in nondiabetic or diabetic rats after RH (Fig. 6B). Interestingly, although epinephrine secretion in control EGFP-injected diabetic rats exposed to RH was similar to that of nondiabetic controls at all time points, VMH Trx-1 overexpression tended to increase epinephrine levels at 60 and 120 min. A similar trend was observed for the epinephrine AUC (Fig. 6C). Norepinephrine secretion after insulin-induced hypoglycemia was significantly higher in the Trx-1–overexpressing STZ rats compared with the other groups at 120 min, and the norepinephrine AUC was greater in these rats compared with STZ rats expressing EGFP alone (Fig. 6D).

Trx-1 overexpression increased the number of VMH GI neurons from STZ rats after RH that were detected in response to a maximal glucose decrease (0.1 mmol/L) to that of RH-exposed nondiabetic controls that received the Trx-1 EGFP or control EGFP vectors. However, unlike our previous results with NAC, Trx-1 overexpression did not increase the number of detectable VMH GI neurons in the nondiabetic controls after RH. Moreover, VMH Trx-1 overexpression had no effect on the response of VMH GI neurons isolated from diabetic or nondiabetic rats to a more physiological glucose decrease (0.5 mmol/L) after RH (Fig. 7A and B, left panels). Detection of VMH GI neurons from the Trx-1–overexpressing diabetic and nondiabetic rats that were not transfected was not different from the EGFP controls in response to either glucose decrease (Fig. 7A and B, right panels). In addition, RH reduced the percentage of KCl-responsive neurons regardless of treatment group, suggesting that RH, but not hyperglycemia, slightly reduced neuronal viability (P < 0.05 [RH vs. saline] two-way ANOVA) (Fig. 7C).

Pretreatment with the glutathione precursor NAC decreased VMH ROS and sGC s-nitrosation after RH (6). NAC pretreatment also preserved normal activation of VMH GI neurons in low glucose and the CRR after RH (6). These observations suggest that increasing glutathione may be a viable treatment for HAAF. However, NAC did not preserve activation of VMH GI neurons in low glucose when RH occurred in rats with type 1 diabetes (Supplementary Fig. 2). This may be because the NAC-driven increase in glutathione is not able to compensate for the combined oxidative stress resulting from RH and hyperglycemia. The Trx system compensates for oxidative stress in diabetes (18,24). Trx-1 and Trx-2 are two isoforms of Trx. Trx-1 is considered the major protein denitrosylase in mammalian cells (25). Trx-1 overexpression alleviated oxidative stress–induced osteopenia in STZ mice (21). We thus hypothesized that VMH Trx-1 overexpression would similarly preserve activation of VMH GI neurons by low glucose and the CRR in type 1 diabetes.

In support of our hypothesis, VMH Trx activity decreased in rats with type 1 diabetes. Moreover, TrxR inhibition by auranofin reduced glycemia recovery after insulin-induced hypoglycemia and the activation of VMH GI neurons in low glucose. These data suggest that the Trx antioxidant system is necessary for glucose sensing by VMH GI neurons and the full CRR. Consistent with this concept, VMH Trx-1 overexpression completely restored glycemic recovery and glucagon release after insulin-induced hypoglycemia in STZ rats to that observed in nondiabetic rats. VMH Trx-1 overexpression also completely normalized the activation of VMH GI neurons in low glucose. Although epinephrine and norepinephrine secretion was elevated in diabetic compared with nondiabetic rats, levels were the same in diabetic rats injected with VMH Trx-1 or EGFP control AAV. Therefore, the restoration of glycemia recovery is likely due to the normalized glucagon response. This is very important, because glucagon secretion is impaired early in the development of type 1 diabetes (26). Moreover, hypothalamic glucose-sensing appears to be normalized only when glycemia is normalized, strengthening the causal relationship between glucose sensing and CRR failure.

The mechanisms underlying the impaired glucagon response in diabetes are not clear. Campanucci et al. (27) and Taborsky and colleagues (28) suggest that reduced sympathetic activation of the islet α-cells plays an important role. Alternatively, inhibiting intrapancreatic zinc (cosecreted with insulin) infusion restored the glucagon response to hypoglycemia in STZ rats, suggesting that the loss of β-cell inhibition may also play a role (29,30). Although these mechanisms could result from local changes caused by autoimmune destruction of the β-cell, the important role VMH plays in the impaired glucagon response in diabetes is becoming increasingly clear. For example, increased VMH γ-aminobutyric acid-ergic (GABAergic) output contributes to defective glucagon secretion in response to insulin-induced hypoglycemia in STZ rats (31). Conversely, inhibiting VMH GABA transmission restored glucagon secretion (31). These observations suggest that despite of β-cell loss, reduced VMH GABAergic transmission can restore the glucagon response to insulin-induced hypoglycemia. Similarly, VMH AMPK activation increased glucagon secretion in response to hypoglycemia in rats with type 1 diabetes (32). AMPK activation is crucial for activation of VMH GI neurons by low glucose (10). Interestingly, Trx-1 is a cofactor for AMPK that maintains this enzyme in its activated state (33). Together, our data suggest that VMH Trx-1 overexpression preserved glucagon secretion and led to normalization of glycemic recovery in STZ rats after insulin-induced hypoglycemia by normalizing VMH hypoglycemia detection. Finally, the observation that epinephrine secretion in diabetic rats 2 h after insulin injection was more than twice that of the nondiabetic rats is consistent with one study that evaluated patients with type 1 diabetes and also STZ rats (34).

VMH Trx-1 overexpression did not restore the glucagon response or glycemic recovery to subsequent insulin-induced hypoglycemia after RH in STZ rats or normalize activation of VMH GI neurons in response to a physiological glucose decrease. As we hypothesized for NAC, one antioxidant system alone may not be able to compensate for the combined oxidative stress of RH and diabetic hyperglycemia; that is, although glutathione and Trx overlap to some extent, they also have unique functions. On the one hand, glutathione can, for example, regulate cellular processes through protein glutathionylation (35). On the other hand, Trx but not glutathione is a cofactor for AMPK that maintains normal activation (33). Thus, glutathione and Trx in combination may possibly be more effective in preventing the deleterious effects of hypoglycemia and hyperglycemia. This suggests that a treatment that enhanced both antioxidant systems may be needed to prevent the impaired CRR and reduced activation of VMH GI neurons in low glucose when RH occurs during diabetes. However, this hypothesis requires further study.

A slight increase in the percentage of detectable VMH GI neurons in response to low glucose was also observed in nontransfected GI neurons from diabetic rats with VMH Trx-1 overexpression. It is possible that these neurons were actually transfected with virus. However, they could have been mistakenly counted as nontransfected neurons because of photobleaching or weak transfection in which EGFP expression was below the detection threshold. Alternatively, NO diffusion from adjacent VMH GI neurons that were transfected may have activated nontransfected GI neurons. Also important to consider is that Trx-1 overexpression was unlikely to be limited to GI neurons and probably occurred in GE neurons as well as in nonglucose-sensing neurons and glial cells.

An unexpected finding with potential high impact was the effect of VMH Trx-1 overexpression on peripheral glucose homeostasis in STZ rats. Although almost all EGFP control rats became diabetic after one STZ injection, a statistically larger proportion (∼27%) of those with VMH Trx-1 overexpression required a second injection. VMH lesions change the expression of genes related to differentiation, proliferation, apoptosis, metabolism, and immune signaling pathways in pancreatic β-cells (36–38). VMH lesions also increase mortality after STZ administration (39) These data suggest that the VMH modulates islet β-cell vulnerability to STZ. VMH Trx-1 overexpression also lowered the insulin requirement in STZ rats without affecting body weight. This is consistent with a recent report showing that VMH activation increases peripheral insulin sensitivity (40).

In summary, NAC pretreatment preserved the activation of VMH GI neurons by low glucose in RH-exposed rats but did not preserve normal glucose sensing when RH occurred during diabetes. VMH Trx-1 overexpression preserved the CRR and activation of VMH GI neurons by low glucose in rats with type 1 diabetes before but not after RH. Our data suggest that increasing both glutathione and Trx antioxidant defenses in the VMH may restore the activation of VMH GI neurons and prevent HAAF during diabetes.

Acknowledgments. The authors are deeply grateful to Dr. Owen Chan, University of Utah School of Medicine, Salt Lake City, UT, and Dr. Xavier Fioramonti, INRA, Universite de Bordeaux, Bordeaux, France, for advice in evaluating the CRR in diabetic rats and also thank Dr. Chan for advice in the care and prevention of severe hyperglycemia in diabetic rats. Dr. Zhenyu Sheng, Rutgers New Jersey Medical School, Newark, NJ, provided assistance with care of the diabetic rats.

Funding. This work was partly supported by JDRF Award 2-SRA-2014-269-M-R and National Institutes of Health grant RO1-DK-081538-01A1.

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

Author Contributions. C.Z. collected the data. C.Z. and V.H.R. designed the study and wrote the manuscript. V.H.R. edited the manuscript. V.H.R. 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.

Prior Presentation. Parts of this study were presented in abstract form at the 77th Scientific Sessions of the American Diabetes Association, San Diego, CA, 9–13 June 2017, and at the American Society for Nutrition Scientific Sessions and Annual Meeting at Experimental Biology, Chicago, IL, 22–26 April 2017.