2-Hydroxyisobutyrylation on histone H4K8 is regulatedby glucose homeostasis in Saccharomyces cerevisiaeJing Huanga,b,1, Zhouqing Luoa,b,1, Wantao Yingc,1, Qichen Caoc, He Huangd, Junkai Donga, Qingyu Wua,Yingming Zhaod, Xiaohong Qianc,2, and Junbiao Daia,b,2

aMinistry of Education Key Laboratory of Bioinformatics, Centre for Synthetic and Systems Biology, School of Life Sciences, Tsinghua University, Beijing100084, China; bCenter for Synthetic Biology Engineering Research, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen518055, China; cState Key Laboratory of Proteomics, National Protein Science Center, Beijing Proteome Research Center, Beijing Institute of RadiationMedicine, Beijing 102206, China; and dBen May Department of Cancer Research, University of Chicago, Chicago, IL 60637

Edited by Jef D. Boeke, New York University Langone Medical Center, New York, NY, and approved July 10, 2017 (received for review February 3, 2017)

New types of modifications of histones keep emerging. Recently,histone H4K8 2-hydroxyisobutyrylation (H4K8hib) was identified asan evolutionarily conserved modification. However, how this mod-ification is regulated within a cell is still elusive, and the enzymesadding and removing 2-hydroxyisobutyrylation have not beenfound. Here, we report that the amount of H4K8hib fluctuates inresponse to the availability of carbon source in Saccharomyces cer-evisiae and that low-glucose conditions lead to diminished modifi-cation. The removal of the 2-hydroxyisobutyryl group from H4K8 ismediated by the histone lysine deacetylase Rpd3p and Hos3pin vivo. In addition, eliminating modifications at this site by alaninesubstitution alters transcription in carbon transport/metabolismgenes and results in a reduced chronological life span (CLS). Further-more, consistent with the glucose-responsive H4K8hib regulation,proteomic analysis revealed that a large set of proteins involvedin glycolysis/gluconeogenesis are modified by lysine 2-hydroxyiso-butyrylation. Cumulatively, these results established a functional andregulatory network among Khib, glucose metabolism, and CLS.

protein posttranslational modifications | lysine acetylation | lysine 2-hydroxyisobutyrylation | chronological life span | histone deacetylase

Posttranslational modifications (PTMs) of proteins influencetheir properties, such as cellular localization, stability, in-

teraction, and enzymatic activity (1–5). Over the past decade,several new types of PTMs have been identified on lysine residues,including propionylation, butyrylation, crotonylation, succinylation,malonylation, glutarylation, and 2-hydroxyisobutyrylation, which,collectively, are termed as lysine acylation (6–11). The possibledonors of these acylations are, presumably, their correspondingacyl-CoAs, the intermediates of many cellular metabolic processes(12). Several recent studies indicated that these new acylations oc-cur on thousands of proteins in various cellular metabolic processesand play important roles in metabolic regulation (11, 13–20).Acylations were also found on histones (6, 8, 10, 21), the major

protein components of chromatin, which, together with a fragmentof DNA, form the basic building block of a eukaryotic genome(22). Modifications on histones have important functions intranscriptional regulation, DNA repair, replication, and chromatincondensation (23). Two major mechanisms have been proposed.One mechanism is to alter the charge state of the modified resi-dues, which might interfere with the histone–histone and/or his-tone–DNA electrostatic interactions, leading to a chromatin statetransition. The other mechanism is to act as “bait” to recruit ef-fector proteins to chromatin (23). A previous study of histonecrotonylation suggested that it is functionally distinct from lysineacetylation (Kac) and marks active testis-specific genes in post-meiotic cells (8). Furthermore, histone crotonylation was provedto stimulate transcription to a greater extent than histone acety-lation in a cell-free transcription system (24).Although quite a few new modifications have been identified,

identifying the enzymes that write or erase the modification haslargely lagged behind. Some histone acetyltransferases (HATs)

and histone deacetylases (HDACs) have been reported to beable to catalyze additional acylation or deacylation reactions inaddition to acetylation or deacetylation (6, 24–26), hinting thatthe regulation and function of these new histone acylations mayalso be similar to or perhaps redundant with histone acetylation.However, until now, little about the regulation and function ofthese new acylations on histones is known.Lysine 2-hydroxyisobutyrylation (Khib) is a newly identified

histone mark conserved from yeast to humans; specifically,H4K8 2-hydroxyisobutyrylation (H4K8hib) has been detected inactively transcribed genes in mouse meiotic and postmeiotic cells(10). Here we report the discovery of a carbon stress-relatedfunction and regulation of H4K8hib in Saccharomyces cerevisiae.We found that H4K8hib was a stress-responsive modification andidentified its modifying enzymes. In addition, we showed that thenonmodifiable substrate mutant, H4K8A, leads to a reducedchronological life span (CLS). Finally, we performed Khib pro-teomics analysis in Saccharomyces cerevisiae and identified1,458 modified sites on 369 proteins, revealing an enrichment ofthis modification in the glycolysis/gluconeogenesis pathway.

ResultsH4K8hib Is Dynamically Regulated by the Availability of a CarbonSource. Histone Khib has been reported as a dynamic markenriched in active chromatin, but its regulatory mechanism re-mains elusive (10). To identify potential modulators and dissectits functions, we monitored changes of this modification underdifferent stress conditions using the H4K8hib-specific antibody

Significance

DNA–histone complexes are packed into the eukaryotic genomeand fit into the nucleus of the cell. One mechanism to access thegenetic information is to disrupt the complexes through post-translational modification of histones. Recently, histone H4K82-hydroxyisobutyrylation (H4K8hib) was identified as an evolution-arily conserved active mark. However, how this modification isregulated and what are the enzymes to modulate this modificationwithin a cell remain a mystery. In this study, we discover that thismodification is regulated by the availability of a carbon source inSaccharomyces cerevisiae and identify the enzymes catalyzing theremoval of this active mark in vivo. This discovery provides insightinto the function and regulation of the histone mark H4K8hib.

Author contributions: Y.Z., X.Q., and J. Dai designed research; J.H., Z.L., W.Y., Q.C., H.H.,and J. Dong performed research; J.H., Z.L., W.Y., H.H., Q.W., Y.Z., X.Q., and J. Dai analyzeddata; and J.H., Z.L., and J. Dai wrote the paper.

Conflict of interest statement: Y.Z. is on the science advisory board of PTM Biolabs.

This article is a PNAS Direct Submission.1J.H., Z.L., and W.Y. contributed equally to this work.2To whom correspondence may be addressed. Email: qianxiaohong@mail.ncpsb.org orjbdai@tsinghua.edu.cn.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1700796114/-/DCSupplemental.

8782–8787 | PNAS | August 15, 2017 | vol. 114 | no. 33 www.pnas.org/cgi/doi/10.1073/pnas.1700796114

Dow

nloa

ded

by g

uest

on

Mar

ch 2

0, 2

021

that was developed, characterized, and reported in a previousstudy (10). We found that H4K8hib exhibited no or little responseupon most treatments such as DNA damage, temperature, andredox stresses (Fig. 1A). In contrast, we detected significant re-duction of H4K8hib after cells were incubated in water (Fig. 1 Aand B). During water treatment, cells were challenged by mul-tiple stresses including osmotic pressure and severe nutrientstarvation. Since 1 M sodium chloride (NaCl, osmotic stress) andsynthetic defined medium lacking nitrogen (SD-N, nitrogenstarvation) treatment had little effect on H4K8hib (Fig. 1A), weasked whether glucose deprivation might be the cause. As shownin Fig. 1B, treating the cells in synthetic complete mediumlacking glucose (SC-D) resulted in a decreased H4K8hib levelcomparable to that after water treatment. In addition, supple-menting glucose in both water and SC-D could restore theamount of H4K8hib (Fig. 1B). Together, these results stronglyargue that the presence of glucose in the medium is the majorfactor required to maintain a normal level of H4K8hib.To study the dynamics of this modification, we monitored the

change of the H4K8hib level during water treatment. As shown inFig. 1C, the amount of H4K8hib decreased gradually and waslargely eliminated after 4 h, suggesting that H4K8hib is a relativelystable mark. On the other hand, the H4K8hib level recoveredquickly in response to glucose supply, and it took only 30 min torestore H4K8hib completely (Fig. 1C). This observation correlateswell with the glucose level in the cell, since it has been shown thatthe cellular glucose level decreases gradually during glucose star-vation, but adding glucose back into the medium could lead to afaster increase of cellular glucose level (27). Therefore, we pro-posed that H4K8hib, as a histone mark, orchestrates glucose levelwithin the cells with chromatin epigenetic state regulation.S. cerevisiae preferentially uses fermentable sugars (such as

glucose and fructose), but it can also use nonfermentable sub-strates (such as glycerol and ethanol) as sole energy and carbonsources. To test whether other carbon sources can also regulate

H4K8hib, we supplied different sugars back into the SC-D me-dium and monitored the recovery of H4K8hib, respectively. Asshown in Fig. 1D, not only glucose, but also fructose, is able torestore the H4K8hib level rapidly after glucose starvation. Incontrast, glycerol, ethanol, and even galactose that is ferment-able but a “secondary” carbon source failed to rescue H4K8hib,suggesting that a preferred fermentable sugar is required for thismodification. Taken together, these data indicate that H4K8hib isa dynamic modification directly regulated by glucose/fructoseavailability and establishes a link between histone modificationand carbon metabolism.

Glycolysis Is Required to Restore H4K8hib but Dispensable for ItsEstablishment and Maintenance. Given that H4K8hib is a type ofglucose-regulated modification, we next asked how glucose isable to modulate H4K8hib. Glucose and fructose are preferen-tially used by many unicellular organisms since they can directlyenter the glycolytic pathway. Therefore, we hypothesized that thepresence of an intact glycolysis pathway is essential for thismodification. Mutations of genes encoding two key enzymes,PFK1 and FBA1, in the pathway were used to test whether therapid recovery of this modification is impaired. PFK1 encodes asubunit of phosphofructokinase that catalyzes the formation offructose 1,6-biphosphate from fructose 6-phosphate (28). FBA1encodes the enzyme converting fructose 1,6-biphosphate intotwo 3-carbon molecules in glycolysis and is essential for sur-vival of yeast (29). Consistent with our hypothesis, deletion ofPFK1 completely blocked regeneration of H4K8hib after re-plenishment of glucose for 30 min (Fig. 1E). It should be notedthat a similar amount of H4K8hib could be identified in bothpfk1Δ and WT strains when they are cultured in YPD medium(Fig. 1E). This observation is consistent with the previous ob-servation that the pfk1 mutant strain still can grow on theglucose-containing medium, depending on the residual fermen-tative activity of Pfk2p in yeast (28, 30), indicating that a fullyfunctional glycolysis pathway is required for quick recovery of

A B

C D

E F

Fig. 1. H4K8hib is dynamically affected by the glucose and glycolysis pathway. (A) H4K8hib level under different stress conditions. The WT (BY4741) cells werefirst cultured in YPD medium to log phase, collected, washed three times with sterile water, and then transferred to each stress condition for 4 h. NaCl, methylmethanesulfonate (MMS), DTT, and Benomyl were added to the YPD medium at the indicated concentration. (B) The restoration of H4K8hib level by glucosealone after water or SC-D treatment. WT (BY4741) cells at log phase were washed three times with sterile ddH2O and then suspended in water or SC-D for 4 h,and harvested directly or after adding 2% glucose to the starved cells for 30 min. (C) The H4K8hib level during water treatment and resuming process. WT(BY4741) cells were treated with water for a different time, and 2% glucose was added after treatment in water for 4 h. (D) Supplying cells with glucose andfructose, but not other carbon sources, can restore H4K8hib level quickly. WT (BY4741) yeast cells were treated with SC-D for 4 h, and then different carbonsources were added to a final concentration of 2% to treat the cell for 30 min. Eth, ethanol; Fru, fructose; Gal, galactose; Glu, glucose; Gly, glycerol. (E)Deletion of PFK1 blocks restoration of the H4K8hib level upon glucose replenishment. The WT (BY4741) and pfk1Δ cells were first cultured in YPD medium tolog phase and then transferred into SC-D medium for 4 h; the glucose was added last and treated for 30 min. (F) The fba1-ts mutant failed to restore theH4K8hib level at restrictive temperature. The WT (BY4741) and fba1-ts mutant was grown at 25 °C in YPD medium and then shifted to 37 °C or washed withsterile ddH2O three times and suspended in water at 37 °C for 4 h. Glucose was added to starved cells, and cells were harvested after 30 min.

Huang et al. PNAS | August 15, 2017 | vol. 114 | no. 33 | 8783

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Mar

ch 2

0, 2

021

H4K8hib after adding glucose. Similarly, using a temperature-sensitive mutant (fba1-ts), glucose failed to restore H4K8hibmodification when cells were shifted to the restrictive tempera-ture (37 °C) simultaneously (Fig. 1F). The loss of the H4K8hibsignal at 37 °C results from the inactivation of Fba1p since underpermissive temperature (25 °C) the signal could be restored asthat in the WT strain (Fig. S1). Therefore, we concluded that thefast regeneration of H4K8hib modification after glucose starva-tion depends on the glycolysis pathway.

Rpd3p and Hos3p Are Required for the Removal of H4K8hib DuringGlucose Starvation. To identify enzyme(s) that could remove the2-hydroxyisobutyryl group from H4K8, we reasoned that the lossof the de-2–hydroxyisobutyryation enzyme would lead to an in-crease of H4K8hib. However, and unfortunately, after screeningthe 10 HDAC deletion mutants in the Yeast Knockout (YKO)Collection, none of them was able to increase the H4K8hib signal,suggesting that either they were not able to remove this modi-fication or the increased modification is not detectable (Fig.S2A). As shown above, treating the cells with water led to a fastdecrease of the H4K8hib signal, indicating that the de-2–hydroxyisobutyryation enzyme(s) must be required during thisprocess. Therefore, we treated the HDAC deletion strains withwater, looking for failure to decrease H4K8hib signal. However,none of the HDAC mutants alone could completely block theremoval of H4K8hib (Fig. S2B). However, among these mutants,we found that the signal was reduced slightly but reproducibly inrpd3Δ (Fig. 2A). Therefore, we hypothesized that multiple en-zymes might be functioning redundantly to remove this modifi-cation, one of which may be Rpd3p. To test this hypothesis, wedeleted RPD3 in each HDAC knockout strain, respectively, togenerate the double mutants. Among these mutants, we foundthat most of them—except the rpd3Δ hos3Δ double mutant—stillfailed to prevent the decrease of the H4K8hib signal (Fig. 2A andFig. S2C). In this strain, compared with that of each single mu-tant, the amount of H4K8hib remained constant upon watertreatment (Fig. 2A). These data strongly suggest that Rpd3 andHos3p orchestrate de-2–hydroxyisobutyryation in H4K8. In ad-dition, based on the H4K8hib signal from each single mutant, it islikely that Rpd3 is the major player during this process.There are three Rpd3p-containing complexes, Rpd3μ, Rpd3L,

and Rpd3S, in budding yeast. To further test which Rpd3 complex isspecifically responsible for this modification, we deleted severalcomplex-specific genes in the hos3Δ strain, respectively, includingRco1p for the Rpd3S complex, Sds3p for the Rpd3L complex, andSnt2p for the Rpd3μ complex (Fig. S2D) and then monitoredchanges in H4K8hib level (31, 32). We found that H4K8hib de-creased in all three strains upon water treatment, similar to that ofWT (Fig. S2E), suggesting that deletion of a single Rpd3 complex(in combination with hos3Δ) is not enough to prevent the removalof H4K8hib modification.In addition to these complex-specific proteins, Sin3p is a core

subunit of both Rpd3L and Rpd3S complexes (31). We deletedSIN3 in the hos3Δ strain and found that the H4K8hib level wasresistant to glucose deprivation in the double mutant (Fig. 2B),similar to what was observed in the rpd3Δ hos3Δ strain. This resultindicates that either Rpd3L or Rpd3S are sufficient to removeH4K8hib and that inactivation of both complexes simultaneously isrequired to prevent the removal of H4K8hib. In addition, we foundthat an Rpd3p mutant (H150A, H151A), which compromisesHDAC activity (33), could not remove 2-hydroxyisobutrylation fromH4K8 when the cells were under glucose deprivation (Fig. 2C). Thisresult not only indicated that a functional Rpd3 is required toremove 2-hydroxyisobutyrylation but also suggested that thedeacetylase and de-2–hydroxyisobutyrylase activities share a com-mon active center. In summary, our data reveal that both Rpd3pand Hos3p are required for the decrease of the H4K8hib level

during glucose starvation and that they orchestrate the catalyzationof histone de-2–hydroxyisobutyrylation reaction in vivo.

H4K8A Alters Transcription of Carbon Transport/Metabolism Genes andReduces the CLS in S. cerevisiae. To dissect the function of H4K8hib,we first mutated the lysine to alanine to eliminate 2-hydrox-yisobutyrylation at this site. Phenotypic analysis of H4K8A showedalmost no difference between WT and the H4K8A mutant underseveral conditions (Fig. S3), consistent with a previous report (34).Since glucose deprivation could affect the H4K8hib level (Fig. 1 Aand B), we tested whether this mutant would behave differently onmedia containing a different amount of glucose. To our surprise, noalteration in growth was observed between the mutant and WT (Fig.3A). In addition, we found that the rate of glucose consumptionduring cultivation was also similar between WT and the H4K8Amutant, as shown in Fig. 3B. These data indicate that the H4K8hibmodification does not affect glucose utilization in yeast. On the otherhand, we also traced the change in the H4K8hib level during culti-vation and found that the H4K8hib level decreased gradually duringthe process and was lost entirely when entering the stationaryphase (Fig. 3B). This prompted us to ask whether H4K8hib wouldaffect the physiological state of cells in the stationary phase; thephysiology state is known to be closely related to CLS in yeast (35).Based on the close connection between CLS and oxidative

resistance of stationary-phase cells (35), we at first tested thesensitivity of WT and H4K8A mutant cells in the stationaryphase to oxidative stress and found that the H4K8A mutant was

A

B

C

Fig. 2. Rpd3p and Hos3p are required for the decrease of H4K8hib levelduring glucose starvation. (A) Water treatment did not decrease H4K8hib inthe rpd3Δhos3Δ strain. The indicated strains were first cultured in YPD me-dium to log phase and then treated with sterile ddH2O for 4 h. The yeastcells with or without water treatment were collected, and Western blot (WB)was performed. (B) Deletion of HOS3 combined with disrupting theRpd3 complex by deleting SIN3 disabled cells to catalyze H4K8 de-2–hydroxyisobutyrylation. The experiment was done similarly to that in A.(C) The inactive form of Rpd3p (Rpd3p-H150A H151A) could not mediateH4K8 de-2–hydroxyisobutyrylation upon water treatment. The rpd3Δ hos3Δstrain was transformed with corresponding plasmids containing WT or aninactive form of Rpd3p. The vector was also transformed as a control.

8784 | www.pnas.org/cgi/doi/10.1073/pnas.1700796114 Huang et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

0, 2

021

more sensitive (Fig. 3C). This result suggested that the H4K8Amutant might have a reduced CLS. Using a standard chrono-logical aging assay (36), we confirmed that CLS in the H4K8Amutant was significantly shorter than WT (Fig. 3D), suggestingthat the modification of the H4K8 site might have an importantfunction in chronological aging.To better understand the change in cellular metabolism and

function for the H4K8A mutant, we performed a transcriptomeanalysis using RNA-seq. A total of 236 genes were identified asdifferentially expressed between H4K8A and WT (P < 0.05, DatasetS1). Of these genes, 26 were up-regulated over twofold in H4K8Aversus WT, whereas 23 were down-regulated. Interestingly, we foundthat genes involved in the carbohydrate metabolic process (P =0.00019) and carbohydrate transport process (P = 0.00098) werehighly enriched. Particularly, genes in glucose transportation, treha-lose metabolism, and gluconeogenesis were repressed, whereasgenes in fatty acid β-oxidation and amino acid deamination were up-regulated. These data are consistent with cellular responses duringglucose starvation, inferring that alternative pathways are used,presumably, to provide more intermediates for glycolysis and thetrichloroacetic acid (TCA) cycle to ensure energy supply (Fig. 3E).Putting all data together, we proposed a regulatory and func-

tional linkage among glucose metabolism, H4K8hib, and CLS asshown in Fig. 3F. When glucose is present, the 2-hydroxyisobutyryl-CoA is abundant and a high H4K8hib level is maintained; whenglucose is absent, the Rpd3p and Hos3p act together to reduce theH4K8hib level to orchestrate the glucose level, which signals the cellsto change their transcriptome and expedite chronological aging.

The Lysine 2-Hydroxyisobutyrylation Proteome Is Intricately Interlinkedto Glucose Metabolism in S. cerevisiae. The identification of H4K8hibas a glucose-responsive epigenetic element prompts us to ask

whether a relationship between Khib and glucose metabolism existsat the proteome level. To map the Khib proteome in S. cerevisiae,an enrichment-based method was applied (Fig. 4A), using Khibpan-antibody–conjugated beads to collect the modified peptides,which were subsequently analyzed by HPLC-MS/MS. Using thismethod, we identified 1,458 Khib sites on 369 proteins (DatasetS2). Further bioinformatics analysis of the modified proteins re-veals a strong enrichment in the ribosome and glycolysis/glyco-genesis pathways, suggesting a possible function of Khib inregulating cellular glucose metabolism (Fig. 4 B and C). In addi-tion, the modified proteins were also enriched in the aminoacyl-tRNA biosynthesis pathway and in some amino acid metabolismpathways, which indicates a possible function of this modificationin coordinating carbon metabolism with nitrogen metabolism.Lysine acetylome and succinylome have been studied exten-

sively in S. cerevisiae recently. These two types of acylation and2-hydroxyisobutyrylation have many features in common. Forexample, they are evolutionarily conserved and can use theircorresponding acyl-CoAs as cofactors for modifying lysine(16, 37). By comparing the published data with ours, we foundthat 206 proteins were modified by all three acylations, sug-gesting a comprehensive cross-talk among them (Fig. 4D andDataset S3). The Kyoto Encyclopedia of Genes and Genomes(KEGG) pathway enrichment analysis of the 206 proteins re-veals a close relationship among the three acylations with ri-bosome and glycolysis/gluconeogenesis pathways, which furtherenhances the correlation of these three acylations with glucosemetabolism (Fig. 4E) (38, 39). Although many proteins aremodified by these three modifications, there are some proteinsthat are modified by only one or two types of acylations (Fig.4D). Interestingly, we found that each part of the electrontransfer chain complex I–V contains at least one protein that is

A

B

C

F

D

E

Fig. 3. Modification of H4K8 is required for CLS. (A) The H4K8 mutation does not affect growth in different glucose concentrations. The log-phase cells werespotted onto different plates in a 10-fold series dilution. (B) The dynamics of H4K8hib and glucose level in medium during normal culture condition. Theovernight-cultured BY4741 cells were diluted to OD600 = 0.1 using YPD medium, and cells were collected at each time point. Growth curve and glucoseconcentration data are represented as mean ± SEM. (C) Stationary-phase H4K8A mutant is more sensitive to H2O2 stress. Cells cultured in Sc medium forchronological aging assay at day 3 were washed with sterile water twice and then suspended in a 0.1-M K3PO4 (pH 6.0) buffer containing the indicatedconcentration of H2O2 for 1 h. Finally, the treated cells were spotted onto YPD plate in a 10-fold series dilution. (D) The H4K8A mutation leads to a shortenedCLS. The survival rate is represented as mean ± SEM. (E) Transcriptome changes in the carbon transport and metabolism process in the H4K8A mutant. Thedown-regulated genes are in purple; up-regulated genes are in red. (F) A model for the actions of H4K8 site modifications.

Huang et al. PNAS | August 15, 2017 | vol. 114 | no. 33 | 8785

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Mar

ch 2

0, 2

021

2-hydroxyisobutyrylated specifically (Fig. S4A). Site-directedmutagenesis analysis showed that these sites are critical fornormal yeast growth (Fig. S4B).Given that histone H4K8hib correlates with the availability of

glucose, we tested if the entire 2-hydroxyisobutyrylation pro-teome was affected similarly using Western blot and stable iso-tope labeling by amino acid in cell culture (SILAC). We foundthat nearly all enzymes involved in the glycolysis pathway were 2-hydroxyisobutyrylated (Fig. S5A), and the Khib level was differ-entially regulated under different glucose concentrations (Fig. S5B and C and Dataset S4). Therefore, by analyzing the lysine 2-hydroxyisobutyrylation proteome, we revealed an intricate linkbetween Khib and glucose metabolism.

DiscussionProtein posttranslational modification is a key regulatory mecha-nism used by the cell to fine-tune protein functions. Recently, manynew acylation forms have been identified on histones (6, 8, 10, 21).Compared with the functions of histone N-tail acetylation, verylittle is known about the function and regulation of these newmodifications. Using S. cerevisiae as a model organism, we dem-onstrated that the newly identified histone H4K8hib is a histonemark responsive to carbon starvation. We further showed that onlythe preferred carbon sources (glucose and fructose) could restorethe modification rapidly through a pathway depending on glycol-ysis. And interestingly, the fully active glycolysis pathway is notrequired for its maintenance in YPD medium. The selectivity ofcarbon source and glycolysis-dependent rapid restoration of thismodification further strengthen the previously proposed link be-tween protein acylations and metabolism (3, 12, 40, 41), suggestinga delicate active regulation of these new acylation forms by cells.Previous findings have shown that glucose availability affects

histone acetylation (42, 43), which very likely affects the pro-duction and cellular concentration of acetyl-CoA and, therefore,influences histone acetylation and cell proliferation or differen-tiation (44–46). Similarly, it is possible that the decreasedH4K8hib level as glucose deprivation may result from the re-duced production of 2-hydroxyisobutyl-CoA, the potential donorfor Khib. Given that glucose metabolism is the center of carbonand energy metabolism in a cell and the closed relationship be-tween acy-CoA and energy metabolism (12), other acylation may

also be influenced in a similar manner. However, other possi-bilities, such as regulation of the activities of acyltransferase ordeacylase, cannot be ruled out, and more efforts are needed todetail the regulation mechanism. Overall, the stress responseof H4K8hib gives us a good model to explore the regulation andfunction details of these new histone acylations.HPLC-MS/MS–based proteome analysis is a widely used

method for dissecting the possible functions of protein modifica-tion. The proteome study of lysine acetylation and succinylation(Ksucc) suggests a broad function in cellular metabolism and sig-naling (7–9, 11). Since all types of acylation require the corre-sponding acyl-CoAs as donors, it is not surprising that a stronglinkage existed between these new acylations and metabolism (3,12, 40, 41). However, many of the metabolic enzymes are heavilyacylated, making it difficult to study the exact functions of indi-vidual modification. Given that the H4K8hib is also regulated byglucose availability and the glycolysis pathway, we speculate thatthe link between Khib and glucose metabolism exists at both epi-genetic and proteome levels. A similar link may also exist for othertypes of lysine acylations, such as succinylation and crotonylation.Our study shows that the removal of H4K8hib during glucose

starvation requires both Rpd3p (class I HDACs) and Hos3p (classII HDACs) in budding yeast. Consistently, the mammalianHDAC3 (class I HDACs) was capable of carrying out the de-2–hydroxyisobutyrylation reaction in vitro (10). In addition, severalstudies have demonstrated that Sirt5 (class III) is a de-succinylase,de-malonylase, and de-glutarylase both in vitro and in vivo (7, 11,26). Furthermore, at least one HAT, p300, in human cells hasbeen reported as a propionylase, butyrylase, and crotonylase (6,24). Together, these results suggest that many HATs and HDACsmight function promiscuously and catalyze different acylations.How cells decide to add or remove a particular type of acyl

group onto or from the same residue and differentiate them willbe of great interest. A possible explanation is that this will help adelicate regulation of gene expression in response to the complexmetabolic state of the cell and environmental nutrient change.Since the different acyl groups have distinct structures and chargestates, a differential transcription response is expected based onthe known epigenetic models (23). As reported previously, theH4K8hib may be a better indicator of high transcriptional activity,and an additive effect is seen between H4K8hib and H4K8ac. The

A B

C

D

E

BY4741 cells at log phase in YPD medium

Total protein extrac�on

Tryp�c diges�on

Khib pep�de enrichment

HPLC-MS/MSIden�fying Khib pep�des

Fig. 4. Landscape of the lysine 2-hydroxyisobutyrylation proteome. (A) Schematic representation of the workflow used for HPLC-MS/MS–based Khib siteidentification in S. cerevisiae. The log-phase WT (BY4741) yeast cells cultured in the YPD medium were harvested and lysed mechanically. The total proteinsextracted were then trypsin-digested, and the 2-hydroxyisobutyrylated peptides were enriched using Khib pan-antibody–conjugated beads. The purifiedpeptides were analyzed by HPLC-MS/MS. (B) Gene ontology (GO) enrichment analysis of Khib proteins and all yeast proteins using Saccaromyces GenomeDatabase (SGD) GO Term Finder. The top five enriched GO terms are shown with their corresponding P value. All of the enriched GO terms can be found inDataset S2. (C) KEGG pathway enrichment analysis of the Khib proteins using DAVID software. All pathways with a P value < 0.05 are shown. Detailed in-formation can be found in Dataset S2. (D) Venn diagram showing the overlaps among Kac, Ksucci, and Khib proteins. The Kac and Ksucci proteome data are fromtwo previously published papers (19, 37). (E) KEGG pathway enrichment analysis of the common proteins with Kac, Ksucci, and Khib using DAVID software. Thetop three enriched pathways are shown. Detailed information can be found in Dataset S3.

8786 | www.pnas.org/cgi/doi/10.1073/pnas.1700796114 Huang et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

0, 2

021

cellular crotonyl-CoA regulates histone crotonylation throughp300, and the histone crotonylation up-regulates transcription to agreater extent than histone acetylation (10, 24). It will also beimportant to evaluate whether some specific “readers” exist todistinguish different acylation forms.Our data clearly show a correlation between H4K8A and re-

duced CLS. However, we cannot attribute the effect to a particularmodification since H4K8 is known to be acetylated as well.Acetylation on H4K8 (H4K8ac) recruits the SWI/SNF complex, achromatin-remodeling complex involved in gene transcription,and correlates with opening the chromatin domain (47, 48).Similarly, H4K8hib is also reported as an active marker (10),suggesting a general role of H4K8 modification in transcriptionactivation. In addition, both H4K8ac and H4K8hib decreased afterentering the stationary phase (42) and are closely related to glu-cose metabolism, which implies that the calorie restriction af-fecting CLS may be partly through modifying the modifications onH4K8. Identifying the “readers” that distinguish these acylationforms will help to dissect the functions of these histone acylations.

Materials and MethodsStrains and Antibodies Used in This Study. Strains used in this study are listed inTable S1, except those from the yeast YKO library (49). Standard methods forgene disruption and transformation were applied. Both H4K8hib and pan-Khib antibodies were purchased from PTM Biolabs (catalog no. PTM-805 for

H4K8hib antibody, catalog no. PTM-804 for the pan-Khib antibody). TheH4K8Ac antibody was purchased from Abcam (catalog no. ab15823). Allantibodies were used according to the product manuals.

Stress Resistance Assay and Chronological Life Span Assay. Both assays wereconducted by following the protocols in previously reports (36). A detaileddescription of the assays is included in SI Materials and Methods.

Stable Isotope Labeling and Mass Spectrometry Analysis. The yeast strainZLY018 (arg4::kanMX4 leu2Δ0 lys2Δ0 ura3Δ0 his3Δ1) was inoculated in syntheticcomplete medium containing 2% glucose plus arginine (arg) (13C6) and lysine (lys)(13C6) or in synthetic complete medium containing 0.2% glucose plus arg (12C6)and lys (12C6). After 24 h, the cells were reinoculated into the same medium andgrown until log phase before harvesting and subsequently disrupted with glassbeads. The total proteins were isolated by 20% TCA and digested by trypsin. The2-Hydroxyisobutyrylated peptides were enriched with pan-anti-Khib–conjugatedresin (PTM BioLabs, catalog no. PTM-804) and subjected to HPLC-MS/MS analysis.

ACKNOWLEDGMENTS.We thank the following agencies for financial support:the National Key Research and Development Program of China (Grant2017YFA0505103); the Research Fund for the Doctoral Program of HigherEducation of China (Grant 20120002110022); the National Key Program forBasic Research of China (Grants 2012CB910603 and 2014CBA02001); and theNational Natural Science Foundation of China (Grants 31471254, 81530021,31100591, and 21235001). Y.Z. was supported by National Institutes of HealthGrants GM105933, DK107868, and GM115961.

1. Glozak MA, Sengupta N, Zhang X, Seto E (2005) Acetylation and deacetylation ofnon-histone proteins. Gene 363:15–23.

2. Shiloh Y, Ziv Y (2013) The ATM protein kinase: Regulating the cellular response togenotoxic stress, and more. Nat Rev Mol Cell Biol 14:197–210.

3. Choudhary C, Weinert BT, Nishida Y, Verdin E, MannM (2014) The growing landscape oflysine acetylation links metabolism and cell signalling. Nat Rev Mol Cell Biol 15:536–550.

4. Huang H, Sabari BR, Garcia BA, Allis CD, Zhao Y (2014) SnapShot: Histone modifica-tions. Cell 159:458–458 e451.

5. Bond MR, Hanover JA (2015) A little sugar goes a long way: The cell biology ofO-GlcNAc. J Cell Biol 208:869–880.

6. Chen Y, et al. (2007) Lysine propionylation and butyrylation are novel post-translational modifications in histones. Mol Cell Proteomics 6:812–819.

7. Peng C, et al. (2011) The first identification of lysine malonylation substrates and itsregulatory enzyme. Mol Cell Proteomics 10:M111 012658.

8. Tan M, et al. (2011) Identification of 67 histone marks and histone lysine crotonylationas a new type of histone modification. Cell 146:1016–1028.

9. Zhang Z, et al. (2011) Identification of lysine succinylation as a new post-translationalmodification. Nat Chem Biol 7:58–63.

10. Dai L, et al. (2014) Lysine 2-hydroxyisobutyrylation is a widely distributed active his-tone mark. Nat Chem Biol 10:365–370.

11. Tan M, et al. (2014) Lysine glutarylation is a protein posttranslational modificationregulated by SIRT5. Cell Metab 19:605–617.

12. Lin H, Su X, He B (2012) Protein lysine acylation and cysteine succination by inter-mediates of energy metabolism. ACS Chem Biol 7:947–960.

13. Colak G, et al. (2013) Identification of lysine succinylation substrates and the succinyla-tion regulatory enzyme CobB in Escherichia coli. Mol Cell Proteomics 12:3509–3520.

14. Park J, et al. (2013) SIRT5-mediated lysine desuccinylation impacts diverse metabolicpathways. Mol Cell 50:919–930.

15. Rardin MJ, et al. (2013) SIRT5 regulates the mitochondrial lysine succinylome andmetabolic networks. Cell Metab 18:920–933.

16. Weinert BT, et al. (2013) Lysine succinylation is a frequently occurring modification in pro-karyotes and eukaryotes and extensively overlaps with acetylation. Cell Reports 4:842–851.

17. Nishida Y, et al. (2015) SIRT5 regulates both cytosolic and mitochondrial proteinmalonylation with glycolysis as a major target. Mol Cell 59:321–332.

18. Goudarzi A, et al. (2016) Dynamic competing histone H4 K5K8 acetylation and bu-tyrylation are hallmarks of highly active gene promoters. Mol Cell 62:169–180.

19. Li Y, et al. (2016) Molecular coupling of histone crotonylation and active transcriptionby AF9 YEATS domain. Mol Cell 62:181–193.

20. Xie Z, et al. (2016) Metabolic regulation of gene expression by histone lysineβ-hydroxybutyrylation. Mol Cell 62:194–206.

21. Xie Z, et al. (2012) Lysine succinylation and lysine malonylation in histones. Mol CellProteomics 11:100–107.

22. Luger K, Mäder AW, Richmond RK, Sargent DF, Richmond TJ (1997) Crystal structureof the nucleosome core particle at 2.8 Å resolution. Nature 389:251–260.

23. Kouzarides T (2007) Chromatin modifications and their function. Cell 128:693–705.24. Sabari BR, et al. (2015) Intracellular crotonyl-CoA stimulates transcription through

p300-catalyzed histone crotonylation. Mol Cell 58:203–215.25. Cheng Z, et al. (2009) Molecular characterization of propionyllysines in non-histone

proteins. Mol Cell Proteomics 8:45–52.26. Du J, et al. (2011) Sirt5 is a NAD-dependent protein lysine demalonylase and de-

succinylase. Science 334:806–809.

27. Bermejo C, Haerizadeh F, Takanaga H, Chermak D, FrommerWB (2010) Dynamic analysisof cytosolic glucose and ATP levels in yeast using optical sensors. Biochem J 432:399–406.

28. Arvanitidis A, Heinisch JJ (1994) Studies on the function of yeast phosphofructokinasesubunits by in vitro mutagenesis. J Biol Chem 269:8911–8918.

29. Schwelberger HG, Kohlwein SD, Paltauf F (1989) Molecular cloning, primary structureand disruption of the structural gene of aldolase from Saccharomyces cerevisiae. Eur JBiochem 180:301–308.

30. Lloyd D, James CJ, Maitra PK (1992) Mutations in phosphofructokinases alter the controlcharacteristics of glycolysis in vivo in Saccharomyces cerevisiae. Yeast 8:291–301.

31. Carrozza MJ, et al. (2005) Histone H3 methylation by Set2 directs deacetylation ofcoding regions by Rpd3S to suppress spurious intragenic transcription. Cell 123:581–592.

32. Baker LA, et al. (2013) The yeast Snt2 protein coordinates the transcriptional responseto hydrogen peroxide-mediated oxidative stress. Mol Cell Biol 33:3735–3748.

33. Kadosh D, Struhl K (1998) Histone deacetylase activity of Rpd3 is important fortranscriptional repression in vivo. Genes Dev 12:797–805.

34. Dai J, et al. (2008) Probing nucleosome function: A highly versatile library of synthetichistone H3 and H4 mutants. Cell 134:1066–1078.

35. Fabrizio P, Longo VD (2003) The chronological life span of Saccharomyces cerevisiae.Aging Cell 2:73–81.

36. Wei M, et al. (2008) Life span extension by calorie restriction depends on Rim15 andtranscription factors downstream of Ras/PKA, Tor, and Sch9. PLoS Genet 4:e13.

37. Henriksen P, et al. (2012) Proteome-wide analysis of lysine acetylation suggests its broadregulatory scope in Saccharomyces cerevisiae. Mol Cell Proteomics 11:1510–1522.

38. Kanehisa M, Goto S (2000) KEGG: Kyoto Encyclopedia of Genes and Genomes. NucleicAcids Res 28:27–30.

39. Huang W, Sherman BT, Lempicki RA (2009) Systematic and integrative analysis oflarge gene lists using DAVID bioinformatics resources. Nat Protoc 4:44–57.

40. He W, Newman JC, Wang MZ, Ho L, Verdin E (2012) Mitochondrial sirtuins: Regulatorsof protein acylation and metabolism. Trends Endocrinol Metab 23:467–476.

41. Newman JC, He W, Verdin E (2012) Mitochondrial protein acylation and intermediary metabo-lism: Regulation by sirtuins and implications for metabolic disease. J Biol Chem 287:42436–42443.

42. Friis RM, et al. (2009) A glycolytic burst drives glucose induction of global histoneacetylation by picNuA4 and SAGA. Nucleic Acids Res 37:3969–3980.

43. Friis RM, et al. (2014) Rewiring AMPK and mitochondrial retrograde signaling for metaboliccontrol of aging and histone acetylation in respiratory-defective cells. Cell Reports 7:565–574.

44. Takahashi H, McCaffery JM, Irizarry RA, Boeke JD (2006) Nucleocytosolic acetyl-coenzyme asynthetase is required for histone acetylation and global transcription.Mol Cell 23:207–217.

45. Cai L, Sutter BM, Li B, Tu BP (2011) Acetyl-CoA induces cell growth and proliferationby promoting the acetylation of histones at growth genes. Mol Cell 42:426–437.

46. Moussaieff A, et al. (2015) Glycolysis-mediated changes in acetyl-CoA and histone acety-lation control the early differentiation of embryonic stem cells. Cell Metab 21:392–402.

47. Chen WY, Townes TM (2000) Molecular mechanism for silencing virally transduced genes in-volves histone deacetylation and chromatin condensation. Proc Natl Acad Sci USA 97:377–382.

48. Agalioti T, Chen G, Thanos D (2002) Deciphering the transcriptional histone acetyla-tion code for a human gene. Cell 111:381–392.

49. Winzeler EA, et al. (1999) Functional characterization of the S. cerevisiae genome bygene deletion and parallel analysis. Science 285:901–906.

50. Kim SC, et al. (2006) A clean, more efficient method for in-solution digestion ofprotein mixtures without detergent or urea. J Proteome Res 5:3446–3452.

51. Cox J, Mann M (2008) MaxQuant enables high peptide identification rates, in-dividualized p.p.b.-range mass accuracies and proteome-wide protein quantification.Nat Biotechnol 26:1367–1372.

Huang et al. PNAS | August 15, 2017 | vol. 114 | no. 33 | 8787

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Mar

ch 2

0, 2

021