Abstract
Genetic changes resulting in increased life span are often positively associated with enhanced stress resistance and somatic maintenance. A recent study found that certain long-lived Caenorhabditis elegans mutants spent a decreased proportion of total life in a healthy state compared with controls, raising concerns about how the relationship between health and longevity is assessed. We evaluated seven markers of health and two health-span models for their suitability in assessing age-associated health in invertebrates using C elegans strains not expected to outperform wild-type animals. Additionally, we used an empirical method to determine the transition point into failing health based on the greatest rate of change with age for each marker. As expected, animals with mutations causing sickness or accelerated aging had reduced health span when compared chronologically to wild-type animals. Physiological health span, the proportion of total life spent healthy, was reduced for locomotion markers in chronically ill mutants, but, surprisingly, was extended for thermotolerance. In contrast, all short-lived mutants had reduced “quality-of-life” in another model recently employed for assessing invertebrate health. Results suggest that the interpretation of physiological health span is not straightforward, possibly because it factors out time and thus does not account for the added cost of extrinsic forces on longer-lived strains.
An assumption underlying the desire for increased life span is that it will also improve quality of life by helping ameliorate the impact of diseases and diminution of health associated with aging. This possibility is generally supported by evidence from model organisms showing that enhanced somatic maintenance, frequently measured by increased survival under various forms of stress, coincides with increased longevity (1–3). Other measurements of health have shown that prolongevity interventions maintain physiology in a youthful state, including preservation of locomotion, feeding behavior, and accumulation of autofluorescent pigments in cells (4–6). Furthermore, when genetic models of longevity are combined with age-associated models of disease, results often indicate amelioration or delay in the onset of pathology (7,8). Together, the evidence indicates that health span can be extended through evolutionarily conserved genes and pathways.
Although many of the studies linking enhanced longevity with increased stress resistance have come from investigations utilizing Caenorhabditis elegans model system, a recent study provided evidence suggesting that the actual percent of life spent in a frail state, referred to as physiological gerospan, was increased among long-lived C elegans mutants compared with wild-type animals (9). In their investigation, they examined several markers of health that changed with age in well-characterized long-lived mutants and compared them with wild-type animals. While generally confirming enhanced chronological health span, their results suggested that the proportion of total life spent healthy in long-lived mutants, termed physiological health span, was shorter than that of wild-type animals. A later study demonstrated that the reduced physiological health span of long-lived daf-2 mutants was a result of altered foraging behavior caused by tracking unstimulated locomotion in the presence of food (10). However, the reduced physiological health span of other long-lived mutants remains unexplained. Additionally, assessment of health span in mouse using markers for body composition, energetics, and movement showed little association with mortality suggesting an uncoupling of health span and life span in more complex systems (11). These provocative studies have raised concerns about the way in which we examine health in animal systems and the applicability of such results to human health. The importance of considering health span in the context of life span was made clear in the recent convening of the Geoscience Network, which deemed extension of life span alone as insufficient evidence of delayed aging (12).
Obstacles to the characterization of invertebrate health span include the choice of which strain and markers to use. For the former, we used short-lived C elegans mutants displaying signs of accelerated aging and/or chronic illness, allowing assessment of candidate health markers and recently reported models that assess age-associated health. For the latter, we wanted to make a distinction between health markers strongly dependent on biological aging and those which only have a temporal relationship or are weakly dependent on the aging process. Given that our chosen short-lived strains displayed signs of chronic illness, we adopted definitions from Brody and Schneider (13) used for chronic human diseases to make this distinction for health markers. In their manuscript, only disorders with mortality rates that increase with age were defined as age dependent. Applying a similar logic to the present study, markers which failed sooner chronologically in short-lived mutants were defined to be aging dependent, as the earlier failing of both health and longevity is presumably regulated by similar processes. Conversely, markers were considered time dependent if they changed with age but not in a manner consistent with life span.
For short-lived strains, we used animals with mutations in daf-16 (abnormal dauer formation), mev-1 (abnormal methyl viologen sensitivity), hsf-1 (heat shock factor), and sir-2.1 (silent information regulator). daf-16 encodes a FOXO transcription factor that acts in the insulin/insulin-like growth factor signaling (IIS) pathway and is required for increased life span and enhanced resistance to stress when the IIS pathway is attenuated (14). mev-1 encodes the C elegans ortholog for human mitochondrial succinate dehydrogenase cytochrome b560 subunit, the mutation of which is associated with significantly decreased life span, increased reactive oxygen species (15), and severely impaired reproduction (16). hsf-1 encodes the heat shock transcription factor important for mediating the response to unfolded proteins in the cytoplasm and signals a series of prolongevity genes (7,17). Its expression is required for life span and stress-resistance phenotypes associated with attenuated IIS (17) and loss-of-function leads to defects in development and reproduction (18,19). sir-2.1 encodes a NAD-dependent deacetylase important for metabolic homeostasis, with lost function resulting in shortened life span and increased sensitivity to stress (20).
Using mutants for these genes, we determined which previously used markers of health span were aging dependent and thus informative candidates for health-span determination linked to life span. Additionally, we found maximum bending amplitude, which is an indicator of muscle integrity (21), to be a new aging-dependent health-span marker with possible associations with longevity. To determine when the healthy period of life ends, we posited an empirical method based on the rate of change for a given marker. Although chronological health span was reduced in the chosen short-lived mutants, physiological (time normalized) health span was actually greater for thermotolerance in short-lived strains in this model. Although the apparent simplicity of the physiological health-span model suggests a straightforward interpretation, these unexpected results indicate additional consideration is required, such as the cost of extrinsic forces on aging, which is not taken into consideration when time is factored out.
Materials and Methods
Nematode Culture and Strains
C elegans strains were cultured at 20°C and maintained on nematode growth media (NGM) plates seeded with the Escherichia coli strain OP50. N2 wild-type, CF1038 daf-16(mu86) I, XA8203 sir-2.1(ok434) IV, PS3551 hsf-1(sy441) I, and TK22 mev-1(kn1) III were obtained from the Caenorhabditis Genetics Center (CGC). All strains were permitted to grow at least three generations under normal, unstressed conditions prior to use in experiments. Except for thermotolerance assays, all experiments were carried out at 20°C.
Life Spans
Survival was scored starting from the first day of adulthood using 100 or more worms per strain divided among three seeded NGM plates containing 200 µM fluorodeoxyuridine (FUDR) to prevent contamination by progeny. Survival status was determined by touch provocation. Worms with vulval protrusion were censored from the assay. Differences in survival curves were calculated using the log-rank test in the Graphpad Prism 6.0 software. Maximum life span was calculated by taking the average life span of the top 10 longest-lived worms.
Development and Fecundity
Several dozen gravid adults were allowed to lay eggs on spotted NGM plates for 1 hour and removed. Time to egg-laying adulthood was started from this point. Eggs were allowed to hatch and 20–22 animals were randomly picked and separated onto individual plates. Plates bearing single young adults were checked every hour for the presence of eggs. Animals from the development assay were transferred to fresh plates every 24 hours following initiation of egg laying. Hatched progeny and unfertilized oocytes were counted for each plate 48 hours after removal of the egg-laying adult.
Health-Span Assays
Speed, bending angle, and worm size were assayed for each time point by sampling from a population of age-synchronized animals on seeded NGM plates. Worms were tracked in real time after gently tapping plates to stimulate movement and distinguish the ability to move from food-seeking behavior (10). Forward and backward speed were recorded with a Stingray F504B ASG digital (Allied Technologies GmbH) camera equipped with an AF Micro-Nikkor 60mm f/2.8D lens (Nikon). 30-second intervals of continuous (unbroken) tracking were analyzed using Wormlab ver 3.0 (MBF Bioscience). For pharyngeal pumping, worms were observed individually with a Leica M165FC stereo microscope and digitally recorded (VLC, ver. 2.1.3) on seeded NGM plates. Pumps were counted manually by playing 30 seconds of video at a reduced speed. For autofluorescence measurements, worms were transferred to 1% agar pads containing 20-mM sodium azide spotted on glass slides. Autofluorescence was measured on a Leica M165FC stereo microscope in both the green fluorescent protein (GFP) (GFP narrow band filter set, excitation ET470/40 nm, emission ET510/10 nm) and the 4′,6-diamidino-2-phenylindole (DAPI) (ET DAPI filter set, excitation AT350/50x nm, emission ET460/50m) channels using a 3-second exposure. Quantification of autofluorescence was performed using ImageJ for mean pixel intensity and corrected for background fluorescence. Thermotolerance assays were conducted with 100 worms (aged as indicated) and distributed among two plates at 35°C. Survival was measured every hour by touch provocation. Resistance to oxidative stress was tested using worms distributed to individual wells (12 worms/well, 4 wells/strain) of a 24-well culture plate containing 7.5-mM hydrogen peroxide in S basal buffer. Survival was measured every other hour by touch provocation. All health-span assays were independently repeated three times.
Statistical Analysis
All statistics other than log-rank tests were performed in the software package R (ver 3.1.3). For each health marker, analysis of variance was used to estimate effects of genotype, time, and the interaction of the two. Replicate experiment was used as a blocking factor. Post hoc Tukey’s honest significant difference was used to compare means between genotypes and time points. The rate of change in each aging-dependent health marker was modeled using linear regression. The linear regression estimated the effect of time with replicate experiment as a blocking factor. As maximum speed and maximum amplitude did not decrease until Day 5 for most genotypes, Day 1 was omitted from the linear regression analysis for these parameters. For similar reasons, heat stress survival on Day 2 was omitted from the regression analysis. To determine the onset of rapid health decline in wild type, linear regression was used as above but with a sliding window of 3 consecutive time points. The time by which each genotype entered gerospan was imputed using the regression function from the linear models. Because measurements were not taken on Day 19 for thermotolerance, the missing values were extrapolated for each genotype using the linear regression from the other time points. Cumulative quality-adjusted survival was determined for each genotype as described by Hahm and colleagues (10) by first calculating the normalized health metric score and then the quality-adjusted survival for each aging-dependent trait.
Results
Reduced Whole-Life Health and Accelerated Aging in Short-Lived Mutants
Testing was carried out to confirm the short-lived and impaired stress response status of daf-16(mu86), sir-2.1(ok434), hsf-1(sy441), and mev-1(kn1) mutants. Life-span survival curves are shown in Figure 1A and median life span ranged from 26% to 38% below that of N2 wild type (17.0 days for sir-2.1(ok434), 16.5 days for mev-1(kn1), 15.5 days for daf-16(mu86), and 14.3 days for hsf-1(sy441)). Information pertaining to statistics and replicates are reported in Supplementary Table 1. Survival under heat or oxidative stress was also reduced in these strains (Figure 1B and C), indicating defects in somatic maintenance.
Short-lived Caenorhabditis elegans mutants demonstrate deficiencies in health during development and adulthood. (A) Representative survival curves of N2 wild-type and short-lived mutants from four biological repeats (Supplementary Table 1). Survival curves were compared using Mantel-Cox log-rank tests and all short-lived mutants had significantly (p < .05) shorter median life spans than wild-type N2 (mean survival difference for daf-16(mu86) was −33%, mev-1(kn1) was −29%, hsf-1(sy441) was −37%, and sir-2.1(ok434) was −26%). (B) Mean survival at 35°C on Day 6 of adulthood was reduced in daf-16, mev-1 and hsf-1, but not in sir-2.1 (p = 8.0E-8, 2.4E-11, 2.4E-11, and 0.79, respectively, calculated using Tukey’s test). Data were pooled from three independent experiments, n = 300/strain. (C) Mean survival was reduced in all short-lived strains in response to 7.5-µM hydrogen peroxide at Day 5 of adulthood (p < .0001, Tukey’s test). Data were pooled from three independent experiments, n = 138/strain. (D) The size of selected strains based on imaged area on Day 5 of adulthood showed that only mev-1 were smaller compared with wild type (p = 2.8E-13, Tukey’s test). Data were pooled from 3 independent experiments, n = 45/strain. Data were pooled from 3 independent experiments. E) Time to development for each strain as measured from time to egg laying to first day of adulthood. Compared with wild type, mev-1 and hsf-1 had a longer development time while daf-16 and sir-2.1 were similar (p = 1E-16, p = 1.0E-16, p = .088 and p =.049, respectively Tukey’s test). Data were pooled from 4 independent experiments, n = 70/strain. F) Fecundity of each selected strain as measured by the number of eggs hatched. The mutants mev-1 and sir-2.1 had reduced hatched progeny compared with wild type (p = 1.0E-12 and 1.4E-12, respectively in Tukey’s test). Data were pooled from four independent experiments, n = 64/strain. Error bars indicate SEM. ***p < .001.
In long-lived strains, increased life span and resistance to stress often coincide with tradeoffs in growth and reproduction. Such defects in short-lived strains would indicate illness at all life stages. Growth was assessed by analysis of nematode size quantified on Day 5 of adulthood, by which time all strains were completely grown. Only mev-1 mutants were significantly smaller than wild-type mutants (106.2 mm2 vs 154.7 mm2, p = 2.87E-13, Tukey’s test; Figure 1D). Development was assessed by measuring the time required for each strain to develop from an egg to an egg-laying adult (Figure 1E). Development was delayed in hsf-1 (92.8 h compared with 89.2 h in wild type, p = 6.85E-06, Tukey’s test) and severely delayed in mev-1 (112.5 h, p = 1.0E-16, Tukey’s test). Reproductive health was assessed by counting hatched progeny (Figure 1F), and unfertilized oocytes (Supplementary Figure 1), as well as by measuring the distribution and duration of reproduction (Supplementary Figure 2). Significantly reduced number of progeny were produced in mev-1 (129 compared with 259 in wild type, p = 1.02E-12, Tukey’s test) and sir-2.1 (199, p = 1.44E-12) mutants (Figure 1F). Interestingly, a significant number of unfertilized oocytes were laid in the daf-16 (78.5, p = 1.03E-12, Tukey’s test) and hsf-1 (41.7, p = 1.07E-12) backgrounds compared with wild-type N2 (0.7; Supplementary Figure 1) near the end of the reproductive period, indicative of dysregulated ovulation. However, there was not a significant impediment to the production of viable progeny in daf-16 (254, p = .95) and hsf-1 (246, p = .34) compared with wild type (259) in the conditions tested (Figure 1F). Finally, the bulk distribution of reproduction in sir-2.1 animals was shifted to after the second day of egg laying, unlike in the wild-type background where more than 50% of viable progeny were laid within the first 48 hours of the onset of egg laying (Supplementary Figure 2).
Together, the data for longevity, stress resistance, development, and reproduction suggested that the hsf-1, mev-1, and sir-2.1 mutants used here model sickness due to their deficits in early- and late-life metrics of health. While the high number of unfertilized oocytes in daf-16 animals suggests anomalous regulation of ovulation, total fecundity was not significantly reduced. Conversely, daf-2 mutants produce fewer unfertilized oocytes compared with wild type (22), presumably due to increased daf-16 activity. Thus, with the absence of major developmental or reproductive deficiencies, daf-16 mutants may be considered a model of accelerated aging, although lack of functional DAF-16 means these animals have chronically reduced stress resistance. With these characterizations in mind, we analyzed the lifelong performance of health markers concerning motility, feeding, autofluorescence, and stress resistance.
Locomotion and Thermotolerance Are Aging-Dependent Markers of Health
Locomotion is dependent on the integrity of muscle mass, connective tissue, and neuronal signaling (21,23–26). We investigated changes in movement with age using digital video recording and nematode tracking software (Wormlab 3.1, see Methods). Mean and maximum speed declined over time in all strains tested (Figure 2A and B). Short-lived mutants were typically slower than wild type for both mean and maximum speed. By Day 5 of adulthood, the maximum speeds of daf-16, sir-2.1, mev-1, and hsf-1 were slower than wild type by 30%, 32%, 48%, and 51%, respectively. Differences in maximum speed increased by Day 15 (slower by 60%, 70%, 59%, and 58%, respectively). The decline in mean speed began after Day 1 for the short-lived mutants and after Day 5 for wild type, with similar results for maximum speed except that the decline began after Day 5 for daf-16 and hsf-1. As mean and maximum speed both declined with age sooner in all short-lived mutants compared with wild-type animals, we categorized these traits as aging dependent. Of the markers tested in the present study, maximum speed had the highest correlation with survival rate (Table 1).
Locomotion and thermotolerance were reduced sooner chronologically in short-lived mutants. Mean and maximum speed, maximum bending amplitude, and thermotolerance throughout life are shown in A–D. Error bars indicate SEM. Data were pooled from three independent experiments. n = 45/strain for locomotion parameters. n = 300/strain for thermotolerance. *p < .05, ***p < .001, as compared with N2 wild type on the same day by analysis of variance with post hoc Tukey’s test.
Survival Rate Is Highly Correlated With Maximum Speed and Well Correlated With Other Health Markers. Spearman Correlations Between Survival Rate and Each Health-Span Parameter
| Speed | Max Speed | Max Amp | Pumping | DAPI | GFP | Thermotolerance | |
|---|---|---|---|---|---|---|---|
| Coefficient | 0.88 | 0.95 | 0.69 | 0.89 | −0.74 | −0.37 | 0.89 |
| p Value | 9.7E-11 | 1.3E-15 | 2.5E-05 | 7.3E-11 | 3.7E-06 | 0.045 | 7.9E-11 |
| Speed | Max Speed | Max Amp | Pumping | DAPI | GFP | Thermotolerance | |
|---|---|---|---|---|---|---|---|
| Coefficient | 0.88 | 0.95 | 0.69 | 0.89 | −0.74 | −0.37 | 0.89 |
| p Value | 9.7E-11 | 1.3E-15 | 2.5E-05 | 7.3E-11 | 3.7E-06 | 0.045 | 7.9E-11 |
Note: DAPI = DAPI channel autofluorescence; GFP = GFP channel autofluorescence; Max Amp = maximum bending amplitude; Max Speed = maximum movement speed; Pumping = pharyngeal pumping; Speed = mean movement speed.
Survival Rate Is Highly Correlated With Maximum Speed and Well Correlated With Other Health Markers. Spearman Correlations Between Survival Rate and Each Health-Span Parameter
| Speed | Max Speed | Max Amp | Pumping | DAPI | GFP | Thermotolerance | |
|---|---|---|---|---|---|---|---|
| Coefficient | 0.88 | 0.95 | 0.69 | 0.89 | −0.74 | −0.37 | 0.89 |
| p Value | 9.7E-11 | 1.3E-15 | 2.5E-05 | 7.3E-11 | 3.7E-06 | 0.045 | 7.9E-11 |
| Speed | Max Speed | Max Amp | Pumping | DAPI | GFP | Thermotolerance | |
|---|---|---|---|---|---|---|---|
| Coefficient | 0.88 | 0.95 | 0.69 | 0.89 | −0.74 | −0.37 | 0.89 |
| p Value | 9.7E-11 | 1.3E-15 | 2.5E-05 | 7.3E-11 | 3.7E-06 | 0.045 | 7.9E-11 |
Note: DAPI = DAPI channel autofluorescence; GFP = GFP channel autofluorescence; Max Amp = maximum bending amplitude; Max Speed = maximum movement speed; Pumping = pharyngeal pumping; Speed = mean movement speed.
Locomotion is also characterized by maximum bending amplitude, which depends on thick filament organizing centers that transmit the force of muscle contraction into locomotion (21). Maximum bending amplitude decreased after Day 5 in all strains (Figure 2C). By Day 12 and beyond, values for all short-lived strains were significantly reduced compared with wild type, in accordance with an aging-dependent marker. To our knowledge, this is the first characterization of maximum bending amplitude with age in C elegans on solid media. Thus, locomotion markers in Figure 2A–C are all aging dependent, making them strong candidates for health-related diagnostics of aging.
Heat stress can disrupt cellular homeostasis though protein misfolding. Thermotolerance decreased with age in all strains tested, and short-lived mutants had significantly reduced thermotolerance compared with wild-type worms. The strain-specific reduction was greatest for hsf-1, which was 34% (p = 3.0E-09), followed by 29% for mev-1 (p = 1.3E-06), 28% for sir-2.1 (p = 4.1E-06), and 24% for daf-16 (p = 2.6E-04) on Day 13 (Figure 2D; Tukey’s test). Results indicate that thermotolerance is an aging-dependent marker.
Autofluorescence and Pharyngeal Pumping Are Time-Dependent Markers of Health
Both DAPI and GFP channel autofluorescence have been used as aging metrics in C elegans (27,28). To better understand the relationship between autofluorescence and age, we measured DAPI channel (350/50ex,450/50em) and GFP channel (480/20ex, 510/20em) autofluorescence throughout adulthood. DAPI channel fluorescence was seen to increase with age in N2 wild-type, mev-1 and hsf-1 strains (Figure 3A). Surprisingly, daf-16 DAPI autofluorescence stopped increasing after Day 12, whereas sir-2.1 remained steady throughout life. Although GFP channel autofluorescence increased with time for all strains, values in daf-16 and sir-2.1 mutants were never greater than wild-type and were significantly lower by Day 19 (p = .039 and .002 respectively, Tukey’s test, Figure 3B). Conversely, mev-1 mutants displayed significantly higher GFP autofluorescence by Day 19 (p = .0002). hsf-1 mutants had significantly higher GFP autofluorescence on Day 12 of adulthood (p = .006) but not on Day 15 (p = .99). Based on results for the strains used, DAPI and GFP channel autofluorescence are supported as time-dependent markers of health.
Autofluorescence and pharyngeal pumping were not altered sooner in short-lived mutants. (A) DAPI channel autofluorescence, (B) GFP channel autofluorescence, and (C) pharyngeal pumping of wild-type and short-lived mutants throughout life. Insufficient numbers of hsf-1 mutants survived to Day 19 to take complete measurements for that time point. Pumping was not significantly different at any time point for any strain compared with wild type. Error bars indicate SEM. Data were pooled from three independent experiments. n = 45/strain for autofluorescence measurements, n = 200/strain for pharyngeal pumping. *p < .05, **p < .01, ***p < .001, as compared with wild type on the same day by analysis of variance with post hoc Tukey’s test.
The ability of the worms to feed is based on the pumping rate of their pharynx, which they modulate according to availability of food (29) and which has been used as a marker for health in C elegans (4,5,27). Pumping rates were measured on Days 1, 5, 8, 12, and 15, after which no appreciable pumping was detected. Although pumping declined over time (Figure 3C), rates were similar or higher in short-lived mutants compared with wild type, indicating that pumping is time dependent. This time dependency was also observed for long-lived mutants (9) where pumping was not detected after 15 days.
Chronological Health Span Is Reduced in Short-Lived Strains
To further investigate the relationship between health span and life span, we calculated the amount of time spent healthy in wild-type animals for each marker. Because candidate aging markers do not always decrease (eg, autofluorescence; Figure 3) nor change by more than 50% (maximum bending angle; Figure 2), we implemented a method that uses the most dynamic period of change taking place in a wild-type animal to determine when health falters for a given marker. Thus, this method is customized to the natural transition for the health marker being used. This was determined by measuring rates of change over a sliding window of three time points, with the last time point in the window that demonstrates the steepest slope indicating the first time point of the geriatric state (see Methods). Based on this method, which we refer to as Sliding Window Analysis of Vigor, the onset of gerospan for aging-dependent markers in N2 wild-type animals was Day 12 for speed and maximum speed, Day 13 for thermotolerance, and Day 19 for maximum bending amplitude (Table 2; Supplementary Figure 3). With these transition points, the percent changes from peak values at the time of gerospan onset were near middle- to late-middle age for average speed, maximum speed, and thermotolerance (36.3%, 48.3%, and 43.6%, respectively; Table 2). However, the onset of gerospan for maximum bending amplitude was 72.8%, indicating that it is the only aging-dependent marker with a transition point that occurs in late life, closer to the time humans are typically characterized as entering a geriatric state. In order to determine analogous transitions into the geriatric state in the short-lived strains, interpolation using linear regression models (Supplementary Figure 4 and Supplementary Table 2) was carried out based on wild-type transition values.
Empirical Determination of the Geriatric State for Individual Health Markers
| Marker | Day | Mean | Max | % Max |
|---|---|---|---|---|
| Speed (µm/s) | 12 | 75.2 | 207.3 | 36.3 |
| Max speed (µm/s) | 12 | 188.7 | 390.2 | 48.3 |
| Max amplitude (µm) | 19 | 150.6 | 206.9 | 72.8 |
| Thermotolerance (hours) | 13 | 2.3 | 5.3 | 43.6 |
| Marker | Day | Mean | Max | % Max |
|---|---|---|---|---|
| Speed (µm/s) | 12 | 75.2 | 207.3 | 36.3 |
| Max speed (µm/s) | 12 | 188.7 | 390.2 | 48.3 |
| Max amplitude (µm) | 19 | 150.6 | 206.9 | 72.8 |
| Thermotolerance (hours) | 13 | 2.3 | 5.3 | 43.6 |
Note: % Max = the percentage of the maximal value when geriatric; Day = day of adulthood geriatric onset for wild-type N2 worms; Max = maximum value each marker observed in wild-type over life span; Max Amp = maximum bending amplitude (µm); Max Speed = maximum movement speed (µm/s); Mean = mean value for each marker when N2 becomes geriatric; Speed = mean movement speed (µm/s).
Empirical Determination of the Geriatric State for Individual Health Markers
| Marker | Day | Mean | Max | % Max |
|---|---|---|---|---|
| Speed (µm/s) | 12 | 75.2 | 207.3 | 36.3 |
| Max speed (µm/s) | 12 | 188.7 | 390.2 | 48.3 |
| Max amplitude (µm) | 19 | 150.6 | 206.9 | 72.8 |
| Thermotolerance (hours) | 13 | 2.3 | 5.3 | 43.6 |
| Marker | Day | Mean | Max | % Max |
|---|---|---|---|---|
| Speed (µm/s) | 12 | 75.2 | 207.3 | 36.3 |
| Max speed (µm/s) | 12 | 188.7 | 390.2 | 48.3 |
| Max amplitude (µm) | 19 | 150.6 | 206.9 | 72.8 |
| Thermotolerance (hours) | 13 | 2.3 | 5.3 | 43.6 |
Note: % Max = the percentage of the maximal value when geriatric; Day = day of adulthood geriatric onset for wild-type N2 worms; Max = maximum value each marker observed in wild-type over life span; Max Amp = maximum bending amplitude (µm); Max Speed = maximum movement speed (µm/s); Mean = mean value for each marker when N2 becomes geriatric; Speed = mean movement speed (µm/s).
The transition to chronological gerospan using aging-dependent markers, based on interpolation of linear regression data, occurred earlier in short-lived mutants and is presented graphically in Figure 4A–D. The transition for daf-16, mev-1, hsf-1, and sir-2.1 for maximum speed was 8.1 ± 0.3, 5.6 ± 0.5, 4.6 ± 0.5, and 8.7 ± 0.2 days, respectively, which were all reduced compared with wild type (13.4 ± 0.3 days; Figure 4B). Qualitatively similar results were obtained for mean speed (Figure 4A). The onset of gerospan for maximum bending amplitude was 9.6 ± 0.5 days for daf-16, 4.5 ± 0.9 days for mev-1, 7.6 ± 0.5 days for hsf-1, and 9.4 ± 0.4 days for sir-2.1, earlier than wild type (18.6 ± 2.0 days) (Figure 4C). The onset of gero span for thermotolerance (Figure 4D) was also earlier than wild type (15.3 ± 0.2 days) in daf-16 (13.3 ± 0.1 days), mev-1 (13.5 ± 0.1 days), hsf-1 (12.1 ± 0.3 days), and sir-2.1 (14.0 ± 0.2 days). Results were in line with expectations for intrinsically health-deficient animals.
Physiological health span is decreased for locomotion markers and increased for thermotolerance in short-lived mutants. Chronological health span and gerospan for movement speed, maximum speed, maximum bending amplitude, and thermotolerance markers were calculated in A–D, respectively. (E,H) Same as in A–D, but for physiological health span and gerospan. Health span and gerospan are indicated by lighter and darker hues, respectively. Error bars indicate SEM for the prediction of the onset of gero span. The cumulative quality-adjusted survival, at metric sensitive to both the total quality and duration of life was calculated for speed, maximum speed, maximum amplitude, and thermotolerance in I–L.
Physiological Comparisons Reveal That Health Span Is Nonproportional With Life Span
We calculated the physiological health span for each aging-dependent marker in Figure 4E–H. Health span for maximum speed was reduced in daf-16 (40.9 ± 1.6%), sir-2.1 (25.8. ± 2.2%), mev-1 (23.6. ± 2.5%), and hsf-1 (41.1 ± 1.2%) compared with wild type (49.7 ± 1.3%; Figure 4F), with similar results for mean speed (Figure 4E). Physiological health span for maximum bending amplitude was also reduced in daf-16 (48.5 ± 2.4%), hsf-1 (20.8 ± 4.0%), and sir-2.1 (44.8 ± 1.8%) as compared with wild type (35.2 ± 1.4%; Figure 4G). Surprisingly, physiological health span for thermotolerance was extended in all of the short-lived mutants (daf-16: 60.8 ± 0.5%, mev-1: 55.0 ± 0.4%, hsf-1: 55.6% ± 0.5%, and sir-2.1: 58.3 ± 0.4%) compared with wild type (47.0 ± 0.4%; Figure 4H). As the observed maximum life span of a strain may be dependent on the size of the population, we additionally calculated physiological health span for each strain normalized by their median life span (Supplementary Figure 5). However, no qualitative differences in the physiological health span were evident between the two methods. These results make interpreting previously reported decreases in physiological health span markers in long-lived mutants (9) challenging and suggest that the apparent simplicity of health-span metrics normalized to remove time may be misleading.
For comparison, we also evaluated strains using a cumulative quality-adjusted survival metric (10,30) (Figure 4I–L). In this method, survival rate and health performance at each time point are first normalized to the maximum value from wild type and then multiplied together to yield a survival curve adjusted by the quality of life experienced for each strain. The area under the quality-adjusted survival curve between each consecutive time point is then calculated and cumulatively summed to provide a diagnostic of both total quality and duration of life. Based on this analysis, wild-type worms have the highest quality of life for all aging-dependent markers. A similar reduced total quality of life for daf-16 was reported for maximum speed, as compared with wild type (10). Unlike physiological health span, the quality-adjusted metric does not factor out time, which may be important if extrinsic factors that influence aging play a significant role in health among differently lived strains.
Discussion
Evaluation of Health-Span Markers
Three major categories of health are frequently used to investigate aspects of aging in C elegans. They include muscle function/integrity, cellular accumulation of autofluorescent pigments, and resistance to acute perturbation of homeostasis, particularly proteostasis. In the clinic, muscle function and integrity with age are judged by muscle wasting, or sarcopenia, which is diagnosed according to gait speed, grip strength, and muscle mass (31). In C elegans, movement speed is akin to human gait speed and maximum movement speed has been suggested to be the one of the most informative metrics of health in C elegans due to its high correlation with longevity and ability to predict remaining life span (10). In agreement with this, maximum speed was the highest correlated with survival out of the markers tested in the present study. We also considered maximum bending amplitude as a marker akin to grip strength that is distinct from movement speed, as C elegans mutants deficient in components of force-transmitting sarcomeres exhibited differences in bending amplitude but not locomotion (21). Out of the aging-dependent markers, bending amplitude was found to fail latest in life. A previous characterization of the locomotion of C elegans while swimming reported an increase in the maximum difference in curvature with age. These two opposing results suggest that bending amplitude on solid media and curvature while swimming measure different aspects of health and behavior. Supporting this notion, the kinematics of crawling has been shown to be distinct from that of swimming and is characterized by a greater amplitude and smaller frequency (32). Finally, although the pharyngeal muscle may show variable deterioration among strains (not tested), cessation of pumping was largely invariant among wild-type and short-lived strains, similar to findings in long-lived mutants (9), suggesting that this marker is related to longevity mainly by virtue of being dependent on the passage of time.
Cellular autofluorescence, a second category with the potential to be a diagnostic of aging, can occur due to accumulation of endogenous fluorescent particles such as lipofuscin, advanced glycation end products, flavins, dihydronicotinamide-adenine dinucleotide phosphate, elastin, and collagen (6,33), each with distinct but sometimes overlapping spectral excitation and emission properties. Of these particles, the accumulation of lipofuscin, the oxidized bi-products of lysosomal degradation, has been a focus in mammals as a marker of aging (34). In worms, lipofuscin accumulation has been proposed as a marker of aging by measuring changes in autofluorescence using a DAPI channel filter (27). However, Coburn and colleagues (28) demonstrated that the presence of DAPI channel fluorescence did not increase with induced oxidative stress nor with age in C elegans. Instead, they showed that this autofluorescence occurs as a burst upon death and concluded that the signal was due to anthranilate acid esters originating from gut granules. Although that study and the present study were able to detect significant increases in GFP channel autofluorescence with age, they were not consistently greater in short-lived strains and so failed the criteria to be considered an aging-dependent marker. However, it is important to note that there are some differences in the way in which channel-specific fluorescence is measured (eg, see DAPI filter settings in (6,35) compared with (9,28), and ones used in the present study). Thus, autofluorescence cannot be ruled out as a possible aging-dependent marker, but will require standardization in approach and further characterization of what is actually measured at specific wavelengths in C elegans.
The third category of health used to investigate aspects of aging in C elegans involves proteostasis and the ability to maintain somatic function, measured by survival, in the presence of acute stress. Mechanisms contributing to proteostasis are necessary for survival under heat stress (36) and decline with age (37). On the first day of adulthood, proteostatic maintenance undergoes a dramatic transition in C elegans that can be measured in hours (38,39). As one of the most robust aging-dependent markers, we find a second transition that occurs in midlife for wild type that is delayed in short-lived strains relative to their life span. Although the result is surprising, the ability to maintain proteostasis may be more influenced by extrinsic aging factors than other markers.
Considering a Role for Extrinsic Factors in Health Span
Mortality is determined by a combination of intrinsic and extrinsic factors. Intrinsic aging is thought to be dependent on somatic maintenance and senescence, whereas extrinsic aging is driven by environmental conditions (40). Although intrinsic and extrinsic factors can affect longevity and health, the relative contribution on each is still largely unknown. Our assessment of aging-dependent health parameters included the binary separation of health span and gerospan based on the physiological age of each strain. When comparing health span physiologically, time is factored out, but this discounts the impact of extrinsic factors in longer-lived animals, which in the current study is the N2 wild-type strain. For example, analysis of thermotolerance indicated physiologically extended health span in all short-lived mutants compared with wild type (Figure 4H). This disproportionality between health span and life span could be due to a requirement of the passage of time to allow extrinsic factors, like background radiation (41), to damage proteins to the point where thermotolerance is impacted. Thus, although physiological comparisons have limited use in accessing the overall quality of life a particular intervention may provide, they may be beneficial in estimating the susceptibility of a particular health marker to extrinsic forces.
The Relative Importance of Life Span and Health Span
The question remains, is the disproportionality between life span and health span cause for concern in the translation of aging research? On one hand, the extended period of frailty observed in C elegans long-lived mutants could have “disastrous economic and social consequences if applied to humans” (9). On the other hand, we could apply reduction ad absurdum to this notion with our results from short-lived mutants. If longer life span without a proportional increase in health span is disastrous, then conversely, shorter life span with a disproportionately longer health span, like we observed for thermotolerance, would be advantageous. Although it may be true that there are economic benefits of having a shorter-lived, healthier population, such a proposition brings to mind images of a dystopian world akin to that portrayed in Logan’s Run, where the aged are culled on their 30th birthday before they can burden society. Additionally, this view of longevity research is missing one important consideration, that time itself is a precious commodity which scientists and economists alike have strived to appraise (42–44). Therefore, while chronological versus physiological comparisons are useful to evaluate the effect environment plays in the deterioration of health, evaluating the utility of pathways that affect aging for therapeutic potential is best considered using metrics that account for both health and time.
Funding
This work was supported by grants from the National Institute on Aging of the National Institutes of Health (R00AG037621) and by the Ellison Medical Foundation (AG-NS-1087-13), both to A.R. Additionally, this project was supported by Institutional Development Awards (IDeA) from the National Institute of General Medical Sciences of the National Institutes of Health (grant numbers P20GM0103423 and P20GM104318, respectively).
Acknowledgments
The authors thank George Sutphin and Santina Snow for discussions and critical comments. Some strains were gratefully provided by the CGC, which is funded by the NIH Office of Research Infrastructure Programs (P40 OD010440).
References
Author notes
Address correspondence to Aric Rogers, PhD, Davis Center for Regenerative Biology and Medicine, Mount Desert Island Biological Laboratory, 159 Old Bar Harbor Road, Bar Harbor, ME 04609. E-mail: [email protected]
Decision Editor: Rafael de Cabo, PhD
