Autofluorescence as a measure of senescence in C. elegans: look to red, not blue or green

Autofluorescence in C. elegans is spatially and spectrally complex

We first sought to determine whether autofluorescence in C. elegans is due to a homogenous material. To qualitatively characterize this material, we performed an excitation/emission scan of a live, aged C. elegans using confocal microscopy with a tunable laser for excitation and a detector capable of resolving emission spectra at each image pixel. The results, shown in Figure 1b, demonstrate that there are two main sources of autofluorescence in C. elegans: the intestine and the uterus. In animals of this age the uterus is often filled with unfertilized oocytes containing yolk proteins, which are known to be autofluorescent. Therefore we speculate that some of the uterine fluorescence observed may be due to yolk accumulation. Regardless, the autofluorescence in this organ is distinctly heterogeneous in terms of emission wavelengths from any given excitation, and is distinct from the autofluorescence of the intestine. The material in the intestine is also qualitatively not homogenous: for any given excitation, the material fluoresces with more blue-shifted or more yellow-shifted emissions in different spatial regions. These differences are more subtle in the intestine than uterus, however.

Red autofluorescence increases linearly with time, blue peaks near death only

Overall, autofluorescence in C. elegans appears to be due to a spatially heterogeneous mix of substances with distinct spectral properties. We next sought to determine how this autofluorescence changes in time. To do so, we used a culture apparatus in which each individual animals are isolated and can be imaged under a widefield fluorescent microscope in situ without having to be transferred to a microscope slide [18]. In this way, we were able to collect longitudinal timecourses of autofluorescence for 40 individual animals from hatching to death in a minimally invasive fashion.

Images at 10× magnification in the three fluorescence ranges illustrated in Figure 1a were collected for each individual every eight hours. Animals were assumed to be alive after each time-point if there was visible movement following blue-light stimulus. This produced a detailed survey of C. elegans autofluorescence across space and time, and among individuals.

Figure 2 shows selected images throughout the life of two specific individuals from our study. Several major trends are clear. First, increases in blue autofluorescence are almost exclusively a near-death phenomenon, appearing from 6–12 hours before each individual's death. As previously described, this autofluorescence is predominantly intestinal in nature [17]. While we observe some blue autofluorescence in the gonad and uterus, this material is co-localized with red autofluourescence in that organ, and does not appear to increase near death. Green autofluorescent material, in contrast, is observable throughout life, primarily in the intestine. Near death, green autofluorescence increases dramatically in the intestine and also peripheral tissues. Finally, red autofluorescent material is observable in both the intestine and gonad/uterus throughout life, and does not noticeably increase near death. Interestingly, though both red and green materials are visible in the intestine, there is limited spatial overlap within this tissue. This suggests that these materials are biologically distinct from one another, in agreement with the observations of spatial inhomogeneity in autofluorescence from the spectral imaging shown in Figure 1.

Figure 2b demonstrates that temporal trends in autofluorescence are similar regardless of whether the mean whole-body fluorescent intensity is (solid lines) or 95^th-percentile intensity (dashed lines) is plotted. The 95^th-percentile intensity, calculated across all body pixels, is an outlier-resistant proxy for maximum fluorescence intensity. According to both measures, blue autofluorescence remains static until a dramatic peak near death, red autofluorescence increases roughly linearly with time, and green autofluorescence appears to combine both trends.

Population averages of blue autofluorescence are confounded by dead and dying individuals

While Figure 2 presents data from only two individuals, the same measures averaged across the 40-animal population are shown in Figure 3. Mean and 95^th-percentile fluorescence were calculated for each individual at each time-point, and the mean of these measures across the whole population is plotted as a function of time. This analysis shows that, when centered at the time of death (Figure 3a), the population trends are very similar to the individual trends described above: a consistent increase of red autofluorescence over time, a peak in blue autofluorescence only near death, and a combination of both of these patterns in green.

Without individual-level longitudinal analysis, however, it is not possible to recenter time trends around each individual's death. As such, population-average fluorescence measurements usually appear as in Figure 3b, where average autofluorescent trends are plotted against time of hatching. In this analysis, blue autofluorescence appears to increase somewhat over time in the population. However, this effect is driven almost entirely by dead and near-dead animals, as shown in Figure 3c, which excludes animals within 12 hours of death from the trend. Therefore, we suspect that previous observations of increasing blue autofluorescence in aging C. elegans were effectively measures of the fraction of dead and dying animals in the study populations.

Red autofluorescence is the best predictor of future lifespan in C. elegans

As discussed above, previous studies have used blue autofluorescence to compare aging, aging rates, and health between different populations of C. elegans. In general, longer-lived populations have been found to have lower blue autofluorescence at any given time-point. As before, our results suggest that this simply reflects a smaller proportion of near-dead animals, rather than a slowed aging rate per se. (These two parameters are obviously quite related, however. Thus, results of analyses of blue fluorescence remain meaningful, though their precise interpretation must be revised.)

In order to shed light on which autofluorescent wavelengths are most reflective of individual and population senescence, we asked which measures best correlate with an individual's remaining lifespan. While “senescence” is a difficult concept to define, it is clear that among animals of the same age, those with more days to live are in a meaningful sense less senescent than those with fewer days remaining. Therefore we adopt this measure as a plausible, though certainly imperfect, proxy for senescence.

Figure 4 shows remaining lifespan plotted against autofluorescence for each individual in our population, at each time-point measured. Across all ages, blue autofluorescence is weakly correlated with remaining lifespan, and this correlation is driven entirely by animals within a day of death (Figure 4a). Moreover, for animals in late adulthood (after the reproductive span but before the onset of mortality), blue autofluorescence is entirely unrelated to remaining lifespan (Figure 4b). In contrast, red autofluorescence is highly predictive of remaining lifespan in the overall population, in late adulthood, and in older populations (Figure 4d,e). Only in highly aged animals in which nearly the whole population is within a day of demise does the level of blue autofluorescence become more predictive of remaining lifespan than that of red autofluorescence.

These results confirm our previous findings [18] that the 95^th-percentile of whole-body red autofluorescence is well correlated with an individual's future lifespan. The difference in future lifespan between individuals with high vs. low levels of red autofluorescence is quite pronounced. This difference peaks at day 10 post-hatch (a time when <5% of the population has died). At this age, individuals with lower-than-average red autofluorescence go on to live an average of 1.66 days longer than those with higher-than-average red autofluorescence. (For scale, the standard deviation of lifespans in our dataset is 1.4 days.) From days 7–12 post-hatch (inclusive), the difference in mean remaining between brightly autofluorescent and dimly autofluorescent animals is over 1 day.

Based on this new analysis, red autofluorescence (mean or 95^th percentile) is the best proxy measure for individual and population health. Blue auto-fluorescence, in contrast, predicts essentially only incipient demise. As green autofluorescence appears to combine properties of both blue and red auto-fluorescence (increase over time as well as a peak near death), we believe it is difficult to interpret fluorescence levels in these wavelengths, and did not include it in this analysis. Moreover, green autofluorescence has been previously reported to not correlate with future lifespan [19].

Oxidative damage does not increase auto-fluorescence in C. elegans

As lipofuscin is believed to consist of highly oxidized proteins and lipids, it is unsurprising that previous observations of mammals and mammalian cells have found that conditions in which oxidative damage is increased also raise lipofuscin levels. In particular, both high O₂ concentrations and supplementation with redox-active iron increase endogenous autofluorescence in multiple in vivo and in vitro experiments in different mammalian models(reviewed in ref. [7]). In C. elegans, media supplemented with 15 mM ferric ammonium citrate has been shown to increase oxidative damage (as measured by protein carbonylation) [17,23] and shorten lifespan [23]. However, such supplementation was previously reported to not increase blue auto-fluorescence [17]. As mammalian lipofuscin is often reported to fluoresce in green to red wavelengths when viewed in vivo [10], we hypothesized that, though blue autofluorescence in C. elegans appears to be distinct from mammalian lipofuscin [17], green or red autofluorescence might increase following oxidative damage. We therefore repeated the analysis of C. elegans autofluorescence with redox-active iron supplementation from ref. 17, and extended it to green and red fluorescent wavelengths (Figure 5). We confirmed that iron supplementation shortens lifespan dramatically (Figure 5a); however autofluorescence is not much altered at any time in any wavelength (Figure 5b). Indeed, all significant changes between control and treatment groups are in the direction of reduced auto-fluorescence. Examining autofluorescence specifically in the intestine or gonad produced identical results (data not shown). While inter-individual variability in autofluorescence is increased in iron-treated animals at very late timepoints, there is no evidence of wholesale increases in blue, green, or red autofluorescence as a result of these pro-oxidation conditions. These results corroborate previous observation of no increase in blue autofluorescence in mutant C. elegans strains that produce high levels of reactive oxygen species [16], and stand in marked contrast to the behavior of lipofuscin as characterized in mammals.

Spectral imaging

Seven day-post-hatch individuals (4^th day of adulthood at a culture temperature of 23.5°C) were mounted to a 2% agarose pad on a glass slide and immobilized under a coverslip with 10 mM levamisole. The animals were imaged on a Zeiss LSM 510 Meta confocal microscope using a tunable laser (Chameleon, Coherent Inc.) as an excitation source. The instrument was configured for two-photon excitation and the Meta monochromator was used to collect emission wavelengths in ∼10-nm bins. The resulting spectral images were computationally straightened as described previously [18]. For each excitation wavelength, the image of emissions from each wavelength bin was false-colored according to the wavelength using Bruton's algorithm, and combined using the screen image compositing operator.

Individual time-course imaging and quantification

Using culture techniques previously described [18], we reared 40 individual animals on gel pads embedded in a glass slide, such that each individual was trapped on the gel surface between the pad and a thin layer of PDMS silicone acting as a flexible, gas-permeable, and optically transparent “coverslip”. These individuals, bearing the spe-9(hc88) temperature-sensitive sterility mutation (resulting in otherwise normal lifespans and physiology [24,25]), were deposited on the pads as embryos and reared at the restrictive temperature of 23.5°C. Every eight hours, the slides were placed on a Zeiss Axioskop microscope, each animal was manually located and brought into focus under 10× magnification, and fluorescence images were obtained using DAPI, GFP, and TRITC filtersets (Chroma Technology; filtersets 31000v2, 41017, and 41002c, respectively). At each time, images of autofluorescent beads (Molecular Probes) were also obtained to control for temporal variability in light-source intensity. Similarly, flat-field images using uniformly autofluorescent acrylic slides (Chroma Technology) were collected to control for spatial variability in fluorescence illumination. Fluorescent images were dark-current and flat-field corrected and divided by the average per-bead fluorescence, in order to produce fluorescence values comparable across space and between time-points. The same exposure time was used for each fluorescence channel across the full experiment.

Images of each individual were computationally straightened using custom software [18] and times of death were manually annotated. Autofluorescence intensities were quantified as follows. First, background autofluorescence due to the culture medium was estimated by calculating median fluorescence in a region adjacent to the position of the animal in each image. After subtracting this background value, the mean and 95^th percentile of intensity in the image pixels belonging to the animal's body were calculated and recorded. (Our findings are not substantially altered by omitting this background subtraction; data not shown.)

Iron supplementation and imaging

15 mM ferric ammonium citrate (FAC) plates were prepared as previously described [23], and supplemented with 5-fluoro-2′-deoxyuridine (FUDR) to a final concentration of 40 μM to prevent reproduction [26,27]. Gravid wild-type (N2) C. elegans were hypochlorite treated to isolate age-synchronized first-larval stage offspring, which were reared on standard plates until the 4^th larval stage at which point individuals were transferred to FAC/FUDR plates and FUDR-only control plates. For lifespan assays, 50 individuals were transferred to each of four FAC and four control plates. For image acquisition, four additional FAC and control plates with 50 individuals each were prepared. At 0, 5, 10, and 15 days post transfer to FAC, individuals were mounted microscope slides with agarose pads and levamisole as above, and fluorescence images obtained. At days 0 and 5, 20 individuals were measured for each group; at day 10, 20 FAC and 19 control animals were examined; at day 15, 14 FAC and 13 control animals were examined.