Early‐onset aging and mitochondrial defects associated with loss of histone acetyltransferase 1 (Hat1)

Abstract Histone acetyltransferase 1 (Hat1) is responsible for the acetylation of newly synthesized histone H4 on lysines 5 and 12 during the process of chromatin assembly. To understand the broader biological role of Hat1, we have generated a conditional mouse knockout model of this enzyme. We previously reported that Hat1 is required for viability and important for mammalian development and genome stability. In this study, we show that haploinsufficiency of Hat1 results in a significant decrease in lifespan. Defects observed in Hat1+/− mice are consistent with an early‐onset aging phenotype. These include lordokyphosis (hunchback), muscle atrophy, minor growth retardation, reduced subcutaneous fat, cancer, and paralysis. In addition, the expression of Hat1 is linked to the normal aging process as Hat1 mRNA and protein becomes undetectable in many tissues in old mice. At the cellular level, fibroblasts from Hat1 haploinsufficient embryos undergo early senescence and accumulate high levels of p21. Hat1+/− mouse embryonic fibroblasts (MEFs) display modest increases in endogenous DNA damage but have significantly higher levels of reactive oxygen species (ROS). Consistently, further studies show that Hat1−/− MEFs exhibit mitochondrial defects suggesting a critical role for Hat1 in mitochondrial function. Taken together, these data show that loss of Hat1 induces multiple hallmarks of early‐onset aging.

Hat1 (or Kat1) was the first lysine acetyltransferase identified and serves as a paradigm for the study of type B histone acetyltransferases (Kleff, Andrulis, Anderson, & Sternglanz, 1995;Parthun, Widom, & Gottschling, 1996). Type B histone acetyltransferases are responsible for the acetylation of newly synthesized histone H4 prior to assembly into chromatin (Nagarajan et al., 2013). Loss of this modification on newly synthesized histone H4 translates into numerous alterations in nascent chromatin structure during DNA replication (Agudelo Garcia et al., 2017). In addition, high levels of spontaneous DNA damage, sensitivity to DNA damaging agents, and genome instability are highly conserved phenotypes in Hat1 −/− cells that are observed in fungi, chicken, and mammalian cells (Barman et al., 2006;Benson et al., 2007;Ge et al., 2013;Ge, Wang, & Parthun, 2011;Nagarajan et al., 2013;Qin & Parthun, 2002, 2006Tscherner, Stappler, Hnisz, & Kuchler, 2012;Yang et al., 2013). Hat1 is also essential for viability in the mouse, as pups lacking Hat1 display neonatal lethality. In addition, neonates lacking Hat1 exhibit impaired lung F I G U R E 1 Haploinsufficiency of Hat1 reduces lifespan. (a) Kaplan-Meier survival curves of Hat1 +/+ and Hat1 +/− mice. The percentages of survival are plotted as a function of age in weeks. Animals were monitored for tumors, morbidity or spontaneous death over a period of 120 weeks. All mice were of C57Bl6J background. The survival curves of the Hat1 +/+ and Hat1 +/− mice were significant (p < 0.001). Survival curves were analyzed with GraphPad prism software using log-rank test to determine statistically significant difference between survival curves of the two groups.  (Nagarajan et al., 2013).
To obtain a more complete view of the biological function of Hat1 in mammals, we have analyzed mice that are heterozygous for Hat1 (Hat1 +/− ). Strikingly, Hat1 +/− mice show a significant decrease in lifespan. Hat1 +/− mice display a number of pathologies that suggest this loss of viability is the result of early-onset aging. Consistent with this, at the cellular level, Hat1 +/− cells display an increase in senescence markers. Further analyses indicate that, in addition to previously shown defects in chromatin structure and DNA repair, loss of Hat1 results in impaired mitochondrial function. Hence, Hat1 +/− mice are a novel model of premature aging that can provide insight into the role of protein acetylation in the regulation of multiple cellular processes.

| Haploinsufficiency of Hat1 reduces lifespan
Mice with a complete loss of Hat1 are not viable (Nagarajan et al., 2013 Figure 1a). Mice that were found dead or were euthanized due to poor body score were assessed by overall body condition and by examination of major internal organs. Approximately 40% of the Hat1 +/− animals that died spontaneously could not be analyzed due to tissue lysis. Figure 1b lists the spectrum of phenotypes observed at the time of death. The most common phenotypes observed were lordokyphosis ( Figure 1c), hindlimb paralysis, tumors, and muscle atrophy.
Tumors were observed primarily in the liver, spleen, and kidney. The variety of abnormalities observed in Hat1 +/− animals suggested that they were not succumbing to a single disease state but was consistent with the premature appearance of phenotypes associated with aging.

| Hat1 expression decreases during normal aging
To determine whether the expression of Hat1 was linked to the normal process of aging, Hat1 +/+ mice were sacrificed at 3, 12, and 30 months of age (n = 3 animals at each time point). Hat1 mRNA levels were measured from several tissues at each of the time points. As seen in Figure 2a, Hat1 mRNA was variable but easily detectable in many tissues at 3 and 12 months of age. There was a significant decrease in Hat1 mRNA levels in thymus, lung, muscle, and brain at 30 months of age.
We determined the levels of Hat1 protein by immunofluorescence staining of sections of several different tissues from young and old Hat1 +/+ mice (3 months and 31 months, respectively). We also analyzed the levels of histone H4 lysine 5 and lysine 12 acetylation, the primary known targets of Hat1, to determine whether they correlated with Hat1 protein levels. In general, an age-dependent loss of Hat1 was observed that largely mirrored the results seen for Hat1 mRNA expression. There was Hat1 protein in most cells of the thymus in 3-month-old mice, while the number of Hat1-positive cells decreased dramatically in 30-month-old mice ( Figure 2b). The levels of histone H4 lysine 5 and lysine 12 acetylation closely mirrored that of Hat1, suggesting that Hat1 may be the primary enzyme responsible for these modifications in the thymus.
In the lung of young animals, Hat1 protein was present in only scattered nuclei with the exception of cells surrounding the alveoli, where most of the cells showed a high level of Hat1 protein ( Figure 2c). Consistent with the Hat1 mRNA analysis, Hat1 protein was largely absent in the lung of old mice. The patterns of H4 lysine 5 and lysine 12 acetylation were not identical to the pattern of Hat1. H4 lysine 5 and lysine 12 acetylation was more widespread throughout the lung of young mice than Hat1. However, the levels of H4 lysine 5 and lysine 12 acetylation were also highest in the cells immediately surrounding the alveoli. Interestingly, in old mice, while much of the H4 lysine 5 and lysine 12 acetylation remained stable, the alveoli-proximal staining was significantly reduced.
Hat1 protein was expressed in cells throughout the liver, and this expression was largely stable with age ( Figure 2d). Histone H4 lysine 5 and lysine 12 acetylation was also widespread throughout the liver and was largely stable with age.
Interestingly, in the intestine, Hat1 protein was found specifically in cells of the crypt where intestinal stem cells are located ( Figure 2e).
The presence of Hat1 protein in these cells was not altered by age.
The pattern of H4 lysine 5 and lysine 12 acetylation was complex. In young mice, both sites of acetylation were found evenly throughout all cells of the intestine. In old mice, the level of acetylation of both sites decreased dramatically and became predominantly localized to cells of the crypt.
An identical analysis was performed with tissues from young and old Hat1 +/− mice, 3 months and 24 months, respectively (we were not able to obtain Hat +/− mice older than 24 months). The results in the Hat1 +/− mice were similar to the Hat1 +/+ except that the overall level of Hat1 expression in heterozygous mice was typically lower than in the wild-type mice ( Figure S1). Immunoblot analysis of Hat1, H4 lysine 5 acetylation, and H4 lysine 12 acetylation levels in these tissues in young and old mice was also performed ( Figure S2). The immunoblot analyses largely mirror the results observed in the immunofluorescence images. Taken together, the results indicate that Hat1 protein expression is highly tissue-specific and that normal aging can be accompanied by a marked loss of Hat1 mRNA and protein from some tissues. In addition, the acetylation state of H4 lysine 5 and lysine 12 acetylation can also vary with age but does not always correlate with Hat1 expression.

| Hat1 +/− animals exhibit agingrelated phenotypes
Loss of body weight can occur as mice reach old age (Alhurani et al., 2016;Goodrick, Ingram, Reynolds, Freeman, & Cider, 1990;Samorajski et al., 1985). Hat1 +/− mice displayed a reduced size at birth that became less pronounced by 5 weeks of age (Nagarajan et al., 2013). However, after 40 weeks, the Hat1 +/− mice began to exhibit weight loss and animals that survived to 80 weeks had a pro- To obtain a more complete view of the effect of Hat1 haploinsufficiency in mice, age-matched Hat1 +/+ and Hat1 +/− mice were randomly selected and sacrificed. The mice were between 54 and 80 weeks of age, which brackets the average lifespan of the Hat1 +/− mice. Tissue samples were isolated from a variety of organs, and an independent pathologist, blinded to the genotype of the tissue samples, performed a histological analysis. Table 1 summarizes the results of these analyses, and representative images are shown in histiocytic sarcoma in a high percentage of these animals (50% and 33%, respectively) (Table 1, Figure 4a,b). Interestingly, histiocytic sarcoma is a malignancy typically found in old mice (Blackwell, Bucci, Hart, & Turturro, 1995;Lacroix-Triki et al., 2003).
Comparison of skin from Hat1 +/+ and Hat1 +/− animals revealed that the Hat1 +/− mice had a nearly complete lack of subcutaneous fat (Table 1, Figure 4c). There were also differences in visceral fat in the Hat1 +/− animals. The adipocytes from Hat1 +/− mice were much smaller, indicative of fat depletion in these cells (Table 1, Figure 4d).
This dramatic loss of fat is another characteristic that links Hat1 loss with early-onset aging.
The histological analyses also indicated the presence of muscle atrophy in the Hat1 +/− animals (Table 1, Figure   F I G U R E 2 Tissue-specific decreases in Hat1 expression occur during aging. (a) Hat1 mRNA levels were determined in the indicated tissues by digital droplet PCR. Tissues were harvested from Hat1 +/+ animals sacrificed at 3, 12, or 30 months of age, as indicated (n = 3 animals per time point; ns = not significant; *p < 0.05; **p < 0.01). Levels of Hat1 mRNA are plotted relative to cyclophilin. (b-e) The indicated tissues were isolated from young (3 months) and old (31 months) mice. Tissue sections were stained with DAPI and with antibodies against Hat1, histone H4 lysine 5 acetylation, and histone H4 lysine 12 acetylation, as indicated. Images were taken at 10× magnification | 7 of 17 NAGARAJAN et Al.

| Loss of Hat1 increases DNA damage and oxidative stress
The accumulation of DNA damage is clearly an important driver of the aging process. In addition, defective DNA damage repair is an evolutionarily conserved phenotype observed in Hat1-depleted cells in a wide variety of organisms (Barman et al., 2006;Benson et al., 2007;Qin & Parthun, 2002;Tscherner et al., 2012;Yang et al., 2013), including Hat1 −/− MEFs, which have been shown to exhibit increased levels of γ-H2AX staining and genome instability (Nagarajan et al., 2013). Therefore, we investigated whether Hat1 +/− cells also displayed increased levels of DNA damage. Using comet assays to detect DNA double-strand breaks, we found that Hat1 +/− cells showed a modest, but statistically significant increase in DNA damage, con- To comprehensively assess whether Hat1 heterozygosity results in alterations in gene expression that might be linked to the observed cellular defects, we performed RNA-Seq analysis.
Transcript levels were compared between primary Hat1 +/+ and Hat1 +/− MEFs at passage 3. As described above, these Hat1 +/− cells had increased levels of senescence and elevated ROS. As seen in Figure 6d, there was very little change in gene expression in the Hat1 +/− cells. There is one gene, Igfrbp1, whose expression increased by greater than twofold, and there is one gene, Mgll, whose expression decreased by greater than twofold. Alterations in the expression of these genes are unlikely to result in increased cellular senescence or ROS. Hence, large-scale changes in gene expression do not result from haploinsufficiency of Hat1 in MEFs.
F I G U R E 3 Haploinsufficiency of Hat1 results in decreased body and muscle mass. (a) Total body weight of Hat1 +/+ and Hat1 +/− mice at the indicated ages (n = 6 for each genotype  Loss of mitochondrial membrane potential is associated with uncoupled oxidative phosphorylation that arises from either defect in the ATP synthase or to increased permeability of the mitochondrial inner membrane (Sack, 2006).  specific defects in the recombinational repair of DNA double-strand breaks (Qin & Parthun, 2002). Hat1p appears to function directly in the repair process as it is recruited to sites of double-strand breaks, influences histone H4 acetylation near the break sites, and promotes chromatin reassembly following repair (Ge et al., 2011;Qin & Parthun, 2006). Similar results were observed in a human tissue culture model (Yang et al., 2013). Hat1-dependent sensitivity to DNA damage has also been observed in S. pombe, chicken DT40 cells, and C. albicans (Barman et al., 2006;Benson et al., 2007;Tscherner et al., 2012). There are multiple mechanisms through which Hat1 may influence mitochondrial function. A previous report suggested that Hat1 was required for the expression of several nuclear genes important for mitochondrial biogenesis and function (Marin et al., 2017).

| D ISCUSS I ON
However, this may be a cell type-specific phenomenon, as we see no evidence for Hat1-dependent expression of these genes in our A more intriguing possibility is that Hat1 may affect mitochondrial function more directly through the acetylation of mitochondrial proteins. Acetylomic studies have shown that the majority of mitochondrial proteins are acetylated and that the acetylation state of mitochondrial proteins can be a key regulator of protein function (Baeza, Smallegan, & Denu, 2016;Wang et al., 2010;Zhao et al., 2010). There are a limited number of protein acetyltransferases in mammalian cells, and the majority of these enzymes are strictly localized to the nucleus. Hat1 is one of the few protein acetyltransferases that is not restricted to the nucleus. Whether Hat1 can localize to the mitochondria remains to be determined, but Hat1 may also influence mitochondrial protein acetylation through the acetylation of mitochondrial proteins in the cytoplasm prior to their import into the mitochondria.
Protein acetylation has been clearly linked to aging through the Sirtuin family of proteins. The Sirtuins are NAD + -dependent protein deacetylases. A role for Sirtuins in regulating aging was first identified in S. cerevisiae, where overexpression of Sir2p increases lifespan and sir2Δ mutants show decreased lifespan (Kaeberlein et al., 1999). There are multiple Sirtuins in mammalian cells, several of which influence aging processes. For example, SirT1 is a nuclear deacetylase whose substrates include a number of proteins important in aging, such as p53, PGC1a, Nf-kB, and fork head proteins (Haigis & Guarente, 2006).
While knockout of SirT1 leads to neonatal lethality in mice, overexpression has an anti-aging effect on a number of tissues (Alcendor et al., 2007;Kim et al., 2007;Leibiger & Berggren, 2006;Sommer et al., 2006;Wang et al., 2016). SirT3 is a mitochondrial protein deacetylase that is responsible for regulating the acetylation state of many mitochondrial proteins. Mutations in SirT3 in mice result in defects in many of the same processes implicated in aging, such as increased incidence of cancer, cardiovascular disease, metabolic syndrome, and neurodegenerative disease (McDonnell et al., 2015). Another Sirtuin, Sirt6, is also a nuclear protein deacetylase. Sirt6 −/− mice display rapid aging and typically die by one month of age (Mostoslavsky et al., 2006). The spectrum of phenotypes observed in Sirt6 −/− mice is strikingly similar to the phenotypes we observe in Hat1 +/− , including lordokyphosis, loss of subcutaneous fat, and genome instability.
It seems counterintuitive that loss of Sirtuin protein deacetylases in mice would share aging-related phenotypes with loss of the histone acetyltransferase Hat1. However, this has been observed in other systems. Deletion of SIR2 in yeast results in loss of telomeric silencing (Grunstein, 1997). Similarly, deletion of HAT1, in combination with mutations in histone H3, results in loss of telomeric silencing (Kelly, Qin, Gottschling, & Parthun, 2000). Explanations for similar phenotypes resulting from loss of protein deacetylase and acetyltransferase activities include the possibility that Hat1 is a regulator of Sirtuin activity or that intact cycles of acetylation and deacetylation are necessary for the function of common targets of Hat1 and the Sirtuins.
Mice have served as an important mammalian model system for aging research (Harkema, Youssef, & Bruin, 2016;Koks et al., 2016;Quarrie & Riabowol, 2004;Yuan, Peters, & Paigen, 2011). There are a number of models that demonstrate an extremely rapid onset of aging and that model human diseases of aging. These include mutations in the LMNA gene that model Hutchinson-Guilford Progeria (Osorio et al., 2011;Yang, Andres, Spielmann, Young, & Fong, 2008). Other mouse models of extreme aging include mutations in the Werner syndrome helicase (WRN) (Lebel & Leder, 1998). These models typically have a lifespan of 1 to 3 months. Other mouse models that have very short lifespans  Trifunovic et al., 2004;Walston et al., 2008;Weeda et al., 1997).
There are relatively few mouse models of premature aging that have the type of intermediate phenotype seen with the Hat1 +/− animals. A particularly interesting example is mice heterozygous for p53. P53 +/− mice have a lifespan similar to that of Hat1 +/− mice and have a number of phenotypes in common. Interestingly, heterozygosity for p53 significantly increases the lifespan of some progeroid-like mouse aging models and Sirt6 −/− mice (Ghosh et al., 2018). This suggests that p53 signaling can have a profound influence on the aging process.
Metformin may provide an intriguing link between Hat1 and aging.
Metformin, which is used to lower blood sugar, has shown potential as an anti-aging therapeutic. Metformin functions by inducing AMPK activity. While AMPK down-regulates many factors involved in epigenetic regulation, it phosphorylates and activates Hat1 (Bridgeman, Ellison, Melton, Newsholme, & Mamotte, 2018). A primary target of Hat1 acetylation, histone H4 lysine 12, has also been linked to aging.
Increased acetylation of H4 lysine 12 is associated with learning and memory in the hippocampus of mice. With age, the acetylation of H4 lysine 12 is lost along with the ability to form memories. Importantly, increasing H4 lysine 12 acetylation through treatment with HDAC inhibitors restored learning and memory in older mice (Peleg et al., 2010). These results suggest that modulation of Hat1-dependent acetylation may be a therapeutic anti-aging strategy.

| Generation of Hat1 mutant mice
Hat1 +/− mice were generated as described previously (Nagarajan et al., 2013). Heterozygous mice were backcrossed to a C57BL6 background for more than ten generations. All mice were bred and housed in a pathogen-free facility, and all studies were conducted in accordance with guidelines of the institutional animal care and use committee and university laboratory animal resources at The Ohio State University under protocol number 2007A0094. Early passage cells were seeded at 25% confluency in 6-well plates and transfected with 2 μg of expression vector using Fugene reagent (Roche). Cells were harvested and seeded into 100-mm dishes after 48 hr of transfection. The cells were split at 1 in 10 dilutions until passage 5.

| Histology analyses
Mouse tissues were fixed in 10% formalin phosphate buffer, embedded in paraffin, sectioned, and stained with hematoxylin and eosin.
Tissue sections were read by an independent pathologist (Histowiz, Inc), who was blinded to the groups being studied.

| Body fat analysis
EchoMRI were used to measure body fat in live Hat1 +/+ and Hat1 +/− mice. Mice were placed in a plastic cylinder without sedation or anesthetic agent. The cylinder was then inserted into a tubular space in the EchoMRI™ system. EchoMRI was performed at OSU small animal imaging core laboratory (SAIC).

| Imaging
Hat1 +/+ and Hat1 +/− mice skeletal images were taken using Small Animal Radiation Research Platform (SARRP) by Xstrahl. Mouse cropped protocol entails: 180 projections at 60 kV 0.8mA fine focus w/0.5 Al filter for 1.2cGy. Images were analyzed by Muriplan imaging software. Slides were transferred to electrophoresis system which contained prechilled alkaline electrophoresis solution and run at 1 V/cm, 300 mA for 45 min at 4°C. The slides were rinsed twice with sterile water for 5min. and washed in 70% ethanol for 5 min. Slides were stained with 100μl of SYBR Green I for 5-10 min in the dark, and slides were analyzed under Zeiss Axiophot fluorescence microscope.

| Comet assay
Images were taken using Metavue software version 6.3r2 software, and comet tails were analyzed using OpenComet by ImageJ. Dotplot was generated by SigmaPlot 12.0. Volcano plot was created using ggplot2 (Wickham, 2009). For labeling purposes, transcripts were considered statistically significant if they exhibited greater than twofold change in abundance and a p value of less than 0.05.

| Immunoblotting
Whole-protein lysates from MEFs and mouse tissues were prepared by using RIPA buffer (Research product International-R26200-100 mM

| Immunofluorescence (IF) on paraffin embedded mouse tissue sections
Mouse tissues were fixed in 10% buffered formalin phosphate solution (cat.# SF100-4; Fisher Scientific) for 48 hr and transferred to 70% EtOH. Tissues were processed, embedded in paraffin, and cut in 8 micron thickness section on positively charged slides. Paraffinembedded tissues were deparaffinized using a xylene-ethanol series and subjected to antigen retrieval using Target Retrieval solution (cat.# S1700; Dako North America, Inc.) in a steamer for 60 min. The sections were then cooled to RT, blocked for 60 min. at RT in 1% Antibody dilution buffer (10% ADB in PBSpH7.4:3% BSA, 10% Goat serum, 0.05% Triton X-100), and then incubated overnight with primary antibodies (Hat1; H4K5 and H4K12-Abcam; γ-H2AX ab26350, Abcam) in antibody dilution buffer at 4°C. Next day, slides were washed three times in 1% ADB and incubated with Alexa Fluor ® 594-conjugated secondary antibody (Jackson immunoresearch Lab. Inc) diluted with ADB in the dark for 60min. Following three times PBS washes, the slides were mounted in Vectashield mounting medium with DAPI (cat.# H1200, Vector Laboratories, Burlingame, CA).
Photographs were taken by ANDOR microscope by Metamorph 64bit Software at The Ohio State University Neuroscience Microscopy Core.

| Detection of reactive oxygen species
Cells were incubated with 25 μM DCFDA for 30 min at 37°C in the dark. ROS generation was assessed using fluorescence microscopy.

| Determination of mitochondrial membrane potential
Cells were incubated with 1 μg/mL JC-1 (Life Technologies) for 30 min at 37°C in the dark. Cells were then washed twice with PBS.
Changes in mitochondrial membrane potential were determined using fluorescence microscopy.

| Determination of cell viability
Cells were plated in a 96-well plate. After 24 hr, cells were starved by changing to media containing 10 mM galactose. Cell viability was determined by MTS assay using the CellTiter 96 Aqueous One Solution according to the manufacturer's protocol (Promega, Madison, WI, USA). Absorbance was read at 490 nm.

| Metabolic measurements
Oxygen consumption measurements were performed on the XF24 analyzer (Seahorse Bioscience) following the manufacturer's protocol. Briefly, cells were plated in 24-well plates and incubated overnight at 37°C. The next day, cells were incubated in assay media for at least 1 hr before measurements. Wells containing cells were sequentially injected with 1 μM oligomycin, 2 μM FCCP, and 0.5 μM antimycin A and rotenone.

| Statistical analysis
Mouse survival curve analysis was calculated by the Kaplan-Meier method, and differences were determined by the log-rank test. All statistics were performed using sigmaplot 12.0. Results are presented as mean ± SE. Comparisons of experimental groups were carried out with Student's t test. Differences among multiple groups were analyzed by a one-way ANOVA. p < 0.05 was considered to be significant when analyzing statistical differences between Hat1 +/+ , Hat1 +/− , and Hat1 −/− .

ACK N OWLED G M ENTS
This work was supported by a grant from the National Institutes of Health (GM062970 to M.R.P.). The Ohio State Neuroscience Microscopy core is supported by NIH/NINDS P30NS104177. We would like to thank Dr. Michael Freitas for assistance with the statistical analyses, Dr. Anne Stroheker for helpful discussions, and Dr.
Dan Gottschling and Dr. Maria Mihaylova for critical reading of the manuscript.

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