HMBA ameliorates obesity by MYH9‐ and ACTG1‐dependent regulation of hypothalamic neuropeptides

Abstract The global epidemic of obesity remains a daunting problem. Here, we report hexamethylene bisacetamide (HMBA) as a potent anti‐obesity compound. Peripheral and central administration of HMBA to diet‐induced obese mice regulated the expression of hypothalamic neuropeptides critical for energy balance, leading to beneficial metabolic effects such as anorexia and weight loss. We found that HMBA bound to MYH9 and ACTG1, which were required for the anti‐obesity effects of HMBA in both NPY‐expressing and POMC‐expressing neurons. The binding of HMBA to MYH9 and ACTG1 elevated the expression of HEXIM1 and enhanced its interaction with MDM2, resulting in the dissociation of the HEXIM1–p53 complex in hypothalamic cells. Subsequently, the free HEXIM1 and p53 translocated to the nucleus, where they downregulated the transcription of orexigenic NPY, but p53 and acetylated histone 3 upregulated that of anorexigenic POMC. Our study points to a previously unappreciated efficacy of HMBA and reveals its mechanism of action in metabolic regulation, which may propose HMBA as a potential therapeutic strategy for obesity.


Introduction
Obesity is considered a pandemic worldwide (Swinburn et al, 2011;Bluher, 2019).The main etiological factor of obesity is an energy imbalance between excessive calorie intake and reduced energy expenditure, which is often associated with impairments in the hypothalamic integration of peripheral signals and/or extrahypothalamic structures (Schwartz et al, 2000;Myers et al, 2021).The hypothalamus plays a key role in maintaining whole-body energy homeostasis through multiple neuronal networks and circuits (Myers & Olson, 2012;Myers et al, 2021).In particular, the arcuate nucleus (ARC) is a crucial hypothalamic nucleus that regulates feeding, energy expenditure, and overall energy homeostasis by sensing and responding to metabolic signals such as nutrients, hormones, and metabolites from the peripheral circulation (Myers et al, 2021).ARC contains neural circuits composed of two functionally distinct neuronal populations.One population expresses orexigenic neuropeptide Y (NPY) and agouti-related peptide (AgRP) that promote appetite and decrease energy expenditure, and the other population expresses anorexigenic proopiomelanocortin (POMC) and cocaineand amphetamine-regulated transcript (CART) that promote satiety and increase energy expenditure.These neuropeptides play critical roles in energy balance by regulating appetite, thermogenesis in brown adipose tissue, and glucose homeostasis in the liver and muscle directly or indirectly by controlling the neuroendocrine and autonomic nervous system (Stanley & Leibowitz, 1985;Poggioli et al, 1986;Brady et al, 1990;Garcia de Yebenes et al, 1995;Huszar et al, 1997;Myers et al, 2021;Vohra et al, 2022).Since the ARC neural system has been suggested to be impaired in the obese state, restoring or regulating the function of the ARC is important for the improvement of obesity and its associated pathologies.Pharmacological interventions including agonists, antagonists (Roseberry et al, 2004;Brothers & Wahlestedt, 2010), hormone analogs (Huang et al, 2021), and natural (Matias et al, 2008;Li et al, 2018) or synthetic compounds (Loftus et al, 2000;Shimokawa et al, 2002;Hu et al, 2003;Tavares et al, 2007Tavares et al, , 2013;;Anderson et al, 2008) that enable modulation of neuropeptide expression in the ARC are considered promising strategies, and understanding the mechanisms of action of these compounds will provide therapeutic directions for the development of obesity treatment.

of 23
The gene of HEXIM1 was initially cloned from human vascular smooth muscle cells and was predicted to encode a nuclear protein, which was upregulated after HMBA treatment (Kusuhara et al, 1999).According to the Human Protein Atlas (Uhlen et al, 2015), HEXIM1 is expressed at the highest level in the heart, kidney, and brain, whereas HEXIM2, a paralog of HEXIM1, is expressed mainly in the skeletal muscle and testis.HEXIM1 inhibits the activity of positive transcription elongation factor b (P-TEFb), preventing transcription elongation by RNA polymerase II (Pol II) (Yik et al, 2003;Dey et al, 2007).HEXIM1 also interacts with p53 in MCF7 breast cancer cells, and overexpression of HEXIM1 enhances the interaction and inhibits p53 ubiquitination, resulting in enhanced p53 stability and activation of p53 target gene expression required for DNA repair, apoptosis, and differentiation (Lew et al, 2012(Lew et al, , 2013)).Although p53 is a well-known potent tumor suppressor, its role in maintaining energy balance is also emerging (Yahagi et al, 2003;Quinones et al, 2018).Selective ablation of p53 in AgRP neurons (AgRPp53KO) of diet-induced obese (DIO) mice fed a high-fat diet (HFD) for a long period increased food intake, promoted weight gain, and reduced energy expenditure and thermogenic activity of brown adipose tissue in comparison with wild-type DIO mice (Quinones et al, 2018).In contrast, POMCp53KO DIO mice showed no changes in the overall metabolic phenotypes including feeding behavior, energy expenditure, and body weight (Quinones et al, 2018).Furthermore, obese Hexim1 heterozygous mice fed an HFD showed increased adipogenesis in adipose tissue and enhanced glucose transporter 4 expression for glucose disposal in muscle as well as enhanced leptin signaling in the hypothalamus (Dhar-Mascareno et al, 2016).However, it has not been investigated yet whether HMBA has any metabolic effects or how HEXIM1 is involved in energy homeostasis.
Herein, we report that HMBA mitigates obesity by regulating the expression of the Npy and Pomc genes.HMBA-induced HEXIM1 downregulated Npy expression and upregulated Pomc expression in the hypothalamus of DIO mice, leading to reduced appetite, elevated energy expenditure, and weight loss.We propose that HMBA, which has long been investigated in oncology, can be repurposed for the control of appetite and body weight with the aim of extending its use to metabolic disorders.Understanding the action of HMBA and finding its biological targets in the regulation of energy homeostasis will provide not only mechanistic information but also therapeutic strategies relevant to metabolic diseases such as obesity.

Peripheral administration of HMBA ameliorates obese phenotype in DIO mice
We searched the Broad Institute Connectivity Map database (Lamb et al, 2006) to select small-molecule candidates with potential efficacy in the regulation of metabolism.A comparison of gene expression signatures between oleoylethanolamide, a known anorexic endocannabinoid, that reduces food intake and body weight by regulating lipid and glucose metabolism (Fu et al, 2003;Guzman et al, 2004;Gonzalez-Yanes et al, 2005), and 2,429 compound profiles registered in the TouchStone v1.5 of dataset (Lamb et al, 2006) identified HMBA as one of the top 10 compounds with a high connectivity score (Appendix Table S1).We chose HMBA as a suitable candidate for this study since its effects on metabolic diseases such as obesity are not known yet at all among the listed compounds.Although the roles of otenzepad and valacyclovir on metabolism are unknown, these two drugs were excluded from the candidates due to their adverse effects on the central nervous system, nephrotoxicity, or tissue-specific response in previous studies (Mickala et al, 1996;Broadley & Kelly, 2001;Zhang et al, 2016;Brandariz-Nunez et al, 2021).
◀ Figure 1.Peripheral administration of HMBA ameliorates obesity in DIO mice.

A
Chemical structure of HMBA.HMBA has a symmetric structure with amides at both ends (CC(=O)NCCCCCCNC(=O)C).B Design of intravenous and intraperitoneal injections in male DIO mice.Triangles represent the days of each injection.C, D Changes in (C) daily body weight (grams) (left panel) and body weight (percentages) on day 10 relative to day 0 (right panel), and (D) food intake in male DIO mice injected intravenously (i.v.) with HMBA.n = 10 per group.Arrows represent the days of each injection.E, F Changes in (E) daily body weight (grams) (left panel) and body weight (percentages) on day 10 relative to day 0 (right panel), and (F) food intake in male DIO mice injected intraperitoneally (i.p.) with HMBA.n = 5 per group.Arrows represent the days of each injection.

G
The amounts of HMBA in plasma and hypothalamus.n = 5 per group.H, I Relative mRNA levels of hypothalamic neuropeptides and Hexim1 in (H) i.v. group and (I) i.p. group.RNA was isolated 4 h after injection.n = 4, saline (i.v.); n = 6, HMBA (i.v.); n = 3, saline (i.p.) or HMBA (i.p.).J Fat and lean mass 2 days after the last dose of HMBA.n = 8 per group.K Representative infrared thermography images and interscapular brown adipose tissue temperature.n = 10, Saline (i.v.); n = 11, HMBA (i.v.).The temperature of brown adipose tissue was determined as the average skin temperature of the same area in each thermography image (one image per mouse).In the color key, the temperature decreases from top to bottom.L Relative mRNA levels of Ucp1 of interscapular brown adipose tissue.n = 10, Saline (i.v.); n = 11, HMBA (i.v.To investigate the effects of HMBA on obesity, male DIO mice fed an HFD for 4 weeks were injected with HMBA either intravenously (i.v.) or intraperitoneally (i.p.) as indicated in Fig 1B .Intravenous injections every other day (Fig 1C and D) or daily i.p. injections (Fig 1E and F) significantly reduced weight gain and food intake in comparison with those in saline-injected control groups.
To determine the effect of HMBA on food intake and body weight in females, we injected HMBA i.v. and i.p. to female DIO mice as the same experimental design in Fig 1B .Our data showed that each injection of HMBA had very similar effects on the reduction in body weight and food intake in females as seen in HMBA-injected DIO males.It is likely that HMBA has the same effects on food intake and body weight regardless of gender.Hence, all subsequent experiments were consistently performed in male DIO mice.
To explore whether the peripherally injected HMBA is delivered to the hypothalamus and exerts a central effect, we analyzed and quantified HMBA in plasma and hypothalamus 4 h after the last injection using liquid chromatography with tandem mass spectrometry (LC-MS/MS).We detected 6.59 AE 0.71 lmoles of HMBA in plasma and 0.79 AE 0.09 lmoles in the hypothalamus per mouse.The concentrations of HMBA in plasma and hypothalamus were higher after i.v.injection than after i.p. injection (Fig 1G).Given these results, we tested whether peripherally delivered HMBA regulated whole-body metabolism by modulating orexigenic or anorexigenic neuropeptides.Both i.v. and i.p. administration of HMBA decreased the mRNA levels of Npy and Agrp and increased those of Pomc and Cart in the hypothalamus of DIO mice, whereas that of Hexim1 was elevated in the hypothalamus of both HMBA-injected groups (Fig 1H and I).
To assess the effect of HMBA on body composition, we analyzed fat and lean masses using in vivo nuclear magnetic resonance 2 days after the last dose.Fat mass was significantly reduced, while lean mass was not changed (Fig 1J).We measured thermogenesis in brown adipose tissue.HMBA increased interscapular surface temperature adjacent to brown adipose tissues and the mRNA level of Ucp1, indicating greater thermogenesis (Fig 1K and L).HMBA led to marked induction in energy expenditure compared to saline-injected group, indicating that HMBA increased metabolic rates independently of the body mass (Fig 1M).Calorimetry analysis showed that HMBA increased metabolic consumption of oxygen and generation of carbon dioxide (Fig 1N -P) and increased heat production (Fig 1Q) without changes in locomotor activity (Fig 1R and S).Taken together, the data indicated that peripheral administration of HMBA ameliorated the obese phenotype in DIO mice.

Central administration of HMBA prevents obesity in DIO mice
To confirm the central effect of HMBA, we daily injected 400 nmoles of HMBA into the third ventricle (3 V) of male and female DIO mice, which resides near the hypothalamus (Fig 2A).This intracerebroventricular (i.c.v.) injection of HMBA significantly decreased body weight and food intake.After dosing was stopped, body weight remained low despite the recovery of food consumption in both sexes (Fig 2B and C,and Appendix Fig S1E and F).All subsequent experiments were conducted in male DIO mice.
Central administration of HMBA to mice fed a normal chow diet (NCD) also led to weight loss and reduced appetite (Appendix Fig S2).
Consistent with the effects of peripheral injection of HMBA, its central injection reduced the mRNA levels of Npy and Agrp and increased those of Pomc, Cart, and Hexim1 in the hypothalamus (Fig 2D).

HMBA did not cause sickness behaviors in DIO mice
To investigate whether HMBA exhibited sickness behaviors, we monitored various behavioral tests in DIO mice.First, we observed that none of the mice displayed any noticeable signs of sickness such as vomiting, diarrhea, inactivity, and hair loss during HMBA treatment.Our metabolic cage data also showed no negative effect of HMBA on locomotor activity compared with saline-injected mice.Second, in order to demonstrate the health status of the mice more accurately, we employed the laboratory animal behavior observation registration and analysis system (LABORAS) (Fig EV1 ) after the last dosing of HMBA (i.v., i.p., or i.c.v.).The LABORAS data demonstrated that eating behavior was decreased by HMBA treatment, however, no significant differences in the other behaviors were observed compared to saline-injected groups (Fig EV1A , B, D, E, G,  and H).Lastly, we performed a conditioned taste aversion (CTA) test to assess the effect of HMBA on a taste aversion associated with an illness over sucrose preference (Fig EV1).The CTA test showed that HMBA did not induce a taste aversion (Fig EV1C,F and I).Collectively, our data demonstrated that HMBA did not exhibit sickness-associated or aversive appetite suppression and weight loss under our experimental conditions.

HMBA is metabolized in hypothalamic tissue and cells
HMBA in the body is sequentially metabolized into N-acetyl-1,6diaminohexane (NADAH), 1,6-diaminohexane (DAH), and 6aminohexanoic acid (AmHA) (Fig 3A), and the metabolic rate varies depending on cell type or dosage (Callery et al, 1986;Egorin et al, 1988).We simultaneously analyzed these metabolites with LC-MS/MS by obtaining high-resolution peaks of each metabolite through multiple reaction monitoring (Fig 3B).Using this approach, we determined the metabolic flow of HMBA in the hypothalamus and plasma of DIO mice harvested 4 h after i.c.v.injection of HMBA.High levels of HMBA and NADAH were detected in the hypothalamus, and DAH and AmHA were also readily detectable.The molar concentrations of HMBA (47%) and NADAH (47.6%) were much higher than those of DAH and AmHA (Fig 3C).However, the i.c.v.injection of HMBA did not affect plasma HMBA concentration (Appendix Fig S3).
We also screened HMBA and its metabolites in hypothalamic cell lines expressing Npy (mHypoE-N41) or Pomc (mHypoE-N43/5) treated with 0.1 mM HMBA for 1-8 h (Fig 3D and E).The concentration of 0.1 mM HMBA did not reduce cell survival and was nontoxic (Appendix Fig S4).HMBA gradually decreased from 1 to 8 h in both cell lines.NADAH decreased by 4 h, probably due to its breakdown to DAH or AmHA.DAH and AmHA continuously increased from 1 to 8 h.The molar concentration of HMBA was much higher (80.7% in mHypoE-N41 and 82.2% in mHypoE-N43/5) than those of its metabolites.These data suggest that HMBA remains mainly intact, although it is metabolized to some extent in the hypothalamic cell lines tested.

HMBA regulates neuropeptide expression in a HEXIM1dependent manner
To elucidate the mechanism by which HMBA regulates the neuropeptide expression in vitro, we treated hypothalamic cells with HMBA for up to 24 h and examined the changes in neuropeptide expression.In HMBA-treated cells, the mRNA level of Npy declined, with the lowest value at 2 h, and remained significantly lower than the control until 24 h (Fig 3F), whereas that of Pomc was significantly higher than in the control at 0.5 h and peaked at 2 h (Fig 3G).The time point (2 h) and the concentration of HMBA (0.1 mM) that showed the strongest anorexigenic effect (Fig 3H and I) were selected for the following experiments.
We hypothesized that HMBA-induced HEXIM1 regulates Npy and Pomc expression since Hexim1 was the dominant isoform in

HMBA binds to MYH9 and ACTG1 in hypothalamic cells
To identify the proteins binding HMBA in hypothalamic cells, we applied a chemical proteomics approach wherein biotin-conjugated HMBA (hereafter referred to as HMBA-biotin) was used as a bait for HMBA targets.HMBA-biotin was synthesized by reacting the primary amine group of NADAH with the N-hydroxysulfosuccinimide group of biotins (Fig 4A  S2).Treatment with HMBA-biotin for 2 h was sufficient to induce HEXIM1, confirming that HMBA-biotin retained the biological activity of HMBA (Fig 4D).HMBA-binding proteins were identified by biotin-neutravidin pull-down assay and peptide analysis using LC-MS/MS (Fig 4E).Two protein bands from Coomassie blue gels were identified as MYH9 (non-muscle myosin heavy-chain IIA, 230 kDa) and ACTG1 (c-actin, 42 kDa) based on the highest mascot scores (Fig 4F).We predicted the structures of the protein-HMBA complexes, including the HMBA-binding pockets using an autodocking approach (Forli et al, 2016).HMBA was successfully docked at the best poses of MYH9 (Fig EV2A ) and ACTG1 (Fig EV2B).In the docking model of HMBA-MYH9, a hydrogen bond was formed between the hydroxyl group (acceptor O atom) at the amide of HMBA and the guanidinium side chain of Arg424 (donor H atom), and another one was formed between the amine group at the symmetrical amide of HMBA (donor H atom) and the carboxyl group of Leu611 (acceptor O atom).Trp428 was revealed as a residue in the hydrophobic pocket potentially capable of hydrogen bonding with HMBA (Figs 4G and EV2C).In the docking model of HMBA-ACTG1, a hydrogen bond was formed between the hydroxyl group (acceptor O atom) of HMBA and the amine group of the side chain of Lys213 (donor H atom), and another one between the symmetrical hydroxyl group (acceptor O atom) of HMBA and the amine group of Gly15 (donor H atom). HMBA potentially had hydrogen bonds and hydrophobic interactions with Met16, Asp157, Glu214, and Tyr306 of ACTG1 (Figs 4H and EV2D).These data suggest that MYH9 and ACTG1 directly bind HMBA.

HEXIM1 and p53 differentially regulate neuropeptide expression
Recent studies have shown that HEXIM1 interacts with p53 (Lew et al, 2012(Lew et al, , 2013) ) S6. Source data are available online for this figure.
Npy and Pomc but was required for HMBA-induced changes in their expression.To further delineate the cooperative action of HEXIM1 and p53 in neuropeptide gene regulation in the presence of HMBA, we first assessed the effect of HMBA on HEXIM1 and p53 protein levels.In both cell lines, HMBA increased the levels of both proteins at the 0.5, 1, and 2 h time points (Fig 6E and H).HMBA also increased the mRNA levels of Hexim1 but not those of p53 (Trp53), suggesting that HMBA regulates p53 at the protein level only (Fig 6F and I).Notably, the basal protein level of p53 was significantly increased by Hexim1 deletion (Fig 6G and J).Therefore, p53 is under the control of HEXIM1 at basal state.The HMBA-induced increase in p53 protein level was abolished by Hexim1 deletion, indicating that HMBA regulates p53 in a HEXIM1-dependent manner (Fig 6G and J).Despite an increase in basal p53 protein level, our cell lines did not undergo apoptosis, as indicated by the absence of increases in cleaved forms of caspases 3 and 9 (apoptosis markers) upon HMBA treatment or Hexim1 KO (Appendix Fig S8).Taken together, these results suggest differential roles of HEXIM1 and p53 in mediation of the anorexigenic effects of HMBA.HEXIM1 regulates the expression of p53, Npy, and Pomc at basal state as well as upon HMBA treatment, whereas p53 participates in the gene regulation of Npy and Pomc when HMBA is administered.To delineate the role of HEXIM1 and p53 in the nucleus, we measured the levels of neuropeptide expression after treatment with 50 ng/ml leptomycin B (LMB), an inhibitor of nuclear export, to ensure protein retention in the nucleus.LMB alone did not affect the basal mRNA levels of Npy or Pomc.However, the HMBA-induced decrease in the expression of Npy and increase in that of Pomc were further exacerbated by LMB (Fig EV5).It was unclear whether there was any further increase in nuclear localization of HEXIM1 and p53 by LMB since they were located mainly in the nucleus in both cell lines (Fig EV5).These data suggest that nuclear HEXIM1 and p53 are important for HMBA to regulate the gene expression of Npy and Pomc.

HMBA induces translocation of HEXIM1 and p53 to the nucleus
The amounts of proteins bound to HEXIM1 were analyzed by immunoprecipitating HEXIM1 from cytoplasmic and nuclear fractions (Fig 7D).Although majority of HEXIM1 and p53 proteins were present in the nucleus, their interaction occurred mostly in the cytoplasm; this complex, however, dissociated upon HMBA treatment.In contrast, the abundance of the HEXIM1-p53 complex in the nucleus was low regardless of HMBA treatment (Fig 7D).Overexpressed HEXIM1 was reported to inhibit p53 degradation by binding to p53, which disturbs the interaction between p53 and E3 ubiquitin-protein ligase mouse double minute-2 (MDM2, called HDM2 in humans), which binds to p53 along with ubiquitin (Lew et al, 2012).HEXIM1 competes with MDM2 for binding to the negative regulatory domain of p53 (Lew et al, 2013).In the cytoplasm of mHypoE-N41 and mHypoE-N43/5 cells, HMBA increased HEXIM1 binding to MDM2 but decreased HEXIM1 binding to p53 (Fig 7D).Although cytoplasmic MDM2 was markedly induced by HMBA, MDM2 and HEXIM1 did not compete for binding to p53, but rather bound to each other.Interestingly, the p53-MDM2 complex was not formed either at basal state or upon HMBA treatment (Fig 7D).MDM2 expression was also governed by MYH9 and ACTG1, as silencing of Myh9, Actg1, or both prevented MDM2 induction upon HMBA treatment (Appendix Fig S10).Immunoprecipitation results suggest that HMBA boosts the competitive binding of HEXIM1 with MDM2 rather than with p53 in the cytoplasm.Overall, the above results indicate that HMBA treatment results in the disassembly of the HEXIM1-p53 complex and translocation of free HEXIM1 and p53 to the nucleus.

HMBA regulates neuropeptide transcription
To ascertain whether nuclear HEXIM1 and p53 directly regulate transcription of neuropeptide genes, we performed chromatin immunoprecipitation assay and analyzed the amounts of HEXIM1, p53, and histone modification markers bound to the promoter regions of Npy (Titolo et al, 2008;Melas et al, 2013) and Pomc (Wu et al, 2014).We observed binding of HEXIM1 and p53 to promoter Epigenetic modifications such as histone acetylation regulate Npy and Pomc (Benite-Ribeiro et al, 2016).In obese rats displaying increased Npy and decreased Pomc expression, the binding of histone 3 acetylated at lysine 9 (H3K9Ac) to the Npy promoter is increased but that to Pomc promoter is decreased in comparison with non-obese rats, whereas there is no change in H3K9 methylation in either gene (Mahmood et al, 2013).We assessed H3K9Ac in the Npy and Pomc promoters.We also evaluated histone 3 trimethylated at lysine 4 (H3K4me3) and at lysine 27 (H3K27me3), which indicate transcriptional activation and repression, respectively (Voigt et al, 2013).HMBA reduced the binding levels of H3K9Ac and H3K4me3 to the Npy promoter and elevated that of H3K9Ac to the Pomc promoter.The binding levels of H3K27me3 were marginal and were not changed by HMBA in either (Fig 7E and F).Our data suggest that HMBA induced differential binding of HEXIM1 and H3K9Ac to the Npy and Pomc promoters, leading to opposite outcomes in gene expression.
Overall, our data suggest the following sequential events: (i) HMBA binds to MYH9 and ACTG1, inducing HEXIM1 and MDM2 expression; (ii) in the cytoplasm, HEXIM1 competitively binds to MDM2 rather than to p53, resulting in a release of HEXIM1 and p53 from the HEXIM1-p53 complex, and the free cytoplasmic HEXIM1 and p53 translocate to the nucleus; (iii) in the nucleus, p53 and HEXIM1 inhibit the transcription of Npy, whereas p53 and H3K9Ac activate that of Pomc, which results in the anorexigenic and antiobese effects of HMBA (Fig 7G).

Discussion
Since the first synthesis of HMBA more than 4 decades ago (Reuben et al, 1976), research on this compound has focused on its applications in oncology, but it failed in a phase II clinical trial as an anticancer drug due to toxicity at high dosages (Andreeff et al, 1992;Conley et al, 1992).HMBA has a short biological half-life, as it is rapidly excreted into the urine and is quickly eliminated from the body, which limits the duration of its anti-cancer efficacy (Egorin et al, 1987a,b;Litterst et al, 1985).Treatment of tumors required a high concentration of HMBA to induce apoptosis or cell differentiation, for example, patients were recommended an intravenous infusion of 4,000 mg/kg for 10 consecutive days, repeated for at least 7 cycles with 4-week intervals (Young et al, 1988;Andreeff et al, 1992).Our study investigated the anti-obesity efficacy of peripheral and central injection of HMBA in DIO mice under milder conditions, that is, lower dosage and a shorter period of injection than in cancer studies.
Our results showed that HMBA-injected mice were resistant to the obesogenic effects of an HFD.Every other day i.v. or daily i.p. injections of HMBA (≤1,000 mg/kg; 4 or 7 injections, respectively) were effective in suppressing appetite and weight gain, while concentrations below 500 mg/kg were not (Appendix Fig S12).Daily i.c.v.injections of HMBA (400 nmoles; 7 injections) improved the obese phenotype and abnormal metabolic parameters, and even at a lower dose of 80 nmoles, food intake and body weight were reduced (Appendix Fig S12).Consistent with the in vivo data and unlike the high concentrations of HMBA (5 mM) required for cell differentiating effects in vitro (Reuben et al, 1976;Guilbaud et al, 1990;Kondo et al, 1990), relatively low concentrations of HMBA (0.01-0.1 mM) were sufficient to change the mRNA levels of Npy and Pomc in hypothalamic cells.The amounts of HMBA accumulated in the hypothalamus after i.v., i.p., or i.c.v.injections indicate that not only centrally but also peripherally delivered HMBA reached the hypothalamus and served as an anti-obesity compound, highlighting the importance of HMBA as a potential small-molecule targeting the hypothalamus for the control of metabolic homeostasis.
In breast cancer cells, 70-kDa heat-shock protein (Lama et al, 2017) and lysine demethylase 5B (Montano et al, 2019) have been identified as proteins binding HMBA.In hypothalamic cells, we found robust binding of HMBA to MYH9 and ACTG1.MYH9 and ACTG1 contribute to the regulation of tumorigenesis-associated gene expression via multiple signaling cascades (Dong et al, 2018;Wang et al, 2019;Ye et al, 2020;Xiao et al, 2021).However, their roles in the hypothalamus or functions in the metabolism have not been studied.Our findings provide new insights into the function of MYH9 and ACTG important for mediating the effects of HMBA on metabolic regulation.Notably, we found that HMBA-induced HEXIM1, which is crucial for regulating energy balance, requires MYH9 and ACTG1 in the NPY-expressing and POMC-expressing neurons.Although the precise mechanism(s) by which MYH9 and ACTG1 regulate HEXIM1 and MDM2 expression is not discovered yet, it might rely, at least in part, on p53 stabilization through an increase in the level of the HEXIM1-MDM2 complex, which regulates Npy and Pomc expression.
HEXIM1 was proposed as a mediator of HMBA actions in cancer cells (He et al, 2006;Turano et al, 2006;Yoshikawa et al, 2012;Ketchart et al, 2013Ketchart et al, , 2016;;Lama et al, 2017).Indeed, we observed that HMBA induced HEXIM1 in the hypothalamus, and genetic ablation of Hexim1 abolished the effects of HMBA on neuropeptide expression.At the molecular level, HMBA enhanced the formation of the HEXIM1-MDM2 complex rather than the HEXIM1-p53 complex in the cytosol, released p53 from HEXIM1, and prevented MDM2-mediated degradation of p53.Although the knockdown results clearly suggest HEXIM1, not p53, as the main regulator of the transcriptional control of neuropeptides under both basal and HMBA-treatment conditions in the hypothalamic cells, distinct regulation of Npy and Pomc may require other nuclear factors as well.
Upon HMBA treatment, the dissociated HEXIM1 and p53 translocated to the nucleus, where p53 coordinated the downregulation of Npy gene expression by binding to the promoter and probably forming a complex with a corepressor, for example, histone deacetylase 3 (Kim et al, 2021), which would decrease binding of H3K9Ac to the Npy promoter.The reduction in H3K4me3 levels may also contribute to the downregulation of Npy expression.In contrast, p53 likely upregulates Pomc gene expression by binding to its promoter and inducing acetylation of H3K9, given the observation that p53 can induce acetylation of histone 3 or 4 (Barlev et al, 2001).Intriguingly, binding of HEXIM1 to the Pomc promoter was decreased after HMBA treatment despite an increase in nuclear HEXIM1, suggesting that HEXIM1 may facilitate Pomc transcription without binding to its promoter.HMBA induces HEXIM1 phosphorylation by activating the PI3K/Akt pathway, and this phosphorylation releases P-TEFb from the inactive complex with HEXIM1 and 7SK small nuclear RNA and leads to subsequent recruitment of P-TEFb to the HIV promoter to stimulate transcription in HIV-infected cells (Contreras et al, 2007).Hence, further studies are warranted to investigate whether other factors are involved in the regulation of Npy and Pomc genes by HMBA.
Several pharmacotherapies targeting gut-derived peptides, intestinal lipase, 5-hydroxytryptamine receptor, and b3 adrenergic receptor have been developed to prevent or treat obesity (Bhat & Sharma, 2017).Despite efforts to lessen the burden of metabolic diseases, however, the prevalence of overweight or obesity is increasing globally.This study points to a previously unappreciated effect of HMBA in the regulation of hypothalamic neuropeptides by which it changes whole-body energy homeostasis.Our understanding of the mode of action of HMBA and its mechanisms underlying the phenotypic changes in appetite, energy expenditure, and overall energy balance might provide ideas and platforms for identification and development of a new chemotype of anti-obesity compounds.

Animal studies
C57BL/6N male and female mice at 4 weeks of age were purchased from Koatech (Pyeongtaek, Republic of Korea).NPY-hrGFP (B6.FVB-Tg(Npy-hrGFP)1Lowl/J), POMC-hrGFP mice (FVB-Tg (Pomc1-hrGFP)1Lowl/J), AgRP-IRES-Cre (AgRP Cre ; Agrp tm1(cre)Lowl / J), and POMC-Cre (POMC Cre ; Tg(Pomc1-cre)16Lowl/J) male mice were obtained from The Jackson Laboratory.NPY-hrGFP and POMC-hrGFP mice were maintained hemizygous for the hrGFP transgene by crossing with C57BL/6J mice.Adult male and female mice were housed in a specific pathogen-free animal facility in individually ventilated cages under a 12-h light/dark cycle (light: 7:00 am to 7:00 pm; dark: 7:00 pm to 7:00 am) at 22 AE 3°C and humidity controlled at 50 AE 10%.Mice had ad libitum access to water and were fed an NCD (12% kcal from fat; Lab Supply, 5053) or HFD (60% kcal from fat; Envigo, 06414) for 4 weeks.Food intake per cage and individual body weight were measured daily.All animal studies followed the National Institutes of Health guidelines and were approved by the Institutional Animal Care and Use Committee at the Daegu Gyeongbuk Institute of Science and Technology Laboratory Animal Resource Center (approval number DGIST-IACUC-21112513).

Gene knockout by CRISPR
The CRISPR-Cas9 system was used to delete the Hexim1 gene from hypothalamic cells.The small guide RNA for Hexim1 gene inactivation (5 0 -GGCACCCGTTCCTCCGCGCTTGG-3 0 ; 23 bp, reverse direction) was designed in exon 1 and was synthesized by ToolGen (Seoul, Republic of Korea).Hypothalamic cells were transfected with the Cas9 and small guide RNA plasmids (1:4 ratio) using TurboFect (Thermo Scientific, R0534).Homogenous Hexim1 knockout was achieved by selection using 1 mg/ml hygromycin B (Enzo Life Science, ALX-380-306-G001) at 36 h after transfection, followed by subculture in fresh media.To validate the knockout, genomic DNA was extracted using a G-spin Total DNA Extraction Mini Kit (iNtRON Biotechnology, 17046).The gDNA concentration was determined using a NanoDrop spectrophotometer (Thermo Scientific).PCR was performed with 100 ng of gDNA using Ex Taq DNA polymerase (TaKaRa, RR001C).Primer sequences are listed in Appendix Table S3.The PCR products were sequenced at Bionics.Multiple-sequence alignment was performed using blastn at NCBI and 4peaks software (Nucleobytes).The Hexim1 knockout efficiency was verified using western blotting.
Intraperitoneal (i.p.) or intravenous (i.v.) injections of HMBA HMBA (diluted in saline; 1,000 mg/kg) or saline was injected into mice daily i.p. for 7 days.In the case of i.v.injection, mice were initially injected with 500 mg/kg of HMBA into the tail vein on day 1, followed by three injections of 1,000 mg/kg bi-daily.Administration was performed before the beginning of the dark cycle (4:30 pm).After HMBA injection, body weight and food consumption changes were measured over 10 days.Mice were housed in individually ventilated cages and had ad libitum access to water and food.To determine relative mRNA expression and HMBA amounts using LC-MS/ MS, the hypothalami and plasma were harvested 4 h after injection.Blood samples were taken from the facial vein using a lancet.

Cannula implantation
Mice at 8 weeks of age were anesthetized with an i.p. injection of a mixture of Zoletil 50 (30 mg/kg body weight; Virbac) and Rompun (10 mg/kg body weight; Bayer Korea) in saline.Mice were placed in a stereotaxic frame (Stoelting Co.), and a 26-gauge guide cannula (Plastic One, Dusseldorf, Germany) was implanted into the third ventricle (3 V) according to the following coordinates: À1.3 mm posterior to the bregma and À5.4 mm below the surface of the skull.The guide cannula was secured to the skull by dental cement (Dentsply Sirona, LAD010) and covered with a dust cap.Mice were allowed to recover for 10 days after surgery.Body weight and food intake were measured daily after recovery.Intracerebroventricular (i.c.v.) injection of HMBA HMBA (2 ll, diluted in saline; 400 nmoles) or saline were i.c.v.injected at a rate of 0.8 ll/min into the 3 V using a Hamilton syringe pump (KD Scientific, Legato series) before the beginning of the dark cycle (4:30 pm).Mice were housed in individually ventilated cages and had ad libitum access to water and food.Mice fed an NCD were injected daily for 10 days, and mice fed an HFD were injected daily for 7 days.To determine relative mRNA expression and HMBA amounts using LC-MS/MS, the hypothalami were harvested 4 h after injection.Plasma was collected from the facial vein.

Stereotaxic microinjection of adeno-associated virus (AAV)
Ultra-purified Cre-dependent AAV-FLEXon-shRNAs (shControl or shMyh9 + shActg1)-EGFP PHP.eB serotype viruses were purchased from VectorBuilder (Guangzhou, China) for silencing both MYH9 and ACTG1 genes of the AgRP Cre or POMC Cre DIO mice.Viral vectors are listed in Appendix Fig S7 .AgRP Cre or POMC Cre DIO mice fed an HFD for 4 weeks were anesthetized by an i.p. injection of a mixture of Zoletil 50 (30 mg/kg body weight; Virbac) and Rompun (10 mg/kg body weight; Bayer Korea) in saline.Mice were placed in a stereotaxic frame (Stoelting Co.), and then the 2 ll of viruses were stereotaxically injected into the ARC using a 10 ll Hamilton syringe mounted to a microinjection pump (KD Scientific, Legato series) at a rate of 0.2 ll/min.Injection coordinates were 1.3 mm caudal to the bregma, 0.4 mm bilateral to the midline, and 5.8 mm below the pial surface.Mice were fed an HFD ad libitum during the 2-week recovery.After the recovery, mice were i.p. injected with HMBA daily for 7 days and body weight and food intake were monitored for a total of 10 days starting from the first day of injection.The transduction efficiency of the viral vectors was demonstrated via the expression of EGFP in the ARC using immunohistochemistry.

Body composition and indirect calorimetry
Body composition was measured using a nuclear magnetic resonance MiniSpec LF50 analyzer (Bruker) prior to metabolic measurements.Indirect calorimetry was performed when body weight monitoring after HMBA injections was completed.For indirect calorimetry measurements, mice were housed individually with free access to water and HFD in metabolic chambers of the Comprehensive Lab Animal Monitoring System (CLAMS; Columbus Instruments).Mice were allowed to acclimatize in metabolic chambers for 24 h.After acclimation, mice were injected with HMBA once i.v., but not i.c.v., and then metabolic parameters were collected over the next 24 h.Oxygen consumption (VO 2 ), carbon dioxide production (VCO 2 ), heat production, and locomotor activity were determined using an Oxymax System (Columbus Instruments).VO 2 , VCO 2 , and heat production were normalized to body weight.The respiratory exchange ratio (RER) was calculated as VCO 2 /VO 2 .Locomotor activity was determined by measuring interruptions of the infrared beams (X-and Z-axes of total ambulatory counts).Energy expenditure was analyzed by ANCOVA using body weight as covariate as previously described (Tschop et al, 2011;Muller et al, 2021).

Brown adipose tissue thermogenesis
The surface interscapular brown adipose tissue temperatures of DIO mice were recorded 3 days after the last HMBA injection using a FLIR infrared camera (FLIR Systems, E6).Infrared pictures of nonanesthetized mice were taken from a distance of 30 cm.Temperature was quantified using FLIR analysis software.

Glucose and insulin tolerance test
Glucose and insulin tolerance tests were performed in DIO mice after monitoring body weight and food intake from the last i.c.v.injection of HMBA.Mice were fasted overnight for glucose tolerance tests and 6 h for insulin tolerance tests.Glucose (2 g/kg; Sigma-Aldrich, G7021) or human insulin (1 U/kg; Sigma-Aldrich, I9278) was administrated i.p.Blood was collected from the tail vein and blood glucose was determined at 0, 15, 30, 60, 90, and 120 min after glucose or insulin injection.Glucose levels were determined using an Accu-Check II glucometer (Roche).

LABORAS
HMBA (diluted in saline; 1,000 mg/kg for i.p., 400 nmoles for i.c.v.) or saline was injected into mice daily i.p. or i.c.v. for 7 days.In the case of i.v.injection, mice were injected with 500 mg/kg of HMBA on day 1, followed by three injections of 1,000 mg/kg bi-daily.After the last dosing of HMBA (on day 7 for i.p. or i.c.v.; day 8 for i.v. at 4:30 pm), mice were placed in every single LABORAS cage and recorded for 24 h starting from the dark cycle at 7:00 pm.Mice activities including locomotion, climbing, rearing, grooming, eating, and drinking were recorded and automatically analyzed using the Laboratory Animal Behavior Observation Registration and Analysis System (LABORAS, Metris).

CTA test
Conditioned taste aversion (CTA) test was performed as previously described (Rosenblum et al, 1993;Berman & Dudai, 2001;Eisenberg et al, 2003;Akirav et al, 2006;Slouzkey et al, 2013).Male DIO mice were housed in individually ventilated cages and were pre-trained for 2 consecutive days to get their water rations for 20 min once a day.On the conditioning day, water was replaced with 0.3 M sucrose during the 20 min access session (referred to as pre).Twenty minutes later, mice received lithium chloride (LiCl, 0.05 M, 2% body weight; i.p.) to induce malaise, or received HMBA (i.v., i.p., or i.c.v.) as unconditioned stimulus.On the next day, mice were given a retrieval session (referred to as post) entailing 20 min access to 0.3 M sucrose.All water bottles were weighted before and after each session (pre/post) to measure the total amount (g) of sucrose consumption.

LC-MS/MS analysis
Hypothalami were extracted with a mixture of 400 ll of 80% methanol and 300 ll of 80% acetonitrile and were homogenized using a MagNA Lyser (Roche).Samples were sonicated (10 cycles, 40 s each, medium intensity) using a Bioruptor sonicator (Diagenode) and centrifuged at 16,000 g for 15 min.Plasma (50 ll) was extracted with 800 ll of methanol:chloroform:water, 2:1:1 (v:v).The aqueous phase was evaporated to dryness under nitrogen gas (TurboVap, Biotage).Dried samples were reconstituted with 200 ll of buffer A (10 mM ammonium acetate with 0.1% formic acid in water) and filtered through a 0.2 lm syringe filter.LC-MS/MS analysis was carried out on an Agilent 1290 ultra-performance liquid chromatography system coupled with an Agilent 6490 triple quadrupole mass spectrometer.Samples (2 ll) were separated on a Phenomenex Synergi HydroRP column (150 mm × 2 mm, 4 lm, 80 A) maintained at 45°C with 10 mM ammonium acetate with 0.1% formic acid in water as buffer A and acetonitrile as buffer B. The flow rate was 0.3 ml/min and the total run time was 9 min.The gradient was as follows: 0-2 min at 100% buffer A; 2-5 min, linear gradient to 90% buffer B; 5-7 min, 90% buffer B; 7-7.1 min, return to the initial conditions; and 7.1-9 min, re-equilibration.The system was run-in multiple-reaction monitoring mode using optimized collision energy with positive electrospray ionization mode.The precursor-to-fragment ion transitions were selected according to Smith et al (2011), and NADAH, DAH, and AmHA were selected on the basis of our mass spectrometry settings (Appendix Table S2).

GC-MS/MS analysis
Serum (50 ll) underwent sonication using cold methanol:chloroform:water solution in a ratio of 2:1:1 (v/v/v), which included an internal standard (20 nM D2-oleate).Following sonication, the samples were subjected to centrifugation at 4°C for 10 min at 12,000 g.The organic phases were subsequently dried in a TurboVap evaporator (Biotage).To reconstitute the samples, they were dissolved in hexane containing 0.5 M KOH-MeOH and allowed to incubate at 25°C for 10 min.BCl 3 methanol (12% w/w) was then introduced, and the samples were heated at 70°C for 10 min, followed by rapid cooling on ice for 5 min.Next, a mixture of hexane and water in a 2:1 (v/v) ratio was added to the samples, which were then vortexed.Finally, the supernatants were transferred into autosampler vials.To quantify free fatty acids in serum, GC-MS/MS analysis was carried out in an Agilent 7000B gas chromatography system coupled with an Agilent 7000C triple quadrupole mass spectrometer.
Samples were separated on an Ultra HP-5 ms capillary column (30 m × 0.25 lm, i.d., 0.25 lm film thickness, Agilent J&W Scientific).The instrument temperature was set as follows: inlet, 250°C; transfer line, 290°C; ion source, 230°C; and quadrupoles, 150°C.The capillary voltage was set at 70 eV.The gradient was run at a flow rate of 1.2 ml He/min and 4°C/min with 50°C up to 240°C.Metabolomics data were processed using the Mass Hunter software package (Agilent Technologies).

Triglyceride analysis
Serum triglyceride levels were measured using the Triglyceride Quantification Colorimetric/Fluorometric Kit (BioVision, K622-100) after the completion of body weight and food intake measurements after the last i.c.v.injection of HMBA.Mice were fasted overnight for 16 h.Blood was collected through the facial vein.Serum (2 ll) was mixed with 50 ll of assay buffer containing 0.4 ll/well of the probe and 2 ll of TG enzyme mix.Lipase reaction was allowed to proceed at room temperature for 60 min.Fluorescence was measured at Ex/Em = 545/590 nm.

Biotin-neutravidin pull-down assay
Cells were sonicated (5 cycles, 15 s each, high intensity) in lysis buffer described above containing 1× protease and phosphatase inhibitor cocktails but no Tris-HCl at 4°C using a Bioruptor sonicator.The cell lysate was incubated with biotinylated HMBA at room temperature for 1 h and then incubated with neutravidin resin (Thermo Scientific, 53150) for 30 min in a disposable column (Thermo Scientific, 29922).The mixture was centrifuged at 400 g for 1 min.The resin was washed with phosphate-buffered saline five times to remove non-specific binding.Proteins bound to biotinylated HMBA were eluted and denatured in 2× Laemmli sample buffer (Bio-Rad, 1610737) supplemented with 10 mM dithiothreitol, and electrophoresed on a 4-15% gradient SDS-PAGE (Bio-Rad, 4561084).The gel was visualized with InstantBlue Coomassie protein-staining solution (Abcam, ab119211).

LC-MS/MS for peptide analysis of HMBA-binding proteins
Coomassie blue-stained bands were in-gel digested with trypsin.Nano-LC-MS/MS analysis was performed with an Easy-nLC system connected to an LTQ Orbitrap XL mass spectrometer (Thermo Fisher San Jose) equipped with a nano-electrospray source.Samples were separated on an Agilent C18 Nanobore column (150 mm × 0.075 mm, 3 lm) with 0.1% formic acid and 3% acetonitrile in water as buffer A and 0.1% formic acid in acetonitrile as buffer B. The flow rate was 1 ll/min and the total run time was 100 min.The linear gradient was as follows: 0-80 min, 0 to 40% buffer B; 80-84 min, 40 to 60% buffer B; 84-88 min, 60 to 95% buffer B; 88-94 min, 95 to 100% buffer B; and 94-100 min, re-equilibration.Mass spectra were acquired using data-dependent acquisition with a full mass scan (m/z 350-1,200) followed by 10 MS/MS scans.For MS1 full scans, the orbitrap resolution was 15,000 and the automatic gain control value was 2 × 10 5 .For MS/MS in the LTQ, the automatic gain control value was 1 × 10 4 .Protein identification through peptide sequence matching was performed using the Mascot algorithm (Matrix Science).The criteria for database search were as follows: taxonomy, Mus musculus; fixed modification, carbamidomethylated cysteine residues; variable modification, oxidized methionine residues; maximum allowed missed cleavages, 2; MS tolerance, AE 10 ppm; MS/MS tolerance, 0.8 Da; and peptide filtration, a significance threshold of P < 0.05.

Computational protein-ligand docking modeling
The interactions between MYH9 or ACTG1 and HMBA were predicted by computational docking using AutoDock free opensource software (Forli et al, 2016).AutoDock was run using the AlphaFold predictions of MYH9 and ACTG1 (Q8VDD5 and P63260, respectively; taxonomy, Mus musculus) from UniProt (http://www.uniprot.org),and a chemical table file (hexamethylene bisacetamide, CID 3616) from PubChem (NCBI), with the following parameters: Lamarckian genetic algorithm runs, 300; the maximum number of energy evaluations, 30 million; and grid map setting points and spacing, 126 × 126 × 126 and 1 A, respectively, in all dimensions.The top candidate docking model was used in this study.Additionally, we verified and visualized three-dimensional protein-ligand complexes using BIOVIA Discovery Studio (Dassault Syst emes, Release 2022), PLIP (Adasme et al, 2021), Protein Plus (Schoning-Stierand et al, 2020), and PyMOL 2.5.2; the two-dimensional diagrams of these complexes were generated by LIGPLOT + (Wallace et al, 1995).

Quantitative real-time PCR
Samples were prepared as previously reported (Park et al, 2019).Briefly, total RNA was isolated from mouse hypothalamus, brown adipose tissue, or cells with Trizol reagent (Invitrogen, 15596018).The RNA pellet was dissolved in nuclease-free water (Promega, P1195), and total RNA concentration was determined using a NanoDrop spectrophotometer (Thermo Scientific).cDNA was synthesized from 3 lg of RNA using GoScript Reverse Transcriptase (Promega, A5001).mRNA levels were quantified by qRT-PCR with a TB (SYBR) Green Premix Ex Taq II (TaKaRa, RR820L).Primer sequences are listed in Appendix Table S3.The mRNA levels were normalized to those of Gapdh.

Chromatin immunoprecipitation-quantitative PCR
Cells were cross-linked with 1.4% formaldehyde for 15 min at room temperature and then quenched with 125 mM glycine for 5 min.After harvesting the cells, cells were centrifuged at 16,000 g for 5 min.Cell pellet was resuspended in lysis buffer (150 mM NaCl, 50 mM Tris-HCl pH 7.5, 5 mM EDTA, 0.5% v/v NP-40, and 1% v/v Triton X-100) containing 1× protease and phosphatase inhibitor cocktails.To shear DNA, the lysate was sonicated (15 cycles, 30 s each, high intensity) at 4°C using a Bioruptor sonicator.After sonication, the samples were centrifuged at 16,000 g for 10 min and the supernatants were incubated overnight at 4°C with 2 lg of primary antibodies or control IgG and then reacted with Dynabeads Protein G (Invitrogen, 10007D) at 4°C for 2 h.DNA was eluted with elution buffer (1% SDS, 50 mM Tris pH 8.0, and 10 mM EDTA), and cross-linking was reversed with 200 mM NaCl, RNase A (10 mg/ml), and Proteinase K (20 mg/ml) at 65°C overnight.DNA was purified using a phenol:chloroform:isoamyl alcohol mixture (25:24:1, v/v) (Thermo Scientific, 15593031) and was quantified by qRT-PCR.Primer sequences for each promoter are listed in Appendix Table S4.The expression levels were normalized to the amount of input sample.

Immunofluorescence
Cells were fixed in 4% paraformaldehyde for 15 min at room temperature, washed three times with PBS, permeabilized in 0.05% Triton X-100 for 15 min at room temperature, rewashed with PBS, and blocked in 5% normal donkey serum with 1% bovine serum albumin for 1 h at room temperature.Cells were incubated overnight at 4°C with primary antibodies (HEXIM1, 1:150; p53, 1:150) in blocking buffer.Cells were washed three times with PBS, incubated with secondary antibodies (Cy3 to detect HEXIM1 and Alexa488 to detect p53; both 1:400) in blocking buffer for 1 h at room temperature, washed with PBS, and stained with Hoechst 33342 for 5 min at room temperature.Fluorescence images were captured using a Carl Zeiss LSM 780 or LSM 800 confocal laser-scanning microscope and analyzed in Zen software (Carl Zeiss).The list of antibodies is provided in Appendix Table S5.To measure the signal intensities of HEXIM1 or p53-stained cells, the nucleus/cytoplasm signals of individual cells per slide were analyzed using ImageJ software (National Institute of Health).

Immunohistochemistry
Mice were anesthetized with an i.p. injection of a mixture of Zoletil 50 (30 mg/kg body weight; Virbac) and Rompun (10 mg/kg body weight; Bayer Korea) in saline.They were then perfused with PBS followed by 4% paraformaldehyde in PBS (pH 7.4) to fix the brain tissue.(i) Cryopreservation: the brains were extracted and postfixed in 4% paraformaldehyde for 16 h at 4°C, then transferred to 30% sucrose in PBS.The brains were embedded in optimal cutting temperature compound (Leica) frozen on dry ice, and stored at À80°C.The brains were sliced (thickness: 40 lm), blocked in 5% normal donkey serum with 1% bovine serum albumin for 1 h at room temperature, washed three times with PBS, permeabilized in 0.05% Triton X-100 for 15 min at room temperature, rewashed with PBS, and blocked in 5% normal donkey serum with 1% bovine serum albumin for 1 h at room temperature.The brain slices were incubated overnight at 4°C with primary antibody (HEXIM1, 1:400) in blocking buffer.After washing, brain slices were incubated with secondary antibody (Cy3 to detect HEXIM1; 1:500) in blocking buffer for 2 h at room temperature and then washed and stained with Hoechst 33342 for 5 min at room temperature.(ii) Paraffin embedding: the brains were extracted and post-fixed in 4% paraformaldehyde for 9 h at 4°C, then dehydrated at 70-100% gradient ethanol for 15 min, washed with ultra-pure water for 10 min, and blocked in 5% normal donkey serum with 1% bovine serum albumin for 1 h at room temperature.The brain slices were incubated overnight at 4°C with primary antibody (MYH9 and ACTG1, 1:250) in blocking buffer.After washing, brain slices were incubated with secondary antibody (Alexa flour 647 to detect ACTG1 and 1:400; Cy3 to detect MYH9, 1:500) in blocking buffer for 2 h at room temperature and then washed and stained with Hoechst 33342 for 5 min at room temperature.The antibodies are listed in Appendix Table S5.Fluorescence images were taken with a Carl Zeiss LSM 800 confocal laser-scanning microscope and analyzed in Zen software (Carl Zeiss).To measure the numbers and intensities of GFP-positive or HEXIM1-stained cells, brain slices per mouse were analyzed using Fiji of ImageJ 2 software (National Institute of Health).

Western blotting
Cells were lysed in a buffer solution composed of the following components: 50 mM Tris-HCl (pH 7.4), 250 mM sucrose, 5 mM sodium pyrophosphate, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100 (obtained from Sigma-Aldrich, T8787), 0.1 mM benzamidine (obtained from Sigma-Aldrich, B6506), 1 mM dithiothreitol (obtained from Sigma-Aldrich, 43816), 0.5 mM phenylmethylsulfonyl fluoride (obtained from Sigma-Aldrich, P7626), 50 mM sodium fluoride, a protease inhibitor cocktail (purchased from Millipore, 535140), and a phosphatase inhibitor cocktail (obtained from Sigma-Aldrich, P5726).To determine protein concentrations, the Pierce BCA assay kit from Thermo Fisher Scientific (23228) was used.Next, 10 lg of lysates were separated using SDSpolyacrylamide gels and transferred onto polyvinylidene difluoride membranes from Millipore (IPVH00010).The membranes were subjected to a 35-min blotting process at 20 V in transfer buffer consisting of 25 mM Tris base and 192 mM glycine.Following this, the membranes were blocked for 1 h using a solution of 5% skim milk in 1× TBST buffer (20 mM Tris, 125 mM NaCl, and 0.1% Tween 20, obtained from Sigma-Aldrich, P1379) at pH 7.4.Primary antibodies were then applied and incubated either at room temperature for 1 h or at 4°C overnight.Subsequently, the membranes underwent three washes with 1× TBST and were subsequently incubated with secondary antibodies linked to horseradish peroxidase.The visualization of proteins was accomplished using either the Supersignal West Pico Chemiluminescent Substrate from Thermo Fisher Scientific (NCI4080KR) or the WesternBright ECL system from Advansta (K-12045-D50), following the manufacturer's guidelines.The antibodies are listed in Appendix Table S5.Blots were quantified using ImageJ software (National Institutes of Health), and the level of each protein was normalized to that of GAPDH.

Nuclear and cytoplasmic fractionation
Nuclear and cytoplasmic fractions of cells were obtained using an NE-PER Nuclear and Cytoplasmic Extraction Reagent Kit (Thermo Scientific, 78835).

Immunoprecipitation
Nuclear or cytoplasmic lysates (200 lg total protein) were diluted with lysis buffer (150 mM NaCl and 25 mM Tris-HCl pH 7.2) containing protease and phosphatase cocktail inhibitors.The samples were incubated overnight at 4°C with 2 lg of primary antibodies and then reacted with Pierce Protein A agarose beads (Thermo Scientific, 20333) at 4°C for 6 h.Bound proteins were eluted with 4× Laemmli sample buffer (Bio-Rad, #161-0737) containing 10% 2-mercaptoethanol (Sigma-Aldrich, M6250) and incubated at 95°C for 5 min.The supernatants were separated by SDS-PAGE, and the subsequent procedure was the same as for western blotting.

The paper explained Problem
Obesity is considered a pandemic with an increasing prevalence worldwide.Lack of new knowledge of key components and mechanisms regulating metabolism is one of the obstacles to therapeutic designs for the treatment of obesity.Hypothalamic neuropeptides such as NPY and POMC are pivotal in regulating energy balance.Therefore, pharmacological interventions that can modulate their expressions are considered one of the promising strategies.

Results
Hexamethylene bisacetamide (HMBA) prevents diet-induced obesity by inducing hypophagia and increasing energy expenditure, leading to weight loss in mice.Of note, HMBA modulates the expression of hypothalamic neuropeptides NPY and POMC toward the direction of negative energy balance.HMBA binds to MYH9 and ACTG1 proteins, inducing HEXIM1 expression in NPY and POMC neurons.HMBA facilitates the interaction of HEXIM1 with MDM2, dissociating the HEXIM1-p53 complex, which leads to the downregulation of NPY and upregulation of POMC expression.

Impact
Our findings propose the molecular mechanisms by which HMBAinduced HEXIM1 regulates neuropeptide expression to affect systemic energy balance.In addition, our study may advance our understanding of hypothalamic control through the intervention of HMBA as an anti-obesity compound and suggest a new therapeutic strategy for the treatment of metabolic diseases such as obesity and diabetes.

Statistical analysis and drawing chemical structures
GraphPad Prism software (v.10.0.2) was used for graphing and statistical analysis.For comparison between two groups, datasets were analyzed by a two-tailed unpaired Student's t-test.Multiple comparisons were analyzed by one-or two-way ANOVA (Tukey or Bonferroni post hoc test) to determine the statistical significance between groups.Differences in energy expenditure were calculated by ANCOVA using body weight as covariate.All quantitative data were represented as mean AE standard error of the mean (SEM).A value of P < 0.05 was considered significantly different.ChemDraw software (v.20.0) was used for drawing chemical structures.The sample size estimate was not determined.Animals were assigned to the groups randomly and no mice were excluded from the experiments.No blinding was done in the analysis. Figure1.

(
Fig 2F and G).Consistent with i.v.injection of HMBA, i.c.v.injection of HMBA significantly increased metabolic rates independently of the body mass (Fig 1H).Consumption of oxygen, generation of carbon dioxide (Fig 2I-K), and heat production (Fig 2L) were elevated by HMBA without changes in a day/nighttime locomotor activity (Fig 2M and N).

▸Figure 3 .
Figure 3. HMBA regulates neuropeptide expression in a HEXIM1-dependent manner.A Metabolic pathway of HMBA.B Chromatogram peaks of LC-MS/MS.C Metabolite concentrations in the hypothalamus and relative percentage of molar concentrations.Metabolomic analysis was performed 4 h after i.c.v.injection of 400 nmoles of HMBA.n = 6 per group.D Metabolic flow of HMBA in mHypoE-N41 and relative percentage of the area under the curve (AUC).n = 3 per group.E Metabolic flow of HMBA in mHypoE-N43/5 and relative percentage of the area under the curve (AUC).n = 3 per group.F Relative mRNA levels of Npy upon HMBA treatment up to 24 h.n = 3 per group.G Relative mRNA levels of Pomc upon HMBA treatment up to 24 h.n = 3 per group.H, I Relative mRNA levels of (H) Npy and (I) Pomc.Cells were treated with 10 À3 , 10 À2 , or 10 À1 mM HMBA for 1 or 2 h.n = 3 per group.J Heatmap of HEXIM isoforms' quantification cycle (cq) values in the hypothalamus and cell lines.n = 3 per group.K, L Relative mRNA levels of Hexim1 upon HMBA treatment up to 24 h in (K) mHypoE-N41 and (L) mHypoE-N43/5.n = 3 per group.M, N Relative basal mRNA levels of (M) Npy and (N) Pomc in Hexim1 KD (si-Hexim1).n = 3 per group.O, P Relative mRNA levels of (O) Npy and (P) Pomc in Hexim1 KO (sgHexim1) after HMBA treatment for 2 h.n = 3 per group.Data information: (D-G) and (K-P), Cells were treated with 0.1 mM HMBA.(A) Data represent six biological replicates.(D, E, J) Data represent three technical replicates.The datasets in qPCR experiments were comprised of three biological replicates and each biological replicate was an average of three technical replicates.(F-L) Statistical significance was determined by two-way ANOVA followed by a post hoc Bonferroni test.*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.(M-P) Statistical significance was determined by two-way ANOVA followed by a post hoc Tukey test.*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns or unless otherwise stated, no significance; nd, not detected.Data are mean AE SEM.The exact P-values are reported in Appendix TableS6.Source data are available online for this figure.
Figure 4. HMBA regulates neuropeptide expression through binding to MYH9 and ACTG1.A Schematic diagram of NADAH biotinylation (HMBA-biotin, biotin-conjugated HMBA).B L C -MS/MS chromatogram peaks of NADAH (Àbiotin, pink) and HMBA-biotin (+biotin, green), and a bar graph of percentage peak integration values before and after biotinylation.C A400-MHz 1D 1 H NMR spectrum of HMBA-biotin.Chemical shifts are described in the Materials and Methods section.D Western blot analysis of HEXIM1 after HMBA-biotin treatment.Blot quantifications are the mean of two biological replicates.E Schematic of the experimental design of chemical proteomics study including biotin pull-down using neutravidin and LC-MS/MS peptide analysis.F Coomassie blue-stained gel of samples after biotin pull-down (left panel) and Mascot score plots of MYH9 and ACTG1 (right panel).In the stained gel, blue ellipses indicate MYH9 (230 kDa) and pink ellipses indicate ACTG1 (42 kDa).G, H Docked conformation of (G) the MYH9-HMBA complex and (H) the ACTG1-HMBA complex.Raw data generated by AutoDock were verified and displayed through Protein Plus (left panels) and PLIP (right panels).HMBA is shown as gray (left panels) and yellow (right panels) sticks.Amino acid residues are shown as cyan sticks.Hydrogen bonds are indicated by solid gray lines with the distance ( A) from the donor atom to the acceptor atom indicated, and hydrophobic interactions are indicated by dashed yellow lines.N and O atoms are shown in blue and red, respectively.Residues in black are involved in major hydrogen bonding, while those in gray are involved in hydrophobic interactions.I, J Relative mRNA levels of (I) Hexim1 and (J) Npy in si-Myh9 of mHypoE-N41.n = 3 per group.K, L Relative mRNA levels of (K) Hexim1 and (L) Npy in si-Actg1 of mHypoE-N41.n = 3 per group.M, N Relative mRNA levels of (M) Hexim1 and (N) Pomc in si-Myh9 of mHypoE-N43/5.n = 3 per group.O, P Relative mRNA levels of (O) Hexim1 and (P) Pomc in si-Actg1 of mHypoE-N43/5.n = 3 per group.Data information: Cells were treated with 0.1 mM HMBA or HMBA-biotin for 2 h.The datasets in qPCR experiments were comprised of three biological replicates and each biological replicate was an average of three technical replicates.Statistical significance was determined by two-way ANOVA followed by a post hoc Tukey test.**P < 0.01, ***P < 0.001, ****P < 0.0001; ns or unless otherwise stated, no significance.Data are mean AE SEM.The exact P-values are reported in Appendix TableS6.Source data are available online for this figure.

Figure 5 .
Figure 5. Anti-obesity effects of HMBA are abolished by ablation of Myh9 and Actg1 in AgRP or POMC neurons of DIO mice.A-D Immunohistochemistry analysis for the fluorescence intensity of NPY, POMC, and HEXIM1 in the ARC of (A, B) NPY-hrGFP DIO mice and (C, D) POMC-hrGFP DIO mice after HMBA treatment.The fluorescence intensity of NPY and POMC expression was determined based on the native fluorescence of hrGFP, while the fluorescence intensity of HEXIM1 (Cy3) which co-localized with hrGFP-positive cells was determined.NPY-hrGFP and POMC-hrGFP DIO mice fed an HFD for 4 weeks were injected with HMBA (i.p., 1,000 mg/kg), and then brains were harvested 4 h after injection.n = 32 per group: eight brain slices were obtained from each of the four mice per group for determining hrGFP intensity in panels (A) and (C); n = 16 per group: four brain slices were obtained from each of the four mice per group for determining HEXIM1 intensity in panels B and D. Scale bars, 25 lm.E Schematic diagram of studies on double knockdown of Myh9 and Actg1 in AgRP or POMC neurons.AgRP Cre and POMC Cre DIO mice fed an HFD for 4 weeks were injected with AAV-FLEXon-shRNAs (shControl or shMyh9 + shActg1)-EGFP PHP.eB serotype viruses into the ARC.After recovery, mice were injected daily with HMBA for 7 days (i.p., 1,000 mg/kg).F, G Immunohistochemistry analysis for the knockdown efficiency of MYH9 and ACTG1 in the ARC of (F) AgRP Cre ::AAV (shControl or shMyh9+shActg1) DIO mice and (G) POMC Cre ::AAV (shControl or shMyh9+shActg1) DIO mice after HMBA treatment.The fluorescence intensity of MYH9 (Cy3) or ACTG1 (Alexa 647; Blue) which co-localized with EGFP-positive cells was determined.n = 25 per group: A total of 25 EGFP-positive cells were randomly counted across eight brain slices obtained from two mice per group.Scale bars, 25 lm.H, I Immunohistochemistry analysis for the fluorescence intensity of HEXIM1 in the ARC of (H) AgRP Cre ::AAV (shControl or shMyh9+shActg1) DIO mice and (I) POMC Cre :: AAV (shControl or shMyh9+shActg1) DIO mice after HMBA treatment.The fluorescence intensity of HEXIM1 (Cy3) which co-localized with EGFP-positive cells was determined.n = 24 per group: eight brain slices were obtained from each of the three mice.Scale bars, 25 lm.J, K Relative mRNA levels of hypothalamic neuropeptides Npy and Pomc in (J) AgRP Cre ::AAV (shControl or shMyh9+shActg1) DIO mice and (K) POMC Cre ::AAV (shControl or shMyh9+shActg1) DIO mice after HMBA treatment.RNA was isolated 4 h after HMBA treatment.n = 6 per group.L, M Changes in (L) daily body weight (grams) (left panel) and body weight (percentages) on day 10 relative to day 0 (right panel), and (M) food intake for AgRP Cre :: AAV (shControl or shMyh9+shActg1) DIO mice after HMBA treatment.n = 7, shControl group; n = 6, shMyh9 + shActg1 group.N, O Changes in (N) daily body weight (grams) (left panel) and body weight (percentages) on day 10 relative to day 0 (right panel), and (O) food intake for POMC Cre :: AAV (shControl or shMyh9+shActg1) DIO mice after HMBA treatment.n = 6, saline-injected group; n = 7, HMBA-injected group.Data information: (J-O) Data represent different numbers (n) of biological replicates.Data are represented in box and whisker plots where the central band denotes the median value, box contains interquartile ranges, while whiskers mark minimum and maximum values in panels (L) and (N).Statistical significance was determined by a two-tailed unpaired Student's t-test in panels (A-D) and (H-K), and two-way ANOVA followed by a post hoc Bonferroni test in panels (F) and (G), and two-way ANOVA followed by a post hoc Tukey test in panels (L-O).*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 vs. between HMBA treatment, and # P < 0.05, ## P < 0.01, ### P < 0.001, #### P < 0.0001 vs. saline.Data are mean AE SEM.The exact P-values are reported in Appendix TableS6.Source data are available online for this figure.
To elucidate the mechanism by which HMBA regulates neuropeptide expression, we first assessed changes in the localization of HEXIM1 and p53 after 2 h HMBA treatment using confocal microscopy imaging and subcellular fractionation.While HMBA increased HEXIM1 levels in both the cytoplasm and nucleus, p53 was increased in the nucleus only (Fig 7A-C and Appendix Fig S9).HMBA-induced increase in nuclear p53 was abolished by Hexim1 deletion (Fig 7A-C and Appendix Fig S9).

▸Figure 7 .
Figure 7. HMBA regulates the transcription of Npy and Pomc through HEXIM1, p53, and H3K9Ac.A, B Immunofluorescence analysis of HEXIM1 and p53 after HMBA treatment in sgControl and sgHexim1 of (A) mHypoE-N41 and (B) mHypoE-N43/5, and signal intensity in the cytoplasm and nucleus.n = 10, sgControl group of both cell lines; n = 7, sgHexim1 group of both cell lines.Scale bars, 10 lm.C Western blot analysis of HEXIM1 and p53 in the cytoplasmic and nuclear lysates of sgControl or sgHexim1 cells treated with HMBA.D Immunoprecipitation (IP) analysis of cytoplasmic and nuclear lysates.E Chromatin immunoprecipitation (ChIP) analysis of the Npy promoter.n = 3 per group.F ChIP analysis of the Pomc promoter.n = 3 per group.G Schematic diagram suggesting the mechanism of the regulation of neuropeptide transcription by HMBA; HMBA binds to MYH9 and ACTG1, inducing HEXIM1 and MDM2 expression.HEXIM1 competitively interacts with MDM2 rather than p53, resulting in dissociation of the HEXIM1-p53 complex in the cytoplasm.The dissociated HEXIM1 and p53 translocate to the nucleus.Nuclear p53 and HEXIM1 bind to the Npy promoter, which negatively regulates Npy transcription, whereas nuclear p53 and H3K9Ac bind to the Pomc promoter, which positively regulates Pomc transcription.Data information: Cells were treated with 0.1 mM HMBA for 2 h.For IP or ChIP analysis, 200 lg of total protein was used.The datasets in qPCR experiments were comprised of three biological replicates and each biological replicate was an average of three technical replicates.The numbers under western blots indicated relative quantitative mean values of three biological replicates.Statistical significance was determined by two-way ANOVA followed by a post hoc Tukey test.*P < 0.05, ****P < 0.0001; À or nd, not detected.Data are mean AE SEM.The exact P-values are reported in Appendix TableS6.Source data are available online for this figure.

). HMBA increased thermogenesis in brown adipose tissue ▸ Figure 2. Central administration of HMBA has anti-obesity effects in DIO mice.
< 0.0001 vs. saline.Data are mean AE SEM.The exact P-values are reported in Appendix TableS6.Source data are available online for this figure.