NELF‐A controls Drosophila healthspan by regulating heat‐shock protein‐mediated cellular protection and heterochromatin maintenance

Abstract NELF‐mediated pausing of RNA polymerase II (RNAPII) constitutes a crucial step in transcription regulation. However, it remains unclear how control release of RNAPII pausing can affect the epigenome and regulate important aspects of animal physiology like aging. We found that NELF‐A dosage regulates Drosophila healthspan: Halving NELF‐A level in the heterozygous mutants or via neuronal‐specific RNAi depletion improves their locomotor activity, stress resistance, and lifespan significantly. Conversely, NELF‐A overexpression shortens fly lifespan drastically. Mechanistically, lowering NELF‐A level facilitates the release of paused RNAPII for productive transcription of the heat‐shock protein (Hsp) genes. The elevated HSPs expression in turn attenuates the accumulation of insoluble protein aggregates, reactive oxidative species, DNA damage and systemic inflammation in the brains of aging NELF‐A depleted flies as compared to their control siblings. This pro‐longevity effect is unique to NELF‐A due to its higher expression level and more efficient pausing of RNAPII than other NELF subunits. Importantly, enhanced resistance to oxidative stress in NELF‐A heterozygous mutants is highly conserved such that knocking down its level in human SH‐SY5Y cells attenuates hydrogen peroxide‐induced DNA damage and apoptosis. Depleting NELF‐A reconfigures the epigenome through the maintenance of H3K9me2‐enriched heterochromatin during aging, leading to the repression of specific retrotransposons like Gypsy‐1 in the brains of NELF‐A mutants. Taken together, we showed that the dosage of neuronal NELF‐A affects multiple aspects of aging in Drosophila by regulating transcription of Hsp genes in the brains, suggesting that targeting transcription elongation might be a viable therapeutic strategy against age‐onset diseases like neurodegeneration.


| INTRODUC TI ON
Controlled release of paused RNA polymerase II (RNAPII) at the promoter-proximal region is one of the most critical regulatory steps in transcription (Chen et al., 2018;Core & Adelman, 2019).
Yet how it may influence important aspects of animal physiology like aging remains unclear. RNAPII pausing occurs at a huge fraction of eukaryotic genes (Core et al., 2008;Min et al., 2011) and is mediated by the functionally conserved negative elongation factor (NELF) complex comprised of four subunits (A, B, C/D, and E; Narita et al., 2003;Wu et al., 2005). NELF complex interacts with 5,6-dichloro-1β-d-ribobenzimidazole sensitivity-inducing factor (DSIF) and the hypophosphorylated form of RNAPII to induce its pausing at ~30-60 nucleotides downstream of the transcriptional start site (TSS; Wada et al., 1998;Wu et al., 2003;Yamaguchi et al., 1999). Phosphorylation of NELF complex and RNAPII by the positive transcription elongation factor-b (P-TEFb) evicts NELF complex, thereby releasing RNAPII for productive elongation (Lis et al., 2000).
Interestingly, perturbations of different NELF subunits impact transcriptional output and expression patterns in cell-types specific manner (Gilchrist et al., 2008;Wang et al., 2010). In vivo studies showed that NELF-mediated promoter-proximal stalling of RNAPII ensures robust and coordinated gene activation in response to different types of signals and external stimuli (Gilchrist et al., 2010(Gilchrist et al., , 2012Saunders et al., 2013;Williams et al., 2015).
In Drosophila S2 cells and larvae, genes affected by NELF depletion are enriched for heat-shock and immune responses (Gilchrist et al., 2008(Gilchrist et al., , 2012. As many of these genes are linked to the process of aging, it is plausible that NELF may play a role in regulating animal lifespan.
We found that the level of NELF-A, but not other subunits, regulates adult fitness and lifespan in fly. Both heterozygous

NELF-A mutants and neuronal-depleted NELF-A flies exhibited
increased locomotor activity and lifespan as compared to their control siblings. Halving the level of NELF-A reduces RNAPII pausing at heat-shock protein (Hsp) genes, leading to significant upregulation of their expression. Elevated level of HSPs in the brain and head tissues suppresses the accumulation of protein aggregates and reactive oxygen species (ROS), attenuates systemic inflammation, and enhances animal resistance against external oxidative stress during aging. Similarly, we observed enhanced stress resistance in human cells depleted of NELF-A, whose N-and C-terminal region is highly conserved to Drosophila NELF-A protein (Wu et al., 2005). In heterozygous NELF-A mutant flies, reduced DNA damage and increased S-adenosyl methionine (SAM) concentration coincide with the maintenance of the repressive H3K9me2-marked heterochromatin and the repression of specific classes of TEs during aging. Taken together, NELF-A-mediated RNAPII pausing influences animal healthspan by regulating HSP-mediated cellular protection and H3K9me2dependent maintenance of genome integrity.

| Neuronal expression of NELF-A regulates lifespan and locomotor activity
To examine the regulatory roles of NELF genes in Drosophila lifespan, we first examined the highly expressed NELF-A gene and backcrossed heterozygous mutant Nelf-A KG09483 /Tm3 line (Chopra et al., 2009;Tsai al., 2016;Wang et al., 2010) to w 1118 male for at least 10 consecutive generations. Compared to their control siblings, backcrossed heterozygous male Nelf-A KG09483 /+ (herein NelfA*/+) flies expressed less than half the level of NELF-A mRNA ( Figure 1a) and protein (Figures 1b and S1A,B). Both the male and female NelfA*/+ flies showed approximately 10% and 20% increase in their lifespan, respectively (Figures 1c and S1C), suggesting that animal longevity may be affected by NELF-A expression level. To test this, we ectopically expressed low level of NELF-A protein by crossing UAS-NelfA-6myc line to Hsp70-Gal4 driver. When cultured at 30°C, their F1 offspring expressed higher level of NELF-A mRNA (Figure 1a), produced exogenous myc-tagged NELF-A protein (Figures 1b and S1A), and had significantly shorter lifespan than their control siblings (Figures 1c and S1C). This indicates that the level of NELF-A expression is inversely correlated to animal lifespan.
Drosophila undergoes age-dependent decline in their locomotor activity (Jones & Grotewiel, 2011). Negative geotaxis assays showed that heterozygous NelfA*/+ male flies consistently exhibited faster locomotor activity than their control siblings at different ages in adulthood (Figure 1d), suggesting that lowering the level of NELF-A expression may improve overall fitness or healthspan of the animals.
We next used GeneSwitch (GS) system to determine the organs by which NELF genes may regulate animal healthspan. As depletion of NELF genes affects the expression of immune-response genes in Drosophila S2 cells (Gilchrist et al., 2012), we examined the effect of RNAi-mediated depletion of NELF gene in the fat bodies from the brain (Bun-GS) and abdomen (106-GS; Hwangbo et al., 2004). In addition, NELF genes were also knockdown (KD) in the neurons using Elav-GS line given that NF-κB immune signaling in the brain regulates Drosophila lifespan (Kounatidis et al., 2017). To mimic the heterozygous NelfA*/+ flies, the level of RU486 administrated in fly food was optimized to induce ~50% KD of NELF-A expression in the brain Surprisingly, despite substantial reduction in the level of NELF-B and NELF-E expression by tissue-specific RNAi KD, the male flies did not F I G U R E 1 Lower neuronal level of NELF-A improves animal lifespan and locomotor activity. (a) Quantification of NELF-A mRNA in NelfA*/+ (left) and HspGal4/UAS-NelfA-6myc (right) fly heads as compared to their respective control siblings. Data presented as mean ± SEM, n = 3, unpaired two-tailed t test. (b) Immunoblot of endogenous and myc-tagged (red arrowhead) NELF-A protein in the heads harvested from male NelfA*/+ flies (left) and Hsgal4/UAS-NelfA-6myc flies (right), respectively. β-tub: β-tubulin, Lam-B: Lamin-B1. (c) Lifespan assay of male NelfA*/+ flies (9.8%, n = 180, p < 0.0001; left) and HspGal4/UAS-NelfA-6myc flies (−44.1%, n = 160, p < 0.0001; right) as compared to their control siblings. Log-rank test. (d) Climbing assay of male NelfA*/+ flies and their control siblings at different ages (D20: 38.8%; D30: 52.9%; D50: 34.5%, n = 105 per group). Percent calculated based on median height, Mann-Whitney test. (e) Pan-neuronal NELF-A depletion in RU486-treated ElavGal4. Switch/NelfA-RNAi male flies improved lifespan and locomotor activity. Left: Quantification of NELF-A mRNA in the brains harvested from EtOH or RU486 treated flies. Data presented as mean ± SD, n = 2, unpaired two-tailed t test. Middle: Lifespan assay (8.45%, n = 280 flies after combining two independent cohorts, p = 0.0001, Logrank test). Right: Climbing assay (n = 100-120 flies, Mann-Whitney test) show significant change in their adult lifespan as compared to the control siblings ( Figure S1G). Similarly, overexpression of UAS-Nelf-E-6myc gene by Hsp70-Gal4 driver also did not reduce their lifespan as seen with the NELF-A overexpressing flies ( Figure S1H). Taken together, these results suggest that the healthspan of Drosophila is highly dependent on the expression level of NELF-A, rather than other NELF subunits, within the neuronal populations.

| NELF-A reduction leads to pro-longevity transcriptional signatures
To identify downstream molecular pathways that may be responsible for the improved healthspan observed in NelfA*/+ flies, RNAsequencing was performed on the mRNA isolated from the heads of Day 50 (herein D50) male heterozygous mutants and their control F I G U R E 2 Pro-longevity transcriptome in NELF-A depleted fly heads. (a) Heatmap of differentially expressed genes (DEGs) in the heads of D50 male NelfA*/+ flies and their +/+ control siblings. Rep: Biological replicate. (b) GO analysis of DEGs identified in NelfA*/+ and their control siblings fly heads (adjusted p < 0.05, with Benjamini-Hochberg correction). (c) RT-PCR validation in the heads of D50 NelfA*/+ flies and their control siblings. Data presented as mean ± SEM, n = 3 biological replicates, unpaired two-tailed t test. (d) Quantification of AMP genes expression in the heads of D15 HspGal4/UAS-NelfA:6myc flies and their control siblings. Data presented as mean ± SD, n = 2 biological replicates, unpaired two-tailed t test. (e) Quantification of AMP genes expression in the brains isolated from either male NelfA*/+ flies or pan-neuronal NELF-A KD flies (RU486-treated ElavGal4. Switch/NelfA-RNAi) and their respective control siblings. Data presented as mean ± SD, n = 2 biological replicates, unpaired two-tailed t test siblings. Differential gene expression analysis using DESeq2 package revealed two cluster of genes that were highly elevated in either control siblings (+/+) or heterozygous mutants (NelfA*/+; Figure 2a, Table S1). Gene ontology (GO) analysis of the differentially expressed genes in the Drosophila heads revealed transcriptional changes in several stimulus-responsive genes that resembled NELF-B/E depleted S2 cells (Gilchrist et al., 2008(Gilchrist et al., , 2012 Table S2).
Like NELF-B/E depleted S2 cells, NelfA*/+ flies had reduced expression in genes that are enriched for defense response and response to bacterium ( Figure 2b). In addition, the NelfA*/+ mutants also showed upregulation of genes that are involved in different signaling pathways, heat-shock response, and protein folding (Figures 2b and S2).
During aging, elevated expression of immune-response pathway genes such as the induced immune molecules (IM) and antimicrobial peptides (AMPs) may trigger excessive inflammation that leads to morbidity in older flies (Badinloo et al., 2018;Cao et al., 2013;Kounatidis et al., 2017). In accordance, overexpression of Rel (NF-κB) transcription factor can induce AMP genes in the absence of upstream signals; whereas reducing Rel level may attenuate inflammation and promote longevity (Kounatidis et al., 2017). Consistent with these findings, there was a significant increase in the expression of Rel and AMP (AttC, DptA) genes in w 1118 male flies during aging ( Figure S2A). In contrast, elevated expression of genes involved in heat-shock responses and protein refolding has been well-documented to promote animal health and longevity (Biteau et al., 2010;Morrow et al., 2004;Wang et al., 2004). Elevated AMPs level can cause neurodegeneration in old adult fly brains, marked by increased number of vacuoles in the brain sections (Cao et al., 2013;Kounatidis et al., 2017). In accordance, vacuoles (red arrow) were observed in D50 but not D20 w 1118 fly brains ( Figure S2E). Since climbing ability of older flies is impaired by age-dependent neurodegeneration (Kounatidis et al., 2017), we asked if the improved locomotor activity in NelfA*/+ flies was due to reduced AMPs level, and hence lower neurodegeneration as compared to their control siblings. Histological examination of D50 brains harvested from NELF-A*/+, RU486-treated ElavGal4.
Switch/NelfA-RNAi and their respective control sibling flies did not show any morphological difference (data not shown). This unexpected observation might be explained by the significantly lower level of Rel and AMPs expression in NelfA*/+ and their control siblings as compared to age-matched w 1118 flies ( Figure S2A). Nevertheless, the lower AMPs expression in NelfA*/+ flies is indicative of less systemic inflammation (Figure 2b,c), presumably attributed to other pro-longevity molecular changes.

| NELF depletion induces HSPs to protect against oxidative stress
HSPs are molecular chaperones that regulate protein homeostasis in response to environmental stress (Richter et al., 2010). In accordance, ectopic expression of specific Hsp genes could lead to extension in Drosophila lifespan (Morrow et al., 2004;Wang et al., 2004). and S1A) may facilitate the release of paused RNAPII at Hsp promoters, leading to higher transcriptional output that is similar to NELF-depleted S2 cells (Gilchrist et al., 2008). To test this, RNAPII chromatin immunoprecipitation (ChIP) was performed on D50 heads harvested from NelfA*/+ flies and their control siblings.
Site-specific primers on Hsp68 and Hsp83 genes were designed based on RNAPII-ChIP-seq data from KC167 cells ( Figure S3B,C).
IgG-ChIP yielded undetectable signal whereas opposite patterns of RNAPII-ChIP signals were observed at the highly transcribed genes upon H 2 O 2 treatment, although the level of Hsp genes was significantly higher in NELF-A KD cells (Figure 3c). This suggests F I G U R E 3 Elevated HSPs enhances stress resistance and protein homeostasis. (a) Quantification of Hsp genes expression in the heads of D50 male NelfA*/+ flies and their control siblings. Data presented as mean ± SEM, n = 3, unpaired two-tailed t test. (b) RNAPII occupancy at the TSS, GB, and TES of Hsp68 and Hsp83 in the heads of D50 NelfA*/+ flies and their control siblings. Representative ChIP-data presented as mean ± SD from triplicate qPCR reactions. Unpaired two-tailed t test. Data of another biological replicate are presented in Figure S3. (c) Top: NELF genes were KD by dsRNA on Day 0 and 1 (D0, D1) followed by 16 h of 5 mM H 2 O 2 treatment. Bottom: Quantification of mRNA expression on D3 with data presented as mean ± SD, n = 2. *p = 0.005; **p < 0.001 and ***p < 0.0003, unpaired one-tailed t test. (d) βgal, NELF-A or NELF-E KD S2 cells treated with 1 mM H 2 O 2 were harvested for γ-H2AV immunoblot (Top) and RT-PCR of NELF genes (Bottom). The level of γ-H2AV was normalized to β-tubulin and presented as mean ± SD, n = 2, unpaired two-tailed t test. (e) Quantification of Hsp27 expression in brains isolated from pan-neuronal NELF-A and NELF-E KD flies as compared to their respective control siblings. Data presented as mean ± SD, n = 2. *p < 0.03; **p < 0.003, two-tailed t test. (f) Top: Immunoblot of γ-H2AV in heads of aging w 1118 , NelfA*/+ flies and their control siblings. Rep: biological replicate. Bottom: Data presented as mean ± SD, n = 3, unpaired two-tailed t test. (g) Top: Immunoblot of Ref(2)P in total lysates and insoluble fraction of the heads harvested from D50 NelfA*/+ flies and their control siblings. Bottom: Graphs of individual and mean (±SD) level of total and insoluble Ref (2)  This indicates that NELF-A KD confers more robust protection against DNA damage induced by oxidative stress as compared to depleting other NELF subunits. As Hsp27 gene has been shown to suppress ROS (Wyttenbach et al., 2002), we also examined its expression level in the brains harvested from flies where either NELF-A or NELF-E gene has been KD by RNAi (Figures 1e and S1G).

| Reduced DNA damage and protein aggregation in aged NelfA*/+ flies
To gain mechanistic insight into the extended healthspan, we examined several markers of aging in the head/brain tissues harvested from NelfA*/+ flies and their control siblings. Consistent with the notion that Drosophila undergoes age-dependent accumulation of DNA damage, γ-H2Av level in old D50 w 1118 male fly heads is significantly higher than young D10 flies (Figure 3f). In contrast, γ-H2Av level in the heads of D50 NelfA*/+ flies was drastically lower than their control siblings and aged-matched w 1118 flies (Figure 3f). ROS accumulation can disrupt protein homeostasis to cause protein aggregation during aging (Reichmann et al., 2018). Given their roles in mediating proteostasis (Richter et al., 2010), we asked if elevated HSPs observed in NelfA*/+ fly heads may regulate the level of protein aggregation. In Drosophila brains, Ref(2)P marks ubiquitinated protein bodies for autophagic degradation such that the accumulation of insoluble Ref(2)P fraction denotes the formation of protein aggregates (Bartlett et al., 2011;Nezis et al., 2008). Although Ref(2)P protein in the total head lysates was relatively similar across biological replicates, the level of insoluble Ref(2)P fraction is significantly lower in NelfA*/+ fly heads as compared to their control siblings (Figures 3g and S3G). Similarly, the level of insoluble Ref(2)P protein detected in the NELF-A KD brains is also significantly lower than their control siblings (EtOH-versus RU486-treated ElavGal4. Switch/NELFA-RNAi flies; Figure S3H,I). These results suggest that elevated level of HSPs in NelfA*/+ flies may attenuate DNA damage and protein aggregation during aging.

| Lower ROS level is linked to enhanced stress resistance in NelfA*/+ flies
To determine if age-dependent ROS accumulation is substantially different between NelfA*/+ flies and their control siblings, D50 brains were stained with MitoSOX TM Red reagent, a dye that produces red fluorescence when oxidized by superoxide in the mitochondria.
Imaging showed that the normalized MitoSOX signal in NelfA*/+ brains is significantly lower than their control siblings (Figures 4a,b, and S4A). Similarly, MitoSOX fluorescent signal is also reduced in the head, but not in the body lysates prepared from NelfA*/+ flies (Figure 4b), suggesting that NELF-A may modulate ROS level specifically in the head/brain tissues. To determine the effect of knocking down NELF-A gene in neuronal populations, MitoSOX signal was measured in the head and body lysates prepared from male ElavGal4.
Switch/NELFA-RNAi flies. Interestingly, pan-neuronal KD of NELF-A reduces the level of ROS in the head, but not in the body lysates ( Figure S4B). This suggests that manipulating the level of NELF-A in the brains, presumably through HSPs, is sufficient to inhibit ROS accumulation in the heads during aging. To further test if the lower ROS level translates into better protection against environmental stress, D30 NelfA*/+ flies and their control siblings were exposed to 16 hours of 1% H 2 O 2 treatment. Surprisingly, male NelfA*/+ flies lived longer (~10% higher than untreated NelfA*/+) after H 2 O 2 treatment whereas the mean lifespan of control siblings was reduced (~5% lower than untreated control siblings; Figure 4c). In contrast, female flies appeared to tolerate this short-term H 2 O 2 treatment ( Figure S4C). Taken together, halving the level of NELF-A not only lowers the accumulation of ROS, DNA damage, and protein aggregation in the head/brain tissues, it also enhances resistance against environmental stress in male NelfA*/+ flies.

| NELF-A depletion enhances stress resistance in human SH-SY5Y cells
Regulated release of paused RNAPII is crucial for the activation of pro-survival transcriptional programs upon genotoxic stress in human cells (Bugai et al., 2019), suggesting that NELF-A-mediated stress responses might be highly conserved across species. Consistent with this notion, genes with highly paused RNAPII promoters are associated with DNA damage pathways in human SH-SY5Y cells ( Figure   S4D, Table S2). It is plausible that depleting NELF-A gene can facilitate productive transcription of these genes to enhance cellular stress responses like in Drosophila. To test this, NELF-A gene in SH- . Absorbance (Abs) of MitoSOX signals from the head/body lysates normalized to the protein concentration was presented as mean ± SEM, n = 3 (Bottom). Unpaired two-tailed t test. (c) Lifespan of male NelfA*/+ flies and their control siblings that were untreated or exposed to 16 h of 1% H 2 O 2 at Day 30. n = 100-120 flies per group. Log-rank test. (D) NELF-A gene was KD by siRNA on D1 and D2 followed by 14 h of 1 mM H 2 O 2 treatment on D3. Cells were analyzed on D4 or D5 following 24 h recovery (Top). mRNA expression of different genes on D4. Data presented as mean ± SD, n = 2, two-tailed t test (Bottom). showed that Scr KD cells have significantly higher incidence of apoptosis than NELF-A KD cells (Figure 4f). Taken together, our results from Drosophila S2 and human SH-SY5Y cells demonstrate that reducing the level of NELF-A is sufficient to prevent DNA damage and enhance resistance against oxidative stress.  Table S3). More than 70% (6700/9363) of the differential H3K9me2 peaks in NelfA*/+ flies overlap with known TEs; and are located within the promoter, intronic, and distal intergenic regions (Figures   5d and S5C). The expression of Gypsy and Diver in NelfA*/+ flies is significantly lower than their control siblings. Of the TEs bound by H3K9me2, about 25% of the Gypsy and Diver TEs contain differential H3K9me2 peaks that are more enriched in NelfA*/+ flies ( Figure   S5D, Table S3). The differential enrichment of H3K9me2 across specific Diver2 and Gypsy1 loci was furthered validated by ChIP-qPCR ( Figure 5f). In contrast, there is lower percentage of differential H3K9me2 peaks at DMRC1A, whose expression is highly similar between NelfA*/+ flies and their control siblings ( Figure S5D). These results suggest that lower NELF-A level may promote the maintenance of H3K9me2-marked heterochromatin regions, which in turn represses specific TEs like Gypsy and DIVER in the NelfA*/+ flies.

| GNMT may promote H3K9me2 level through SAM in NelfA*/+ fly heads
To understand what contributes to the increased H3K9me2/3 level in old NelfA*/+ flies, we examined the expression pattern of genes that regulate H3K9me2/3 modifications directly or via carbon metabolism (Table S1)

| DISCUSS ION
Through functional analyses and epigenome profiling, we showed that the level of NELF-A in the head tissues plays an important role in regulating Drosophila healthspan. As RNAPII pausing is the ratelimiting step in the transcription of many Hsp genes (Lis et al., 2000;Wu et al., 2003), lowering the level of NELF-A in either NelfA*/+ flies or by pan-neuronal NELF-A KD might facilitate the release of paused RNAPII and increase its productive elongation of these genes (Figure 3a,b). Consistent with their roles in extending lifespan (Biteau et al., 2010;Morrow et al., 2004;Wang et al., 2004) and suppressing neurotoxicity in fly Huntington's disease model (Warrick et al., 1999), Unexpectedly, animal lifespan is only affected by NELF-A perturbation even though depleting other NELF subunits was sufficient to induce the expression of Hsp genes in S2 cells (Gilchrist et al., 2008).
Biochemical and structural studies showed that NELF-A mobile Cterminal domain plays a more important role in stabilizing RNAPII pausing complex than NELF-E subunit (Narita et al., 2003;Vos et al., 2018). Consistent with these findings, there is higher expression of Hsp genes in conjunction with reduced DNA damage in NELF-A KD cells as compared to NELF-E KD cells (Figure 3c-e). In vitro RNA extension assay reveals that NELF-A, and not NELF-E, is required for RNAPII pausing (Vos et al., 2018). It is likely that perturbating NELF-E or other subunits might not elicit sufficient change in the level of HSPs to influence animal lifespan. Moreover, the expression of NELF-A gene in fly heads is ~7-10-fold higher than NELF-E and NELF-B respectively ( Figure S2F), suggesting that NELF-A may play a more dominant role in gene regulation than other NELF subunits in the head tissues.
In aging flies, elevated Rel level in the fat bodies and brains triggers systemic inflammation to cause gut hyperplasia (Chen et al., 2014) and neurological decline, respectively (Kounatidis et al., 2017). Interestingly, NELF depletion has been shown to attenuate immune responses via distinct mechanisms in different model systems. In S2 cells and Drosophila larvae, NELF KD downregulates the basal expression of many components of the immune signaling pathways (including Rel), thereby reducing the activation of AMP genes (Gilchrist et al., 2012). In mouse macrophage, knocking out NELF increases the expression of key anti-inflammatory cytokine IL-10 through AP-1-dependent transcriptional circuit (Yu et al., 2020). Despite having similar level of Rel, the brains/heads of D50 NelfA*/+ flies has significantly lower AMPs expression than their control siblings. This is indicative of less systemic inflammation, a likely consequence of enhanced HSP-mediated protection. The potential pro-longevity effects of other neuropeptides and signaling pathways in NelfA*/+ flies will warrant further studies (Figure 2b).
The pro-longevity role of NELF-A appears to be restricted to the brains since depleting NELF-A in the neuronal populations, but not in fat bodies, affects the animal healthspan ( Figure S1). Similarly, ROS level is significantly reduced only in the brain/head tissues of NelfA*/+ flies as compared to their control siblings (Figure 4b).
In developing mouse cortex, epigenetic profiling of neural progenitor cells (NPC) and differentiated neurons revealed that the promoters of many DNA damage response and repair genes have highly paused RNAPII (Liu et al., 2017). Moreover, GO analysis showed that these genes are associated with molecular functions like NF-κB and HSP binding ( Figure S5D, Table S2), suggesting the release of paused RNAPII may be the rate-limiting step in their transcription. In human cells, activation of P-TEFb by RNA-binding motif protein 7 is necessary to release paused RNAPII for productive transcription of DNA damage response genes upon genotoxic stress (Bugai et al., 2019). Consistent with these findings, NELF-A KD in human SH-SY5Y cells led to better protection against H 2 O 2induced DNA damage and apoptosis than Scr KD cells (Figure 4d,f).
These results suggest that regulation of stress-responsive genes by NELF-A mediated RNAPII pausing is likely to be conserved across different species.
Although it remains to be seen if NELF-A affects lifespan in mammals, several lines of evidence indicate that lowering the level of NELF may have wider implications in regulating human health and disease state. NELF-A is a prognostic marker of liver cancer such that higher level of protein expression correlates with unfavorable survival probability in patients (proteinatlas.org). Conversely, knocking down NELF-B gene in hepatocellular carcinoma cells is sufficient to reduce cellular proliferation and migration (El Zeneini et al., 2017).
In mouse, microRNA-133 has been demonstrated to inhibit cardiac hypertrophy through repressing the translation of RhoA, Cdc42, and NELF-A mRNA (Care et al., 2007). Conversely, overexpression of NELF-A in cardiomyocytes can induce hypertrophic gene program (Care et al., 2007). Further studies on RNAPII pausing in mammalian systems under different contexts might provide insights into its roles in regulating stress responses and aging transcriptome. Investigation of direct NELF targets would complement these studies and potentially lead to the discovery of pro-longevity or specific therapeutic strategies.

| Experimental procedures
Additional experimental procedures are described in supporting information (SI).
Constructs were sequenced and sent to Bestgene Inc to make transgenic fly. Homozygous lines (w 1118 background) are denoted as UAS-NelfA-6myc and UAS-Nelf-E-6myc.
The UAS-RNAi males were mated to GeneSwitch drivers. F1 offspring were cultured in media containing either ethanol or 120 mg/ ml of mifepristone (RU486, Sigma).

| Lifespan assay
Nelf-A KG09483 /Tm3 virgin females were backcrossed to w 1118 males for >10 generations to obtain Nelf-A KG09483 /+ mutants (NelfA*/+) and control siblings (+/+) with similar genetic background. Twenty flies were housed in a vial (diameter 2.5 cm, height 9.5 cm) with 5 ml of standard fly media. Food was changed every two days with daily recording. Data were plotted with Prism GraphPad and statistical significance calculated by log-rank Mantel-Cox's test. There were at least 100 flies per aging assay.

| Negative geotaxis (climbing) assay
About 10-15 age-matched flies were transferred to fresh food a day before assay. On the day of assay, flies were transferred without CO 2 anesthesia into glass cylinder (height 22 cm, diameter 2.5 cm, capacity 50 ml, AS ONE) capped with cotton plug. The flies were tapped to the bottom and distance ascended in 6 s (D20, 30) or 10 s (D50) was captured by digital camera. This was repeated to obtain sizeable number for each genotype. Data were plotted with Prism GraphPad and statistical significance calculated by Mann-Whitney test.

| RNA isolation, RNA-seq, and data processing
Total RNA was isolated from 20 x D50 fly heads using RNAzol®RT (Sigma) protocol and sequenced by BGI (Hong Kong). RNA-seq dataset was processed and analyzed according to previous methods detailed in SI.

| Chromatin Immunoprecipitation (ChIP) and sequencing
ChIP and library amplification were performed according to previous protocol detailed in SI. For each biological replicate, 100 fly heads and 3.11 | S2 cells culture, dsRNA KD, and H 2 O 2 treatment NELF-A and NELF-E dsRNA was generated with Ambion MEGAscript T7 kit using PCR template. On Day 0 (D0), 0.5e 6 of S2 cells were seeded in 0.5 ml of serum-free medium (SFM, Gibco™, 21720024) in a 24-well plate.
Cells were transfected with 4 µg of dsRNA (βgal, NELF-A, and NELF-E) with 4 µl of Cellfectin™ II (ThermoFisher, 10362100) for 2 h. SFM was replaced with 0.5 mL of full media. After 24 h (D1), second round of transfection was performed with 8 µg of dsRNA and 8 µl of Cellfectin. On D2, cells were split evenly into two wells and incubated with either 1 or 5 mM of H 2 O 2 for 16 h. On D3, cells were washed twice with PBS and harvested for analyses.

| Primers list
Primers sequences are listed in Table S4.

ACK N OWLED G EM ENTS
This work was supported by core funding from Temasek Life Sciences Laboratory, Singapore.

CO N FLI C T O F I NTE R E S T
None declared.