In the United States and beyond, obesity has reached a remarkable prevalence. A total of 42% of US adults are obese [1
], and an estimated 88% of adults are considered metabolically unhealthy [2
]. Understandably, the explosion in obesity, and our ongoing failure to broadly address it, has given rise to intensive efforts to better understand adipose tissue physiology. These efforts have revealed, among other things, that humans store fat in two distinct depots that differ by function: white adipose tissue (WAT) and brown adipose tissue (BAT). Whereas WAT primarily acts to store energy for subsequent use with a very low energy expenditure and low mitochondrial content, BAT is enriched with very active mitochondria that manifest a surprisingly high energy expenditure, serving a thermogenic role.
In higher vertebrates, WAT acts primarily for lipid storage, centralized in unilocular lipid droplets within adipocytes that are catabolized and released as fatty acids when necessary. However, obesity is typified by excess WAT accumulation, which poses a meaningful risk for developing insulin resistance and type 2 diabetes mellitus [3
], cardiovascular disease [5
], and some cancers [6
]. In stark contrast, BAT has a limited lipid storage, with multi-locular lipid droplets co-mingled with a substantial population of mitochondria (giving rise to the distinct reddish-brown color). Increased BAT mass and activity are associated with a resistance to obesity and related disorders [7
]. Perhaps the most unique aspect of BAT is the nature of its mitochondria—they express a proteome that fosters energy wasting by uncoupling oxidative phosphorylation from electron transport. In contrast, the mitochondria in WAT are very tightly coupled, which is evidence of a more efficient energy storage profile [9
WAT and BAT rapidly undergo adaptive and dynamic changes in response to both energy and temperature changes [10
]. The initial changes within the first 24 h may only involve an altered expression of proteins. However, after 2–3 days, dietary and environmental stimuli can induce marked tissue remodeling, which results in altered adipose tissue morphology and possibly also modified functional properties [11
]. In particular, WAT is able to manifest the characteristics of BAT, including mitochondrial biogenesis and uncoupling, a process known as “beiging” [12
]. Insofar as obesity may partly derive from reduced BAT activity [1
], inducing its activity or pushing WAT to manifest BAT characteristics could add a new approach to combating the condition.
While ketones such as acetoacetate and its more prevalent metabolite β-hydroxybutyrate (βHB) are known fuel sources for all cells with mitochondria [14
], recent work has detailed their roles as signaling molecules, altering inflammation [15
], cognition [16
], oxidative stress [19
], and more [20
]. Moreover, ketones are a critical intermediate between adipose tissue and brain energy fuel supply, allowing the brain to meet energetic requirements during glucose restriction and starvation [22
]. Furthermore, βHB acts as a signal to directly regulate the metabolism and maintain energy homeostasis during nutrient deprivation [25
Perhaps due to both their energy potential and effects on cellular signaling, ketones appear to facilitate a more favorable metabolic milieu that should be considered a tool in our collective efforts to address human obesity [14
]. Thus, with the complex mechanistic regulation of adipose physiology in mind, and the role of ketones as metabolic intermediates and signaling molecules, we aimed to explore the effect of the prominent ketone βHB on adipose tissue mitochondrial bioenergetics.
Across cell, rodent, and human models, the central conclusion from our studies is that ketones stimulate mitochondrial uncoupling in subcutaneous adipose tissue. As with all nutrients, particularly carbohydrate and fat, ketones are both a source of energy and a signaling molecule. However, ketones appear to elicit a unique effect with regard to energy use in adipocytes. Whereas carbohydrate and fat, independent of their caloric value, activate processes to store energy in adipocytes, such as ChREBP [27
] and PPARγ [28
], respectively, ketones appear to activate processes that spend and even waste energy via mitochondrial uncoupling [29
Our findings of increased mitochondrial uncoupling (i.e., reduced ATP:O2
) in adipose in response to ketone exposure corroborates previous work from the Veech lab, which identified an effect of ketones enhancing the native mitochondrial uncoupling within brown adipose tissue (BAT) [29
]. BAT is found in limited amounts in adult humans, though it can nevertheless be activated via cold exposure and exercise and offer some protection against obesity [30
]. Far more abundant than BAT is subcutaneous adipose (SAT), and while generally more metabolically inert, SAT has the capacity to both induce mitochondrial biogenesis and uncoupling, making it characteristically more similar to BAT [32
]. Though we were unable to quantify UCP1 from our human SAT samples, our evidence of reduced ATP production compared to O2
consumption indicates a greater degree of mitochondrial uncoupling in human SAT when in ketosis. A limitation to this conclusion with the human tissue is the method we used to quantify ATP. Whereas ATP was quantified from cultured adipocytes and rodent adipose without permeabilization, due to constraints on human adipose sample mass, we measured the ATP following permeabilization, which could lead to artificially lower ATP levels due to potential loss through the permeabilized cell membrane. However, we found very little difference in ATP levels in cultured adipocytes and rodent tissue before and after permeabilization, which lends confidence to our findings.
Other studies that have measured mitochondrial coupling have generally utilized one of two methods, either singularly measuring mitochondrial respiration or a combination of respiration with ATP quantification. The former relies on the ATP synthase inhibitor oligomycin. We elected the latter option, as others have done [5
], which we consider a more explicit indicator of the degree of synchronization between electron transport (i.e., respiration) and ATP generation (i.e., oxidative phosphorylation). Because we sought to make a direct comparison of oxygen use to ATP generation, and did not base our findings on respiration alone, we did not use oligomycin in our respiration protocols.
Visceral and subcutaneous WAT both contain the potential towards increased mitochondrial uncoupling, though each appears to respond to unique stimuli. While SAT responds to nitric oxide [7
], visceral adipose responds to catecholamines [8
]. Based on our rodent work, ketones appear to be a unique stimulus to subcutaneous fat depots, including inguinal and interscapular adipose, but not visceral fat. At this point, we can only speculate on the disparate responses. This may be a result of the varying G protein-coupled receptor (GPCR) expression in adipose tissue depots [9
]. GPCRs both respond to βHB [10
] and stimulate mitochondrial uncoupling [11
], making their differential adipose expression a possible explanation for our findings.
When comparing the subcutaneous fat depots across humans and rats, the effect was similar, though the magnitude was greater in human adipose. This could be due to several reasons, including temperature variations (i.e., humans were likely occasionally exposed to colder temperatures) and inter-species differences. Namely, humans appear to have a more robust reliance on ketones for development, and even more rapidly and easily reach ketosis with diet [12
The observation of significantly increased mitochondrial respiration in SAT in response to ketone exposure may play some part in the increased whole-body energy expenditure that has been seen in humans in ketosis. Most notably and recently, Ebbeling et al. [33
] utilized doubly labeled water to determine the energy expenditure in free-living humans. By rotating study subjects across three diets varying in carbohydrates and fats, they found that energy expenditure was the greatest during adherence to the diet lowest in carbohydrates and highest in fat, a macronutrient profile that typifies a ketogenic diet. Furthermore, by scrutinizing energy expenditure in a metabolic ward, Hall et al. [34
] found a similar result—energy expenditure was significantly elevated when the subjects followed a ketogenic diet.
Type 1 diabetes presents an additional and relevant context to our findings. When untreated, the absence of insulin in type 1 diabetes results in unchecked lipolysis and ketogenesis, driving ketones to dangerously high levels, surpassing the plasma buffer capacity and impacting pH. Often, the weight loss of untreated type 1 diabetes is attributed to glucosuria—i.e., the loss of calories from excreted glucose causes weight loss. However, changes in urinary glucose excretion fail to correlate with the weight changes in type 1 diabetes [14
]. Thus, to fully explain the dramatic weight changes in states of treated vs. untreated type 1 diabetes, additional explanation is needed. While renal glucose excretion certainly accounts for a loss of some glucose, and therefore a reduction in available calories, it does not explain the demonstrable differences in energy expenditure with the disease. Over 100 years ago, Joslin and Benedict noted that the metabolic rate in untreated insulin-deficient individuals with type 1 diabetes was roughly 15% higher compared with similar body weight subjects without type 1 diabetes [35
]. Remarkably similar findings were observed decades later by Nair et al. [36
]; energy expenditure was 20% higher than predicted in the absence of insulin therapy and, with the initiation of insulin, rapidly slowed to the predicted values.
While insulin itself was the focus of the aforementioned studies into energy expenditure in type 1 diabetes, one cannot but wonder at the relevance of ketones; in other words, might some of the increased energy expenditure in states of low or deficient insulin be the result of ketone-induced mitochondrial uncoupling? Indeed, a confounding variable and essential consideration underlying our data and other studies that explore the adipocyte-specific, as well as whole-body, metabolic effects of ketones on energy expenditure and mitochondrial physiology is the hormone insulin. Insulin has direct and powerful control over ketogenesis. Briefly, a relative reduction in insulin disinhibits both adipocyte lipolysis and hepatic ketogenesis, thereby providing both substrate and stimulus for ketone production. Additionally, insulin dampens adipocyte mitochondrial uncoupling and, by extension, energy expenditure. We recently found that long-term insulin therapy to induce hyperinsulinemia in rodents was sufficient to reduce the whole-body energy expenditure [9
]. Moreover, insulin inhibited mitochondrial uncoupling, resulting in more tightly coupled mitochondria in both BAT and SAT. Our future research efforts will explore the effect of elevated ketones alone, without the need for low basal insulin, on adipocyte mitochondrial uncoupling in vivo by using exogenous ketones.
These results are thought-provoking when viewed through the lens of starvation. Food restriction is the most rapid stimulus for ketogenesis. Of course, in such a state of energy deprivation, it is difficult to imagine the body wasting energy through ketone-induced adipocyte mitochondrial uncoupling. Interestingly, short-term starvation (i.e., fasting) paradoxically increases energy expenditure. Zauner et al. [37
] found that energy expenditure increased throughout the first two days of a four-day fast and remained elevated until the end. Coincidentally, plasma ketones followed a similar trend. However, the phenomenon has a limit—long-term starvation, such as that seen with anorexia nervosa, is associated with a reduction in the metabolic rate [38
]. There are certainly other factors that mediate some of this effect, such as reduced muscle mass, but a relative insufficiency of adipose, and the lipid substrate for ketogenesis, may also be relevant.
This work adds a novel dimension to our previous findings regarding insulin’s effects on the adipocyte mitochondrial function [15
]. Taken together, our observations that insulin enhances adipocyte mitochondrial coupling, while ketones drive uncoupling, may provide insight into obesity etiology. Specifically, these findings represent a unifying theory of two origins of obesity—calories and hormones. Elevated insulin and, thus, reduced ketones are an essential feature of human obesity [16
]. In addition to inhibiting adipose lipolysis [20
], insulin promotes both lipogenesis and adipogenesis in adipose tissue [21
]. It is tempting to speculate that the contrasting changes in adipocyte mitochondrial respiration rates in states of varied insulin, with corresponding varied ketones, further contribute to adipocyte energy use and size and, by extension, obesity progression or regression.