Scaffolding Activities of Pseudodeacetylase HDAC7

Summary

Histone Deacetylase (HDAC) enzymes remove acetyl groups from acetyllysine-containing proteins, including nucleosomal histones to control gene expression. Beyond fundamental cell biology, HDAC activity is linked to various cancers, with many HDAC inhibitors developed as anticancer therapeutics. Among the eleven metal-dependent HDAC proteins, the four class IIa isoforms (HDAC4, 5, 7, and 9) are “pseudodeacetylases” without measurable enzymatic activity due to mutation of a catalytic tyrosine. Deacetylase-related activities of class IIa HDAC proteins are attributed to scaffolding functions, where recruitment of an active HDAC isoform leads to bound substrate deacetylation. Scaffolding of class IIa proteins beyond simple recruitment of an active HDAC are only starting to emerge. This perspective explores the various scaffolding roles of HDAC7, including recently reported acetylation-mediated reversible scaffolding, which is a form of acetyllysine-binding reader function. Studying the functional roles of HDAC7 will provide molecular insight into normal and pathological conditions, which could facilitate drug design.

Introduction

Acetylation is a well-studied protein post-translational modification (PTM) that governs many cellular processes, such as gene transcription, cell cycle, and protein folding.^1,2,3 Regulation of transcription by acetylation of nucleosomal histone proteins is an illustrative biological example. Specifically, an octamer of histone proteins arranges together to form the nucleosome structure that binds double stranded genomic DNA to form chromatin. Lysines on the N-terminal tails of histones protrude from the core nucleosome structure and interact electrostatically with the phosphate backbone of DNA, which condenses the chromatin to reduce transcription factor binding and repress gene expression.⁴ In contrast, acetylation of the N-terminal lysines relieves the electrostatic interactions with DNA to loosen the nucleosome structure, allowing transcription factors to access gene promoters and promote transcription. Thus, histone acetylation has a direct impact on chromatin accessibility and gene expression, which is a form of epigenetic regulation.

Three major protein families control histone acetylation, namely “writers”, “erasers”, and “readers” (Figure 1).⁵ Enzymes that introduce an acetyl group on N-terminal lysines of histones are writers of acetylation and include histone acetyl transferase (HAT) proteins, which are generally associated with activation of transcription (Figure 1A). Enzymes that remove acetyl groups from histone acetyllysines are erasers of acetylation and include histone deacetylase (HDAC) proteins, which generally repress transcription (Figure 1B). Finally, acetyllysines on histones are bound by reader proteins, including bromodomain-containing proteins, to recruit or scaffold proteins to histones and nearby genes, which influences transcription (Figure 1C).⁶ Taken together, epigenetic modulators govern the activation and repression of transcription in a balanced state for normal cell function. However, when the acetylation equilibrium is unbalanced due to unregulated HAT or HDAC protein activities, unregulated transcription and diseases can occur, as discussed below.

Figure 1. Epigenetic modulators of acetylation.

(A) Histone acetyl transferase (HAT) enzymes are epigenetic “writers” that add acetyl groups to N-terminal lysine residues on histones bound to genomic DNA (depicted as a line) in the nucleosome structure to generally activate transcription. (B) Histone deacetylase (HDAC) enzymes are epigenetic “erasers” that remove acetyl groups from histone acetyllysines, generally causing transcription repression. (C) Bromodomains are epigenetic “readers” that bind to histone acetyllysine to recruit other transcription factors to activate or repress transcription. K= lysine, Ac= acetyl, AcCoA= acetyl coenzyme A, R= reader protein

HDAC proteins

Aberrant expression of HDAC proteins has been linked to countless diseases, such as kidney diseases, cardiovascular diseases, and cancers.⁷ HDAC proteins are involved in various stages of cancer, including proliferation, metastasis, resistance to apoptosis, and cell cycle changes.⁸ For example, Vanaja et al. showed that HDAC proteins were overexpressed in cervical cancers to deacetylate tubulin, leading to alterations in cell cycle.⁹ Kim and colleagues observed elevated HDAC protein levels in gastric cancer tissues.¹⁰ The involvement of HDAC proteins in cancers and other diseases is reviewed elsewhere.^11,12,13

Given the role of HDAC proteins in disease, small molecule HDAC inhibitors have been used to significantly decrease tumor growth in vivo (reviewed elsewhere).¹⁴ Importantly, three HDAC inhibitor drugs have been approved by the US Food and Drug Administration (FDA) as anti-cancer therapeutics; vorinostat (or SAHA, subaroylanilide hydroxamic acid, Figure 2A), belinostat, and romidepsin are clinically used to treat cutaneous and/or peripheral T cell lymphoma.^15,16,17 Moreover, multiple HDAC inhibitors are being tested in clinical trials against a variety of cancers.¹⁸ We also note that small molecule inhibitor drugs are used commonly in biomedical research to discover and understand novel biology of HDAC proteins in normal and pathological conditions.

Figure 2. Comparison of active and inactive HDAC proteins.

(A) Structure of Vorinstat HDAC inhibitor. (B) Class-specific sequence alignment is shown where the conserved and catalytic Tyr (Y, cyan) in class I, IIb, and IV is changed to His (H, purple) in class IIa. (C) The structure of HDAC2 bound to inhibitor Vorinostat (PDB ID: 4LXZ) is displayed, with Vorinostat and Y308 represented as line structures. (D) The structure of HDAC7 bound to Trichostatin A (PDB ID: 4CBT) is shown, with Trichostatin A and H843 represented as line structures. The catalytic Zn metal ion is shown as a gold sphere. The line structures are color coded by atom, with all structures showing C = cyan, O = red, and N = blue, except His in the HDAC7 structure in part D, which has C = purple. Distance measurements are shown in yellow.

Given their connection to disease and therapeutics, the HDAC protein family has been well characterized. The 18 human HDAC isoforms are divided into two large groups based on catalytic mechanism. Classes I (HDAC1, 2, 3, and 8), II (HDAC4, 5, 6, 7, 9, and 10), and IV (HDAC11) require a metal ion for catalytic activity, whereas Class III (SIRT1–7) uses NAD+ as the cosubstrate for deacetylation.¹⁹ Subcategorization of the metal-dependent HDAC proteins (Classes I, II, and IV) is based on isoform size, subcellular localization, and similarity to yeast proteins.⁵ Among the three metal-dependent HDAC classes, class II is further subdivided into class IIa and IIb based on catalytic activity and domain composition.²⁰ Class IIa proteins (HDAC4, 5, 7, and 9) shuttle between the nucleus and cytoplasm and have the characteristic property of weak deacetylase activity.^21,22 In contrast, Class IIb HDACs (HDAC6 and 10) are predominantly cytoplasmic and contain two putative deacetylase domains, suggesting additional functions beyond epigenetics. The focus of this perspective is on the Class IIa metal-dependent HDAC proteins, as further discussed below.

Class IIa HDAC proteins: “Pseudodeacetylases”

The Class IIa isoforms, HDAC4, HDAC5, HDAC7, and HDAC9, contain the highly conserved deacetylase active site, similar to all HDAC family members. However, a key catalytic tyrosine (Tyr) residue is naturally altered to histidine (His) in Class IIa (Figure 2B).^21,22 The catalytic Tyr is required for stabilization of the tetrahedral intermediate formed after metal-mediated hydration of the acetyl carbonyl during deacetylation.²³ The natural mutation of the Tyr to His prevents stabilization of the tetrahedral intermediate, resulting in weak catalytic activity. To show the significant impact of the Tyr to His change on active site organization and interactions, the structures of Class I HDAC2 bound to the HDAC inhibitor Vorinostat (PDB: 4LXZ) and Class IIa HDAC7 bound to the HDAC inhibitor Trichostatin A (PDB: 3C10) were compared. The inhibitors bound in the active sites of HDAC2 and HDAC7 similarly, with interaction of their hydroxamic acid groups with the catalytic Zn metal ion (Figure 2C–D). However, the orientation of catalytic Y308 in HDAC2 and the aligned H843 in HDAC7 were signicantly altered, resulting in different proximities to the bound inhibitor (Figure 2C–D). Assuming that the carbonyl oxygen of the hydroxamic acid in each inhibitor mimics the acetyl carbonyl oxygen of a bound substrate, the distance to Tyr or His was measured. The Y308 oxygen in HDAC2 was located 2.5 Å from the carbonyl oxygen of Vorinostat (Figure 2C), which is within hydrogen bonding distance appropriate for interaction and stabilization during catalysis. In contrast, the 1-nitrogen of H843 in HDAC7 resided 9.0 Å from the carbonyl oxygen of Trichostatin A (Figure 2D), suggesting limited interaction and stabilization. The catalytic Tyr to His change and loss of enzymatic activity distinguishes the Class IIa isoforms from the other active HDAC family members and is a critical factor dictating the epigenetic functions of Class IIa HDAC proteins in cell biology.

Enzymes that retain high sequence alignment with catalytically active members of the same family yet lack a conserved catalytic residue or motif causing a loss of enzymatic activity are termed “pseudoenzymes”.^{24,25, 26} Several enzyme families include pseudoenzyme members, such as pseudokinases, pseudophosphatases, pseudonucleases, pseudotransferases, and pseudoligases.²⁴ Given the single mutated catalytic amino acid and undetectable enzymatic activities, similar to other pseudoenzymes, Class IIa HDAC proteins are accurately described as “pseudodeacetylases”.

Despite the absent or minimal catalytic activity, pseudoenzymes are common in biology and play a variety of essential roles in cellular events. Generally, pseudoenzymes perform three major biological functions: scaffolding, allosteric regulation, and competitive inhibition.^24,26 Given the relevance to Class IIa HDAC isoforms, here we will focus on scaffolding, where the pseudoenzyme bridges together one or more other proteins to build a functional complex that activates downstream biological events.²⁷ As an illustrative example, the well-studied pseudokinases are discussed. Kinases are enzymes that phosphorylate substrates to control cell signaling.²⁸ Pseudokinases contain a catalytic active site but lack several essential residues for enzymatic activity. For example, STRAD (STE20-related adaptor) is a pseudokinase that interacts with and activates the active kinase LKB1 (Liver kinase B1). Specifically, LKB1 relies on binding to the adapter Mo25 for both stabilization and activation, with the STRAD pseudokinase serving as a scaffold to facilitate LKB1-Mo25 association.^29,30 Despite the known biological functions of other pseudoenzyme classes, the mechanistic roles of Class IIa HDAC in biological events are relatively understudied, including HDAC7, which is the focus of this perspective.

HDAC7

HDAC7 maintains tissue specific expression in muscle, bone, brain, heart, the vascular system.^31,32 HDAC7 governs gene expression, which is further discussed below, in addition to multiple normal cellular functions, such as muscle differentiation and T cell apoptosis.^33,34 HDAC7 is also associated with various pathological conditions, including cancers.³¹ In addition to cancer, Haldar and coworkers showed that phosphorylated HDAC7 was stabilized during cardiac stress in rodent models, suggesting HDAC7 was overexpressed in cardiac muscles and involved in cardiovascular diseases.³⁵ Wang and coworkers elucidated that HDAC7 is involved in hippocampal-dependent memory formation, which suggested HDAC7 as a potential therapeutic target for memory loss due to aging or other neurological disorders.³⁶ Involvement of HDAC7 in cancers and other pathological conditions has been largely described and reviewed elsewhere.^37,38 Collectively, HDAC7 is involved in normal and pathological conditions, which makes careful study of HDAC7 function beneficial in future drug design.

Like other Class IIa HDAC isoforms, the structure of HDAC7 consists of several functional domains that dictate activity, localization, and interactions with other proteins. Generally, the HDAC7 structure comprises two important functional domains: an N-terminal interaction domain and a C-terminal pseudo-catalytic domain (Figure 3).³⁹ The N-terminal domain assists in interacting with other proteins, such as kinases, phosphatases, coactivators, and transcription factors.⁴⁰ Specifically, the transcription factor MEF (myocyte enhancer factor) binds HDAC7 via the MEF binding domain (Figure 3, MEF BD) to repress MEF2-dependent transcription.⁴¹ A recent study published by Schuetz et al. identified a novel zinc binding motif (Figure 3, ZBD) adjacent to the pseudodeacetylase domain that was speculated to assist in substrate recognition and protein-protein interactions.⁴² Two regions also regulate shuttling of the protein in and out of the nucleus, namely the nuclear localization signal (NLS, Figure 3) and the nuclear export signal (NES, Figure 3). Subcellular localization is further controlled by phosphorylation of several serines that bind to 14-3-3 chaperone proteins.³⁷

Figure 3. HDAC7 structural domains.

HDAC7 comprises two major functional domains: The N-terminal interaction domain and the C-terminal pseudo-catalytic domain. Each major domain contains multiple regions to interact with different proteins, including the MEF binding domain (MEF BD), nuclear localization signal (NLS), nuclear export signal (NES), zinc binding domain (ZBD), and pseudo-active site. Several serines (S194, S360, and S489) are phosphorylated to enable binding to 14-3-3 chaperone proteins. Amino acid numbering are provided below the schematic

Similar to other Class IIa HDAC isoforms, the HDAC7 pseudo-active site resides in the C-terminal region. Further, HDAC7 physically interacts through the pseudo-catalytic domain with co-repressor proteins, such as NCoR (nuclear receptor corepressor) and SMRT (silencing mediator of retinoic acid and thyroid receptor).⁴³ Although class IIa HDAC proteins are relatively less studied compared to the other metal-dependent HDAC isoforms, physical interactions with NCoR and SMRT are critical for scaffolding, which is the most established function of HDAC7, as described below.

Scaffolding functions of HDAC7

The best characterized mechanistic role of class IIa HDAC proteins, including HDAC7, in cell biology is the recruitment or scaffolding of active HDAC isoforms to acetylated substrates.⁴³ Specifically, HDAC7 binds to HDAC3 through the corepressor proteins NCoR or SMRT,⁴⁴ and HDAC3 requires association with NCoR or SMRT for full deacetylase activity.⁴⁵ The model that emerged suggests that HDAC7 assembles with the NCoR/SMRT-HDAC3 complex to bind and deliver acetylated protein substrates to the active HDAC3 isoform for deacetylation (Figure 4A). Fischle et al. provided the first evidence that the HDAC7-HDAC3 interaction was critical for HDAC7-mediated deacetylation, as demonstrated by deacetylation of peptide substrates only in the presence of HDAC3.⁴⁴ Further, coimmunoprecipitation experiments showed that the HDAC7 C-terminal pseudo-catalytic domain bound the repressor domain 3 (Rd3) of NCoR and SMRT.⁴⁴ Similarly, multiple additional studies documented binding of NCoR, SMRT, and HDAC3 with HDAC7 via the C-terminal pseudo-catalytic domain (Figure 3).^46–49 Altogether, strong experimental evidence supports that HDAC7 functions to scaffold or recruit acetylated substrates to the NCoR/SMRT-HDAC3 complex both in vivo and in vitro.

Figure 4: Models for scaffolding functions of HDAC7.

(A) HDAC7 binds and recruits acetylated substrates to the multiprotein complex comprising corepressor SMRT/NCoR (only NCoR shown) and active HDAC3. HDAC7 positions the acetyllysine of the substrate near HDAC3 for deacetylation. (B) In the context of transcription regulation, the substrates are the N-terminal acetyllysines on nucleosomal histones. Interactions between HDAC7 and a DNA binding transcription factor (TF) position the acetyllysines of the histones in proximity of HDAC3, which results in repression of transcription. AP= associated protein, K= lysine, Ac=acetyl.

Corroborating the evidence for HDAC7 scaffolding, HDAC4 is the prototype class IIa HDAC (see other reviews)^33,50 and has the best characterized scaffolding activity. Similar to HDAC7, HDAC4 binds to active HDAC3 through the repressor proteins SMRT or NCoR for substrate deacetylation (Figure 4A).⁴³ As evidence, Guenther et al. identified that SMRT and NCoR interact with the HDAC4 C-terminal catalytic domain using GST-HDAC4 pull-down assays. SMRT alone did not elevate HDAC4 enzymatic activity, strongly suggesting recruitment of active HDAC3.⁴⁵ Fischle et al. reported that HDAC4 enzymatic activity was significantly elevated after binding to HDAC3, compared to HDAC4 alone, and NCoR/SMRT bridged the binding between HDAC3 and HDAC4.⁴³ Further, deletion of HDAC3 led to loss of enzymatic activity of HDAC4, suggesting that HDAC4 recruits co-repressor complexes that contain HDAC3 for deacetylation.⁴³ Finally, a peptide that contained a consensus motif to Rd3 of NCoR/SMRT interacted with recombinant catalytic domains of all four class IIa HDAC proteins.⁵¹ Combined, the data provide further evidence that the NCoR/SMRT-HDAC3 complex is recruited by both HDAC4 and HDAC7 to deacetylate bound substrates.

Based on this model of scaffolding (Figure 4A), the goal of this perspective is to detail the known substrates, molecular interactions, mechanisms, and cellular events related to the scaffolding roles of HDAC7. While data is still emerging, the known scaffolding functions of HDAC7 can be organized by the substrate being deacetylated, including histone and nonhistone proteins. In addition to simple recruitment of substrates to HDAC3 for deacetylation, two alternative scaffolding mechanisms involving HDAC7-mediated protein-protein interactions are discussed, including reversible scaffolding and scaffolding without deacetylation. Finally, cellular activities of HDAC7 that are not yet clearly linked to scaffolding are also presented.

HDAC7 scaffolding for substrate deacetylation

HDAC7-mediated deacetylation of nucleosomal histones

Based on the scaffolding model (Figure 4A), the best studied substrates of HDAC7-NCoR/SMRT-HDAC3 deacetylation are nucleosomal histones, where deacetylation of acetyllysines on the N-terminal tails of histones generally leads to transcriptional repression (Figure 4B). Barneda-Zahonero et al. observed that HDAC7 bound the transcription factor MEF2C, as well as SMRT and HDAC3, in Burkitt lymphoma-derived Namalwa cells.⁵² Importantly, HDAC7 expression caused down regulation of key leukemic oncogenes, such as MYC (myc proto-oncogene), TERT (telomerase reverse transcriptase), and AICD (amyloid precusor protein intrcellular domain),⁵² which provided a link between HDAC7, the SMRT-HDAC3 complex, and transcriptional repression. More recently, Zhou and colleagues reported that the HDAC7 C-terminal pseudo-catalytic domain interacted with the transcription factor Aiolos to repress T-helper 17 (Th17) cell negative regulator genes via NCoR/SMRT-HDAC3.⁵³ Specifically, HDAC7 knockout elevated histone H3 lysine acetylation and reduced NCoR-HDAC3 localization on the IL-2 (interleukin-2) promoter in Th17 cells.⁵³ Collectively, the available data are consistent with a model where HDAC7-mediated nucleosomal histone deacetylation occurs through NCoR/SMRT-HDAC3 to repress transcription (Figure 4B).

Many studies also established a role of HDAC7 in transcriptional repression, although without clear ties to HDAC3 and NCoR/SMRT. As examples, Gao et al. documented that HDAC7 overexpression reduced mRNA transcription of the FOXA2 (forkhead box protein 2A) gene, whereas HDAC7 knockdown increased FOXA2 mRNA.⁵⁴ In another study by Nam et al., HDAC7 knockdown increased ACO2 (aconitase-2), SUCLG1 (succinate-CoA ligase [ADP/GDP-forming] subunit alpha, mitochondrial), and SDHC (succinate dehydrogenase cytochrome b560 subunit, mitochondrial) mRNA levels in renal cell carcinoma, suggesting histone deacetylation. Further, proteins levels of ACO2 and SUCLG1 were also elevated in HDAC7 knockout in CAKI-1 cells, which correlated with the knockdown studies.⁵⁵ As a final example, Ye et al. found that HDAC7 overexpression caused hypoacetylation of histone H3 and H4 on the CNT2 (concentrative nucleoside cotransporter 2) gene promoter to repress transcription in colorectal cancer.⁵⁶ While the role of HDAC7 in gene expression is clear, additional experimental evidence is needed to substantiate the involvement of NCoR/SMRT-HDAC3 in HDAC7-mediated transcriptional repression.

HDAC7-mediated deacetylation of non-histone substrates

In addition to HDAC7-mediated transcriptional repression, HDAC7 is associated with deacetylation of non-histone substrates. Although not directly tested, we speculate that HDAC7 scaffolds and recruits the SMRT/NCoR-HDAC3 multi-protein corepressor complex to deacetylate these non-histone substrates (Figure 4A). Guo and colleagues reported that the FOXP1 (fork head box protein P1) transcription factor was deacetylated by HDAC7, which facilitated self-renewal of mesenchymal stem cells.⁵⁷ HDAC7 was the only class IIa HDAC isoform to bind and deacetylate FOXP1. Truncated HDAC7 containing only the pseudo-catalytic domain was sufficient for both interaction and stabilization of FOXP1.⁵⁷ Given that HDAC7 binds to NCoR/SMRT through its C-terminal pseudo-catalytic domain, scaffolding could account for the influence of HDAC7 on FOXP1 and self-renewal of mesenchymal stem cells.

In addition to the FOXP1 transcription factor, kinases are deacetylated by HDAC7. For example, Reid and coworkers proposed a model where deacetylation of PKM2 (pyruvate kinase M2) by HDAC7 drives proinflammatory IL-1β expression.⁵⁸ TMP269, a class IIa HDAC selective inhibitor, disrupted complex formation of PKM2 with HDAC7 and elevated PKM2 acetylation levels in HEK293T cells.⁵⁸ PKM2 interacted directly and selectively with the C-terminal pseudodeacetylase domain, but not the N-terminal domain, of HDAC7, which is consistent with scaffolding to NCoR/SMRT-HDAC3. In a second kinase example, catalytic subunits of IKK (inhibitory kappa B kinase, IKKα and IKKβ) coimmunoprecipitated with HDAC7, and decreased levels of acetylated IKK were observed with overexpression of HDAC7 cells, suggesting HDAC7-mediated deacetylation of IKK.⁵⁹ Although beyond the scope of this perspective, we note that HDAC4-mediated deacetylation of kinases is also reported.⁶⁰ A future direction is to explore the involvement of the NCoR/SMRT-HDAC3 complex in HDAC7-mediated deacetylation of kinases.

Several additional reports documented HDAC7-mediated deacetylation, although without established involvement of the pseudo-catalytic domain in HDAC7. Tao and coworkers uncovered that HDAC7 deacetylated STAT3 (signal transducer and activator of transcription 3), which led to transcriptional repression and lung tumorigenesis.⁶¹ GST (glutathione S-transferase)-pulldown assays showed direct interaction between HDAC7 and STAT3. Also, HDAC7-depleted mice models and cancer cells showed reduced STAT3 acetylation levels. In another example, Guo et al. reported HDAC7-mediated deacetylation of β-catenin in NSCLC (non-small cell lung cancer cells), leading to nuclear translocation.⁶² Overexpression of HDAC7 decreased, while downregulation of HDAC7 increased acetyl-β-catenin levels. Similarly, Wu et al. documented that Hsp70 (heat shock protein 70) was deacetylated by HDAC1 and HDAC7 independently in breast cancer cells using knockdown experiments.⁶³ Moreover, HDAC1/HDAC7-mediated deacetylation enhanced cancer cell survival and autophagy inhibition. Future work is needed to investigate the role of NCoR/SMRT-HDAC3 and the pseudo-catalytic domain in HDAC7 in deacetylation of these substrates.

Alternative scaffolding mechanisms of HDAC7

Reversible scaffolding of HDAC7- acetyllysine-binding “Reader” function

While the evidence for HDAC7 scaffolding in substrate deacetylation is substantial and growing, the purpose of the pseudo-active site (Figure 3) in HDAC7 function remains unclear. The current model of scaffolding to recruit the NCoR/SMRT-HDAC3 complex to substrates (Figure 4) does not require the pseudo-active site. One hypothesis proposed over a decade ago is that the pseudo-active sites of class IIa HDAC proteins act as readers or binders of lysine acetylation.^64,65 As introduced earlier, epigenetic readers bind to acetyllysine-containing proteins, particularly N-terminal acetyllysines on nucleosomal histones, to modulate transcription (Figure 1C). One mechanism for reader proteins is to recruit transcriptional modulators to histone acetyllysines to influence transcription of the nearby gene. As illustrative examples, Dhalluin et al. and Manning et al. observed that the bromodomain of acetyl transferase p300 bound acetylated histone H3 to act as a coactivator of transcription.^66,67 Onder and coworkers found that bromodomain inhibition of p300 or CBP (CREB binding protein) reduced histone acetylation levels and transcription of fibroblast specific genes.⁶⁸ As a final example from Roe et al., acetylated hematopoietic transcription factors and histone H4 recruited BRD4 (bromodomain-containing protein 4) to regulate gene expression in acute myeloid leukemia (AML).⁶⁹ As key players in gene expression regulation, bromodomain-containing proteins are linked to cancer development and metastasis, and several bromodomains have been targeted for drug design and therapeutics.^70,71 If HDAC7 maintains reader activity, similar to bromodomain-containing proteins, then new avenues of biomedical research and drug design become possible.

A prerequisite of reader function is high affinity acetyllysine binding. In the context of bromodomains, Jacobson et al. applied isothermal titration calorimetry (ITC) to observe 5 ± 0.2 μM binding of the double bromodomain in TAF_II250 (TATA binding protein-associated factors) with an acetylated histone 4 (H4) peptide.⁷² In another study, Wright and colleagues reported a K_D of approximately 900 μM for the GCN5 (general control non-depressible 5) bromodomain and an acetyllysine-containing peptide using NMR titration.⁷³ Based on bromodomain binding studies, the class IIa HDAC pseudo-active sites must also bind acetyllysine in the μM range. Our group found that the K_D of HDAC7 bound to an acetyllysine-containing peptide was roughly 1.0 ± 0.1 μM using biolayer interferometry (BLI).⁴⁶ HDAC7 displayed similar or better binding affinity to an acetyllysine-containing peptide compared to bromodomain-containing reader proteins,^72,73,74 which endorses the possibility of HDAC7 reader function.

Another prerequisite of epigenetic reader function is that acetyllysine binding must result in a downstream transcriptional effect. A molecular mechanism for the role of acetyllysine binding by HDAC7 in transcription activation was proposed by our group,⁴⁶ where HDAC7 binds acetylated AR (androgen receptor), a steroid receptor belonging to the nuclear receptor (NR) super family of ligand-dependent transcription factors. According to the model, genomic DNA-bound AR interacts with the HDAC7-NCoR/SMRT-HDAC3 multiprotein complex to recruit HDAC3 to nucleosomes and repress transcription through histone deacetylation (Figure 5A), which is similar to HDAC7 scaffolding function (Figure 4B). Upon acetylation of AR at a critical lysine, the acetyllysine from AR binds the HDAC7 pseudo-active site to release the HDAC7-NCoR/SMRT-HDAC3 multiprotein complex. Without bound HDAC3, transcription will be activated due to acetylation of histones (Figure 5B). The model is a form of “reversible scaffolding”, where pseudo-active site binding by acetyllysine regulates recruitment of an active HDAC isoform to HDAC7. Consistent with the model, NCoR binding to HDAC7 was disrupted by an acetyllysine containing peptide.⁴⁶ AR-dependent transcription required an intact HDAC7 pseudo-active site and HDAC3 activity. Finally, mRNA levels of two AR-regulated genes, SPRF5 (serine and arginine rich splicing factor 5) and WNT16 (Wnt family member 16), were controlled by HDAC7 in a pseudo-active site dependent manner, providing the first evidence of acetyllysine-dependent reversible scaffolding reader function by HDAC7.⁴⁶ Like many other examples of deacetylation-dependent functions of HDAC7, future studies are needed to further substantiate the model.

Figure 5. Model for reversible scaffolding or reader function of HDAC7.

(A) HDAC7 bridges a bound unacetylated transcription factor (TF) to NCoR/SMRT-HDAC3 (only NCoR shown) to repress transcription by maintaining nucleosomal histone lysines in the unacetylated state. (B) Upon acetylation of the TF by a separate HAT enzyme, the acetyllysine will bind to the pseudo-active site of HDAC7, leading to dissociation of the NCoR/SMRT-HDAC3 complex. Without the nearby deacetylation activity of HDAC3, the N-terminal lysines on nucleosomal histones will become acetylated by a HAT, leading to activation of transcription. AP= associated proteins, K=lysine, KAc=acetyllysine

HDAC7 scaffolding without deacetylation

In the prior examples, HDAC7 scaffolding to the NCoR/SMRT-HDAC3 multi-protein complex ultimately resulted in deacetylation of a bound substrate, whether non-histone (Figure 4A) or histone (Figure 4B and 5).^43,46 However, prior literature also documented that HDAC7 acts as a scaffold to build multi-protein complexes without direct evidence for deacetylation. Instead, HDAC7 recruited protein modifiers other than HDAC3, including acetyltransferases, chaperone proteins, and ligases to bound substrates (Figure 6A).^75,76,77 Related to acetyltransferases, Hua et al. observed that a complex of HDAC7, AP-1 (activator protein 1) transcription factor, and the HAT p300 elevated expression of the CTGF (connective tissue growth factor) protein.⁷⁵ A model was proposed where HDAC7 acts as a scaffold to recruit p300 to AP-1 for transcriptional activation. As a second example, Kato et al. detailed that HDAC7 enhanced the transcriptional activity of HIF1-α (hypoxia inducible factor 1-alpha) by recruiting p300 under hypoxic conditions.⁷⁸ HIF1-α bound with an HDAC7 C-terminal fragment that contained the C-terminal pseudo-catalytic domain, similar to the scaffolding observed with NCoR/SMRT-HDAC3 (Figure 4).

Figure 6. Models for alternative functions of HDAC7.

(A) HDAC7 binds either directly or indirectly with protein modifiers, including acetyltransferases, chaperone proteins, and ligases. Simultaneous binding to a substrate facilitates processing by the protein modifier activity. (B) Transcription factors (TF) interact with HDAC7 to regulate transcription via the N-terminal interaction domain, without involvement of the C-terminal pseudo-catalytic domain, through an unknown mechanism. We speculate that scaffolding of HDAC7 to multiprotein complexes containing epigenetic modulators other than NCOR/SMRT-HDAC3 (green and yellow proteins) account for this transcriptional effect, which are yet to be uncovered. AP= associated proteins.

In addition to acetyltransferase recruitment, Margariti et al. found that HDAC7 bridged β-catenin to 14-3-3 proteins in the cytoplasm to prevent nuclear localization and proliferation.⁷⁶ As evidence, HDAC7 degradation resulted in localization of b-catenin to the nucleus.⁷⁶ Related to ligase activity, HDAC7 elevated SUMOylation of PML (promyelocytic leukemia protein) in human endothelial cells.⁷⁷ In this case, HDAC7 interacted with both PML and UBC9 (SUMO conjugating enzyme 9) to stimulate PML SUMOylation. While PML bound the C-terminal pseudodeacetylase domain of HDAC7, UBC9 required residues of both the N- and C-terminal domains to stimulate PML SUMOylation. Based on the data, Gao and colleagues proposed a model where HDAC7 promoted PML SUMOylation via scaffolding to UBC9. Notably, recruitment of a class IIa HDAC protein to UBC9 was first documented for HDAC4 by Yao and colleagues.⁷⁹ As a final example, Oesterreich and coworkers reported collaboration between HDAC7, the transcription factor ER-α (estrogen receptor alpha), and the chromatin remodeler FoxA1 (forkhead box protein A1) to repress the RPRM (reprimo, TP53 dependent G2 arrest mediator homolog) gene.⁸⁰ A model was proposed where FoxA1 was recruited by the ER-α-HDAC7 complex to the RPRM promoter, where release of RNA polymerase II lead to transcription repression. Interestingly, NCoR/SMRT knockdown did not affect RPRM repression. Collectively these many studies are consistent with the model that HDAC7 scaffolds a variety of protein modulators to substrates to influence activity (Figure 6A), with other binding pairs likely to be uncovered in future studies.

Orphan scaffolding functions of HDAC7

While the scaffolding of HDAC7 to NCoR/SMRT-HDAC3 and other protein modifiers has growing experimental support, HDAC7 also represses transcription without evidence for protein modifying activity, such as deacetylation. As an example of this “orphan” repressive function, Westendorf and coworkers disclosed that the transcriptional activity of RUNX2 (runt related transcription factor 2) was repressed by HDAC7 in a deacetylation-independent manner.⁸¹ RUNX2 interacted with both N-terminal and C-terminal domains of HDAC7 in vivo and in vitro, but N-terminal domain interaction was sufficient for transcriptional repression. In the second example, Stemig et al. reported elevated expression of MITF (microphthalmia associated transcription factor) target genes in bone marrow derived osteoclast cultures from HDAC7 knockout mice.⁸² Moreover, loss of the C-terminal pseudo-catalytic domain of HDAC7 did not alter MITF repression, suggesting a mechanism involving the N-terminal interaction domain alone. In the final example, the D’Mello group found that HDAC7 possessed neuroprotective activity, and deletion of the HDAC7 pseudo-catalytic domain did not affect neuroprotection.⁸³ Given that the NCoR/SMRT-HDAC3 complex interacts with HDAC7 via the C-terminal pseudo-catalytic domain, these three examples highlight that an alternative repression mechanism involving interactions by the HDAC7 N-terminal domain has yet to be revealed. Based on the emerging scaffolding roles of HDAC7, we speculate that epigenetic modulators of transcription, such as other writers, erasers, or readers (Figure 6B), interact with the HDAC7 N-terminal domain to regulate transcription.

As a final example of HDAC7 function without an established molecular mechanism, the Ince group documented that HDAC7 knockdown decreased mRNA levels of SMAD3 (SMAD family member 5), VEGFA (vascular endothelial growth factor A), and VDR (vitamin D receptor) genes in CSC-like BPLER cells (cancer stem cell), although the transcription factor collaborating with HDAC7 is not yet known.⁸⁴ Here, HDAC7 acted as a transcriptional activator, unlike the prior examples where HDAC7 functioned as a transcriptional repressor (Figure 4B and 5A). Although the molecular mechanism is understudied, we postulate that HDAC7 scaffolding accounts for transcription activation, perhaps using acetyllysine-mediated reversible scaffolding (Figure 5), similar to AR. Jointly, the many examples of HDAC7-mediated events without molecular details suggest that the full range of HDAC7 functions have yet to be discovered.

Conclusions and future directions

In this perspective, an overview of the reported functions and emerging molecular mechanisms of HDAC7 was presented, focusing on the developing role of scaffolding. Multiple laboratories reported protein deacetylation by HDAC7.^57,61,58,59 However, being a weak deacetylase, direct deacetylation by the pseudo-catalytic domain is unlikely. Instead, based on the experimentally established interaction of the C-terminal domains of HDAC4 and HDAC7 with the corepressor complex containing NCoR/SMRT-HDAC3,^{43, 45–49} a model for HDAC7-mediated deacetylation has been proposed. Specifically, HDAC7 acts as a scaffold to recruit the HDAC activity in the NCoR/SMRT-HDAC3 complex to bound substrates (Figure 4A), including nucleosomal histones via DNA-bound transcription factors (Figure 4B).^44,43 In addition to substrate deacetylation by scaffolding to NCoR/SMRT-HDAC3, HDAC7 bridges other epigenetic modulators, such as acetyltransferases and ligases,^75,76,77 to bound substrates, including transcription factors (Figure 6), suggesting a general role for HDAC7 in epigenetic regulation. Although less documented, Yao and colleagues also reported that HDAC4 formed a dual deacetylase complex with SIRT1 to deacetylate MEF2,⁷⁹ suggesting that class IIa HDAC proteins recruit other active HDAC isoforms in addition to HDAC3. Because the model is still developing, experimental evidence is needed to further substantiate the scaffolding role of HDAC7 with multiple substrates, transcription factors, and perhaps collaborating active HDAC isoforms.

Beyond scaffolding alone, recent evidence suggests that HDAC7 can also act as a binder or reader of acetylated proteins through a novel reversible scaffolding mechanism. Specifically, the current model posits that dissociation of the NCoR/SMRT-HDAC3 complex due to binding of acetyllysine to the pseudo-active site of HDAC7 results in transcription derepression (Figure 5).⁴⁶ This model of reversible scaffolding documents a role for the pseudo-active site of HDAC7 in cellular functions for the first time. In addition, acetyllysine binding and reversible scaffolding also provide a mechanistic explanation for the transcriptional activation observed with HDAC7 in prior work.^{46, 84} Given that HDAC activity and deacetylation are typically associated with transcriptional repression, the model of reversible scaffolding expands the role of HDAC proteins in epigenetic regulation. Finally, we note that the acetyllysine-binding function of reversible scaffolding by HDAC7 (Figure 5) is fundamentally distinct from the reader function of bromodomain-containing proteins (Figure 1C). As discussed earlier, bromodomain-containing reader proteins bind acetyllysine to recruit transcription factors and acetyl transferases to modulate gene expression.⁸⁵ In contrast, the reversible scaffolding activity proposed for HDAC7 relies on dissociation of the NCoR/SMRT-HDAC3 complex upon acetyllysine binding (Figure 5). In other words, the dissociation-mediated reversible scaffolding function of HDAC7 is unique and distinct from the recruitment-mediated reader function of bromodomains, although both involve acetyllysine binding. Given the novelty of reversible scaffolding for dissociation of protein-protein interactions after acetyllysine binding, further evidence is needed to substantiate this novel form of reader function by HDAC7.

To further study the acetyllyine-mediated reversible scaffolding activity of HDAC7, we propose that NR family members, in addition to AR, also act as reader clients. The NR super family act as transcription factors to regulate vital functions in diverse biological processes, such as cell metabolism and inflammation.⁸⁶ NR proteins activate transcription through ligand binding, with some members also dimerizing.⁸⁶ In fact, the NR family is categorized into four subclasses based on ligand binding and dimerization status, which include steroid receptors, RXR heterodimers, dimeric orphan receptors, and monomeric orphan receptors (Figure 7).⁸⁷ Most NR members are acetylated (reviewed previously)⁸⁸ and share a common acetylation motif involved in transcription activation (Figure 7).⁸⁹ For example, Fu et al. showed that acetylation of AR at K630 governs transcription activation.⁹⁰ Building on this prior work, the reversible scaffolding observed with HDAC7 and AR required AR K630 acetylation for NCoR-HDAC3 dissociation.⁴⁶ Similar to AR, Kraus and colleagues documented that acetylation of K266 and K268 in ERα is essential for transactivation of the receptor.⁹¹ Sánchez-Pacheco et al. reported that TRα (thyroid receptor alpha) acetylation on three lysines (K128, K132, and K134) was a critical prerequisite for receptor activity and activation.⁹² In a study by Zhao and coworkers, acetylation on K145 elevated the DNA binding ability and transcription activity of RXRα (retinoid X receptor).⁹³ As a final example, Chung et al. observed that PR (progesterone receptor) acetylation at K183 augmented transactivation and DNA binding kinetics.⁹⁴ A common feature of acetylation-dependent transcriptional activation by NR is involvement of the acetyl transferase p300 in NR acetylation.^90,91,93,94 Given the evidence that acetylation-dependent AR transcriptional activation involved HDAC7 and NCoR-HDAC3 dissociation,⁴⁶ we hypothesize that other NR members function as reader clients of HDAC7, with experiments ongoing.

Figure 7. Sequence alignment of acetylation motifs in Nuclear Receptors (NR).

NR proteins are categorized into four major classes: steroid receptors (orange), RXR heterodimers (blue), dimeric orphan receptors (green), and monomeric orphan receptors (purple). Lysines highlighted in red are known to be acetylated and required for transcription. Lysines that could be potential acetylation sites vital for transcription are highlighted in bold. AR-androgen receptor, ER-estrogen receptor, PR-progesterone receptor, GR-glucocorticoid receptor, MR-mineralocorticoid receptor, TR-thyroid receptor, RAR-retinoic acid receptor, VDR-vitamin D receptor, PPAR-peroxisome proliferator-activated receptor, RXR-retinoid X receptor, HNF4-hepatocyte nuclear factor 4, TLX-homolog of the Drosophila tailless gene , ERR-estrogen related receptor, ROR-retinoic acid-related orphan receptor

In addition to collaboration between HDAC7 and NR transcription factors, we also speculate that HDAC4 and HDAC5 act as acetyllysine-dependent reversible scaffolds to derepress transcription. Highly conserved pseudo-active sites and known interaction with NCoR/SMRT-HDAC3 suggest similar functions of HDAC4, 5, and 7.^{46, 53, 95} Further, HDAC4 and 5 interact with transcription factors, such as HIF1-α and STAT3, similar to HDAC7.^96,97 In contrast, HDAC9 does not maintain NCoR/SMRT-HDAC3 interactions necessary for scaffolding,⁴⁶ suggesting that HDAC9 is mechanistically distinct. Further mechanistic studies are required to uncover the scaffolding roles of all class IIa HDAC proteins.

As highlighted here, studying the molecular mechanisms and modified partner proteins of class IIa HDAC isoforms will be critical to fully characterize the roles of HDAC proteins in cellular events and disease states. A challenge in identification of potential scaffolding protein clients or modified substrates of class IIa isoforms is the lack of systematic tools. Reversible scaffolding clients, such as AR,⁴⁶ and modified substrates, such as FOXP1 or kinases,^{57, 58} were uncovered for HDAC7 serendipitously. Systematic methods to identify and validate substrates and scaffolding protein clients are needed to fully characterize the cellular roles of all class IIa HDAC proteins. Related to the epigenetic functions of class IIa HDAC isoforms, another challenge is the low abundance of transcription factors, which often require overexpression to super physiological levels. While overexpression of transcription factors has been the norm when studying reader functions,^98,99 future use of physiological concentrations of transcription factors in relevant cell lines is needed.⁴⁶

Pseudoenzymes, or catalytically-inactive members of an enzyme family, are well characterized across different species and enzyme families, including kinases, phosphatases, and proteases.¹⁰⁰ In addition to scaffolding, discussed previously,^29,30 two additional non-catalytic functions of pseudoenzymes have been reported, allosteric regulation and competitive inhibition, which aid in cell proliferation, differentiation, and signaling.^{24, 26} As an example of allosteric regulation, Littlefield et al. observed that the inactive HER3 (human epidermal growth factor receptor 3) pseudokinase heterodimerized with an active HER receptor (HER1 or EGFR, HER2, and HER4) in cellular signaling.¹⁰¹ In this case, binding of HER3 near the kinase active site of the active HER family member enhanced catalytic activity. As an example of competitive inhibition, Reiterer and coworkers reported that the inactive STYX (serine/threonine/tyrosine-interacting protein) pseudophosphatase bound ERK2 (extracellular regulated kinase 2) to competitively inhibit interaction with DUSP4 (dual specificity phosphatase 4), which influenced cell differentiation.¹⁰² Similar to these pseudoenzyme activities, we speculate that the class IIa pseudodeacetylases might also play roles in allosteric regulation and competitive inhibition. Future research is necessary to uncover if allosteric regulation and competitive inhibition play a role in the full functional profile of HDAC4, 5, 7, and 9.

HDAC7 regulates the function of many proteins, though validated mechanisms are still emerging. As discussed here, scaffolding might explain many of the activities of HDAC7. Specifically, current evidence suggests that HDAC7 regulates transcription not only by binding the well-established NCoR-HDAC3 complex, but also by recruiting other writers and erasers. Future work to identify modified protein partners and novel scaffolding mechanisms, including acetyllysine-mediated reversible scaffolding, is needed to fully uncover HDAC7 biological functions. Beyond HDAC7, the other class IIa HDAC proteins likely also exhibit a variety of scaffolding functions, with future work also needed. Given the many clinical HDAC inhibitors, discovering the molecular mechanisms governing the functions of class IIa HDAC proteins can ultimately assist in drug design.