Reduced cell proliferation may mediate anticarcinogenic effects of caloric restriction (CR). Using heavy water (2H2O) labeling, we investigated the cell proliferation response to CR in detail, including time course, effect of refeeding, and role of intermittent feeding with 5% CR. In the time-course study, 8-wk-old female C57BL/6J mice were placed on a 33% CR regimen (fed 3 times/wk) for varying durations. Compared with responses in controls fed ad libitum (AL), proliferation rates of keratinocytes, mammary epithelial cells, and T cells were markedly reduced within 2 wk of CR. In mice fed 95% ad libitum (C95, fed 3 times/wk), cell proliferation was also reduced in all tissues so that differences from 33% CR were only significant at 1 mo. In the refeeding study, mice were refed a C95 diet for varying durations after 1 mo of 33% CR. Cell proliferation rebounded to a suprabasal rate in all tissues after 2 wk of refeeding and then normalized after 2 mo, although the C95 group again exhibited lower cell proliferation than the AL group. The role of intermittent feeding was studied by comparing 33% CR and C95 animals (both fed intermittently) with animals fed isocalorically either daily or continuously by pellet dispenser. Intermittent feeding had no additive effect on 33% CR but reduced cell proliferation in all tissues at the 95% caloric intake level. In summary, the CR effect on cell proliferation is potent, rapid, and reversible in several tissues, and an intermittent feeding pattern reproduces much of the effect in the absence of substantial CR.
- stable isotopes
- T cell
caloric restriction (CR), defined as undernutrition without malnutrition (45), was first discovered in 1935 to extend maximal life span in rats (29). Since then, similar findings have been reported in mice, fish, flies, worms, and yeast (45). A range of 30–70% extension of maximal life span has been achieved using variations on CR regimens (45), including both early- and adult-onset CR (44, 46). CR also exerts a number of other beneficial health effects, including reduced carcinogenesis, enhanced insulin sensitivity, and reduced cardiovascular disease risk (15). The inhibitory effect of CR on carcinogenesis is of particular interest, because CR effectively inhibits spontaneous tumor formation as well as neoplasias in knockout/transgenic models of cancer- and chemically induced tumorigenesis (18, 19, 45). The mechanisms by which CR extends life span and inhibits carcinogenesis remain unknown (15, 19).
CR could affect several steps in the multistage carcinogenesis model (18). CR may function as an anti-initiator by decreasing carcinogen activation, enhancing carcinogen detoxification, scavenging reactive oxygen species, or enhancing DNA repair (18). CR could also function as an antipromoter by reducing mitoses of initiated cells, altering expression of cancer-related genes, decreasing inflammation, enhancing immune competence, or stimulating apoptosis (18). Reductions in cell proliferation might be expected in view of observed hormonal effects of CR, such as lower growth hormone and IGF-I levels (18, 19), as well as reduced body temperature (torpor) (22, 45).
Previous work supports the hypothesis that CR reduces cell proliferation. Using tritiated deoxythymidine (3HdT) labeling, Lok et al. (26) demonstrated a 30–60% decrease in cell proliferation after 1 mo of 25% CR in mice in the skin, esophagus, bladder, and gastrointestinal tract and a 72% decrease in the mammary gland. In mice with 30% CR and epidermal treatment with carcinogens, 25% lower cell proliferation was observed [by bromodeoxyuridine (BrdU) labeling] compared with mice fed ad libitum (AL) and given the same carcinogens. Increased papilloma latency was also reported (10). Rats with 60% CR that were treated with 1-methyl-1-nitrosourea (MNU) had up to a 31% reduction in mammary cell proliferation compared with MNU-treated AL rats, as measured by BrdU, as well as decreased mammary carcinoma size and density (47). Food reduction and fasting in rats given the hepatomitogen cyproterone acetate resulted in lower DNA replication and increased apoptosis in liver (12).
Many questions remain, however, regarding the effects of CR on cell proliferation. In particular, the details of the response of cell proliferation to CR, including the time course, dose-response relationships, effects of refeeding, and endocrine correlations, have not been established. The effect of using different control groups in the CR field [e.g., AL vs. 5–15% restriction (1, 5, 13, 23, 24, 37, 41, 43, 44, 46)] also has not been systematically explored. Finally, intermittency of feeding of both controls and CR models also has varied in the field (13, 23, 37, 42, 46) and could influence the effects of CR on cell proliferation.
We recently developed a method for measuring cell proliferation in vivo using heavy water (2H2O). This technique has a number of advantages over 3HdT and BrdU labeling (34, 35) and has been used to measure turnover of keratinocytes, mammary epithelial cells, T cells, adipocytes, colonocytes, leukemic cells, and other cells (6, 16, 17, 21, 30, 32, 39, 40). This technique is highly reproducible and simple to apply, thereby allowing numerous groups or hypotheses to be tested at a relatively high throughput. In the present study, we applied this quantitative method to study the details of the response of cell proliferation to CR in mice.
MATERIALS AND METHODS
Mice and CR Regimens
For all studies, 7-wk-old female C57BL/6J mice (Jackson Laboratory, Bar Harbor, ME) were acclimated for 1 wk, during which time they were fed a semipurified AIN-93M diet ad libitum (Bio-Serv, Frenchtown, NJ). Studies were then started in mice at 8 wk of age. Three studies were carried out.
Study 1: Time Course of CR
A 33% CR diet was fed for varying durations of time to the three treatment groups (n = 8 mice/group): 2 wk CR (2W), 1 mo CR (1M), or 2 mo CR (2M) (Fig. 1A). The 2W group received a control diet for 6 wk before onset of the CR diet, and the 1M group received a control diet for 1 mo before onset of the CR diet, so the ages of all three groups were matched at the end of the experiment. Accordingly, mice in all three groups were killed at 16 wk of age. Two different control groups also were used (n = 8 mice/group): mice fed ad libitum (AL) and mice fed 95% of ad libitum intake (C95). These animals also were killed at 16 wk of age. These control groups represent the different types of control groups that have commonly been used in CR studies demonstrating life span extension or reduced carcinogenesis (1, 37, 41, 44, 46). One additional group of mice (n = 4) was placed on CR for a longer period of time (3 mo, 3M), also starting at 8 wk of age. Animals in this group were killed at 20 wk of age.
During non-CR periods, the treatment groups were maintained on the C95 diet regimen. During CR periods, mice were fed 67% of C95 intake, or ∼64% of AL intake, as previously described (37). The CR and C95 groups were fed 3 days a week, such that two times the daily allotment was given on Mondays and Wednesdays and three times the daily allotment on Fridays, a regimen commonly used in previous CR studies (13, 37, 42, 44, 46). AL and C95 mice were fed a semipurified AIN-93M diet, whereas CR mice were fed an enriched AIN-93M diet that contains 33% more protein, minerals, and vitamins per gram of diet (Bio-Serv). All mice were housed individually. Food intake and body weight were monitored weekly.
Study 2: Refeeding
The time course of response to refeeding was also studied. Mice received a 33% CR diet for 1 mo and were subsequently given a C95 diet (n = 8 mice/group) for either 2 wk of refeeding (R2W) or 1 mo of refeeding (R1M) (Fig. 1B). The CR diet for the R2W group started 2 wk into the study (10 wk old), whereas the CR diet for the R1M group started immediately (8 wk old) so that both groups were killed at 16 wk of age. One additional group of mice received a 33% CR diet for 1 mo and was refed for a longer period of time (2 mo, n = 4) (R2M). These mice were killed at 20 wk of age. All mice were housed individually. Food intake and body weight were monitored weekly.
Study 3: Intermittency of Feeding
The role of intermittent food intake was also investigated. Three groups of mice were put on a 33% CR diet, administered via different feeding protocols (n = 6 mice/group): intermittent feeding 3 times per week (CR-INT), as described for study 1 (37); daily feeding (CR-DF); or continuous feeding via an electronic pellet dispenser (CR-PD). Three additional groups of mice were fed 95% of ad libitum diet via the same three feeding protocols (n = 6 mice/group): intermittent feeding 3 times a week (95-INT); daily feeding (95-DF); or continuous feeding (95-PD). The 95-INT, 95-DF, and 95-PD groups also were compared with an AL group fed concurrently. Intermittent feeding was as described for study 1, with two times the daily allotment given on Mondays and Wednesdays and three times the daily allotment given on Fridays. Mice fed daily were given their food allotment for each day every morning. The amount and type of diet (33% enriched or standard AIN-93M) depended on whether the mice were in the 33% CR groups (CR-INT, CR-DF) or the 5% CR groups (95-INT, 95-DF). Continuously fed mice were housed in cages containing an electronic pellet dispenser that delivered a 45-mg pellet of AIN-93M diet (standard for 95-PD, 33% enriched for CR-PD; Bio-Serv) into the cage every 20–30 min, depending on the caloric intake. All mice were housed individually. Food intake and body weight were monitored weekly. Mice were killed at 12 wk of age after 4 wk of treatment.
Heavy Water Labeling Protocol
Cell proliferation was measured by the 2H2O labeling method as described previously (34, 35). Briefly, mice were given an intraperitoneal injection of isotonic 100% 2H2O (0.18 ml/10 g body wt) 2 wk before death to bring their 2H2O content in body water up to 2.5%. Mice subsequently received drinking water ad libitum that contained 4% 2H2O for 2 wk. The mice were killed by cardiac puncture under isoflurane anesthesia, followed by cervical dislocation.
Epidermal Cell (Keratinocyte) Isolation
Dorsal hair was removed postmortem by application of a hair removal lotion (Nair; Carter Products, New York, NY). After the lotion was cleaned off with the use of an alcohol swab, a piece of dorsal skin was dissected (∼3 cm2). The skin was rinsed with phosphate buffer solution (PBS; GIBCO, Grand Island, NY), divided into three smaller pieces, immersed in dispase II (Roche, Indianapolis, IN), and incubated for 3.5 h at 37°C on a shaker. The epidermis was then peeled from the dermis in thin white sheets, as described previously (17).
Mammary Epithelial Cell Isolation
Mammary epithelial cells (MECs) were isolated as described previously (32, 33). Briefly, mammary fat pads were dissected, minced, and treated with collagenase (Worthington Biochemical, Lakewood, NJ) before isolation of MECs by centrifugation in a Percoll gradient (Amersham Pharmacia Biotech, Piscataway, NJ).
Splenic T Cell Isolation
T cells were isolated from spleen. Briefly, the spleen was removed, minced, and filtered through a 30-μm nylon mesh. T cells were isolated by using anti-CD90 microbeads via a magnetic column method (Miltenyi Biotec, Auburn, CA).
Bone Marrow Cell Isolation
Bone marrow was collected from the femur. Marrow cells were flushed out with the use of a needle and syringe containing PBS (GIBCO) as described previously (27).
DNA was extracted from cell preparations by using Qiagen kits (Qiagen, Valencia, CA) and was hydrolyzed to deoxyribonucleosides as described elsewhere (35). In brief, DNA was heated with magnesium chloride and zinc sulfate, followed by incubation for 2 h in a 37°C water bath with DNase, nuclease P1, snake venom phosphodiesterase, and alkaline phosphatase (Sigma, St. Louis, MO). The deoxyribose (dR) moiety was derivatized to pentane tetraacetate, as described (35). Pentane tetraacetate was analyzed by positive chemical ionization GC-MS with a model 5973 mass spectrometer and model 6890 gas chromatograph (Agilent, Palo Alto, CA). Selected ion monitoring was performed with mass-to-charge ratios (m/z) of 245 and 246, representing the M+0 and M+1 ions, respectively (34, 35). The excess fractional M+1 enrichment (EM1) of dR was calculated as where sample and STD represent the analyzed sample and unenriched standards, respectively. Standards of natural abundance (unlabeled) pentane tetraacetate were analyzed concurrently with samples. Abundance matching of samples to standards and other corrections were as described in detail elsewhere (34, 35). Calculation of fractional replacement (f) of cells was done by comparison with nearly fully turned over cells (bone marrow cells), as described previously (7, 34):
Estrus Cycle Status
The presence or absence of estrous cycle in the 1M CR and C95 groups was determined via vaginal smear and analysis of cell morphology. Vaginal smears were taken during four consecutive days, and samples were fixed and stained on slides with hemoxylin and eosin (Histo-Tec Laboratory, Hayward, CA).
Data from the time course and refeeding studies were analyzed by one-way ANOVA with Dunnett's follow-up test, comparing all groups with either AL or C95 controls (P < 0.05 set as significance level). The intermittency of feeding study was analyzed by one-way ANOVA with Tukey's follow-up test (P < 0.05), comparing all pairwise combinations in the CR and 95% of ad libitum intermittency experiments.
Study 1: Time Course of CR
Food intake and body weight.
On average, AL mice consumed 22 g of food per week. Therefore, C95 mice were fed 21 g of food per week and CR mice were fed 14 g of food per week. The body weight of CR mice dropped initially by as much as 30% but stabilized over time (Fig. 2A). Mice then gained weight on CR diets.
Compared with findings in AL mice, proliferation of epidermal cells, MECs, and T cells was significantly decreased in the CR groups at all time points studied (Fig. 3, A–C). Compared with findings in C95 mice, in contrast, cell proliferation in all tissues was not significantly decreased until 1 mo of CR, after which the response was again not significant. At 1 mo of CR, the time of greatest effect of CR, epidermal cell proliferation was 61% of that in AL mice and 76% of that in C95 mice. MEC proliferation was only 11% of that in AL mice and 29% of that in C95 mice at 1 mo, whereas T-cell proliferation was 41% of that in AL mice and 57% of that in C95 mice.
Differences between C95 and AL control groups.
C95 mice exhibited statistically significantly lower cell proliferation than AL mice in all tissues examined (Fig. 3, A–C). After 2 mo on respective diets, epidermal cell proliferation in C95 mice was 81% of that in AL mice, MEC proliferation was 37%, and T cell proliferation was 71%. Thus CR exerted significant effects on proliferation of all three cell types studied, but C95 also had a potent impact that appeared to account for at least part of the CR effect.
On the basis of cell morphology analysis of vaginal cells collected from 1M and C95 mice, CR mice were anestrus (not cycling), whereas C95 mice were actively cycling. The marked reduction in MEC proliferation in the CR groups might therefore, in part, be explained by a reduction in reproductive hormone levels (31), but the substantial effect observed in the C95 groups excludes this as the primary cause of reduced MEC proliferation.
Study 2: Refeeding
Food intake and body weight.
As in study 1, AL mice consumed ∼22 g of food per week. During the CR phase, mice were therefore fed 14 g of food per week, and during the refeeding phase, mice were fed 21 g of food per week. Refeeding resulted in a rapid gain of lost weight (Fig. 2B). Body weights of CR mice had caught up to the body weights of C95 mice by the end of the study, despite the 1-mo period of CR.
Time course of refeeding effects.
Compared with findings in the C95 control group, cell proliferation in all tissues rebounded to a significantly higher rate after 2 wk of refeeding, persisting through 1 mo of refeeding but normalizing after 2 mo of refeeding (Fig. 4, A–C). Compared with findings in the AL group, cell proliferation in all tissues was no longer significantly different after 2 wk of refeeding. Subsequent comparisons revealed tissue-specific differences. After 1 mo of refeeding of the C95 diet, the T-cell proliferation rate was statistically higher than AL levels; this was normalized after 2 mo of refeeding. MEC proliferation was significantly lower than AL levels after 2 mo of refeeding of C95 diet, consistent with the observation that MEC proliferation was substantially lower in C95 mice than in AL mice (Fig. 3B).
Study 3: Intermittency of Feeding
Food intake and body weight.
Throughout this study, all groups of CR mice were fed 14 g of food per week, and all groups of control mice were fed 21 g of food per week. All mice gained weight on their diets (Fig. 2C). Nonsignificant differences in body weight among mice with the same caloric intake but fed by different feeding patterns may be due to the presence or absence of food in the stomach during weighing.
Feeding intermittency effects among groups of CR mice.
In the three tissues studied, intermittency of feeding (i.e., food given 3 times/wk) had no additional effect compared with daily or continuous feeding on cell proliferation when CR was present (Fig. 5, A–C).
Feeding intermittency effects among groups of control mice.
There was lower cell proliferation in all tissues of the group fed intermittently at 95% of ad libitum diet (95-INT) compared with daily feeding (95-DF), continuous feeding (95-PD), or AL feeding, although not all comparisons were statistically significant (Fig. 6, A–C). MEC proliferation was significantly lower in 95-INT mice than in 95-PD mice, whereas T-cell proliferation was significantly lower in 95-INT mice compared with 95-DF and 95-PD mice. Epidermal and T-cell proliferation rates in AL mice were not statistically different from those in 95-DF or 95-PD mice but were significantly greater than those in 95-INT mice. An intermittent feeding regimen (i.e., food given 3 times per week) therefore caused significant reductions in cell proliferation rates compared with isocaloric diets fed by more constant patterns.
We demonstrated in the present study the application of a relatively simple method for measuring cell proliferation in multiple tissues in mice. With the use of this technique, it is clear that cell proliferation rates in mice are extremely sensitive to changes in caloric intake, whether due to CR or feeding pattern.
Previous methods for measuring cell proliferation include cell-cycle indexes such as Ki67 or PCNA staining (28, 38). These techniques do not accurately reveal rate of progression through the cell cycle, however (16). Dynamic measurements, including incorporation of BrdU and 3HdT, also have limitations. DNA incorporation of these precursors occurs via nucleoside salvage pathways and is dependent on a number of variables, including efficiency of cellular uptake, competition with extracellular nucleosides, and other factors, which can differ among cell types (34, 35). Labeled deoxyribonucleosides released after cell death also may be reincorporated into other cells (16). The stable isotope labeling method used in the present study is safe, yields quantitative kinetic information, does not depend on the deoxyribonucleoside salvage pathway, and is not susceptible to artifacts related to reutilization (16, 34, 35).
We showed in the present study that early-onset 33% CR in C57BL/6J mice, administered by a commonly used feeding regimen in this field (i.e., food given 3 times/wk) (13, 37, 42, 44, 46), reduces proliferation of epidermal cells (keratinocytes), MECs, and splenic T cells. When mice were refed after CR, cell proliferation rates were restored within 2 wk to values equal to those in AL controls, and some tissues became transiently hyperproliferative compared with 95% ad libitum fed controls. These data suggest that the effects of CR on cell proliferation are rapid and reversible. Whether these effects on cell proliferation are sustained over extended duration of CR cannot be deduced from these data.
The mediator(s) of the CR effect on cell proliferation in multiple tissues remains uncertain. IGF-I has been hypothesized to mediate the decrease in cell proliferation in response to CR (18, 19). Serum IGF-I levels have been consistently reported to be reduced in CR studies (4, 9, 14, 20), and exogenous replacement of IGF-I has been found to negate the benefits against bladder cancer conferred by CR in p53-deficient mice (9). In addition, modulations in IGF-I signaling have been correlated to life span extension (3, 8, 11). We were unable to accurately compare IGF-I levels among groups because of differences in fasting times before death. A priority for future studies would be to characterize the relationship between changes in cell proliferation and concentrations of potential mediators.
Our data demonstrate that an intermittent pattern of feeding, resulting in periodic fasting, contributes to the antiproliferative effects of CR regimens, along with caloric deficit. We observed that a 5% decrease in total caloric intake, combined with an intermittent feeding pattern (food given 3 times per week), decreased cell proliferation compared with mice fed isocalorically but according to a more constant feeding pattern (daily or continuously). Intermittency of feeding did not appear to have an additive effect in CR mice. In particular, among mice receiving 95% ad libitum caloric intake, intermittent feeding decreased MEC and T-cell proliferation compared with continuously fed mice. Continuously and daily fed mice at 95% ad libitum caloric intake also did not have significantly lower epidermal and T-cell proliferation compared with AL controls, whereas intermittently fed mice at 95% ad libitum caloric intake did, ruling out an effect of the 5% reduction in caloric intake, per se. Recently, intermittent feeding was found to impart greater benefits than daily feeding at a 40% level of CR (2). The intermittent feeding model employed by Anson et al. (2) involved alternating ad libitum feeding and complete food deprivation every other day. Although the mice compensated for food deprivation on the days during which they were fed, they were only able to attain a caloric intake of ∼90% of ad libitum levels. Thus their model, resulting in 10% CR with intermittent feeding, is similar to our C95 group, fed 5% CR intermittently. Anson et al. (2) reported improved insulin sensitivity in this model compared with a daily fed 40% CR model. Both studies therefore suggest that minimal CR in conjunction with intermittent feeding induces health effects similar to that from traditional, much more substantial CR.
Our data do not suggest, however, that the effects of substantial CR can be completely reproduced by intermittency of feeding. Although intermittent feeding with 5% CR (95-INT) resulted in lower cell proliferation than more continuous feeding at the same caloric level, it is worth noting that the degree of hypoproliferation was not as pronounced as in mice fed 33% CR, regardless of feeding intermittency. This result suggests that substantial CR still has a dominant effect over feeding intermittency. Similarly, Lee et al. (25) have shown that mice fed intermittently on 41% CR have greater life span extension and lower tumor incidence than those fed intermittently on 15% CR as controls.
Nelson and Halberg (36) also investigated the role of intermittent feeding and found that 25% CR with six smaller meals vs. one big meal a day both extended life span to the same extent in mice but resulted in a different circadian rhythm such that less frequent meals resulted in lower core body temperature. This finding may be significant, because CR-induced torpor and cell proliferation are linked (22, 45), but cell proliferation was not measured in this study. The finding that 25% CR with increased feeding intermittency did not extend life span beyond daily feeding of 25% CR may suggest that substantial CR overcomes or masks any effect of intermittency on life span. This interpretation is also consistent with our data, because 33% CR groups had the same cell proliferation rates despite different feeding intermittency patterns. There has yet to be a study comparing life span expectancy in animals with minimal CR by using different feeding patterns, however. Such a study would be necessary to investigate the effect of intermittency of feeding apart from caloric deficit on life span extension.
The suggestion that intermittent feeding may produce benefits similar to caloric restriction is potentially of great interest to human applications. Although it may be impractical to maintain humans on substantially calorically restricted diets for their lifetime, intermittent food deprivation may be feasible. If some of the health benefits of CR can be reproduced, including reduction in cancer promotion, this might be a therapeutic strategy worth pursuing. Human CR studies using the techniques described here [e.g., proliferation of skin cells, T cells and mammary epithelial cells (17, 32, 34)] could, in principle, be performed to test this hypothesis.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2005 by American Physiological Society