Spermidine prolongs lifespan and prevents liver fibrosis and hepatocellular carcinoma by activating MAP1S-mediated autophagy

Liver fibrosis and hepatocellular carcinoma (HCC) have worldwide impact but continue to lack safe, low cost, and effective treatments. In this study, we show how the simple polyamine spermidine can relieve cancer cell defects in autophagy, which trigger oxidative stress-induced cell death and promote liver fibrosis and HCC. We found that the autophagic marker protein LC3 interacted with the microtubule-associated protein MAP1S, which positively regulated autophagy flux in cells. MAP1S stability was regulated in turn by its interaction with the histone deacetylase HDAC4. Notably, MAP1S-deficient mice exhibited a 20% reduction in median survival and developed severe liver fibrosis and HCC under stress. Wild-type mice or cells treated with spermidine exhibited a relative increase in MAP1S stability and autophagy signaling via depletion of cytosolic HDAC4. Extending recent evidence that orally administered spermidine can extend lifespan in mice, we determined that life extension of up to 25% can be produced by lifelong administration, which also reduced liver fibrosis and HCC foci as induced by chemical insults. Genetic investigations established that these observed impacts of oral spermidine administration relied upon MAP1S-mediated autophagy. Our findings offer a preclinical proof of concept for the administration of oral spermidine to prevent liver fibrosis and HCC and potentially extend lifespan. Cancer Res; 77(11); 2938-51. ©2017 AACR.


Introduction
Hepatocellular carcinoma (HCC) is the most common form of liver cancers and has become a serious problem in the US in recent decades (1). Most HCC patients also develop liver cirrhosis caused by liver fibrosis; and liver cirrhosis itself is the most common non-neoplastic cause of mortality among hepatobiliary and digestive diseases in the US (1,2). Currently, only a small portion of HCC cases are diagnosed early enough to be treated by surgical resection or liver transplantation, but general patient prognosis is very poor (3). Otherwise, no effective drug for HCC is available. Although hepatitis B virus (HBV) immunization and antiviral therapy against HBV or hepatitis C virus (HCV) in established patients can reduce HCC risk, their cost-effectiveness for HCC prevention in the general population is unknown. Prescribed medications such as statins, metformin and aspirin have shown chemo-preventive effects for HCC, but each simultaneously exhibits other unintended effects (3). Natural dietary components and phytochemicals consumed by humans over their lifetime provide a great window of opportunity to prevent or change the pathogenic course of the disease in a safer and more cost-effective way.
Autophagy is a major pathway for the degradation of dysfunctional organelles and misfolded or aggregated proteins (4). Autophagy defects trigger oxidative stresses and lysosomal rupture which induce different types of cell death including apoptosis, necrosis and pyroptosis (5,6). Pyroptosis is specifically characterized by the activation of caspase-1 and the release of pro-inflammatory cytokines to cause the death of other cells in the environment (7); and it promotes liver fibrosis (8). Liver fibrosis-induced liver regeneration contributes to tumorigenesis with two key events: the generation of on April 28, 2021. © 2017 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.
Author Manuscript Published OnlineFirst on April 6, 2017; DOI: 10.1158/0008-5472.  genome instability by oxidative stress and the propagation of genome instability by active cell division. Although an inhibition of autophagy might promote apoptosis or necrosis to kill cancer cells, the activation of autophagy leads to a suppression of pyroptosis to reduce liver fibrosis and delay or prevent HCC development.
Microtubule-associated protein 1S (MAP1S), previously named as C19ORF5, associates with microtubules stabilized by either chemotherapeutic drug and microtubule stabilizer taxanes or tumor suppressive protein RASSF1A (9,10). The specific accumulation of MAP1S short chain in response to mitotic arresting leads to a collapse of mitochondria on mitotic spindle and causes mitotic cell death (11). As a sequence homologue of the microtubule-associated protein MAP1A and MAP1B, Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.
Author Manuscript Published OnlineFirst on April 6, 2017; DOI: 10.1158/0008-5472.  Spermidine, a chemical originally isolated from semen and enriched in wheat germ, grapefruit and soybean, was reported to change protein acetylation to regulate autophagy and promote longevity in multiple model systems (20-23). Spermidine mimicking caloric restriction activates autophagy to enhance chemotherapy-induced inhibition of tumor growth possibly through boosting T cell-mediated immune responses (24). MAP1S is one of the proteins whose acetylation was significantly increased upon the exposure to spermidine (22). We hypothesized that spermidine may prolong lifespans and suppress the development of liver fibrosis and HCC by increasing the levels of MAP1S to activate autophagy flux. Indeed, here we show that spermidine increases the acetylation and stability of MAP1S and activates autophagy flux by depleting cytosolic HDAC4 to reduce the association between MAP1S and HDAC4.
Spermidine prolongs lifespans, alleviates carbon tetrachloride (CCl 4 )-induced liver fibrosis and suppresses DEN-induced HCC through MAP1S in mice. All the benefits of spermidine depend on the presence of MAP1S and do not show in MAP1S -/mice.

Materials and Methods
Antibodies, plasmids and other reagents.
Antibody against p62 (BML-PW9860) was from Enzo Life Sciences International.   containing 10% FBS and antibiotics for up to ten passages. Since those cell lines were solely used for biochemical and cell biological assays and not for study of cancerrelated biological functions, they were not authenticated and not tested for mycoplasma.
Primary mouse hepatocytes were isolated from 12-week-old male mice following the protocol as described (28). Mice were anesthetized and laparotomized with a U shape incision. The portal vein was cannulated and the inferior vena cava was sectioned. The liver was simultaneously perfused with EBSS medium containing 0.5 mM EGTA, followed by HBSS medium supplied with 0.3 mg/ml type IV collagenase. After perfusion, the liver was cut out and gently squeezed until most of the hepatocytes got out. The cells were filtered through sterile 70-μm-mesh nylon, washed by centrifugation and resuspended in the culture media. Cell viability was assessed by Trypan Blue staining and cells were seeded at 1×10 6 in the 60 mm dish.

Assays of the impact of spermidine on autophagy and HDAC4 in mice.
To test the acute effect of spermidine on autophagy in mice, one month-old male wild-type or MAP1S -/mice were subjected to intraperitoneal injection of a single dose of spermidine of 50 mg/kg body weight as previously reported (22). The liver and other organs of mice including brain, lung and heart were collected 3 hrs later to prepare cell lysates for immunoblot analyses.

Survival analyses of mice and other statistical analyses.
To conduct survival analyses, mice were fed with normal drinking water or water containing 3 mM spermidine and observed to record their survival times when they were found dead or when they were found to be moribund. The overall survivals and median survivals were analyzed by the Kaplan-Meier method. Cox proportional-hazard analysis with univariate or multivariate method was used to explore the effect of variables on overall survivals. Liver tissues collected from 18-month old mice were either frozen or fixed for either immunoblot analysis or staining with dihydroethidine hydrochloride to measure oxidative stress or with Hematoxylin and Eosin (H&E) to analyze sinusoidal dilatation as we previously reported (16). The intensities of protein bands were quantified using ImageJ software and normalized by the levels of loading control β-Actin.

Induction of liver fibrosis.
To investigate the impact of spermidine on liver fibrosis, three months-old mice were selected to be intraperitoneally injected with 5 µl (0 or 10% CCl 4 in corn oil)/g body weight twice a week for 1 month to induce liver fibrosis as described (29). During the same period, mice were fed with either normal water or water containing 3 mM spermidine. At the end of experiment, mice were weighted and sacrificed to collect liver tissues. Liver tissue sections were stained with H&E and lesion area of liver were quantified using ImageJ software. Liver fibrosis was visualized by Sirius Red staining and quantified by the contents of hydroxyproline, and levels of α-SMA were analyzed either by immunoblot or immunostaining as we previously reported (16). To detect the tumor suppressive function of spermidine, 15-day-old male mice were intraperitoneally injected with a single dose of 10 μg/g body weight of DEN dissolved in saline. Mice were allowed to continuously drink either normal water or water containing 3 mM spermidine and sacrificed at 7 months after birth to examine the development of HCC as we previously described (14). Body weights, liver weights, ratios of liver weight to body weight and number of surface tumors were recorded. Liver tissues were frozen or fixed for immunoblotting, immunostaining and H&E staining similarly as we previously described (14,16,30).

Cell transfection, immunoprecipitation, protein stability assay, subcellular fractionation and confocal fluorescent microscopy.
Cell transfection, immunoprecipitation, immunoblot analysis and confocal fluorescent microscopy were performed as previously described (19,26). Heavy membrane pellet, light membrane pellet, and soluble fraction were prepared from HeLa cells as described (10). Protein Lamin B served as a nuclear marker and 14-3-3 and βtubulin served as cytosolic markers.

Spermidine-induced expansion of mouse lifespans depends on MAP1S.
Spermidine has been shown to increase the lifespans of yeast, nematodes and flies in an autophagy-dependent fashion (20,21). It was later confirmed that mice exposed to spermidine at age of 4 or 18 months exhibited a 10% increase of median survivals (23). Spermidine was shown to enhance the acetylation of MAP1S (22).
Acetylation of MAP1S activates autophagy; and MAP1S depletion leads to an elevation in oxidative stress and intensity of sinusoidal dilatation in mouse liver tissues; and wildtype mice live a median lifespan of 28.0 months while MAP1S -/mice live a median lifespans of 22.4 months (16,19). We reasoned that the impact of spermidine on mouse lifespans may depend on MAP1S. We fed both wild-type and MAP1S -/mice continuously for 18 months with either normal water or water containing spermidine immediately after they were weaned, and found that spermidine increased the levels of MAP1S protein in wild-type mice (Fig. 1A,B). As previously reported, similar increases in both levels of oxidative stress (Fig. 1C,D) and intensities of sinusoidal dilatation (Fig.   1E,F) and a similar reduction in lifespans (from 26.9 to 21.9 months) were observed when MAP1S was deleted in mice ( Fig. 1G,H). Spermidine-treated wild-type mice exhibited significantly lower levels of oxidative stress (Fig. 1C,D) and lower intensities of sinusoidal dilatation in liver tissues (Fig. 1E,F), and a 25% increase of median survival times from the normal 26.9 months of untreated wild-type mice to 33.3 months (Fig.   1G,H). Comparing to the reported 10% increase of median survivals in mice treated with spermidine at late stage of life cycle, a significant enhancement of mouse longevity was achieved when mice were exposed to spermidine immediately after weaning. The on April 28, 2021. © 2017 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Spermidine alleviates liver fibrosis through MAP1S.
We reported that MAP1S-activated autophagy promotes the degradation of fibronectin in lysosomes and alleviates stress-induced liver fibrosis (16). We were triggered to examine the impact of spermidine on liver fibrosis in a mouse model. We treated both wild-type and MAP1S -/mice with CCl 4 and fed them with either normal or spermidine-containing water. We found that CCl 4 did not change the body weights but increased the liver weights and liver/body weight ratios of both wild-type and MAP1S -/mice; and spermidine exposure helped the wild-type but not the MAP1S -/mice to counteract the CCl 4 -induced increases of liver weights and liver/body weight ratios ( Fig.   2A-D). Spermidine exposure led to reductions in both liver lesions caused by CCl 4 in wild-type but not in MAP1S -/mice (Fig. 2E,F) and intensities of CCl4-induced liver fibrosis indicated by the conventional Sirius Red staining in wild-type but not in MAP1S -/mice (Fig. 2G,H). The same trends were confirmed by other assays of liver fibrosis such as the contents of hydroxyl-proline (Fig. 2I) and levels of a-SMA as revealed by both immunoblotting and immunostaining (Fig. 2J-M). Therefore, spermidine exposure leads to an alleviation of liver fibrosis through MAP1S. We previously reported that MAP1S-deficient mice exposed to DEN developed more and larger foci of HCC than the wild-type (14). We predicted that spermidine promotes MAP1S-activated autophagy to suppress HCC. We injected mice with DEN and examined the tumor development in 7-month-old mice. It was confirmed that MAP1S suppressed the development of HCC (Fig. 3A). Wild-type mice drinking water containing spermidine after DEN injection exhibited reduced liver weights, liver/body weight ratios and less surface tumors although the body weights were not altered ( Fig.   3A-E). The typical trabecular structure of HCC as displayed by H&E staining were dramatically reduced (Fig. 3F). The levels of MAP1S were significantly elevated to activate autophagy flux so that the levels of total poly-ubiquitinated proteins were significantly reduced ( Fig. 3G-I), and number of cells with DNA double strand breaks represented by positive γ-H2AX signals were decreased (Fig. 3J,K). MAP1S depletion promoted the development of HCC in MAP1S -/mice while further spermidine exposure exhibited no significant reduction in protein aggregates, DNA damages and HCC (Fig.   3). Therefore, suppressive role of spermidine on the development of HCC also acts through MAP1S.

Spermidine-induced activation of autophagy flux depends on MAP1S.
To understand how spermidine impacts on lifespans, liver fibrosis and HCC through MAP1S, we treated HeLa or MEF cells with spermidine and observed increased levels of MAP1S and autophagy marker LC3-II in a time and dosage-dependent way ( Fig. 4A-D). The levels of LC3-II in the presence of lysosomal inhibitor bafilomycin A1 were elevated upon exposure to spermidine (Fig. 4E,F) accelerated autophagy flux. The spermidine-induced activation of autophagy flux was further confirmed by the accumulation of RFP-LC3 punctate foci in spermidine-treated HeLa cells stably expressing RFP-LC3 in the absence or presence of lysosomal inhibitor (Fig. 4G,H). Therefore, spermidine increases the levels of MAP1S and autophagy flux.
Intraperitoneal injection of a single dose of spermidine to wild-type mice led to a similar increase in levels of MAP1S and LC3-II in livers and other organs of mice including brain and heart (Fig. 4I,J). The correlation between levels of MAP1S and autophagy flux upon spermidine treatment triggered our interest to investigate their cause-effect relation. The spermidine-induced increases in LC3-II levels were not observed in different organs or MEFs from MAP1S knockout mice (Fig. 4I-J). The spermidine-induced increase in autophagy flux as indicated by the levels of LC3-II in the presence of bafilomycin A1 were observed in wild-type but not in MAP1S -/hepatocytes or MEFs (Fig. 4M-P). Therefore, the activation of autophagy flux by spermidine requires the presence of MAP1S.

Spermidine-enhanced acetylation and stability of MAP1S depends on HDAC4.
We reported that HDAC4 directly interacts with MAP1S through a specific domain and acts as a deacetylase of MAP1S. Using purified active, heat-inactive HDAC4, inactive HDAC4 mutant and MAP1S, we confirmed that MAP1S is the substrate of HDAC4 deacetylase in vitro (19). We reasoned that spermidine may increase the levels of MAP1S-mediated autophagy by regulating HDAC4. We found that spermidine treatment resulted in increases in both levels of acetylated MAP1S ( 5A-C) and stability of MAP1S protein (Fig. 5D,E). We re-confirmed the interaction between HDAC4 and MAP1S in HepG2 cells (Fig. 5F). Reducing the levels of HDAC4 with a previously certified HDAC4-specific siRNA (19) led to increases of MAP1S levels, and further treatment of HDAC4-suppressed cells with spermidine did not induce the activation of autophagy flux (Fig. 5G-I). Therefore, spermidine enhances acetylation and stability of MAP1S protein to activate autophagy flux through HDAC4.

Spermidine enhances nuclear translocation and cytosolic depletion of HDAC4.
It would be logically plausible if spermidine treatment acted similarly as the HDAC4-speacific siRNA and led to a reduction in HDAC4 levels. However, spermidine treatment led to increases in HDAC4 levels in a dosage and time-dependent way in different cell lines and different organs of mice (Fig. 5J-N). Although HDAC4 was overexpressed, spermidine exposure enhanced the nuclear translocation of HDAC4 ( Fig. 5O-R) and reduced the actual levels of cytosolic HDAC4 in different types of cells ( Fig. 5S,T). Therefore, spermidine causes a depletion of cytosolic HDAC4.

Spermidine reduces the interaction of HDAC4 with MAP1S.
Binding of phosphorylated HDAC4 with 14-3-3 keeps both of them stayed in cytosol (31).
The spermidine-specific lysine residue 520 of MAP1S is important for MAP1S to promote autophagosomal degradation.
Lysine residue 520 (K520) of human MAP1S was identified in a screen for acetylated sites arising in the presence of spermidine (22). We mutated the lysine residue 520 to arginine (K520R). The K520R mutant of MAP1S protein exhibited higher levels of expression but lower degrees of interaction with HDAC4 than the wild-type (Fig.   7A) although it distributed in cytosol similarly to the wild-type (Fig. 7B). However, overexpression of the MAP1S mutant resulted in a blockade of autophagosomal degradation and an accumulation of autophagosomes (Fig. 7C-F). The forced expression of the K520R mutant at levels higher than the wild-type led to the sustaining levels of autophagosomal biogenesis (Fig. 7C-F), suggesting the role of MAP1S in autophagy initiation remained intact despite the mutation. Notably, the amount of punctate foci of RFP-LC3 not associated with LAMP2-labeled lysosomes in cells expressing the MAP1S mutant was higher than in cells expressing wild-type MAP1S (Fig. 7G,H). This confirmed that mutation of K520 in MAP1S reduced the efficiency of 18 K520R mutant (Fig. 7I,J). Thus, spermidine-specific lysine residue 520 of MAP1S is important for MAP1S to promote autophagosomal degradation.

Discussion
Polyamines were found in both prokaryotic and eukaryotic systems after initially being identified in human semen in 1678 (32). Difluoromethylornithine (DFMO) is a specific inhibitor of ornithine decarboxylase (ODC), one of the key enzymes involved in biosynthesis of polyamines including spermidine (33). Since spermidine was found to be dramatically elevated in the urine of many cancer patients in 1971 (34), there have been attempts to use the biosynthesis pathway of polyamines as therapeutic target or use levels of polyamines as prognostic markers for cancer patients. DFMO plus sulindac has been shown to effectively prevent the recurrence of colon polyp in a phase III clinical trial (33). Although polyamine depletion has been suggested for cancer prevention, neither alone nor in combination with other agents was DFMO clinically effective (35). Recently, spermidine was shown to activate autophagy flux to prolong lifespans in model systems through promoting convergent deacetylation and acetylation reactions in the cytosol and in the nucleus, respectively (20-22). Although it was claimed that spermidine induces autophagy by inhibiting the acetyltransferase EP300, knockdown of EP300 or inhibition of EP300 by a specific antagonist at concentrations significantly inducing autophagy activation failed to cause a significant decrease of lysine deacetylation of cytosolic proteins (36). The precise biochemical function of spermidine was considered as one of the remaining mysteries of molecular cell biology (35).
In contrast, here we show that spermidine promotes the nuclear translocation of  (37). Immunoblot analysis here shows that the levels of phosphorylated S246 or S632 HDAC4 are significantly reduced upon exposure to spermidine. Therefore, spermidine may promote the de-phosphorylation and nuclear translocation of HDAC4. Further understanding the mechanism by which spermidine promotes de-phosphorylation of HDAC4 is currently undergoing. It was reported that suppression of LKB1-AMPK dramatically reduces levels of phosphorylated HDAC4 (38). Spermidine treatment exhibited no impact on the phosphorylation of mTOR substrate RPS6KB1, or AMPK substrate, acetyl-CoA carboxylase, so that autophagy initiation through the mTOR pathway was not affected (22). It is possible that spermidine only alters the HDAC4-specific activity of LKB1.
Aging is the process that an individual organism becomes older, and characterized by an increasing morbidity and functional decline that eventually result in the death of an organism (39). Caloric restriction, methionine restriction and rapamycin addition were the few interventions that truly prolong the lifespans of vertebrates (40)(41)(42).
Other interventions such as resveratrol, metformin and sirtuin activators were also reported to extend mouse lifespans (39). However, resveratrol only exhibited effect in mice on high-fat diets (43); metformin slightly increased lifespans when mice were fed with low doses but dramatically reduced the lifespans when mice were fed with high doses (44); and a sirt1 activator had no effect on the maximal lifespans at all (45).
Spermidine-related knowledge for cancer patients, spermidine has been reported to reactivate autophagy to ameliorate myopathic defects (46)  Our results show that mice drinking spermidine-containing water immediately after weaning achieve much more significant improvement of lifespans than those start to be treated with spermidine at the age of 4 or 18 months (23). Although they are successful in mouse models, both voluntary caloric restriction and methioninedeficient vegan diet are hard to gain much popularity for general human population, and rapamycin has additional immunosuppressant functions in humans (47). Since spermidine is a natural component originally isolated from semen and plentiful in a variety of food and agricultural products such as wheat germ, grapefruit and Natto (a Japanese product of fermented soy), and can be chemically synthesized and developed as a drink or a dietary supplement to serve with normal diets, it will be a promising safe and cost-effective intervention that will be readily adopted for long-term intervention by the general human population to extend lifespans. Besides, spermidine suppresses liver fibrosis or HCC in mice exposed to a strong chemical inducer of liver fibrosis or a potent    quantification of (N) of LC3-II levels in hepatocytes isolated from wild-type or MAP1S -/mice that were untreated or treated with spermidine in the absence or presence of BAF.
(O,P) Immunoblot analyses (O) and quantification (P) of LC3-II levels in MEFs developed from wild-type or MAP1S -/mice that were untreated (Ctrl) or treated with spermidine (SPD) in the absence or presence of BAF.