Calorie restriction remains the most highly researched, nongenetic intervention to improve health and increase life span in research organisms ranging from single-celled yeasts to nonhuman primate models.31,32
These health and longevity benefits are proportional to the amount of restriction up to the point of malnutrition33
and are generally independent of the macronutrient content being restricted (fat versus carbohydrates versus protein). The data, considered as a whole, suggest a clear, causal relationship between energy provision and mortality rate, as well as senescence and metabolic-related disease.
The mechanisms of action for these observations remain unclear. One consistent phenotype of CR across animal models is the reduction of body weight and particularly body fat.32
In longevity studies using rodents, many ad lib-fed laboratory rodents develop age-associated obesity, even when fed a presumably healthy, low-fat diet. When adding a high-fat diet, most rodents exhibit a diet-induced obesity response similar to what is expected of humans who over-consume a calorie-rich, high-fat diet. With the switch from an ad lib diet to a restricted feeding paradigm, CR induces a rapid and sustained weight loss associated with the caloric deficit. This “negative energy balance” phase is followed by the establishment of a new equilibrium in which the reduced body weights are matched to the energy provision, i.e., relative energy balance. Therefore, the body weight and body composition changes associated with CR are more long term in nature than is the temporary energy deficit that is experienced during the initiation of CR. Thus, one could posit that the lower body weight or fat induced by CR partially mediates the effect of CR on life span. We tested this in a large sample of Wistar rats (N
The relative contribution of body weight to the CR effect was approximately 11%, thus supporting this hypothesis of partial mediation by low body weight. In a related study, we tested the influence of intentional weight loss by CR on mortality rates in outbred rats after the establishment of obesity. CR resulted in increased longevity compared with that in animals that remained obese, which suggests that weight loss can increase longevity even after the onset of obesity.33
In addition to the rodent model, two long-term, randomized CR studies in nonhuman primates are providing the first evidence that age-related mortality may be significantly lower in CR animals, concomitant with significantly reduced metabolic-related disease.35
Whether similar disease prevention and maximal longevity extension will be present in the CR monkeys is not yet established, but additional interim results for mortality are expected in the near term (1–3 years).
CR and longevity: is perceived energetic uncertainty (independent of immediate energy intake) a mechanism of causation?
One aspect of the CR paradigm is feeding less than would be voluntarily consumed under similar but ad lib conditions. This results in animals that are both acutely and chronically hungry,36,37
which induces a gorging behavior by animals that are normally restricted once food is presented. Whereas ad lib fed animals have constant access to adequate food supplies, CR animals will often consume the majority of their daily food allotment shortly after it is given, resulting in extended periods of fasting during which food is neither available nor perceived by the senses. This means that not only are CR mice energetically restricted, but a host of neuroendocrine signals that coordinate energy intake and expenditure are presumably altered for most of the animal's life span. Because this may be perceived as a stressful situation, it raises the possibility that a small amount of something that is harmful in large doses may in fact be beneficial, i.e., the idea of hormesis.38–40
The idea of hormesis is often discussed in toxicology, where complex J- or inverted U-shaped curves—as opposed to a monotonic response—are observed with a dose-response pattern (i.e., just the right amount is beneficial, whereas either too little or too much is harmful). Whereas homeostasis is the process by which bodily functions are maintained, allostasis is the process by which bodily functions change in response to environmental challenges.41
It is plausible that the perception of energetic uncertainty is a hormetic signal leading to allostatic adaptive responding that both increases fat deposition when sufficient metabolizable energy is available and leads to increased life span if conditions are otherwise permissive.
Regarding CR and hunger, it seems reasonable that organisms must judge the nutrient and energetic state of their environment on the basis of specific cues. If, as in the case of CR, those cues are altered owing to impaired access to food supplies the majority of the time, might this elicit a protective response to promote the maintenance and preservation of the organism until a more favorable nutritional environment is encountered? Although one might expect these cues to be external from the environment, it is possible that internal cues like fat stores might also be coordinated in the response to balance the need versus availability equation of energy perception. Related to this idea, visceral adipose tissue is associated with increased morbidity and mortality and is proposed to play a role in multiple metabolic-related diseases. Importantly, visceral adipose depots are sensitive to energetic needs and are reduced in negative energy balance, that is, the CR state. By surgically removing part of the visceral fat in rats, significant health and longevity benefits have been achieved independent of CR, which suggests a role for fat and particularly visceral adipose tissue in mediating health and longevity.42
Another possible interpretation would suggest that visceral adipose tissue reduction by mechanical means removes a signal of surplus energy stores, which results in a perceived energetically “lean” time and contributes to the health and longevity benefit. Whether such a signal exists and is directly related to the specific fat depots themselves is unknown.
Perception of nutrient availability has also been studied in worms and flies in relationship to survival and mortality kinetics. Drosophila melanogaster
has been used extensively in nutrition and aging research partly because of its relatively short life span and the ease with which researchers can produce and modify diets of known composition. Although it was previously known that CR increases life span in flies, recent studies have shown that mortality is acutely sensitive to nutrient availability.43
This was modeled by switching flies from dietary restriction (DR) to ad lib feeding (or vice versa—ad lib to DR), which meant switching them from being housed with standard, rich media access (ad lib) to a dilute media composition. Within two days of switching the flies to the opposite nutrient state, mortality rates paralleled those of animals continuously exposed to the nutrient condition.43
Although it is possible that dietary factors and intake amounts could produce a biological caloric effect resulting in survival modulations, the acute response suggests that additional factors such as nutrient or environmental perception were also involved beyond the proposed mechanisms of more long-term aging phenotypes such as accumulated cellular damage.43
This idea is further supported by results showing that the extended life span of DR flies could be shortened by sensing (e.g., smelling) live yeast that were present but not available for consumption.44
Importantly, the perception of live yeast did not shorten life span of ad lib fed flies, which suggests that the effect was not generally negative (like a toxic effect), but rather unique to the DR response. Further studies showed that disruption of olfactory receptor modulator (Or83b) alone was sufficient to increase life span and induce a number of stress-resistant phenotypes, thus supporting the role of perception in mediating multiple aspects of the DR related response.45
In agreement with other work in C. elegans
these studies demonstrate a critical role of sensory perception in the longevity response to DR and point to interactions and modulation of perception on complex phenotypes such as life span.
The hunger aspect of hormesis is further supported by work using intermittent feeding (IF) or every other day (EOD) feeding in rodents. Rather than daily restriction, the EOD or IF models permit ad lib feeding, but only for specific periods of time followed by complete fasting (e.g., one-day fed ad lib, one-day fasted, one-day fed ad lib, etc.). Because food intake during the ad lib period may not fully compensate for the fasting period, EOD feeding may result in mild energy restriction. Even in rodent strains that do not lose weight with EOD feeding, life span is increased when the protocol is started at young age.48,49
Comparing strains of rodents for the body weight and longevity response, it appears that life span is not fully predicted from changes in body weight. This suggests that other factors, possibly daily hunger, may be contributing to the benefit.
If the perception of restriction were beneficial, would repeated bouts of weight loss be beneficial?
As mentioned previously, CR is known to have positive effects on health and longevity, whereas obesity and excess adiposity are detrimental. What would happen if all individuals who were overweight or obese lost weight to a normal body size? Although this is not likely to occur, most of those who successfully lost the weight would regain the amount within a few years.50
If, as is commonly believed, excess weight and adiposity are harmful, would it be beneficial (or harmful or have no effect) to go from overweight to normal and back again? Yo-yo dieting and weight loss is certainly observed in humans, and although not normally encountered in laboratory nutrition studies, this question of the potential health consequences with weight cycling from repeated bouts of weight loss and regain is beginning to be addressed. A study is currently underway at the University of Alabama at Birmingham in which high-fat diet feeding (similar to the Western diet macronutrient composition) is utilized to establish an overweight/obese cohort of mice. The mice are subsequently randomized to interventions in which food intake is continued ad lib, restricted sufficiently to achieve a normal body weight, or food intake is restricted and then refed similar to a yo-yo dieting experience. With the CR phase, animals clearly lose weight, and when released to ad lib feeding, usually return to approximately their pre–weight loss size or greater. Thus, not only is this study a model of CR for intentional weight loss (and regain with yo-yo dieting), these animals experience varied amounts and time periods of imposed hunger by CR. It may be that treating overweight may be beneficial to do at regular intervals, even with modest results, similar to the way that oral health benefits from regular plaque removal (i.e., a “dental model of obesity treatment”). Although a person may not be able or motivated to prevent the buildup of plaque at the gum line (which is not viewed as a treatment failure), frequent plaque removal efforts by dental hygienists reduce the long-term, health-damaging effects.
Other evidence for the stress/hormesis hypothesis
Stress comes in a variety of forms and, as proposed in this work, does not necessarily require a specific physical or biological insult but rather may arise from the organism's perception of its relative state or environment. Multiple stressors have been discussed in association with life span modulation, including environmental temperature and osmotic stress.39,51
The extent to which these survival-altering interventions operate through overlapping pathways and mechanisms that are critical for life span extension and that are mediated by a stress response remains to be demonstrated.
What might this imply about calorie restriction mimetics?
Considering that the perception of nutrients through hunger or other pathways may mediate the effect of CR, might it be possible to illicit a similar physiologic response without requiring actual CR? This concept of mimicking the benefits of CR without actual caloric reduction was first proposed by Ingram et al.
, in part, on the basis of their work with multiple compounds that would interact with metabolic pathways implicated in CR.52–54
Although multiple compounds have been identified that induce a similar physiologic and transcriptional response to CR, few compounds have produced increased longevity in animal studies. Two of the earliest proposed CR mimetics were metformin and 2-deoxyglucose (2DG). Multiple rodent longevity studies are reported for metformin supplementation,55–57
and there is one recent longevity report for 2DG.58
One important aspect of these CR mimetics is how comprehensively the mimetic recapitulates the total CR response. Regarding the hunger associated with CR, there is little reason to suspect that metformin would elicit a hunger response. Thus, any hunger-associated hormetic response would be lacking. Similarly, with 2DG, no increase in food intake was observed,58
which suggests a lack of overt hunger. The clear toxic effects of high 2DG make dosing a challenge; therefore, what dose of either compound, if any, could elicit a hunger response similar to CR is unclear. Whether additional CR mimetics should require demonstrated or reported hunger as part of the response is an open question. Current collaborations with the Interventions Testing Program of the National Institute on Aging testing an α-glucosidase inhibitor (acarbose) may provide useful information on a hunger-hormetic mechanism of action. Acarbose is expected to induce a hunger response in addition to metabolic and physiologic improvements. A related approach is being considered using ghrelin, a gut-derived peptide associated with feeding and hunger that could induce a chronic, artificial hunger state despite access to energetic requirements sufficient to meet the metabolic demands of the organism.
Implications if these hypotheses are true
We offer the diagram on the previous page () as an integrated model of the observations and effects we have discussed. Using this model, future efforts to identify leverage points that affect perceptions of energetic uncertainty may be as important as or more important than identifying specific economic leverage points. Furthermore, environmental manipulations that lead to the perception of restricted food availability may have paradoxical effects. On a physiological level beyond conscious perceptions, some CR mimetics may need to work downstream of hunger signals to be effective. Because of the apparent cascade of processes in CR, even discounting side-effects, losing weight with an anorexigen may be less beneficial than losing an equivalent amount of weight with CR. Some CR-like interventions that do not actually reduce total energy intake may also lead to prolonged life. Our proposed model provides a testable framework in which future investigations might illuminate mechanistic pathways.
Hypothesized model of social and environmental influences on energy balance that affect aging.