The hidden costs of dietary restriction: Implications for its evolutionary and mechanistic origins

Abstract

Dietary restriction (DR) extends life span across taxa. Despite considerable research, universal mechanisms of DR have not been identified, limiting its translational potential. Guided by the conviction that DR evolved as an adaptive, pro-longevity physiological response to food scarcity, biomedical science has interpreted DR as an activator of pro-longevity molecular pathways. Current evolutionary theory predicts that organisms invest in their soma during DR, and thus when resource availability improves, should outcompete rich-fed controls in survival and/or reproduction. Testing this prediction in Drosophila melanogaster (N > 66,000 across 11 genotypes), our experiments revealed substantial, unexpected mortality costs when flies returned to a rich diet following DR. The physiological effects of DR should therefore not be interpreted as intrinsically pro-longevity, acting via somatic maintenance. We suggest DR could alternatively be considered an escape from costs incurred under nutrient-rich conditions, in addition to costs associated with DR.

Fig. 1. Schematic of the evolutionary model of DR.

Resource availability is varied from left to right, from very low (where starvation would occur) to very high (where maximum reproduction would occur). The theoretical optimal allocation to somatic maintenance (pink) versus reproduction (yellow) is depicted at a given resource availability. When resource availability decreases, investment in both somatic maintenance and reproduction is reduced until a threshold is met. Below this point, resources are so scarce that investment in reproduction does not yield a fitness return. This could occur when offspring produced cannot recruit into the population due to the harsh resource environment, or because the capital (start-up) costs of breeding cannot be met. Here, investment in reproduction is lost and is wholly allocated to somatic maintenance. It is this evolved resource allocation decision to invest into somatic maintenance under DR conditions that is thought to underlie life-span extension under DR.

Fig. 2. The effect of different dietary regimes on age-specific mortality risk in DGRP-195.

Age-specific mortality risk (A to F) allows an investigation of instantaneous changes in mortality risk upon dietary switches (points) across the different dietary regimes used. Mortality risk at continuous rich (solid red) and restricted diets (dash black) are plotted as lines. The exacerbation of mortality due to switch phenotypes is the difference between mortality at continuous rich diet (red line) and mortality of switch treatment when on a rich diet (red points). The open dots in the switch treatment, the DR condition, should overlay the continuous DR treatment (dash) if the dietary switch does not modulate the effect of the DR diet [or act as a pure control before a single switch, as in (A)]. N = 19,086 females total; 995 to 3769 per treatment. (A) Long-switch. When returning to a rich diet after a long period of DR, mortality is exacerbated compared with flies fed a rich diet continuously. (B) Four-day switch. Switching from a DR to a rich diet repeatedly every 4 days increases mortality on rich diets compared with continuously rich-fed flies. Flies are still able to modulate their mortality in response to DR even when diet fluctuates rapidly. (C) Two-day switch. Mortality on rich diets is only mildly increased and flies still respond to DR even when it is only imposed for 2 days. (D) Short reverse-switch. After a long period on a rich diet, DR for 4 days returns flies to mortality of continuous DR. The x axis of (D) is age adjusted to correct for age differences (1 to 3 days) at the time of the diet switch for illustration purposes only. (E) Four-day DR, 2-day rich switch (4-to-2–day switch). Flies respond to DR but encounter a slightly blunted effect compared with continuous DR. (F) Four-day rich, 2-day DR switch (4-to-2–day switch). The effect of DR is reduced when imposed for 2 days following 4 days on a rich diet. (G) Survival plot of (A) to (C) with associated continuous diet controls. Total survival of both the 4-day switching dietary regime and the long-switch is lowered compared with continuously rich diets, despite flies spending a considerable extent of their lives on restricted diets. Flies on DR outlive all other categories. (H) Survival plot of (E)/(F) with associated continuous diet controls. Despite spending up to two-thirds of their lives on DR in these asymmetrical regimes, survival benefits are modest, compared with continuous DR. Dietary switch treatments contain daily time points (dots) for the dietary switch treatments, as treatments were mirrored and balanced, with half of flies starting on DR and half on rich diets.

Fig. 3. Long-switch treatment in a panel of 11 DGRP genotypes.

(A) 195, (B) 105, (C) 217, (D) 441, (E) 705, (F) 707, (G) 136, (H) 362, (I) 239, (J) 335, and (K) 853. N = 29,702 females total; ~2725 females per genotype; 13,375 for continuous rich treatments, and ~8170 each for the two other treatments. The dietary switch for the long-switch treatment group occurred at 45 to 65% of continuous rich treatment flies. All panels contain daily time points as in Fig. 2. Exposure to a high-nutrient diet after a period of DR resulted in marked increase in mortality compared with a continuous rich diet in all lines (9 of 11 significant). There was genetic variation in this response, with DGRP-136 (G) and DGRP-362 (H) showing the smallest effects. This marked overshoot was not contingent upon DR extending life span. Lines that showed “starvation” on a DR diet still showed significant overshoots when they were switched to a rich diet, where recovery from starvation was expected, even when compared with continuous DR diets (I to K)

Fig. 4. Survival curves of DGRP panel for both dietary regimes.

(A) 195, (B) 105, (C) 217, (D) 441, (E) 705, (F) 707, (G) 136, (H) 362, (I) 239, (J) 335, and (K) 853. Total survival on the different dietary regimes across the genetic panel tested. Rich diets after a period of DR resulted in such an increase in mortality, that total survival of the cohort was lower (or equal to) than those fed a continuous rich diet for their whole life (A to F). N = 37,897 females total; ~3450 females per genotype; 13,375 for continuous rich treatments, and ~8170 for all other treatments.

Fig. 5. Schematic of the current, and alternative, hypotheses of DR.

Reduced resource availability leading to increased investment toward somatic maintenance explains life-span extension under DR (see Fig. 1) in the most commonly supported current evolutionary theory. This increased investment may be absolute or relative to total resource availability. In our alternative model, based on the conclusions from the experiments we present here, the reduction in resource availability simply elicits a correlated reduction in available resources allocated toward reproductive output. The extension of life span observed under DR would then be a similarly passive response: an escape from unidentified costs incurred under a rich diet. These costs may be related to heightened metabolism or arising from direct insults of excessive protein intake. In addition, we propose restricted diets promote the accumulation of unknown costs, which are only observable upon resumption of a rich diet (not depicted here; see discussion). These hidden costs of DR would be responsible for the exacerbation of mortality observed when a rich diet is resumed. We suggest that these costs result from a period of physiological adaptation to a restricted diet, compensating for particular components of a rich diet. Such compensation on the DR diet, essentially maladapting the organisms to rich diet conditions, is directly contrary to current evolutionary theory that suggests investment in somatic maintenance occurs to survive to reap fitness benefits when resources are plentiful again.