Inorganic Nitrate Promotes the Browning of White Adipose Tissue Through the Nitrate-Nitrite-Nitric Oxide Pathway

The diffuse and complex nature of the metabolic syndrome integrates peripheral insulin resistance and visceral obesity with cardiovascular disease, making the discovery of underlying molecular mechanisms that unite the aspects of the metabolic syndrome both challenging and essential. The perturbation of nitric oxide (NO) synthesis and signaling has emerged as a potential modulator of both cardiovascular morbidity and metabolic dysfunction (1,2).

Until recently, inorganic nitrate was considered a nonbioactive product of NO oxidation (3). However, a number of studies have identified nitrate treatment of humans and rodents as having effects similar to NO (4,5). The discovery of antiobesity effects of nitrate in rodents, including weight loss, a reduction of body fat, reversal of lipodystrophy, and an improvement in glucose and insulin homeostasis, may highlight nitrate as having therapeutic potential for the treatment of the metabolic syndrome (6).

Dietary nitrate increases the circulating concentration of cyclic guanosine monophosphate (cGMP) in humans (7), and low nitrate/nitrite diets decrease steady-state concentrations of cGMP in a number of tissues (8). Recently, cGMP has been shown to regulate energy balance in white adipocytes (9). The development of a brown adipocyte–like phenotype in white adipocytes, a process known as browning, includes the induction of thermogenesis, the dissipation of chemical energy to produce heat (10,11). Peroxisome proliferator–activated receptor γ coactivator 1α (PGC-1α) activates key components of the thermogenic program in white adipocytes, including fatty acid oxidation, mitochondrial biogenesis, and increased oxygen consumption (12). The thermogenic process occurs through the activity and increased expression of several brown adipocyte–specific genes, including uncoupling protein 1 (UCP-1), an inner mitochondrial membrane protein that uncouples the mitochondrial proton gradient (13). Cells expressing brown adipocyte–specific genes are interspersed within the white adipose tissue (WAT) of rodents and humans [so-called beige or brite cells (14,15)] and demonstrate antidiabetic and antiobesity effects in rodent models (16–19). The recent discovery of a physiological small molecule activator of browning in WAT highlights metabolites as both potential mediators of the thermogenic response and therapeutics for the metabolic syndrome (20).

In this study, we investigated the effect of nitrate on WAT metabolism in the classical experimental model of browning in vitro, the mouse primary adipocyte, and in vivo in mice and rats to establish whether this may partly explain the antiobesity activity of nitrate. We demonstrate that nitrate increases the expression of brown adipocyte–specific genes and concordant proteins within white adipocytes to confer a brown adipocyte–like phenotype.

Male Wistar rats (6 weeks old; 269 ± 2 g; n = 24) (Charles River Laboratories) were weight matched and received either distilled water or water containing sodium nitrate (NaNO₃) (0.35, 0.7, and 1.4 mmol/L; n = 6/group) (Ultra-pure Water, Sigma-Aldrich) ad libitum for 18 days, with food and water intake monitored. Animals were housed in conventional cages at room temperature with a 12-h light/dark photoperiod. In the hypoxia study, male Wistar rats (6 weeks old) were weight matched and separated into two groups (n = 10/group), housed in either normoxic or normobaric hypoxic environments (hypoxia chamber: 13% O₂ with 20 air changes/h). The rats in each group received either distilled water containing NaCl (0.7 mmol/L; n = 5) or water containing NaNO₃ (0.7 mmol/L; n = 5) for 14 days. All other details are as aforementioned.

Male p129 mice (6 weeks old) received either distilled water containing NaCl (0.7 mmol/L; n = 7) (control) or NaNO₃ (0.7 mmol/L; n = 7) (Ultra-pure Water) ad libitum for 15 days, with food and water intake monitored. Ob/ob mice (n = 10) and C57BL/6 wild-type mice (n = 10) (9 weeks old) (The Jackson Laboratory) received either distilled water containing NaCl (0.7 mmol/L; n = 5) (control) or NaNO₃ (0.7 mmol/L; n = 5) (Ultra-pure Water) ad libitum for 8 weeks, with food and water intake monitored. Animals were housed in conventional cages at room temperature with a 12-h light/dark photoperiod.

All animals had micronutrient levels normalized by a standardized quality controlled diet [RM1 (E) (55% crude carbohydrate, 3% crude fat, 15% crude protein); Special Diets Services, Essex, U.K.] 1 week before study commencement. The nitrate content of this diet is 2 mg/kg, and the nitrite content was undetectable below a threshold of 1 mg/kg. All procedures were carried out in accordance with U.K. Home Office protocols by a personal license holder.

Rats and mice were killed with sodium pentobarbital 200 mg/mL (Vétoquinol UK Ltd.). Blood was obtained by cardiac puncture, collected in tubes containing N-ethylmaleimide/EDTA (10 and 2.5 mmol/L, respectively) and immediately centrifuged to obtain plasma. WAT and interscapular brown adipose tissue (BAT) were removed and flash frozen in liquid nitrogen.

WAT was fixed for 24 h in 10% formalin and washed for 1 h in PBS before being set in wax. The tissue was then cut into 8-μm-thick slices and stained with hematoxylin and eosin.

Plasma nitrate was measured as described previously (21).

Primary white adipose stromal vascular cells were fractionated from 6–10-week-old C57BL/6 male mice as previously described (22). Stromal vascular cells were then cultured and differentiated into adipocytes according to published methods (22,23). During the 6-day differentiation, cells were cultured with saline (control), 25 μmol/L NaNO₃, 50 μmol/L NaNO₃, or 500 μmol/L NaNO₃ (Ultra-pure Water), and during the investigation of the effects of sodium nitrite (NaNO₂) cells were cultured with saline (control), 50 μmol/L NaNO₂, or 500 μmol/L NaNO₂ (Ultra-pure Water). The pharmacological inhibitor studies used 2-phenyl-4,4,5,5-tetramethylimidazoline-1-oxyl 3-oxide (PTIO) (50 μmol/L), l-N^G-nitro-l-arginine methyl ester (l-NAME) (1 mmol/L), ¹H-​[1,​2,​4]oxadiazolo[4,​3-​a]quinoxalin-​1-​one (ODQ) (1 μmol/L) (Sigma-Aldrich), and KT5823 (1 μmol/L) (Santa Cruz Biotechnology). Cells were treated with PTIO, l-NAME, ODQ, or KT5823 with and without 500 μmol/L NaNO₃. NaNO₃ and inhibitors were added at day 1 of differentiation. In the hypoxia study, cells were isolated and differentiated as aforementioned. Hypoxic conditions were achieved using a New Brunswick Eppendorf Galaxy 14 S incubator supplied with nitrogen and set to maintain a 2% O₂ environment.

FlexiTube small interfering RNA (siRNA) against xanthine oxidoreductase (XOR), AllStars negative control siRNA, and HiPerFect Transfection Reagent were purchased from QIAGEN. Transfection of primary adipocytes was carried out per the manufacturer’s instructions (75 ng siRNA, 3 μL transfection reagent per well, 10 nmol/L final siRNA concentration) on days 2 and 4 of differentiation.

Metabolites were extracted from WAT and primary adipocytes as previously described (24).

Palmitate was solubilized using a dialyzed albumin solution (24). At 6 days postdifferentiation, medium was removed from primary adipocytes and replaced with a serum-free medium containing insulin 850 nmol/L, triiodothyronine 1 nmol/L, and rosiglitazone 1 μmol/L and 140 μmol/L U-¹³C–labeled palmitate (Cambridge Isotope Laboratories). After 2 days, cells were collected and metabolites extracted as previously described. During the 8 days of the experiment, cells were cultured with saline (control), 50 μmol/L NaNO₃, or 500 μmol/L NaNO₃.

Total RNA extraction from WAT, BAT, and adipocytes; cDNA conversion; and quantitative RT-PCR were performed according to published protocols (20). All data were normalized to 18S rRNA (mouse WAT, BAT, and adipocytes) or RLPL1 (rat WAT), and quantitative measures were obtained using the ΔΔCT method.

Analysis of UCP-1 and PGC-1α was performed using ELISA per the manufacturer’s instructions (UCP1 Kit SEF557Ra, PGC-1α Kit SEH337Ra; Cloud-Clone Corp., Houston, TX). Immunoblotting for carnitine palmitoyltransferase (CPT1) was carried out as previously described (25).

Citrate synthase was assayed according to Houle-Leroy et al. (26).

Oxygen consumption rates were measured in white adipocytes (250,000 cells) maintained in Krebs-Henseleit buffer at 37°C using Clark-type oxygen electrodes (Strathkelvin Instruments, Strathkelvin, U.K.) as described previously (27).

Dried aqueous and organic phase samples were derivatized using the method described previously (24). Gas chromatography–mass spectrometry (GC-MS) and data analysis were performed according to published methods (24).

Analysis of organic and aqueous phases was carried out as aforementioned and according to published methods (24).

Liquid chromatography–mass spectrometry (LC-MS) was performed on WAT using a Waters Xevo QTof mass spectrometer (Waters Corporation, Manchester, U.K.) operating with electrospray ionization in combination with an Acquity UPLC (Waters Corporation) according to the method described by Roberts et al. (24). LC-MS spectra and chromatograms were analyzed with the MarkerLynx Application MassLynx version 4.1 (Waters Corporation) using published protocols (24).

LC-MS analysis of cGMP was performed using a 4000 QTRAP triple quadrupole mass spectrometer (Applied Biosystems/Sciex) coupled to an Acquity UPLC according to a previously described method (28). The multiple reaction monitoring parameters for cGMP were Q1 = 346.13 charge/mass ratio, Q3 = 152.1 charge/mass ratio, collision energy = 23, declustering potential = 41, and collision cell exit potential = 10.

Error bars represent the SEM. P values were calculated by either one-way or two-way ANOVA as stated with a Dunnett post hoc test when multiple comparisons were made solely to control and a Tukey post hoc test when comparisons were made among all treatment groups.

Rodents treated with dietary nitrate are protected against diabetes and obesity (6). Activation of the browning response in WAT may represent a process underlying this altered systemic energy balance; therefore, we assessed the effect of dietary nitrate on the expression of brown adipocyte–specific genes in WAT in vivo. Wistar rats were treated with 0.35, 0.7, or 1.4 mmol/L NaNO₃ in drinking water for 18 days based on preliminary dose-escalation studies. Plasma nitrate concentrations were found to increase in a dose-dependent fashion, with the basal concentration in control rats at 11.1 ± 0.7 μmol/L increasing to 15.4 ± 2.0, 22.8 ± 4.0, and 32.6 ± 5.0 μmol/L in the rats treated with 0.35, 0.7, and 1.4 mmol/L NaNO₃, respectively. Water and food intake was not significantly different among the groups (Supplementary Table 1). Because subcutaneous WAT has the greatest predisposition to undergo browning and influence energy balance (23), we concentrated the analyses on the inguinal WAT depot. Nitrate dose dependently increased brown adipocyte–specific gene expression in WAT. The expression of UCP-1 was increased, as was the molecular marker of brown-like adipocytes cell death-inducing DFFA-like effector a (CIDEA), which plays a role in the regulation of the thermogenic process (Fig. 1A). Expression of PGC-1α and the mitochondrial electron transport chain component cytochrome c (CYCS) were increased dose dependently. Nitrate also increased expression of β-oxidation genes, including CPT1 and acyl-CoA dehydrogenase, very long chain (ACADvl) (control vs. 0.35 mmol/L NaNO₃ P ≤ 0.01, control vs. 0.7 mmol/L NaNO₃ P ≤ 0.0001, control vs. 1.4 mmol/L NaNO₃ P ≤ 0.001). Specific beige-selective genes, including T-box transcription factor (Tbx1), transmembrane protein 26 (Tmem26), and CD137, distinguish beige fat cells from both classical brown and white adipocytes (29). Nitrate moderately increased the expression of the beige-selective markers Tbx1, Tmem26, and CD137, specifically characterizing these cells as beige adipocytes within the inguinal WAT depot (Supplementary Fig. 1). Nitrate did not significantly change the expression of the canonical adipocyte-specific markers adiponectin (ADIPOQ) and fatty acid–binding protein 4 (FABP4) in the WAT, indicating that adipose tissue exposed to nitrate presents similar levels of adipogenesis (Supplementary Fig. 2). Furthermore, analysis of brown adipocyte–specific gene expression in the visceral (epididymal) WAT of nitrate-treated rats also revealed a moderate increase in the mRNA of several genes involved in the thermogenic and browning process (Supplementary Fig. 3).

To determine whether changes in the expression of key brown adipocyte–specific genes were translated to the level of protein, the concentrations of UCP-1, PGC-1α, and CPT1 protein in the subcutaneous WAT of nitrate-treated rats were analyzed. Nitrate increased the expression of the brown adipocyte–specific proteins UCP-1 and PGC-1α (control vs. 0.35 mmol/L NaNO₃ P ≤ 0.001, control vs. 0.7 mmol/L NaNO₃ P ≤ 0.05, control vs. 1.4 mmol/L NaNO₃ P = 0.01) (Fig. 1B). Nitrate also increased the concentration of the β-oxidation protein CPT1 within WAT (control vs. 0.35 mmol/L NaNO₃ P ≤ 0.05, control vs. 0.7 mmol/L NaNO₃ P ≤ 0.05, control vs. 1.4 mmol/L NaNO₃ P ≤ 0.05) (Fig. 1C). The observed induction of characteristic brown adipocyte genes and increase in the concentration of brown adipocyte–specific proteins suggests a browning of the subcutaneous WAT.

We next examined whether the changes in gene and protein expression induced by nitrate resulted in altered energy metabolism within the WAT. The activity of citrate synthase, a marker of mitochondrial density and tricarboxylic acid (TCA) cycle flux, was significantly increased following nitrate treatment, suggesting increased mitochondrial biogenesis consistent with browning of WAT (control vs. 0.35 mmol/L NaNO₃ P ≤ 0.05, control vs. 0.7 mmol/L NaNO₃ P ≤ 0.05, control vs. 1.4 mmol/L P ≤ 0.01) (Fig. 1D). Because beige/brite cells use fatty acids as fuel for thermogenesis, we investigated whether nitrate affected lipid metabolism in the WAT of treated rats. GC-MS and LC-MS were used to measure total fatty acid and triacylglycerol (TAG) species in the WAT, respectively. Nitrate treatment increased the medium-chain fatty acid (MCFA) (laurate, myristate)/long-chain fatty acid (LCFA) (arachidate, behenate) ratio, indicating increased β-oxidation shortening of fatty acid chain length (control vs. 0.35 mmol/L NaNO₃ P ≤ 0.05, control vs. 0.7 mmol/L NaNO₃ P ≤ 0.05, control vs. 1.4 mmol/L NaNO₃ P ≤ 0.01) (Fig. 1E and Supplementary Fig. 4). LC-MS analysis demonstrated that nitrate also decreased the total TAG content within WAT (0.95-fold, 0.35 mmol/L NaNO₃; 0.94-fold, 0.7 mmol/L NaNO₃; 0.98-fold, 1.4 mmol/L NaNO₃; control vs. 0.35 mmol/L NaNO₃ P ≤ 0.05, control vs. 0.7 mmol/L NaNO₃ P ≤ 0.01) (Fig. 1F). Therefore, metabolic changes induced in WAT in vivo by nitrate treatment were characteristic of a brown adipocyte–like phenotype.

Histological analyses of the inguinal WAT of nitrate-treated rats at ×50 magnification revealed the presence of fascicles of multilocular brown adipocyte–like cells (Fig. 1G). Under greater magnification, the smaller multilocular brown adipocyte–like cells can be observed, confirming morphological changes in the WAT. Together, these data indicate that nitrate induces the browning of WAT.

To ensure that nitrate-induced expression of brown adipocyte–specific genes in rat WAT is not species specific, mice were treated with 0.7 mmol/L NaNO₃ in drinking water. As in rat, brown adipocyte–specific gene expression was increased in inguinal WAT by nitrate treatment (Supplementary Fig. 5A). The expression of thermogenic genes in classical interscapular BAT from nitrate-treated mice was also increased (Supplementary Fig. 5B).

Nitrate may function directly to increase the expression of brown adipocyte–specific genes in WAT or indirectly through metabolic bioactivation in cells distinct from those in the target tissue. Therefore, stromal vascular fraction–derived primary adipocytes isolated from inguinal WAT of mice were treated with nitrate. NaNO₃ concentrations of 25, 50, and 500 μmol/L were chosen. The latter two concentrations correspond to plasma levels in mice exhibiting improved metabolic phenotypes when chronically treated and acutely dosed with 0.1 mmol/kg NaNO₃, respectively (6). Nitrate treatment significantly increased the expression of brown adipocyte–specific genes UCP-1, CIDEA, and PGC-1α. Also increased in expression were CYCS, CPT1, and ACADvl (control vs. 25 μmol/L NaNO₃ P ≤ 0.01, control vs. 50 μmol/L NaNO₃ P ≤ 0.0001, control vs. 500 μmol/L NaNO₃ P ≤ 0.0001) (Fig. 2A). In contrast, nitrate did not significantly affect the expression of a panel of classical mature adipocyte-specific genes, including ADIPOQ, FABP4, and adipsin, suggesting that nitrate does not directly affect adipogenesis per se (Supplementary Fig. 6) (29). These data indicate that nitrate-mediated induction of brown adipocyte–specific gene expression occurs directly at the WAT.

We next investigated whether the transcriptional changes induced by nitrate conferred functional effects on energy expenditure in primary adipocytes. The oxygen consumption rates of adipocytes treated with nitrate (50 and 500 μmol/L) were measured (Fig. 2B). The basal oxygen consumption rate was dose dependently increased in adipocytes treated with nitrate (control 3.1 nmol O₂/min/10⁶ cells, 50 μmol/L NaNO₃ 4.5 nmol O₂/min/10⁶ cells, 500 μmol/L NaNO₃ 5.1 nmol O₂/min/10⁶ cells, P ≤ 0.05). Maximal respiratory rates were probed using excess succinate (20 mmol/L) (27) and were found to increase in adipocytes treated with nitrate (control 4.2 nmol O₂/min/10⁶ cells, 50 μmol/L NaNO₃ 6.6 nmol O₂/min/10⁶ cells, 500 μmol/L NaNO₃ 7.7 nmol O₂/min/10⁶ cells, P ≤ 0.05). These findings indicate that nitrate increases the respiration of adipocytes consistent with the browning response (control vs. 50 μmol/L NaNO₃ P ≤ 0.01, control vs. 500 μmol/L NaNO₃ P ≤ 0.001). GC-MS analysis of the fatty acid metabolism of adipocytes treated with nitrate also highlighted the increase in the MCFA/LCFA ratio, mirroring the effects observed in WAT in vivo (control vs. 50 μmol/L NaNO₃ P ≤ 0.05, control vs. 500 μmol/L NaNO₃ P ≤ 0.05) (Fig. 2C).

To confirm the increased β-oxidation observed in nitrate-treated adipocytes, the stable isotope substrate U-¹³C-palmitate was used to monitor flux through the fatty acid oxidation pathway. Primary adipocytes were incubated in serum-free media containing U-¹³C-palmitate and treated with nitrate. GC-MS analysis was used to determine the relative enrichment of metabolites. Nitrate significantly increased the ¹³C enrichment of palmitate (C16:0) in adipocytes, indicating increased fatty acid uptake (P ≤ 0.01) (Fig. 3A). Enrichment of shorter chain fatty acids myristate (C14:0) and laurate (C12:0) was also increased in nitrate-treated adipocytes (C14:0 P ≤ 0.05, C12:0 P ≤ 0.05) (Fig. 3B and C), consistent with augmented chain shortening through β-oxidation.

The labeled fatty acids are catabolized, releasing labeled acetyl-CoA, which enters the β-oxidation and TCA cycle flux. Nitrate treatment increased the labeling of TCA cycle intermediates citrate (control vs. 50 μmol/L NaNO₃ P ≤ 0.0001, control vs. 500 μmol/L NaNO₃ P ≤ 0.001) (Fig. 3D), succinate (control vs. 500 μmol/L NaNO₃ P ≤ 0.05) (Fig. 3E), and malate (control vs. 500 μmol/L NaNO₃ P ≤ 0.05) (Fig. 3F). Therefore, nitrate confers a functional effect on white adipocytes, increasing flux through fatty acid β-oxidation. Taken together, these data demonstrate that nitrate induces the expression of thermogenic genes and the development of a brown fat–like phenotype in white adipocytes and WAT.

In addition to the classical l-arginine-NO synthase (NOS)-NO pathway, NO can also be generated in vivo from nitrate through serial reduction to nitrite and then to NO through the nitrate-nitrite-NO pathway (30). Therefore, we hypothesized that nitrate functions through NO to produce the browning response in WAT. Primary adipocytes were differentiated in the presence of nitrate and the NO scavenger PTIO (control vs. 500 μmol/L NaNO₃ P ≤ 0.0001, 500 μmol/L NaNO₃ vs. 500 μmol/L NaNO₃ + 50 μmol/L PTIO P ≤ 0.0001) (Fig. 4A). Sequestration of NO by PTIO negated the nitrate-induced expression of brown adipocyte–specific genes, indicating that nitrate indeed functions through NO to induce browning.

To exclude the induction of NO production through the l-arginine-NOS-NO pathway as a mechanism of nitrate-stimulated browning, control experiments were conducted using the nonisoform-selective NOS inhibitor l-NAME. Inhibition of NOS did not affect nitrate-stimulated brown adipocyte–specific gene expression in adipocytes (Fig. 4B) (control vs. 500 μmol/L NaNO₃ P ≤ 0.001, control vs. 500 μmol/L NaNO₃ + 1 mmol/L l-NAME P ≤ 0.0001, 500 μmol/L NaNO₃ vs. 500 μmol/L NaNO₃ + 1 mmol/L l-NAME not significant). Therefore, nitrate-stimulated browning functions through NO but independently of NOS.

An enzymatic mechanism for the reduction of nitrate to NO in mammals, catalyzed by XOR, has been reported (4). XOR is expressed in WAT and plays a role in adipocyte homeostasis (31). Nitrate dose dependently increased the expression of XOR in primary adipocytes (Fig. 4C). Using PTIO, the nitrate-stimulated increase in XOR expression was demonstrated to be mediated through NO (Fig. 4D). The expression of XOR in primary adipocytes was knocked down by ∼80% using siRNA (Fig. 4E). Knockdown of XOR abrogated the nitrate-induced expression of brown adipocyte–specific genes in white adipocytes (Fig. 4F). Consistent with these results, nitrite promoted brown adipocyte–specific gene expression in adipocytes through an NO-dependent mechanism (Supplementary Fig. 7A and B). By examining fold increases of brown adipocyte–specific genes, nitrite at equivalent concentrations to nitrate appeared more potent in inducing the browning response, consistent with the inefficient overall rate of reduction of nitrate to nitrite and eventually NO (4). These results indicate that nitrate-mediated browning of adipocytes requires XOR and the reduction of nitrate to NO.

We next characterized the downstream signaling/effector pathways mediating nitrate-induced browning of WAT. NO activates cGMP signaling through stimulation of soluble guanylyl cyclase (3). cGMP mediates browning of WAT (9), suggesting that cGMP signaling is a potential mechanism underlying nitrate-induced browning. Therefore, we measured the concentration of cGMP in nitrate-treated adipocytes using LC-MS (Fig. 5A). Nitrate increased the concentration of cGMP in adipocytes from 2.7 pmol/10⁶ cells in the control to 5.2 pmol/10⁶ cells and 4.9 pmol/10⁶ cells in the 50 and 500 μmol/L NaNO₃-treated cells, respectively (P = 0.04, control vs. 50 μmol/L NaNO₃ P ≤ 0.05, control vs. 500 μmol/L NaNO₃ P ≤ 0.05) (Fig. 5B). Similarly, analysis of WAT from rats demonstrated an increase in cGMP concentrations in vivo following nitrate treatment (P < 0.05, control vs. 0.35 mmol/L NaNO₃ P ≤ 0.01, control vs. 0.7 mmol/L NaNO₃ P ≤ 0.05, control vs. 1.4 mmol/L NaNO₃ P = 0.07) (Fig. 5C).

The pharmacological inhibitor of guanylyl cyclase ODQ was used to confirm a cGMP-dependent mechanism for nitrate-induced browning. Inhibition of guanylyl cyclase abolished the nitrate-induced expression of brown adipocyte–specific genes in white adipocytes (control vs. 500 μmol/L NaNO₃ P ≤ 0.0001, 500 μmol/L NaNO₃ vs. 500 μmol/L NaNO₃ + 1 μmol/L ODQ P ≤ 0.001) (Fig. 5D).

The cGMP-dependent protein kinase G (PKG) is expressed in the WAT of rodents and is a key regulator of adipocyte function (9,32). Using a pharmacological inhibitor of PKG (KT5823), we investigated the role of this protein kinase in the nitrate-induced browning response. Inhibition of PKG abrogated the nitrate-induced expression of brown adipocyte–specific genes in the adipocytes (control vs. 500 μmol/L NaNO₃ P ≤ 0.0001, 500 μmol/L NaNO₃ vs. 500 μmol/L NaNO₃ + 1 μmol/L KT5823 P ≤ 0.0001) (Fig. 5E). Therefore, the mechanism underlying nitrate-induced browning of WAT functions through the cGMP/PKG signaling axis.

Unlike the l-arginine-NOS-NO pathway, which depends on oxygen, production of NO through the nitrate-nitrite-NO pathway is augmented as oxygen concentrations decrease (33) and may regulate tissue adaptation to hypoxia (34,35). Because the nitrate-induced browning of WAT is mediated through the nitrate-nitrite-NO pathway, we examined the capacity of nitrate to increase brown adipocyte–specific gene expression in adipocytes during hypoxia.

Primary adipocytes were treated with NaNO₃ (50 μmol/L) in low-oxygen conditions (2% O₂). These oxygen concentrations correspond to levels observed in obese WAT (36). In parallel, adipocytes from the same stromal vascular isolation were treated with NaNO₃ (50 μmol/L) under normoxic conditions. Because nitrate-induced browning requires the reduction of nitrate to NO and XOR reduces nitrate to NO (4), we assessed the expression of XOR in the adipocytes in response to changes in oxygen availability. XOR expression was found to increase in adipocytes exposed to hypoxia, an effect enhanced by simultaneous nitrate treatment (Fig. 6A) (normoxia control vs. hypoxia control P ≤ 0.01, hypoxia control vs. hypoxia 50 μmol/L NaNO₃ P ≤ 0.05, normoxia 50 μmol/L NaNO₃ vs. hypoxia 50 μmol/L NaNO₃ P ≤ 0.001). Nitrate also increased the expression of brown adipocyte–specific genes in both the normoxia- and the hypoxia-conditioned cells (normoxia control vs. normoxia 50 μmol/L NaNO₃ P ≤ 0.0001, hypoxia control vs. hypoxia 50 μmol/L NaNO₃ P ≤ 0.0001) (Supplementary Fig. 8A). Hypoxia per se significantly decreased the expression of the brown adipocyte–specific genes compared with normoxia (normoxia control vs. hypoxia control P ≤ 0.01). However, nitrate increased the expression of the brown adipocyte–specific genes and fully restored the expression of UCP-1, CIDEA, and CPT1 under hypoxia to levels greater than that of normoxic controls. The decrease in expression of the genes due to hypoxia alone was corrected for by normalizing the hypoxia-conditioned cells to hypoxia control to facilitate comparison of the fold changes in expression of the browning genes following nitrate treatment (Fig. 6B). The fold-change increase of brown adipocyte–specific gene expression in hypoxic adipocytes treated with nitrate compared with hypoxic control was significantly greater than that observed in normoxic cells (normoxia 50 μmol/L NaNO₃ vs. hypoxia 50 μmol/L NaNO₃ P ≤ 0.0001).

To investigate the enhanced browning effect of nitrate in WAT under hypoxic conditions in vivo, rats were housed in a hypoxia chamber and either received 0.7 mmol/L NaCl or 0.7 mmol/L NaNO₃ in their drinking water for 14 days. A parallel group of rats were identically treated but housed in a normoxic environment. Water and food intake was not significantly different between the groups (Supplementary Table 2). As in vitro, XOR expression was increased in WAT of rats exposed to hypoxia and further enhanced by nitrate treatment (Fig. 6C) (normoxia control vs. hypoxia control P ≤ 0.01, normoxia 50 μmol/L NaNO₃ vs. hypoxia 50 μmol/L NaNO₃ P ≤ 0.01). As expected, nitrate increased expression of brown adipose–specific genes in WAT during normoxia (normoxia control vs. normoxia 0.7 mmol/L NaNO₃ P ≤ 0.01) (Fig. 6D). Brown adipocyte–specific gene expression was also increased in the WAT of rats treated with nitrate under hypoxic conditions (hypoxia control vs. hypoxia 0.7 mmol/L NaNO₃ P < 0.0001). Of note, brown adipocyte–specific gene expression within WAT was significantly increased in the rats treated with nitrate during hypoxia compared with those treated with nitrate and housed in a normoxic environment (normoxia 0.7 mmol/L NaNO₃ vs. hypoxia 0.7 mmol/L NaNO₃ P < 0.0001). The expression of ADIPOQ was not significantly affected by either nitrate or hypoxia (Supplementary Fig. 8B). Overall, these data indicate that the nitrate-induced expression of brown adipocyte–specific genes in WAT is augmented in hypoxia.

The WAT of obese humans and rodents is in a hypoxic state that perturbs metabolism, increasing glycolysis and de novo lipogenesis and decreasing fatty acid breakdown, further contributing to the pathology of obesity (36–39). Therefore, we investigated the effect of dietary nitrate on the expression of brown adipocyte–specific genes in WAT in a rodent model of obesity, the ob/ob mouse. Wild-type C57BL/6 and ob/ob mice received either 0.7 mmol/L NaCl or 0.7 mmol/L NaNO₃ in their drinking water for 8 weeks. Plasma nitrate concentrations were elevated from 59.8 ± 4 and 53.8 ± 11 μmol/L in the chloride-treated C57BL/6 and ob/ob mice, respectively, to 373.1 ± 73 and 323.9 ± 58 μmol/L in their nitrate-treated counterparts. The expression of several brown adipocyte–specific genes in the WAT of ob/ob mice was significantly reduced compared with wild-type mice. Nitrate increased the expression of the brown adipocyte–specific genes, partially restoring the impaired levels of expression in this model of obesity (Fig. 6E).

Nitrate was considered a nonbioactive metabolite of NO and a potentially toxic dietary constituent. However, studies showing that nitrate reduces blood pressure (5,7,40) and the oxygen demand of exercise (41) indicated that this anion may be beneficial for metabolic health. Complementary studies demonstrated that nitrate has antiobesity and antidiabetic effects, independent of increased mitochondrial biogenesis or PGC-1α expression in liver or muscle, in endothelial NOS-deficient mice, a strain prone to a metabolic syndrome–like phenotype (6). Similar antidiabetic effects of nitrate have since been described in Sprague-Dawley rats (42). Diets low in nitrate and nitrite reduce the concentration of cGMP in certain tissues (8). In recent years, a role for cGMP signaling in systemic energy balance has emerged (32,43,44). cGMP was recently demonstrated to induce browning within WAT (9). These beige/brite cells (14,15) have cardiometabolic protective effects in rodents (11,16,18,19,22,23,45). Thus, we hypothesized that nitrate contributes to the improved metabolic phenotype by inducing the browning response in WAT.

In the current study, we demonstrate that nitrate increases the expression of thermogenic genes in BAT and activates a brown adipocyte–like transcriptional and functional phenotype within WAT. Furthermore, we find that nitrate induces the browning of adipocytes through the NOS-independent nitrate-nitrite-NO pathway. Thus, nitrate activates the thermogenic process in a manner distinct from physiological metabolite activators of browning previously described (Supplementary Fig. 9) (20). The present findings suggest a mechanism by which nitrate is reduced to NO, which in turn increases cGMP production through soluble guanylyl cyclase activation. Increased cGMP concentrations activate PKG, increasing the expression of PGC-1α and other key browning genes such as UCP-1 and CIDEA in white adipocytes. Increased expression of brown adipocyte–specific and β-oxidation pathway genes translates to a brown adipocyte–like functional phenotype characterized by increased fatty acid β-oxidation (Fig. 7). Incidentally, the mechanism for natriuretic peptide–induced browning of WAT was also revealed to signal through the cGMP cascade (43), underscoring the importance of this signaling pathway for the physiological activation of thermogenesis in WAT.

We observed that the nitrate-induced browning response in WAT was enhanced in hypoxia. The nitrate-nitrite-NO pathway is significantly augmented during hypoxia, with both the activity and expression of XOR increased in low-oxygen conditions (4). The pathway has also been implicated in the adaptive response of some tissues to hypoxia (30,33,35). It is conceivable that production of NO from nitrate leads to the enhanced nitrate-induced browning of WAT observed in hypoxia. The WAT of obese humans and in genetic and dietary models of obesity in rodents is in a hypoxic state (36,37). Exposure of adipocytes to hypoxia leads to cellular metabolic reprogramming, consisting of increased glycolysis and fatty acid and TAG synthesis and decreased fatty acid catabolism, that contributes to the pathology of obesity (38,39). The identification of an enhanced capacity for nitrate to induce the browning response of WAT in low-oxygen conditions through NO, which will likely also contribute to improved vascular health in obese individuals where NO production is decreased (46), may represent a significant therapeutic opportunity to partly reverse this hypoxia-mediated pathological state. Indeed, the current studies show that the impaired expression of several brown adipocyte–specific genes in the WAT of the ob/ob mouse model of obesity is partially reversed by nitrate treatment.

From a nutritional perspective, humans are exposed to nitrate as part of their daily diet, with green leafy vegetables representing a significant source of the anion. A high vegetable component to diets is consistently shown to play a protective role in the development of metabolic morbidity (47,48). The high nitrate content of vegetables has been suggested to be partly responsible for this association with cardiometabolic protection (30,34). It is worth speculating that increasing the flux through the nitrate-nitrite-NO pathway may be one mechanism through which dietary vegetable consumption confers the associated metabolic protection. Given that the concentrations of nitrate used in the current study are readily achievable through dietary vegetable consumption (6,34), dietary nitrate may also activate cGMP signaling in human WAT. The possibility that this ubiquitous dietary constituent could induce browning of WAT in humans is tantalizing.

It should be noted that the responses of WAT to nitrate were not universally dose dependent. This observation may represent rapid tachyphylaxis and/or resistance to inorganic nitrate, a phenomenon well characterized in the clinical use of organic nitrates, particularly at high doses. Alternatively, cellular uptake of inorganic nitrate may become rate limiting or saturated at high doses. The NO signaling pathway is also subject to counterregulation through a number of disparate mechanisms that include the action of phosphodiesterases (PDEs), which may influence the effect of nitrate in WAT (Supplementary Fig. 10). PDEs such as PDE5 control cGMP concentrations and, therefore, inhibit downstream effects such as PKG stimulation. Indeed, inhibition of PDE5 is known to stimulate the browning of WAT (9).

In summary, we identify nitrate as a novel non-β-adrenergic activator of the browning response in WAT. Furthermore, we highlight that this small anion is a potential dietary mediator of protection from and potential therapeutic modality for the treatment of metabolic disease.

Acknowledgments. The authors thank Steve Jackson and Julia Coates (Wellcome Trust/Cancer Research UK Gurdon Institute, University of Cambridge, U.K.) for the use of a low-oxygen incubator.

Funding. L.D.R. is supported by the Medical Research Councilhttp://dx.doi.org/10.13039/501100000265 Human Nutrition Research Elsie Widdowson Fellowship. This work was supported by grants from the Biotechnology and Biological Sciences Research Councilhttp://dx.doi.org/10.13039/501100000268 (bb/H013539/2, bb/I000933/I), the British Heart Foundation, and the Medical Research Council (Lipid Profiling and Signalling Programme grant number UD99999906).

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

Author Contributions. L.D.R. contributed to the study concept; experimental design; cell culture, mass spectrometry, and isotope labeling experiments; statistical analysis; figure preparation; and writing of the manuscript with input from all coauthors. T.A. and A.O.K. contributed to the concept, design, and performance of the animal studies. S.A.M. performed the cell culture and GC-MS experiments. B.O.F. and M.F. performed the nitrate measurements and contributed to the experimental design. A.J.M. contributed to the concept and design of the animal studies. J.L.G. contributed to the study concept and design. L.D.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.