INTRODUCTION

      The brain, similar to other organ systems, undergoes a progressive decline in functional capabilities during aging, which manifests as decrements in learning and memory, attention, speed of decision-making, and motor coordination (Alexander et al., 2012; Mattson and Arumugam, 2018). Interrogation of the brain at the molecular and cellular levels reveals many of the hallmarks of aging evident in other tissues (Lo ´pez-Otin et al., 2013); mitochondrial dysfunction is particularly evident, as mitochondria isolated from brain tissue reveal numerous age-associated alterations, an important one being impaired function of the electron transport chain (Pandya et al., 2016; Pollard et al., 2016; Yao et al., 2010). Thus, preserving brain mitochondrial integrity and metabolism with age is probably critical for maintaining healthy brain function.

      Dietary supplementation of the polyamine spermidine (Spd) has emerged recently as a protective strategy to increase longevity and delay the aging of several animal model systems including mice (Madeo et al., 2018). Spd supplementation (Spd-S) in aging Drosophilae delayed the age-induced impairment of memory formation (Gupta et al., 2013) and age-protected locomotion performance (Minois et al., 2014). The challenge remains to explore the mechanistic actions of this pleiotropic compound to allow for a more rational design o anti-aging interventions and to investigate the mechanistic basis of neuronal and brain aging in more depth.

      We show here that Spd-S elicits a functional protection of mitochondria in aging brains, obviously achieved via posttranscriptional mechanisms. Spd, apart from other roles, is the amino-butyl group donor for the synthesis of hypusine (Nε -[4- amino-2-hydroxybutyl]-lysine) to a specific lysine (position 51) residue of the eukaryotic translation initiation factor 5A (eIF5A). This modification can tune translational and, consequently, proteomic profiles by favoring the efficient translation elongation of certain mRNAs (Doerfel et al., 2013; Ude et al., 2013). We found here that eIF5A hypusination decreased with age but that until mid-age, it could be augmented by dietary Spd. Several genetic constellations (half dose of deoxyhypusine synthase, half dose of an eIF5A allele carrying an on-locus point mutation blocking hypusination, neuron-specific knockdown of deoxyhypusine synthase) that reduced eIF5A hypusination levels in an agetypical manner also provoked age-typical deficits of mitochondrial respiration and memory formation and accelerated aging of locomotion function. While the likely complex relations among enzymatic hypusination activity, dietary Spd, and effective eIF5A hypusination remain to be worked out, most Spd-S effects were abrogated in the hypusination-attenuated genotypes. Having identified hypusination as an important component and target process of brain aging protection enriches our manipulative and diagnostic possibilities in this important biomedical area and might deepen our understanding of processes driving brain aging.

RESULTS

Spd-S boosts mitochondrial abundance and protects mitochondrial functionality in aging Drosophila brains

      In order to investigate the molecular mechanisms of protection from cognitive decay in aging Drosophilae, we performed isobaric tags for relative and absolute quantification (iTRAQ) proteomics on extracts of mechanically, hand-dissected Drosophila brains, comparing brain extracts of isogenic Drosophilae with or without dietary Spd-S at 15 days of age (at this time point, Spd drops steeply in aging fly heads and age deficits consolidate; Gupta et al., 2013). Notably, mitochondrial proteins (Table S1) were clearly upregulated here under Spd-S in Gene Ontology (GO) analysis (Figure 1A). Semantic protein-protein interaction (PPI) network analysis (Szklarczyk et al., 2015) identified oxidative phosphorylation as strongly induced by Spd-S, while cytosolic translation was downregulated (Figure 1C). Moreover, amounts of mitochondrial DNA normalized to amounts of nuclear DNA were significantly increased by Spd-S (Figure 1D).

      We next asked whether mitochondrial functionality, especially respiratory activity, might be ‘‘age protected’’ by Spd-S. Thus, we established measurements of the oxygen consumption rates (OCRs) from dissected and enzymatically digested Drosophila brains using the Seahorse XF analyzer (Homem et al., 2014), allowing us to measure basal and maximal respiration rates and adenosine triphosphate (ATP) production. All rates were normalized to sample protein content. We first set out to determine which aspect of mitochondrial respiration suffered from brain aging by comparing animals of 3, 15, and 30 days of age. Notably, maximal respiration (maximal OCR of the respiratory chain stimulated by the uncoupler FCCP) showed that it suffered drastically from aging, while basal respiration appeared stable and ATP production decreased only rather mildly (Figures 2A–2D).

      We then measured Spd-S effects on mitochondrial OCRs at 15 days of age (Figures 2E–2H). Notably, Spd feeding enhanced the maximal respiration strongly and significantly, while mitochondrial basal respiration and ATP production appeared elevated rather slightly (Figures 2E–2H).

      We next analyzed transcriptional profiles by RNA sequencing (RNA-seq) analysis (Table S2) of 15-day-old brains (30 hand dissected brains per condition) to explore the underpinning of the enhanced mitochondrial functionality. However, we found that transcripts of genes related to mitochondria, oxidative phosphorylation, and ATP synthesis coupled to electron transport remained relatively unchanged upon Spd-S (Figure S1A). This is in pronounced contrast to the changes observed for the corresponding proteins. Thus, as our effects are not likely to be explained by changes in transcription, we considered posttranscriptional, especially translational, mechanisms.

      Spd-S boosts eIF5A hypusination in Drosophila brains Importantly, Spd has been linked specifically to the functional status of the translational initiation factor eIF5A, the only known protein to contain the unusual amino acid hypusine (Nε -[4-amino-2-hydroxybutyl]-lysine]). Hypusine is formed by the posttranslational modification of a specific lysine residue in two consecutive catalytic steps that involve two highly conserved enzymes: (1) deoxyhypusine synthase (DHS; Drosophila ortholog called CG8005), which uses Spd directly as an amino-butyl donor, followed by the activity of (2) the deoxyhypusine hydroxylase (nero in Drosophila) (Mandal et al., 2013; Patel et al., 2009) (Figure S2A).

      Notably, we detected a robust age-related decrease of hypusine staining intensity in aged wild-type Drosophila brains using an anti-hypusine moiety antibody (Nishiki et al., 2013) (Figures 2I and S2B), while total eIF5A levels remained stable (Figures S2C and S2D), resulting in a significant progressive age decline of the normalized hypusine level (Hyp/eIF5A ratio) (Figure 2J). Under Spd-S, Hyp/eIF5A ratios were robustly boosted in young animals (Figure 2J) and still significantly increased at 15 days, but not 30 days, of age (Figure 2J). Thus, while dietary Spd-S can boost eIF5A hypusination into/until mid-age, and Drosophila brain Spd levels decrease with age (Gupta et al., 2013), other factors apart from sheer Spd availability seemingly become rate limiting for hypusination in aging Drosophila brains as well (see also Discussion). Notably, we also found that neuronal overexpression of Odc1 (ornithine decarboxylase, an enzyme rate-limiting for the synthesis of polyamines) boosted hypusine levels (Figures 2K and S2F) as well as the Hyp/eIF5A ratio (Figures 2L, S2E, and S2G), suggesting that a cell-autonomous increase of Spd exclusively in neurons is sufficient to elevate neuronal hypusination levels. 

Reduction of eIF5A hypusination affects the mitochondrial proteome integrity and respiratory function in Drosophila brains

      We decided to challenge the hypusination system genetically to learn about the consequences of when hypusination became attenuated to a similar degree as we observed in aging Drosophila brains (Figures 2I and 2J).

      We first targeted the highly conserved DHS/CG8005, whose activity also has been suggested to be rate limiting for hypusination (Templin et al., 2011). We identified a loss-of-function allele of Drosophila DHS (CG8005DG05802), which resulted in late larval lethality in homozygosity, consistent with the DHS being essential for yeast cell viability and mice embryo survival (Sasaki et al., 1996; Templin et al., 2011) and with nero null situations provoking late larval lethality in Drosophila (Patel et al., 2009).

      Facing lethality, we explored strategies to attenuate hypusination efficacy genetically; however, this would still allow for robust survival into adult stages and aging. First, we analyzed animals heterozygous for the dhs allele CG8005DG05802 (isogenized to w1118 control background for six generations; thereafter referred to as CG8005/+). Freshly hatched CG8005/+ animals appeared ‘‘healthy’’ at first glance (e.g., they exhibited normal locomotion activity (see Figure 6A). Hypusine levels in brains of young CG8005/+ animals were slightly and significantly reduced in confocal scans of anti-hypusine stainings (Figures 3A and S3A) and also appeared slightly reduced in brain-specific western blots (Figure S3B). Hyp/eIF5A ratios were significantly reduced (Figure 3B) when normalized to eIF5A signals in brain immunostainings (Figure S3C). We then further investigated whether reduced eIF5A hypusination in CG8005/+ animals affected mitochondrial respiration. Interestingly, CG8005/+ animal brains suffered from reduced basal and maximal respiration and ATP production (Figures 3C–3F). In principle, hypusination could occur in neurons or rather dominantly involve the glia component of the Drosophila brain. Most of the hypusine label, however, colocalized with the neuronal label (Figure S3D) in a triple staining using generic glia (Repo) and neuron (Elav) markers together with the anti-hypusine antibody.

      Thus, we next intended to attenuate (but not eliminate) DHS activity in neurons specifically. First, we established a neuronspecific knockdown of dhs, combining the neuron-specific Gal4 driver line (elav-Gal4) with two distinct inducible RNA interference (RNAi) transgenic constructs (CG8005-RNAiKK101539 and CG8005-RNAiHMS02169), targeting different regions of the DHS encoding transcript. These flies were raised at 29 C (instead of at 25 C) throughout the study to enhance RNA interference efficiency. Both RNAi lines, when expressed pan-neuronally, provoked a clear reduction of the hypusine levels and Hyp/ eIF5A ratios in brain immunostainings compared to controls (Figures 3G, 3H, and S3E–S3J), again consistent with the neuron population being dominantly responsible for overall brain hypusination signals. We also tested the CG8005-RNAiKK101539 line in western blots, again exhibiting decreased hypusine signals (Figure S3K). Importantly, both RNAi lines with neuronal knockdown of dhs exhibited reduced mitochondrial respiration (Figures 3I–3L and S3L–S3O).

      Finally, we used a CRISPR/Cas9 strategy to mutate the hypusine site lysine-51 into arginine-51 at the endogenous eIF5A gene locus to specifically test for the role of reduced hypusination directly on the level of eIF5A. Consistent with hypusination of eIF5A being essential for survival, homozygosity of this allele also resulted in lethality at larval stages. Both hypusine (Figures 3M and S3P) and Hyp/eIF5A ratios (Figure 3N) were reduced to about half in immunostainings of adult Drosophila brains heterozygous for the eIF5AK51R allele (isogenized into w1118 control background; thereafter referred to as eIF5A_K51R/+), while eIF5A levels remained stable (Figure S3Q). Notably, the boosting of Hyp/eIF5A ratios via Spd-S was abolished in eIF5A_K51R/+brains (Figures S3R and S3S). Similarly, eIF5A_K51R/+ brains also showed reduced mitochondrial respiration (Figures 3O–3R).

      It is worth noting here that for the neuronal knockdowns, brain respiration was measured at 27℃–29℃ to represent their rearing status, while the two other groups were measured at 23℃–25℃. This also precludes a direct quantitative comparison between the groups here.

      In summary, four distinct genotypes (targeting either the DHS/CG8005 or eIF5A directly) converging on attenuated hypusination efficacy provoked a reduction of mitochondrial respiration. It appeared here as if a rather modest disturbance of hypusination level (Figures 2A–2F) could already reduce mitochondrial respiration, while a stronger reduction in hypusination levels (Figures 2G–2R) did not further aggravate mitochondrial respiration deficits.

      We conclude that sufficient levels of hypusination are obviously crucial to maintain mitochondrial functionality in the fly brain. We went on exploring the mechanistic basis for the obviously critical role of efficient hypusination in mitochondrial function.

Reduced neuronal hypusination affects the brain mitochondrial proteome

      eIF5A-hypusination, in other non-neuronal contexts, was shown to shift the translation landscape of cells (Park and Wolff, 2018). We thus carried out quantitative global proteomics (Table S3) from hand-dissected 5-day-old neuronal dhs knockdown brains (elav/CG8005-RNAiKK101539 versus elav/+) to identify neuronal protein level changes associated with attenuated hypusination and consequent functional deficits. Unbiased GO analysis identified cytosolic translation (‘‘cytosolic ribosome’’ and ‘‘cytosolic large ribosomal subunit’’) as upregulated (Figure 4A), appearing antagonistic to the Spd-S-provoked changes in 15-day-old brains (Figure 1B). Most importantly, ‘‘mitochondrion proteins,’’ as such, and ‘‘mitochondrial respiratory chain complex I components’’ were identified as downregulated in the neuronal knockdown (Figure 4B). A semantic PPI analysis of proteins downregulated after DHS/CG8005 neuronal knockdown also identified oxidative phosphorylation to be significantly downregulated (Figure 4C). Consistent with this, proteins related to the semantic groups ‘‘mitochondria,’’ ‘‘mitochondrial ATP synthesis coupled electron transport,’’ and ‘‘oxidative phosphorylation’’ were also generally more abundant in wild-type than in knockdown brains (Figure 4D).

Efficient hypusination is critical for Spd-mediated protection of mitochondrial functionality

       We next investigated whether effective hypusination would be a precondition for the age-protective effects of dietary Spd, again investigating 15-day-old animals with or without Spd-S. Spd-S increased the Hyp/eIF5A ratios in controls significantly but showed only a very slight tendency to increase the Hyp/eIF5A ratios in DHS/CG8005 neuronal knockdown and the DHS heterozygous mutant animals (Figures 5A–5D and S4A–S4D). Together with the time-course brain immunostainings data (Figure 2J), these findings indicate that Spd-S can efficiently boost eIF5A hypusination in aging fly brains (15 days here), while Spd-S was less efficient in boosting hypusination after genetic attenuation of the DHS/CG8005 (Figures 5A–5D).

      We then analyzed the effects on mitochondrial respiration in the dhs neuronal knockdown and heterozygous mutant animals. At 15 days, Spd-S boosted basal and maximal respiration as well as ATP production in the wild-type controls (Figures 5E–5J). By contrast, brains suffering from inefficient eIF5A hypusination showed reduced levels of basal and maximal capacity as well as ATP production throughout (Figures 5E–5J). Consistent with the inability to stimulate Hyp/eIF5A levels effectively in these hypusination-attenuated backgrounds, Spd-S only slightly and insignificantly boosted these parameters of mitochondrial respiration, at least not to the extent as observed in the wild-type backgrounds (Figures 5E–5L). We conclude that efficient eIF5A hypusination is a critical precondition to allow for the full extent of the age-protective effects dietary Spd-S displays on mitochondrial functionality in Drosophila brain aging (see also discussion concerning the relations among Spd availability, hypusination activity, and biological readout). We finally asked whether eIF5A hypusination efficacy would be equally relevant concerning the age-protective effects of Spd-S on animal fitness and behavior.

Premature aging of memory and locomotion after attenuation of hypusination

       Spd-S can extend lifespan across several species, including mice and Drosophila (Eisenberg et al., 2009; Partridge et al., 2020; Tain et al., 2020). We thus subjected wild-type control and isogenic CG8005/+ animals to survival analysis to probe whether the extension of longevity by Spd-S required effective eIF5A hypusination. Indeed, the Spd-S-mediated lifespan prolongation was abrogated in hypusination-deficient animals (Figure S5A).

      We then also used the Gal80ts Gal4 system (McGuire et al., 2003) to restrict dhs knockdown to the post-hatching stage (by switching the rearing temperature from 18 C to 29 C) in order to avoid confounding developmental effects. Again, the flies experiencing dhs knockdown only in their adulthood also no longer benefited from Spd-S, whereas controls still did (Figure S5B). It should be noted that raising flies here at 29 C (to enhance efficacy of knockdown) sped up the aging process obviously, resulting in overall shortened lifespans (Figure S5B).

      Age-related declines in locomotion serve as a surrogate for brain aging in Drosophila (Jones and Grotewiel, 2011), and Spd-S has been previously shown to alleviate age-related locomotor impairments in Drosophila (Minois et al., 2014). In a first time-course experiment, we found that Spd-S indeed significantly ameliorated the normally pronounced decline of locomotive abilities between 15 and 30 days of age measured via negative gravitaxis (Inagaki et al., 2010) (Figure S6A). Hypusination-attenuated CG8005/+ animals showed an enhanced age-induced decrease of locomotive performance between 15 and 30 days of age, while Spd-S showed only a mild, non-significant trend toward protection in this background (Figure 6A).

      Notably, eIF5A_K51R/+ animals showed the same accelerated aging of locomotion function as CG8005/+ animals and were less responsive to Spd-S effects as compared to the wild-type controls (Figure 6B). Obviously, insufficient eIF5A hypusination is, thus, responsible for this accelerated phenotype and the deficient protection by Spd-S.

       We also intended to investigate whether this role of the hypusination effect was mediated by the nervous system. However, reduced survival precluded efficient testing of the CG8005 neuronal knockdown flies at day 30. Nevertheless, as we raised these flies at 29 C, which, as mentioned above, sped up their aging process, we argued that there might be a chance to observe differences between them and their isogenic controls already at 15 days. Indeed, at 29 C, we found that 15-day-old control flies had ‘‘already’’ benefited from Spd feeding (Figure S6B), while such amelioration was abolished in neuronal DHS knockdown flies (Figure S6B).

      As we have shown previously, dietary Spd-S protected flies aged to 30 days from the normally occurring decay in aversive olfactory memory formation (Gupta et al., 2013). We used the CG8005/+ animals in this assay, which reached the relevant age of 30 days without problems (Figure S5A). Importantly, CG8005/+ animals possessed normal smelling abilities (Figure S6C). Thus, we next performed aversive olfactory learning experiments using CG8005/+ animals and isogenic controls. Significantly lower short-term memory (STM) scores were ‘‘already’’ detected for young CG8005/+ animals (Figure S6D). Furthermore, we found that Spd-S-mediated STM age protection in 30-day-old flies was largely occluded after the genetic attenuation of hypusination (Figure S6E). Again, naive odor avoidance scores remained similar across 30-day-old CG8005/+ animals and isogenic controls with and without Spd-S (Figure S6F). We finally turned to intermediate-term memory (ITM). Similarly, 3-h ITM learning scores were reduced in young CG8005/+ animals (Figure 6C). Again, Spd-S mediated age-protection of ITM in 30-day-old flies was abrogated after the genetic attenuation of hypusination (Figure 6D).

      Taken together, our findings demonstrate that efficient eIF5A hypusination plays a critical role in memory formation per se. Attenuation of hypusination is obviously also largely incompatible with Spd-S protection on brain-mediated functions, namely, locomotion and memory formation (see also Discussion).

      Our recent analysis showed that Spd-S attenuates age-associated deterioration of the mossy-fiber-CA3 synaptic transmission and plasticity in the murine hippocampus memory center (Maglione et al., 2019). So-called granule cell neurons in the hippocampus dentate gyrus form these synapses (granule cell layer [GL]). Indeed, GL hypusine levels of aged mice were clearly boosted by Spd-S (Figures 6E and 6F). Thus, dietary Spd-S is obviously able to boost hypusination levels in the aged mammalian brain. Notably, the mitochondrial status of mossy fiber synapses was also age-protected by Spd-S (Maglione et al., 2019), and Spd-S was found to protect cognitive performance in aging mice (De Risi et al., 2020) (see also accompanying paper in this issue of Cell Reports by Schroeder et al. [2021], their Figure 2C). Our findings described here should thus be of relevance concerning the physiological and molecular basis for the mechanistic exploration and therapeutic usage of Spd-S effects.

      Notably, induction of autophagy has been broadly implicated in the execution of Spd-S effects (Eisenberg et al., 2009; Gupta et al., 2013), and eIF5A hypusination was found to be required for the efficient translation of the key autophagic protein ATG3 in mammalian cell culture and C. elegans (Lubas et al., 2018). Indeed, in brain immunostainings, we found that the Spd-Sinduced increase of ATG3 was abrogated in hypusination-attenuated brains (Figures S6G and S6H). Thus, exploring the exact mechanistic and functional links of hypusination- and autophagy-mediated effects of Spd-S here will mark a next step of understanding. 

DISCUSSION

      Taken together, our analysis of Drosophila brains provides evidence that (1) Spd-S promotes mitochondrial abundance and respiration in the course of brain aging; (2) brain hypusination levels decay with aging but can be boosted until mid-age by Spd-S; (3) comparatively mild genetic attenuation of hypusination significantly affects mitochondrial functionality and protein composition; and (4) already comparatively mild genetic attenuation of hypusination abrogated a spectrum of age-associated Spd-S effects on mitochondrial functionality, locomotion, and memory. Thus, hypusination obviously operates at a nexus of Spd-S and its age-protective effects in the nervous system.

      We observed that eIF5A hypusination was highly sensitive regarding mild reduction of DHS/CG8005 activity (e.g., Figure 3B) but until mid-age could also be boosted by a surplus of its substrate compound, Spd (Figures 2I and 2J). Concerning biological readouts, the age-protective effects of Spd-S were largely eliminated after genetic attenuation of hypusination (but see Figures 5 and 6A–6D) using our two independent genetic strategies. This at least formally suggests a surprisingly steep dependence on both substrate availability as well as enzymatic deoxyhypusine synthase activity here. Along these lines, recent genetic data suggest that an already moderate reduction of deoxyhypusine synthase activity is, in fact, associated with a neurodevelopmental disorder in humans (Ganapathi et al., 2019). Future analysis will have to further explore the potentially complex relations between substrate availability and enzymatic activity, which might well involve more regulatory layers in addition to substrate concentrations. We, in this context, would like to emphasize the complexity of the regulation of polyamine synthesis pacemaker enzyme ornithine decarboxylase, whose activity is controlled on transcriptional, translational, as well as posttranslational levels and via the availability of the polyamine products such as Spd (Dever and Ivanov, 2018; Igarashi and Kashiwagi, 2019; Park and Wolff, 2018). Taking this perspective, mechanistic analysis of biological endpoints such as memory or locomotion measured at advanced age but establishing along lifetime is complex, and the critical, rate-limiting steps defining the ultimate phenotype might even change along lifetime.

      Notably, in this regard, we would like to emphasize that in aged flies (30 days old), Spd-S did not increase brain hypusination levels (Figures 2I and 2L), though our previous measurements showed that Spd-S did increase head net Spd contents also in this advanced age (Gupta et al., 2013). The eIF5A and DHS form a tight complex, and deoxyhypusine synthesis proceeds through a multistep process including Spd cleavage, a DHSimine intermediate, and then transfer of the butylamine moiety from the enzyme to eIF5A (Wolff et al., 1990, 1997). Interestingly, the DHS reaction also requires the nicotinamide adenine dinucleotide (NAD) co-factor (Chen and Dou, 1988; Wolff et al., 1990). This might also explain why Spd-S no longer boosted the Hyp/eIF5A in aged animals (Figures 2I and 2L), as NAD+ levels were found to decline with aging in C. elegans and mice (Fang et al., 2014; Lautrup et al., 2019; Stein and Imai, 2014). Equally, DHS/CG8005 levels were reported to decline with age (Chen and Chen, 1997). It will be interesting to test whether a co-supplementation of Spd together with NAD+ precursors might indeed be able to boost hypusination levels in aged

      Concerning the mechanistic role of eIF5A hypusination on translation, the long, basic side chain of hypusine was recently shown to facilitate hard-to-translate polyproline stretches, non-polyproline-specific sequences, and many other ribosome-pausing sites, thus promoting translation elongation for a specific spectrum of mRNAs (Schuller et al., 2017). We suspect that Spd via hypusination might promote the translation of mitochondrial proteins and autophagy-relevant factors.

      Recently published work has shown that deficits in eIF5A hypusination in immune cells interfered with efficient expression of a subset of mitochondrial proteins involved in oxidative phosphorylation (Puleston et al., 2019). Several of these proteins harbor mitochondrial targeting sequences that partially conferred an increased dependency on hypusinated eIF5A. Consistently, we found the semantic terms ‘‘mitochondrion proteins’’ and ‘‘mitochondrial respiratory chain complex I components’’ to be downregulated in the hypusination-compromised fly brains (Figure 4B). The fact that in vivo Spd-S supports hypusination in the brain and that this can promote oxidative phosphorylation and protect from brain aging suggests a fundamental relationship between hypusination and mitochondrial functionality of therapeutic relevance. Future analysis will have to mechanistically dissect the details of how hypusination controls mitochondrial protein composition and mitochondrial functional status as, probably, a direct consequence thereof. The identification and manipulation of hypusination dependence mediating sequence motifs in the open reading frames of target factors should be the focus here.

      Previous analysis showed that maintaining/restoring autophagy in aging brains is an essential part of Spd-S effects (Liang and Sigrist, 2018; Madeo et al., 2019). Indeed, a tripeptide motif in ATG3 causes its dependency on eIF5A hypusination for efficient translation (Lubas et al., 2018). In our contexts, we found in brain immunostainings that the Spd-S-induced increase of ATG3 was abrogated under hypusination-attenuated conditions (Figures S6G and S6H). Furthermore, a recent study identified the autophagy transcription factor TFEB as important in mediating hypusination effects in immune cells (Zhang et al., 2019). Thus, future detailed evaluations of the potential and functional relation of hypusination and autophagy after Spd-S as processes delaying aspects of brain aging are warranted.

STAR+METHODS

Detailed methods are provided in the online version of this paper

and include the following:

● KEY RESOURCES TABLE

● RESOURCE AVAILABILITY

 Lead contact

 Materials availability

 Data and code availability

● EXPERIMENTAL MODEL AND SUBJECT DETAILS

 Fly strains

 Mice

● METHOD DETAILS

 Fly stocks and rearing conditions

 Extraction of 15-day brains for iTRAQ global proteomics

 LC-MS/MS analysis following isobaric tags for relative and absolute quantitation (iTRAQ) labeling

 iTRAQ data analysis and interpretation

 Label-free global proteomics of 5-day-old ‘‘elav/ CG8005-RNAiKK101539 versus elav/+’’ brain sample preparation

 Label-free global proteomics data analysis

 Measurement of mitochondrial respiration using seahorse XFe96 analyzer

 Mitochondrial DNA measurement

 Whole-mount immunostaining, confocal imaging and quantification

 Spermidine supplementation in mice

 Immunohistochemistry on murine brain sections and image analysis

 Western blotting

 iTRAQ proteomics analysis

 Extraction of RNA from fly brains for RNaseq analysis

 Transcriptome analysis

 Negative geotaxis measurements

 Drosophila lifespan experiments

 Olfactory aversive learning

● QUANTIFICATION AND STATISTICAL ANALYSIS

SUPPLEMENTAL INFORMATION

Supplemental information can be found online at https://doi.org/10.1016/j. celrep.2021.108941.

AUTHOR CONTRIBUTIONS

Conceptualization, Y.L. and S.J.S.; methodology, Y.L., C.P., C.B.B., D.T., B.B., J.L., L.K., and S.J.S; investigation, Y.L., C.P., C.B.B., D.T., M.M., and J.C.K.S.; writing – original draft, Y.L. and S.J.S.; writing – review & editing, Y.L., L.K., A.S., F.M., and S.J.S.; funding acquisition, S.J.S.; resources, S.H., T.O.F.C., U.K., F.L., A.S., and S.J.S.; supervision, F.M. and S.J.S.

DECLARATION OF INTERESTS

F.M. and S.J.S. have equity interests in TLL (The Longevity Labs), a company founded in 2016 that develops natural food extracts.

Received: June 3, 2019

Revised: February 11, 2021

Accepted: March 12, 2021

Published: April 13, 2021

This article is excerpted from the Cell Reports 35, 108941, April 13, 2021 by Wound World.