Autophagy drives epidermal deterioration in a Drosophila model of tissue aging

Age-related morphology changes in the adult epidermis

We first tested if Drosophila adult epidermal morphology changes with age. The membranes of the ventro-lateral abdominal epidermis (see Figure 1A) were visualized with a Fasciclin (Fas) III antibody (Figure 1C-F). In 1d old flies the epidermis was a uniform monolayer of mononuclear cells (Figure 1C). Surprisingly, although laboratory-reared flies can live for several months, loss of epidermal membrane labeling was apparent within a few days of eclosion (Figure 1D) and progressed steadily so that very little epidermal membrane labeling remained by six weeks (Figure 1F). A comparable deterioration was also observed using a FasIII-GFP fusion construct or anti-Coracle labeling (data not shown). To quantify the morphological changes in control flies we defined four classes reflecting an increasing loss of epidermal membrane labeling (see Figure 1C-F). All pair wise comparisons of class distributions at 1, 3, 7, 14, and 42 days were significant (Figure 1B). Co-labeling of epidermal nuclei revealed that these initially persisted despite the loss of epidermal membranes (Figure 1D, E). In contrast, by six weeks of age, epidermal nuclei were also strongly decreased in number (Figure 1F and G).

We next examined epidermal changes at the structural level by transmission electron microscopy (TEM) in control flies (Figure 2). On day one, a continuous epidermal monolayer (average thickness 1.34 μm, n = 4) (Figure 2A) was observed distinct from the underlying muscles. A strong loss of cytoplasmic volume was detected as early as 3 d (average thickness 1.03 μm, n = 5), followed by a further gradual reduction at later time points (Figure 2B-D). From 3 to 14 days epidermal nuclei were condensed, while at 42 days clearly defined nuclei were often absent. Despite these other markers of epidermal deterioration, a thin but continuous epidermal layer and its basal lamina were maintained even at 42 d. Together, these fluorescence and TEM data suggest that progressive changes in the epidermis occur as the fly ages, and may represent one of the earliest reporters of insect tissue aging.

A gene that prolongs lifespan decelerates epidermal aging

We next asked if epidermal aging would be altered in a mutant affecting Drosophila lifespan. Decreasing insulin signaling prolongs lifespan in diverse organisms [3, 22-26].

In Drosophila, flies that are homozygous mutant for the insulin-like receptor substrate (chico) have increased lifespan [27-29]. Therefore, we assessed markers of epidermal aging in chico¹ flies. On day one, epidermal morphology in chico¹ mutants was indistinguishable from controls (Figure 3A, E). However, epidermal membrane labeling over time was strongly preserved (Figure 3A-C) in comparison to controls (compare Figure 3B and C to Figure 1E and F). Morphological class distributions at 14 and 42 days were significantly different from age-matched controls (Figure 3E). Indeed, the 14 d old chico¹ mutant epidermis most closely resembled 3 d old controls and the 42 d old chico¹ mutant epidermis still looked younger than that of 14 d old controls. By TEM, the decrease in cytoplasmic volume and epidermal thickness also appeared to be decelerated (Figure S1A-C). These results are consistent with the idea that chico¹ mutants, which have an extended lifespan [27] also retain the characteristics of a younger epidermis.

A mutant with shortened lifespan exhibits accelerated epidermal aging

Conversely, we investigated if a mutation that shortens lifespan would affect epidermal aging. In humans, disorders that resemble premature aging, including atypical Werner syndrome and Hutchinson-Gilford progeria syndrome, have been linked to mutations in genes encoding lamin proteins [30-32]. The premature aging phenotype in these disorders is accompanied by a reduced lifespan. Whether laminopathy disorders truly represent accelerated aging is vigorously debated [33] but it is clear that many of these diseases are characterized by an early onset of gene expression changes and morphological skin changes resembling those of normal aging [34]. We thus investigated if a mutation in a Drosophila lamin gene would alter epidermal morphology. Although many lamin mutant alleles do not survive to the adult stage, adult escapers of a B-type lamin allele, lam^G262[35], have a shortened lifespan [36]. In 1 d old lam^G262 flies epidermal morphology resembled controls (Figure 4A), suggesting that development of the adult epidermis during pupariation proceeded normally. By 3 days and 7 days, the lam^G262 mutant epidermis showed more morphological deterioration than control epidermis of the same age (Figure 4B and C) and the classification of lam^G262 mutant epidermal pictures (Figure 4D) supported an accelerated epidermal deterioration at these time points compared to controls (Figure 4E). In 14 d old lam^G262 flies epidermal morphology did not look significantly different. Finally, TEM revealed strong nuclear condensation at 3 days compared to control nuclei at this time point (Figure S2B). Although there was no obvious difference in epidermal thickness in 3 d old lam^G262 mutants compared to age-matched controls, TEM samples of lam^G262 mutants exhibited altered cellular morphology. In summary, these results suggest that epidermal aging is accelerated at early time points in lam^G262 mutants compared to control flies.

Autophagy correlates with and drives epidermal deterioration

The dramatic loss of membrane and cytoplasm in aging epidermal cells suggested that autophagy might drive the observed changes in morphology. We thus examined the aging epidermis for markers of autophagy. Consistent with the hypothesis that autophagy was responsible for the loss of membrane and cytoplasm, we found large autophagosomes in 14 d old control flies (Figure 5A). We next quantified epidermal autophagosomes at different ages and in different genotypes using a transgene, UAS-LC3-GFP, that labels autophagosomal membranes. We compared 3 d and older flies to avoid high autofluorescence levels in the 1 d old epidermis. Autophagosome numbers increased significantly from 3 d to 42 d in controls (Figure 5B and C). As expected, autophagosome numbers remained low in Atg7^d77 flies with disrupted autophagy [37] and in chico¹ mutants regardless of age. By contrast, autophagic activity on day 3 (but not day 7) was higher in lam^G262 flies than in controls. In conclusion, an early rise in autophagy levels (lamin mutant) compared to controls correlates with accelerated epidermal aging. Conversely, a decrease in autophagy levels (chico or Atg7 mutants) correlates with decelerated epidermal aging.

These findings prompted us to test if autophagy drives age-related epidermal morphology changes since autophagy can drive developmental [38] and physiological remodeling of tissues [39]. Like chico¹ and lam^G262 mutants, the initial epidermal morphology of Atg7^d77 flies resembled controls (Figure 6A). Interestingly, at 14 d loss of epidermal membrane labeling was strongly decelerated in comparison to controls (compare Figure 6B to Figure 1E and Figure 6C to Figure 1B). Additionally, TEM analysis revealed a thicker epidermal layer in 14 d old Atg7^d77 flies compared to control epidermis of the same age (Figure S4).

To test if Atg7 controls epidermal aging in a tissue-autonomous manner, we used NP2108-GAL4 to express an RNAi targeting Atg7. NP2108-GAL4 is expressed in the adult epidermis. It is also expressed transiently in the larval fat body at eclosion, but this expression persists less than 24 hours and is therefore unlikely to affect epidermal aging at later stages (Figure S5). No expression in the adult fat body was observed. Epidermal knockdown of Atg7 resulted in decelerated epidermal aging similar to the mutant, suggesting a tissue-autonomous role for autophagy (Fig. 6D). Together, these results suggest that changes in epidermal morphology with age are driven by autophagy.

Complex Roles for Autophagy in Tissue Aging

In the context of aging of the organism as a whole, it is generally believed that autophagy is primarily protective [20]. Increased autophagy correlates with increased lifespan in worms [41] and normal levels of systemic autophagy are required for the lifespan extension observed upon reduction of insulin signaling [42]. In Drosophila, maintenance of a sufficiently high level of autophagy in the nervous system and muscles seems to be particularly important for lifespan extension since neuronal-specific induction of autophagy leads to healthier neurons and lifespan extension [16, 43]. The role of autophagy in other aging tissues such as the epidermis has not been investigated until now. Unexpectedly, we observe that autophagy both correlates with and drives morphological deterioration of an aging adult tissue. Our results highlight the complex relationship between autophagy and the aging of individual tissues. Given the contrasting roles of autophagy in neuronal and epidermal tissue aging we hypothesize that the relationship between tissue-aging and lifespan may be unique for each organ. In other words, the net effect of autophagy on lifespan is likely determined by its particular role in those tissues most critical for survival. Specifically, we propose that in tissues whose healthspan is essential for setting lifespan, such as the nervous system, autophagy is protective, while some other tissues whose healthspan is not critical, such as the epidermis, may be degraded by autophagy.

While it is expected that all organs will deteriorate with age, it is surprising that this seems to be an actively regulated process in the case of the adult fly epidermis. Why? One possibility is that once the cuticle barrier structure is formed shortly after eclosion the adult epidermis becomes dispensable for survival. Our TEM data suggests that cuticle secretion stops after two to three days into adulthood (data not shown) and the expression of known adult cuticle proteins ceases around this same time [44]. The majority of epidermal degradation we observed occurred after cuticle secretion was complete, suggesting that aging, rather than development, drives the degradation process. We hypothesize that the adult fly normally degrades its epidermis as a nutrition source and that autophagy orchestrates this process. This controlled destruction is reminiscent of the autophagic elimination of certain organs that become superfluous at the onset of metamorphosis [38, 45] as well as the elimination of wing epidermal cells in the first hours of adult life [46]. A requirement for autophagy in the epidermis is consistent with previously demonstrated roles of autophagy in cellular remodeling [17, 18]. This proposed role is not inconsistent with a potential protective role of autophagy in the epidermis since a residual epidermis is maintained even in old flies.

Finally, our results imply that in the particular case of autophagy epidermal aging can be uncoupled from lifespan. The short-lived Atg7^d77 mutant [37] maintains a young skin. Conversely, the Progeria-like lam^G262 mutant, which has a short lifespan and higher epidermal autophagy, rapidly develops a very old skin. Interestingly, Progeria mice also exhibit an elevated autophagy level in the heart and other tissues [47]. In both flies and mice it remains unclear if the elevated autophagy levels are a cytoprotective reaction to Progeria or if autophagy is part of the pathology driving organ deterioration.

Parallels between Drosophila and vertebrate epidermal aging

In contrast to the epidermal monolayer in flies, vertebrate skin consists of a multilayered epidermis replenished by basal stem cells and separated from an underlying dermis. The vertebrate epidermis is continuously needed to maintain a functional permeability barrier, and thus is likely to have a higher contribution to overall lifespan than the epidermis of the fly. Given these differences, how then do the changes observed in the aging Drosophila adult epidermis relate to vertebrate skin? A decreased integrin expression in the basal epidermis has been reported for aging human skin [13] and epidermal thinning is one of the hallmarks of aged human skin [12]. Relative to average lifespan, epidermal morphology changes in the fly have an even earlier onset than age-related changes in human skin. This difference may be explained by the fact that fly skin cells are not replenished by stem cell-mediated replacement.

Despite the morphological differences between fly and vertebrate skin, are there parallels on the molecular level that control the skin aging program? Intriguingly, mice bearing an insulin receptor substrate null mutation exhibit delayed skin aging [48] and progeroid mice with a mutant lamin exhibit hallmarks of accelerated skin aging [49]. Whether autophagy correlates with and plays a causal role in the cellular or molecular control of vertebrate skin aging in vivo remains to be tested. However, senescent keratinocytes in cell culture undergo autophagic cell death [21], suggesting that autophagy may play a conserved role in controlling the morphological progression of vertebrate skin aging.

A major need in the aging field is the identification of readouts of “healthspan”, which has been defined as the length of time that an individual maintains good health [6]. Drosophila has proven itself an important model for the identification of genes that regulate lifespan [50]. However, by contrast Drosophila tissue aging has only been examined in a few organ systems and behaviours including the heart, gut, muscles, nervous system, and sleep patterns [8-10]. The changes in the epidermis observed here arise earlier, can be detected using a simple visual assay, and represent a chronometer for the healthspan of an individual organism. Our model thus complements more commonly used population-based longevity assays. In addition, the epidermal aging model provides a simple platform for testing the impact of extrinsic factors such as UV radiation or caloric restriction. We expect that further molecular dissection of the epidermal aging program will reveal other regulators of tissue healthspan and how these genes are connected to organismal lifespan.

Fly stocks and husbandry

Aging Drosophila were reared at 25 °C on standard cornmeal medium under a 12 h light-dark cycle. For consistency, only virgin females were analyzed. Aging flies were switched to new food every two to three days to avoid starvation effects.

Except where noted, w¹¹¹⁸ was used as a control strain for the rate of epidermal aging. In some cases, a Fasciclin III (FasIII)-GFP fusion transgene (YD0853; [51]) was used to visualize epidermal cell membranes. Epidermal morphology was also analyzed in long-lived chico¹ mutants [27], short-lived lam^G262 mutants [35], and Atg7^d77 mutants, which have reduced autophagic activity [37]. The GAL4/UAS system [52] was used for tissue-specific expression of transgenes under UAS control. The NP2108-GAL4 driver (strain 112783; DGRC Japan) expresses in the larval and adult epidermis, as well as the persistent larval fat body. UAS-Atg7-IR (CG5489-IR #45558 from VDRC) was used for knockdown of Atg7. UAS-DsRed2-Nuc and UAS-LC3-GFP allowed labeling of epidermal nuclei and autophagosomes, respectively. chico¹/CyO;NP2108 -GAL4/TM6B and chico¹/CyO;UAS-LC3-GFP were crossed to each other to visualize autophagosomes in the chico¹ mutant background. Similarly, w;lam^G262/ CyO;NP2108-GAL4/TM6B crossed to w;lam^G262/CyO;UAS-LC3-GFP and w;Atg7^d77/CyO;NP2108-GAL4/ TM6B crossed to w;Atg7^d77/CyO;UAS-LC3-GFP were used to examine epidermal autophagosomes in the lam^G262 and Atg7^d77 mutant backgrounds, respectively.

Dissection of the adult abdominal epidermis

Flies were anaesthetized and the head and appendices removed. The thorax and abdomen were pinned dorsal side up on a Sylgard (Dow Corning, Midland, Michigan, United States) dissection plate with 0.1 mm diameter dissection needles (Fine Science Tools). We covered the specimen with 1x phosphate-buffered saline (PBS), flushed out attached air bubbles, and used dissecting scissors (Fine Science Tools #15000-02, Foster City, California, United States) to make an incision along the abdominal dorsal midline. The two dorsal halves were pinned to the sides, the thorax and inner organs removed to expose the ventral abdominal epidermis. Finally, the epidermis was flattened by careful repositioning of the dissection needles. Samples were fixed for 1 to 3 h in 3.7% formaldehyde in PBS. After washing with PBS, the needles were detached and the samples transferred to 1.5 ml microtubes for subsequent staining. A more detailed protocol is available on request.

Analysis of NP2108-GAL4 driver expression

¹¹¹⁸/+; NP2108-GAL4 or UAS-2x eYFP/+; NP2108-GAL4/+ animals were collected immediately after eclosion and aged for 1, 7, or 14 days. Fat bodies were dissected and fixed in 3.7% formaldehyde in PBS for 30 minutes at room temperature. Fat bodies were then washed in PBS, mounted in Vectashield, and imaged with a Leica MZ16FA microscope with a Planapo 1.6x objective.

Immunofluorescence

Immunostaining of adult epidermal whole-mounts was performed as described earlier for larval samples [53]. Primary antibodies were mouse anti-Fasciclin III [54] diluted 1:50, mouse anti-Coracle monoclonal antibody 9C [55] diluted 1:500 and rabbit anti-GFP antibody (A11122, Invitrogen) diluted 1:500. Secondary antibodies were Fluorescein (FITC)-conjugated goat anti-mouse (1: 250), Cy3-conjugated goat anti-mouse (1:1000) and FITC-labeled goat anti-rabbit antibody (1:250), all obtained from Jackson ImmunoResearch Laboratories Inc. All primary and secondary antibodies were diluted in PHT buffer (phosphate-buffered saline containing heat-inactivated normal goat serum and 3% Triton X100). Immuno-labeled epidermal whole mounts were imaged using a Leica MZ16FA microscope and a Planapo 1.6x objective. Photos were taken with a color Leica DFC350FX digital camera.

Quantification of epidermal morphology

To quantify epidermal deterioration pictures of epidermal samples of different genotypes and time points, we defined four classes I to IV reflecting distinct stages of epidermal membrane labeling loss. Class I was characterized as a sample with 0 to 5% epidermal deterioration in respect to the total epidermal area, class II corresponded to 5-20%, class III to 20-50% and class IV to over 50% epidermal deterioration. Four independent individuals were asked to assign ten individual pictures for each genotype-time point combination into these pre-defined classes. The individual groupings of the four judges were then averaged for each genotype-time point combination. Subsequently the classification numbers for different genotypes and time points were compared using an exact one-sided analysis contingency table (Chi square test). Nuclear numbers and LC3-GFP signals were compared using a single-factor Anova test in Microsoft Excel.

Autophagosome quantification

LC3-GFP-labeled epi-dermal whole mounts were visualized using an Olympus Fluoview 500 confocal microscope. Green channel settings were kept constant to allow for quantitative comparison. For each sample a Z series of optical sections in 0.7 μm increments spanning the entire depth of the epidermis and covering an area of 210 by 210 μm in the XY plane was collected and merged into a Z-stack. The resulting images were further analyzed as follows using ImagePro Analyzer 6.2. LC3-GFP signals, apparent as small punctae, were first converted to grayscale and then counted. To exclude background signals from cuticular autofluorescence, only bright punctae were counted (those >30 pixels on a 0-255 range of selected areas). Since the lam^G262 allele contains a GFP insertion that labels epidermal nuclei, in lam^G262 samples GFP-labeled nuclei were substracted from overall GFP counts in one of two ways: 1. By setting a size threshold that excluded large (area threshold here) labeled objects. 2. By then manually subtracting the remaining labeled objects that by morphological criteria and comparison to the original image seemed to result from nuclear labeling. Because some autophagosomes appear to be tightly associated with nuclei, the autophagosome counts for lam^G262 samples likely represent an undercount of the actual autophagosome numbers. The autophagosome numbers from all individual pictures were averaged for each genotype and time point and compared using the Student's t-test in Microsoft Excel.

Transmission Electron Microscopy (TEM)

For transmission electron microscopy whole fly abdomens were individually dissected in 0.2 M sodium phosphate buffer (pH 7.2) and then immediately fixed for 4 h at room temperature in 3% glutaraldehyde, 2% paraformaldehyde and 2.5% dimethylsulfoxide in this same buffer. Samples were post-fixed in 1% Osmium tetroxide (Electron Microscopy Services, EMS, Hatfield, PA), briefly rinsed with water, stained with 2% aqueous uranyl acetate (EMS), dehydrated in graded ethanols, and embedded in Poly-Bed 812 resin (Polysciences Inc., Warrington, PA). Ultrathin sections were cut on a Leica EM UC6RT ultramicrotome (Leica Microsystems Inc., Bannockburn, IL), stained with lead citrate, and viewed in a JEM 1010 transmission electron microscope (JEOL, USA, Inc., Peabody, MA) at an accelerating voltage of 80 kV and 10,000x magnification. Digital images were obtained using the AMT Imaging System (Advanced Microscopy Techniques Corp, Danvers, MA). TEM pictures shown in Figure 2 and Figure S1, S3 and S4 consist of several overlapping pictures merged in Adobe Photoshop CS4. The cuticle-epidermis boundaries could be clearly distinguished by color, while the muscle-epidermis boundaries were defined by the intervening basal laminae.