ITF3756

Histone deacetylase 6 (HDAC6) deacetylates extracellular signal-regulated kinase 1 (ERK1) and thereby stimulates ERK1 activity

ABSTRACT
Histone deacetylase 6 (HDAC6), a class IIb HDAC, plays an important role in many biological and pathological processes. Previously, we found that ERK1, a downstream kinase in the MAPK signaling pathway, phosphorylates HDAC6, thereby increasing HDAC6-mediated deacetylation of α-tubulin. However, whether HDAC6 reciprocally modulates ERK1 activity is unknown. Here, we report that both ERK1 and 2 are acetylated and that HDAC6 promotes ERK1 activity via deacetylation. Briefly, we found that both ERK1 and 2 physically interact with HDAC6. Endogenous ERK1/2 acetylation levels increased upon treatment with a pan-HDAC inhibitor, an HDAC6-specific inhibitor or depletion of HDAC6, suggesting that HDAC6 deacetylates ERK1/2. We also noted that the acetyltransferases CBP and p300 both can acetylate ERK1/2. Acetylated ERK1 exhibits reduced enzymatic activity toward the transcription factor ELK1, a well-known ERK1 substrate. Furthermore, mass spectrometry analysis indicated Lys-72 as an acetylation site in the ERK1 N-terminus,adjacent to Lys-71, which binds to ATP, suggesting that acetylation status of Lys-72 may affect ERK1 ATP binding. Interestingly, an acetylation-mimicking ERK1 mutant (K72Q) exhibited less phosphorylation than the WT enzyme and a deacetylation- mimicking mutant (K72R). Of note, the K72Q mutant displayed decreased enzymatic activity in an in vitro kinase assay and in a cellular luciferase assay compared with the WT and K72R mutant. Taken together, our findings suggest that HDAC6 stimulates ERK1 activity. Along with our previous report that ERK1 promotes HDAC6 activity, we propose that HDAC6 and ERK1 may form a positive feed- forward loop, which might play a role in cancer.

Histone deacetylases (HDACs) and histone acetyltransferases (HATs) are the enzymes that regulate core histones and non-histone proteins by deacetylation and acetylation, respectively (1). HATs acetylate proteins via adding acetyl groups to lysine residues, while HDACs catalyze a reverse reaction by removing the acetyl group from lysine residues. Up to now, a total of 18 HDACs are identified in humans and grouped into four classes based on their sequence similarity to yeast orthologs (2). Class I HDACs are homologous to yeast reduced potassium dependency 3 (Rpd3) and include HDACs 1, 2, 3 and 8. Class II HDACs are homologous to yeast histone deacetylase 1 (HdaI) and are further divided into class IIa and class IIb. Class IIa contains HDACs 4, 5, 7 and 9, and class IIb includes HDACs 6 and 10. HDAC11 is the only member in class IV. The deacetylase activity of HDAC classes I, II and IV is zinc-dependent. Class III HDACs, also knowns as sirtuins, are homologous to yeast silent information regulator 2 (Sir2), and the deacetylase activity of this class is oxidized nicotinamide adenine dinucleotide (NAD+)-dependent.HDAC6 belongs to class IIb HDACs. Its structure is quite unique among all HDACs, in that it contains two functional deacetylase domains in tandem and a zinc finger domain in the C-terminus (2,3). HDAC6 participates in numerous biological and pathologic processes, such as cell migration, DNA damage response and oncogenesis, through modulating its substrates (4-7). For example, HDAC6 deacetylates cytoskeleton proteins and their associated proteins, such as -tubulin and cortactin, to regulate cell mobility (4,5). HDAC6 also deacetylates and ubiquitinates the DNA mismatch repair protein MSH2, in order to regulate MutS homeostasis, DNA mismatch repair, and DNA damage response (6). In addition, HDAC6 deacetylates cell signaling regulators K-Ras and -catenin, leading to altered oncogenic activity and nuclear localization, respectively (8,9).

Mitogen-activated protein kinases (MAPKs) are a conserved family of serine/threonine protein kinases connected to various essential cellular processes (10). To date, a total of 14 MAPKs have been isolated in humans, all of which fall into the seven classes that follow: the extracellular signal-regulated kinase class, p38 class, c-Jun N-terminal kinase (JNKs) class, and ERK5 class belong to the conventional group of MAPKs, all of which have been studied extensively, whereas the Nemo-like kinase (NLK) class, ERK3/4 class, and ERK7 class belong to the atypical group of MAPKs, all of which have been studied inadequately (10,11). In general, the MAPK pathway contains at least three tiers: a MAPK kinase kinase (MAP3K), a MAPK kinase (MAP2K), and a MAPK. These MAPK pathways participate in transducing signals from the surface to the interior of the cell. Being triggered by extracellular stimulus, first tier MAP3Ks are activated to phosphorylate MAP2Ks, which subsequently phosphorylate MAPKs. These MAPK pathways have their own unique primary kinases in different tiers, but they also share some minor activators (10,11). All MAPKs, except ERK3/4 and NLK, contain a conserved Thr-X- Tyr motif in their kinase domain. Phosphorylation of both Thr and Tyr residues in this motif is a critical step for MAPK activation (10).
Human ERK1 (also known as MAPK3 or p44MAPK) and ERK2 (also known as MAPK1 or p42MAPK) are 84% identical in sequence; they share many functions (12). Thus they are often referred to as ERK1/2. Among these MAPKs, ERK1/2 are associated with cell proliferation, cell growth, cell mobility, and cell survival (13), and the Ras-Raf (MAP3K)-MEK (MAP2K)-ERK1/2 (MAPK) signal transduction cascade can be activated by growth factors, osmotic stress, and cytokines (11,14,15). To date, more than 160 substrates of ERK1/2 have been discovered from the nucleus and cytosol to the cell membrane (16).

Post-translational modifications (PTMs) have long been documented as a critical means of regulating ERK1/2 activity. Compared with the decades of studies of ERK1/2 phosphorylation, especially at the Thr202/Tyr204 sites in ERK1 and Thr185/Tyr187 in ERK2 of the Thr-X-Tyr motif, the studies of other PTMs of ERK1/2, including acetylation, methylation, and ubiquitination, are just emerging. Recently, two lysine sites at the ERK1 C-terminus, Lys302 and Lys361, have been revealed to be tri-methylated, and methylation of ERK1 enhances its phosphorylation (17). Arg309 of ERK1 has also been reported as a methylation site via a proteome-wide analysis, yet the function of this site is not clear (18). Ubiquitination of ERK1/2 has been reported from three different proteome- wide analyses, but the role of ubiquitination in ERK1/2 still remains elusive (19-21). Likewise, one ERK1 acetylation site has been identified by SILAC assays, but the function of this site still remains to be determined (22). Lately, several reports have shown that when cell lines including A549, MB361, BT474, MV4-11, PC-3, SKBR-3, HN-9 and SQ20B were treated with HDAC6 inhibitors, the level of phosphorylated ERK1/2 decreased (23-26), suggesting that acetylation of ERK1/2 compromises their activities and that HDAC6 inhibition may down-regulate ERK1/2 activities. However, the mechanisms underlying this observation are not clear.Our previous study showed that ERK1 phosphorylates HDAC6 at its Ser1035 site, and phosphorylation of this site increases HDAC6’s activity toward -tubulin and stimulates cell migration (27). In this study, we have determined that ERK1/2 are acetylated proteins, and that ERK1/2 are novel substrates of HDAC6. Both ERK1/2 show the ability to physically interact with HDAC6 in vitro. Also, both CBP and p300 acetylate ERK1/2. One novel acetylation site, Lys72, was identified in ERK1 via mass- spectrometry analysis, and the acetylation- mimicking mutant of ERK1 exhibits reduced kinase activity, suggesting that the acetylation status of ERK1 plays an important role in regulating ERK1 enzymatic activity.

RESULTS
ERK1/2 interact with HDAC6 directly – We previously showed that ERK1/2 interact with HDAC6 endogenously (27). However, how these proteins interact with each other was unknown. To determine whether ERK1/2 interact with HDAC6 directly or through other proteins, we performed in vitro GST pull-down assays with bacterially-purified HDAC6 and ERK1/2. As shown in Figure 1A, GST-HDAC6, but not GST, efficiently pulled down His-ERK1. Similarly, as shown in Figure 1B, GST-HDAC6 also pulled down His-ERK2. Therefore, ERK1/2 physically interact with HDAC6.
Inhibition or depletion of HDAC6 increases ERK1/2 acetylation – Previously, we demonstrated that ERK1 phosphorylates HDAC6 at the Ser1035 site (27). Given the fact that HDAC6 is a deacetylase, we interrogated HDAC6 to see if it deacetylates ERK1, in other words, whether ERK1 is a substrate of HDAC6. To this end, we set out to determine whether ERK1 is acetylated. Before using the anti- acetylated lysine (AcK) antibodies to examine ERK1 acetylation, we tested two commercial anti-AcK antibodies from Cell Signaling Technology. As shown in Figure 2, both antibodies specifically recognized acetylated BSA but not no-acetyalted BSA. We then used these two antibodies in the following experiments. To determine whether ERK1 is acetylated, mammalian expression vector GST-ERK1 was transfected into HEK293T cells. Then the transfected cells were treated with 0, 50, 100, 200, 400, or 600 ng/ml of pan-HDAC inhibitor, Trichostatin A (TSA) 12 hours prior to harvest. As shown in Figure 3A, after normalizing with the total ERK1, the level of acetylated GST- ERK1 was increased as the dosage of TSA was increased. 600 ng/ml TSA increased the level of acetylated GST-ERK1 by four-fold as compared to a vehicle. We then treated the GST-ERK1- transfected HEK293T cells with 600 ng/ml TSA at 0, 2, 4, 8, 12 and 24 hours. As shown in Figure 3B, the level of acetylated GST-ERK1 increased as the TSA treatment time increased. At the 24- hour time point, the longest time point, the level of acetylated GST-ERK1 is nearly two-fold as compared to that at the 0-hour time point. To ensure that TSA is functional (TSA inhibits class I, II, and IV HDACs including HDAC6) we also detected the level of acetylated -tubulin, which is a well-known HDAC6 substrate (4). Using a similar approach, we have shown that ERK2 is acetylated upon TSA treatment (Figure 3C and 3D). Taken together, TSA increases ERK1/2 acetylation in a dose- and time-dependent manner. Next, we set out to determine whether specific inhibition of HDAC6 would increase ERK1/2 acetylation. We treated GST-ERK1- or GST-ERK2-transfected HEK293T cells with an HDAC6-specific inhibitor, ACY-1215, at the concentrations of, 0, 0.5, 1, 2, 4, and 6 g/ml, for 12 hours. As shown in Figures 4, ACY-1215 increased the level of acetylated GST-ERK1 and GST-ERK2 by 9-fold and 4-fold, respectively.

To determine whether endogenous ERK1/2 are acetylated, we treated 293T cells with the pharmacological inhibitor TSA or ACY-1215 and examined the level of acetylation of ERK1/2. As shown in Figure 5, both inhibitors increased the acetylation levels of endogenous ERK1/2, suggesting that inhibition of HDAC6 increases the acetylation of endogenous ERK1/2. To further confirm the role of HDAC6 in regulating ERK1/2 deacetylation, we compared the acetylation level of endogenous ERK1/2 in HDAC6 wild-type and HDAC6 knockout MEFs. As shown in Figure 6A, the acetylation of endogenous ERK1/2 increased significantly in HDAC6 knockout MEFs as compared with HDAC6 wild-type MEFs. Likewise, the level of endogenous ERK1/2 increased in HDAC6 knockdown A549 cells as compared with that of control A549 cells (Figure 6B). These results validate that inhibition or depletion of HDAC6’s enzymatic activity is responsible for the increase of ERK1/2 acetylation.
CBP and p300 acetylate ERK1/2 in vivo and in vitro – To determine which histone acetyltransferase (HAT) acetylates ERK1/2, we tested five HATs, which belong to the following three families: Gcn5-related N-acetyltransferase (GNAT) family (PCAF), MYST family (TIP60 and HBO1), and p300/ CREB-binding protein (CBP) family (p300 and CBP) (28-30).

These HATs were co-expressed with GST-ERK1 in 293T cells. When GST-ERK1 was co-expressed with CBP, the acetylation level of GST-ERK1 was much higher than that with empty vector and other HATs (Figure 7A). p300 also weakly acetylated ERK1 (Figure 7A). To further confirm the results, we co-expressed the increasing amounts of CBP or p300 with GST-ERK1. As shown in Figures 7B and 7C, CBP and p300 acetylated GST-ERK1 in a dose-dependent manner. To further confirm CBP and p300’s effect on endogenous ERK1, we overexpressed p300 in HEK293T cells and tested the acetylation level of endogenous ERK1. As shown in Figure 7D, exogenous p300 increased the acetylation of endogenous ERK1. To eliminate the potential influence of endogenous HATs and HDACs in the cells on ERK1 acetylation, we executed in vitro acetylation assays to confirm that CBP is an ERK1 acetyltransferase. As shown in Figure 7E, recombinant CBP indeed acetylated bacterially purified GST-ERK. Similarly, we have shown that both CBP and p300 promote ERK2 acetylation in cells and that CBP acetylated ERK2 in vitro. (Figure 8A-D).
HDAC6 deacetylates ERK1 in vivo and in vitro – Because of the low basal acetylation level of ERK1, to confirm whether HDAC6 deacetylates ERK1, we co-expressed CBP to increase the ERK1 acetylation level. As shown in Figure 9A and 9B, as expected, both CBP and p300 increased acetylation of GST-ERK1, whereas overexpression of HDAC6 reduced CBP- or p300-mediated ERK1 acetylation. Then we tested whether HDAC6 is able to deacetylate in vitro acetylated GST-ERK1 purified from bacteria. As shown in Figure 9C, HDAC6 purified from 293T cells could efficiently deacetylate acetylated ERK1 in vitro, suggesting that ERK1 is a bona fide substrate of HDAC6.

Acetylation of ERK1 reduces ERK1’s enzymatic activity -We next examined whether acetylation of ERK1 affects its enzymatic activity. Because HDAC6 deacetylates ERK1, we then hypothesize that in the absence of HDAC6, ERK1 acetylation would be increased which may alter ERK1’s enzymatic activity. To test this hypothesis, the ERK1 plasmid was transfected into wild-type 293T cells or HDAC6KO 293T cells. Then ERK1 was isolated from these two type of cells and ERK1’s kinase activity was examined using recombinant ELK1 as a substrate. ELK1 is a member of Ets transcription factor family, and several serine and threonine sites of ELK1 can be phosphorylated by ERKs (31). Among these sites, phosphorylation status of ELK1 Ser383 is pivotal for ELK1 transcriptional activation (32). Because of this reason, we used ELK1 as the substrate to execute non-radioactive kinase assays to measure the ability of ERK1 to phosphorylate the Ser383 site in ELK1. As shown in Figure 10A, ERK1 purified from HDAC6KO cells displayed significantly lower activity than that from wild-type cells. Then we also directly acetylated ERK1 using CBP and investigated the enzymatic activity of vehicle-incubated ERK1 versus CBP-incubated ERK1. As shown in Figure 10B, CBP-acetylated ERK1 harbored lower enzymatic activity. In summary, we concluded that acetylation of ERK1 decreases its enzymatic activity.

ERK1 Lysine 72 is acetylated – To detect the acetylation site of ERK1, we prepared samples for mass spectrometry analyses. GST- ERK1 and CBP were co-expressed in HEK293T cells for 36 hours. Cells were harvested and lysed in lysis buffer. GST-ERK1 was then pulled-down by glutathione-agarose and resolved on SDS- PAGE. The SDS-PAGE gel was stained by coomassie blue, and the specific bands were excised. Samples were further digested with chymotrypsin and Lys-C endoproteinase sequentially and subjected for LC-tandem mass spectrometry analysis. Lys72 was identified as a novel acetylation site of ERK1 (Figure 11A). Lys72 is located on 3-strand of ERK1’s N-lobe and is very close to the glycine-rich loop (Figure 11B). To show the conservation of Lys72 in ERK1, ERK1 sequences from human to nematode were compared by the T-Coffee alignment program. The alignment results showed that this mass spectrometry-identified ERK1 Lys72 was highly conserved among mammals and even in Zebrafish (Danio rerio), Drosophila, and C. elegans (Figure 11C), indicating that Lys72 plays an important role in ERK1 function.Acetylation mimetic mutant of ERK1 abolishes ERK1 kinase activity toward ELK1 – To test whether acetylation status of Lys72 in ERK1 affects its kinase activity, Lys72 was mutated to glutamine (an acetylation mimetic mutant, K72Q) or arginine (a deacetylation mimetic mutant, K72R), and the resulting mutants were tested by their phosphorylation status in cells followed by kinase assays. As shown in Figure 12A, K72Q, but not K72R, displayed a reduced level of phosphorylation as compared with wild-type ERK1, implying that the K72Q mutant exhibits a diminished enzymatic activity. To confirm this notion, the kinase assay using the most thoroughly studied ERK1 substrate, ELK1, was performed. As shown in Figure 12B, the acetylation mimetic mutant, ERK1 (K72Q) displayed a significantly reduced kinase activity toward ELK1 as compared with the deacetylation mimetic mutant, ERK1 (K72R) and the wild-type of ERK1.

To further demonstrate the impact of ERK1- K72 acetylation in vivo, we monitored activity of ELK1 in a reporter assay. The luciferase construct, (ELK1)2-TATA-Luc, being used in this assay was a kind gift from Dr. Manohar Ratnam. In this construct, the cis-element preferred by ELK1 was placed as two tandem repeat elements upstream of a minimal TATA- dependent Firefly luciferase promoter (33). Wild- type, K72Q and K72R of ERK1 were examined by their ability to activate ELK1-mediated transcription by luciferase assays in HeLa cells. The pRL plasmid encoding Renilla luciferase was also transfected in HeLa cells together with the above constructs. The Renilla reading was then used to normalize the Firefly luciferase reading. As shown in Figure 12C, the ELK1-dependent promoter activity is significantly lower in acetylation mimetic mutant ERK1(K72Q) transfected cells than in wild-type or deacetylation mimetic mutant ERK1(K72R) transfected cells, indicating that acetylation at K72 site reduces ERK1’s activity to activate ELK1-mediated transcription.In order to determine how acetylation/deacetylation of Lys72 regulates ERK1 kinase activity, we examined ERK1’s crystal structure. Lys72 is located near the ATP binding site and stabilizes one wall of the ATP binding site via intramolecular contacts. In particular, Lys72 forms a salt bridge with Asp117 and links to Tyr119 with a hydrogen bond. Acetylation mimetic Lys72 would break the contacts to both Asp117 and Tyr119, and might change the conformation of the ATP binding site leading to reduced enzymatic activity of ERK1 (Figures 8C and 8D).

DISCUSSION
In this study, we have demonstrated that the acetylation/deacetylation status of Lys72 in ERK1 regulates its enzymatic activity. For the first time, we have revealed that ERK1 and ERK2 are acetylated proteins and are novel substrates of the deacetylase HDAC6 and the acetyltransferases CBP and p300. We have also discovered that Lys72 of ERK1 is a novel acetylation site. Furthermore, we have shown that the ERK1 Lys72 acetylation-mimicking mutant (K72Q) displayed reduced kinase activity as compared with the wild-type and deacetylation- mimicking mutant (K72R). Overall, our results suggest that HDAC6/CBP and p300 govern ERK1’s kinase activity via deacetylation/acetylation of Lys72.Although we have provided strong evidence that HDAC6 deacetylates ERK1/2, we cannot rule out the possibility that other Class I, II and IV HDACs can deacetylate ERK1/2. In addition, we tested whether Class III HDACs, also called sirtuins, can deacetylate ERK1/2. We found that the Sirtuin inhibitor, nicotinamide, increased ERK2, but not ERK1, acetylation (data not shown), suggesting that one or multiple sirtuins may deacetylate ERK2. Further investigations are warranted to study whether other HDACs and sirtuins regulate ERK1/2.In addition, we have identified a conserved lysine, Lys72, which is adjacent to a critical ATP- binding site, Lys71, as a novel acetylation site in ERK1, and the acetylation status of Lys72 significantly decreases ERK1’s enzymatic activity toward a well-known ERK1 substrate, ELK1. According to the structural analysis, the acetylation mimetic mutant of ERK1(K72Q), but not the deacetylation mimetic mutant ERK1(K72R), would block the formation of the salt bridges to Asp117 and Tyr119, leading to decreased stability of the 3-strand, diminished ATP binding, and reduced ERK1 kinase activity. Our study is the first to report that the acetylation /deacetylation of a conserved lysine in subdomain II of ERK1 could influence its enzymatic activity. It would be interesting to speculate whether the acetylation/deacetylation of Lys53 in ERK2, which is equivalent to Lys72 in ERK1, regulates ERK2’s kinase activity, although Lys53 has not been identified as an acetylation site yet.

More than two decades ago, it was demonstrated that Lys71 within subdomain II is critical for ATP binding (34,35). Substitution of Lys to Arg at this site therefore abolishes ERK1 kinase activity (35). Interestingly, we found that Lys71 can be acetylated by mass spectrometry analysis (data not shown). It was expected that the replacement of Lys with any other amino acid would ablate ERK1 kinase activity. Because of this reason, Lys to Arg (the deacetylation mimetic mutation) or Lys to Glu (the acetylation mimetic mutation) substitution of Lys71 would not tell us how deacetylation/acetylation regulates ATP binding and ERK1 enzymatic activity. Future studies using a special t-RNA synthetase capable of binding N-acetyl lysine to synthesize ERK1 with acetylated Lys72 may elucidate the role of the acetylation of this site in ERK1 function.However, it is intriguing that acetylation of Lys53, a homologous site of ERK1’s Lys71, in p38 augments p38’s kinase activity (36). Moreover, Lys52 in ERK2 and Lys55 in JNK1 and JNK2 are also homologous to ERK1’s Lys71 (36,37), but whether these sites are acetylated remains to be determined. It is tempting to hypothesize that the acetylation/deacetylation of the two conserved lysines in subdomain II (in the case of ERK1, K71 and K72), which either bind to ATP or form salt bridges to affect ATP binding, is a strategy employed by HATs and HDACs to fine-tune the enzymatic activities of MAPKs.As the main moderator in the downstream of the MAPK pathway, ERK1/2 are emerging as alternative targets, especially when inhibitors of the upstream kinases become resistant to patients
(38). There are several ERK1/2-specific inhibitors being used to combat the resistance to EGFR, Raf or MEK inhibitors in clinical trials. HDAC6-specific inhibitors are also being tested in clinical trials. Most of these trials were conducted with other anti-cancer drugs (33,39- 47). The combination of HDAC inhibitors and ERK1/2 pathway inhibitors have shown synergistic cell killing (48-52). Here we show that inhibition of HDAC6 down-regulates ERK1’s enzymatic activity, suggesting that the combination of HDAC6 inhibitors and ERK1/2 inhibitors may be a promising strategy to overcome the resistance to EGFR, ITF3756 Raf or MEK inhibitors.