Inhibitors of DHHC family proteins

Abstract

Protein S-acylation is a prevalent post-translational protein lipidation that is dynamically regulated by “writer” protein S-acyltransferases and “eraser” acylprotein thioesterases. The protein S-acyltransferases are comprised of 23 Aspartate-Histidine-Histidine-Cysteine-containing (DHHC) proteins, which transfer fatty acid acyl groups from acyl-Coenzyme A (acyl-CoA) onto protein substrates. DHHC proteins are increasingly recognized as critical regulators of S-acylation-mediated cellular processes and pathology. As our understanding of the importance and breadth of DHHC-mediated biology and pathology expands, so too does the need for chemical inhibitors of this class of proteins. In this review, we discuss the challenges and progress in DHHC inhibitor development, focusing on 2-bromopalmitate (2BP), the most commonly used inhibitor in the field, and N-cyanomethyl-N-myracrylamide (CMA), a new broad-spectrum DHHC inhibitor. We believe that current and ongoing advances in structure elucidation, mechanismal interrogation, and novel inhibitor design around DHHC proteins will spark innovative strategies to modulate these critical proteins in living systems.

Introduction

Post-translational modification (PTM) of proteins through the addition of lipids is a common mechanism to tune protein function [1]. The lipid PTM S-acylation (often referred to as S-palmitoylation) involves the covalent addition of a long chain fatty acid, usually palmitate, on cysteine (Cys) residues of proteins through a thioester bond. S-acylation is among the most prevalent of lipid PTMs, observed on up to 10% of the human proteome [2], including therapeutically relevant proteins and hubs of cell growth and proliferation, such as RAS and mTOR1 [3,4]. Additionally, S-acylation plays critical roles in diverse cellular processes, including the membrane fusion of viruses onto host cells [5], making this particular lipid PTM an attractive therapeutic target.

Enzymatic reversibility is another remarkable feature of S-acylation [6] (Figure 1A). Most lipid PTMs are static and primarily serve to anchor proteins to membranes [1]. Uniquely among lipid PTMs, S-acylation can be dynamically regulated by lipid “eraser” acylprotein thioesterases (APTs) and lipid “writer” protein S-acyltransferases (PATs), which are referred to as DHHC proteins or DHHC-PATs, due to the highly homologous DHHC domain required for S-acyltransferase activity [7–9]. In humans, there are 23 DHHC-PATs and at least 5 APTs [8,10–14]. Perturbation of either side of the S-acylation dynamic cycle can lead to abnormal phenotypes, highlighting the importance of tight regulation of this lipid PTM [1,15,16].

Figure 1.

Structural and mechanism of DHHC-PATs. (A) Scheme of dynamic regulation of S-acylation. S-acylation of the cellular proteome is regulated by the ‘writer’ DHHC-PATs and the ‘eraser’ APTs. (B) Two-step catalytic mechanism of DHHC-PATs. (C) The crystal structure of 2-bromopalmitate-bound human zDHHC20 (PDB: 6BML). 2-Bromopalmitate and zinc ions are shown as purple stick and gray sphere, respectively. (C) Structure of 2-bromopalmitate (purple) bound to active cite cysteine (C156) of human zDHHC20. The covalent bond (green) between Cys156 of zDHHC20 with C2 of 2BP is highlighted by green arrow. Critical residues for acyl chain recognition and selectivity shown (teal sticks), as verified by in vitro mutagenesis studies.

A growing body of evidence indicates that DHHC proteins have broad physiological significance. For example, zDHHC4 and zDHHC5 have been shown to be required for correct localization and fatty acid uptake activity of the fatty acid scavenger receptor CD36, and ZDHHC4-null or adipose-specific DHHC5 knockout (ZDHHC5-AKO) mice showed resistance to weight gain under both normal chow and high fat diet feeding conditions [17]. Studies also showed that the epithelial growth factor receptor (EGFR) undergoes zDHHC20-mediated S-acylation in various cell lines, the inhibition of which leads to suppression of phosphoinositide 3-kinase (PI3K) signaling and growth of KRAS mutant-dependent cancer cells [18]. Moreover, other studies suggest DHHCs are potential drug targets for viral infections and autoimmune diseases (for recent reviews on therapeutic potential of DHHC proteins, see: [5,19,20]).

Unfortunately, the available toolbox for studying DHHC protein function is limited and comprised largely of time-consuming S-acylation detecting assays (e.g., acyl-biotin exchange, or ABE), in combination with overexpression of individual DHHC proteins or antagonism using inhibitors that are known to be toxic or minimally selective [21].

Historically, the study of S-acylation has been hindered by a dearth of quality antibodies and chemical inhibitors to selectively disrupt the function of DHHC proteins-of-interest. Adding to this, new knowledge and appreciation of the functional redundancy and complicated intrafamilial regulatory networks amongst DHHC-PATs has complicated the parsing out of individual roles for each family member [22,23]. An increasing number of reports, however, has shown that knockdown or knockout of individual DHHCs can lead to partial reduction in the S-acylation of proteins-of-interest and impairment of related cellular processes [16,24,25]. Moreover, antibodies towards some endogenous DHHCs have now been developed [23,26,27]. Despite these advancements, chemical inhibitors remain both scarce and highly desired, as they can complement genetic manipulation, especially where temporal control is critical to function or in biological contexts where genetic manipulation is challenging. Chemical inhibitors will also permit the assessment of DHHC proteins’ potential as drug targets for a range of indications.

In this review, we examine currently identified DHHC protein inhibitors, beginning with a brief discussion of the known structures and mechanisms of DHHC proteins, and how this knowledge stands to inform the design and discovery of inhibitors. We then review recently reported attempts at DHHC inhibitor discovery. Finally, we focus in on two DHHC inhibitors: 2-bromopalmitate (2BP), currently the most widely used inhibitor, and N-cyanomethyl-N-myracrylamide (CMA), a newly developed inhibitor featuring a novel scaffold with some potential advantages over 2BP.

Structure and mechanism of DHHC proteins

Thus far, all eukaryotic proteins identified with protein S-acyltransferase activity so far are transmembrane proteins with a conserved domain containing a characteristic Aspartate-Histidine-Histidine-Cysteine (Asp-His-His-Cys, or DHHC) motif, with the exception of mammalian zDHHC13, which possesses DQHC instead of DHHC [28]. Studies have demonstrated that the Cys residue is critical for enzyme activity, as mutation to Alanine or Serine completely abolishes S-acyltransferase activity of various DHHC proteins in vitro and in cellulo [29–32].

Unlike other known acyl transferases, DHHC proteins are thought to follow a two-step mechanism for catalysis (Figure 1B). First, the active-site Cys reacts with acyl-Coenzyme A (acyl-CoA) and is autoacylated. Then, the acyl chain is transferred from the DHHC protein to a Cys residue on the substrate protein [31,33]. While mutagenesis and biochemical studies provided important clues about DHHC proteins, a lack of structural data had, until recently, impeded a full mechanistic understanding, and inhibitor development.

In a major 2018 breakthrough, the first crystal structures of DHHC proteins, human zDHHC20 and zebrafish zDHHC15, were reported, revealing the structural basis for autoacylation [32]. The structure of zDHHC20 in complex with 2BP (the most widely used covalent inhibitor in the field, discussed later in this review) (Figure 1C, D), suggests that a long cavity formed by four transmembrane (TM) helices is essential for acyl chain binding. Furthermore, the structure points to key hydrophobic residues on the TM helices and DHHC domain that likely determine acyl chain recognition and chain-length selectivity. In addition, the active site Cys is located precisely at the entrance of the acyl-binding groove, ostensibly facilitating nucleophilic attack to the acyl-CoA donor substrate. The structure of zDHHC20 also suggests the existence of a basic patch that binds with the Coenzyme A (CoA) moiety, but how this putative CoA binding pocket impacts protein function and mechanism is not yet known. Notwithstanding, these crystal structures provide valuable mechanistic information and critical clues for disrupting the autoacylation activity of zDHHC20, and, by virtue of their conserved active domains, other DHHC proteins as well.

While the investigation of the autoacylation mechanism of DHHCs has progressed, the next step of the reaction – recognition of a protein substrate and transfer of the acyl chain – remains largely unknown. No crystal structure of a full-length DHHC-substrate complex has yet been determined, except for the complex of Snap25b(aa 111–120) and the zDHHC17’s ankyrin-repeat (AR) domain, the domain critical for zDHHC17’s binding to Huntingtin [34]. It has been reported that subgroups of DHHCs have preference towards peptides with certain features, such as N-terminal myristoylated peptides [23,35–38], however more thorough studies mapping the key motifs and/or domains for the recognition of protein substrates and potential broader modes of specificity regulation are needed. Understanding protein substrate recognition by DHHCs will aid interrogation of DHHC-mediated biology and will also provide insights into the development of selective inhibitors against individual DHHC proteins.

Assays for the discovery of DHHC inhibitors

Although 20 years have passed since the original annotation of DHHC proteins and several structures are now available, DHHC inhibitor development remains nascent. One of the major obstacles is a paucity in high-throughput assays for DHHCs, which has been hindered by technical challenges, including the difficulty in DHHC protein purification. Nevertheless, several cell-based and in vitro screening assays capable of detecting DHHC-PAT activity have been developed.

Current cell-based assays often monitor the acylation status of a substrate, and can be imaging-based – monitoring changes in the subcellular localization of a fluorescently-tagged reporter substrate – or liquid chromatography-mass spectrometry (LC-MS)-based – directly quantifying the palmitoylation levels of the substrate [39–41]. For example, an imaging-based assay was used to screen for inhibitors of the S-acylation of Dual leucine-zipper kinase (DLK). In this assay, disruption in the S-palmitoylation-dependent localization of a green fluorescence protein (GFP)-fused DLK to the Golgi apparatus was used to identify putative inhibitors [41]. Both an advantage and disadvantage of such cell-based, substrate-dependent assays is that one is potentially measuring a collection of DHHC activities.

In vitro assays, which use purified DHHC proteins, offer direct assessment of putative inhibitors on a DHHC-of-interest, and can monitor DHHC autoacylation (e.g., in-gel radiolabeling assay or couple-enzyme assay [42–44]) or the acylation of the acyl chain acceptor (e.g., peptide-based HPLC assays [31,40]). In a fluorescence-based assay, the production of Coenzyme A, the common by-product of autoacylation in principle shared by all of DHHCs, was monitored by coupling to the α-ketoglutarate dehydrogenase (α-KDH)-catalyzed reduction of nonfluorescent NAD⁺ to fluorescent NADH. This coupled enzyme assay has not only been utilized on biochemical studies of a wide panel of DHHCs, including yeast Erf2/Erf4, human zDHHC9/GCP16, and later human zDHHC20 and zebrafish zDHHC15 [32,43], but also applied for Erf2/Erf4 inhibitor screening [45]. However, the inability to screen for molecules that disrupt the acyl transfer to substrate – instead of autoacylation – likely limits the scope of information obtained from this assay. It is therefore of interest to complement such assays with the acyl chain acceptor-based assays. Recently, our group developed a fluorescence polarization-based assay for zDHHC20 based on Acyl-cLIP [46] using a 5-carboxyfluorescein-tagged peptide from human NRas (aa177–189), which, when S-acylated, interacts with detergent micelles to yield a change in fluorescence polarization [47]. In our lab this assay achieves a z-value of ~0.7, suggesting the potential capacity for this assay to screening on a larger scale while being adaptable for other DHHCs.

As the demand for DHHC inhibitors increases, assays to characterize collections of DHHCs (e.g. zDHHC5 and 8) will also be needed. Therefore, fundamental cellular, biological, and biochemical understanding of the DHHCs, in addition to innovation in assay development, are needed to develop robust methods for both the study of DHHC proteins and for inhibitor screening and optimization.

Overview of DHHC inhibitors

Only a handful of DHHC inhibitors have been reported. Lipid-based covalent inhibitors, which feature an aliphatic chain that presumably mimics the binding mode of DHHC proteins with their acyl-donor substrate (Figure 2A) and an electrophilic warhead for covalent cysteine labeling, are among the earliest discovered inhibitors in the field. For example, 2BP [48], tunicamycin [49], and cerulenin [50] (Figure 2B) are all reported to inhibit DHHC. Of these lipid-based inhibitors, 2BP is by far the most widely used, and is therefore discussed in greater detail in the following section. In general, however, these lipid-based inhibitors suffer from low potency, lack of specificity, and cytotoxicity [51].

Figure 2.

Summary of DHHC substrate and inhibitors. (A) Chemical structures of acyl-CoA donor substrates of DHHC-PATs. Among the acyl-CoAs with different acyl chain lengths, palmitoyl-CoA (n=11) has the highest activity with zDHHC20. (B) Chemical structures of lipid-based DHHC inhibitors. (C) Chemical structures of DHHC inhibitors identified from library screening. (D) Chemical structure of CMA and its alkyne-containing analogue, compound 8.

Early attempts at unbiased in vitro screening have yielded several DHHC inhibitor candidates with diverse scaffolds. For example, in 2006, Compound V was identified from a screen of 8000 compounds using an in cellulo imaging-based assay and an in vitro peptide-based LC-MS assay against PATs targeting C-farnesylated and N-myristylated peptides [40] (Figure 2C). Compound V was then shown to inhibit zDHHC2 and zDHHC9/GCP16 using an in vitro radiolabeling assay in 2009 [42]. No in cellulo evaluation of Compound V has yet been reported, however, the encouraging in vitro data indeed calls for further evaluation of this molecule.

Several inhibitors with diverse scaffolds have been identified from screens targeting S-acylation of specific protein substrates. For example, two potential hits that inhibited NRas acylation at 50 μM were found from an in vitro click-based assay [52]. Ketoconazole, an FDA-approved anti-fungal drug, emerged as a hit compound that inhibits the S-acylation of Dual leucine-zipper kinase (DLK) [41]. Several peptides were also shown to inhibit S-acylation of specific targets in cells [53,54]. It is worth noting that the origin of the inhibitory effect of these molecules remains unknown, as the reduction of S-acylation on protein substrate may not be exclusively due to the direct inhibition of DHHC proteins in cells. Nevertheless, it is of interest to test for direct DHHC engagement of these molecules and their derivatives by further in vitro or in cellulo assays.

Recently, our group identified CMA, a new broad-spectrum DHHC inhibitor that functions both in vitro and in cellulo [47] (Figure 2D). The remainder of this review will be dedicated to 2BP and CMA, the only two DHHC inhibitors whose S-acyltransferase inhibition have been cross-validated in vitro and in cells with clear evidence of on-target activity in phenotypic experiments.

2-Bromopalmitate: a widely-used but problematic broad-spectrum DHHC inhibitor

Since 2-bromopalmitate (2BP) was firstly recognized as an inhibitor of Fyn kinase S-acylation in 2000 [48], this molecule is generally considered to be a pan-inhibitor of S-acylation and has been shown to perturb various S-acylation events in cells. For example, inhibition of PSD95 S-palmitoylation by 2BP led to decreased density of PSD95 clusters and a loss of synaptic AMPA receptors [55]. In 2009, 2BP was found to irreversibly inhibit S-acylation by disrupting the autoacylation of purified human zDHHC2, zDHHC9/GCP16, and their homologues in yeast [42]. The in vitro PAT inhibition of 2BP was further confirmed by in cellulo labeling with alkyne/azide-containing 2BP analogues as well as the crystal structure, which suggested that 2BP inhibits the autoacylation of DHHC proteins by occupying the acyl chain binding cavity and covalently labeling the Cys in the DHHC motif [32,56,57] (Figure 1D).

However, the drawbacks of 2BP — cytotoxicity and promiscuity — though widely known, are simultaneously often overlooked. High concentrations and prolonged treatment times are often required to inhibit S-acylation in cells, but intolerance for long-term treatment with 2BP, and its azide-containing analogue, at or near its working concentration has been observed in several cell lines, including Jurkat-T cells (100 μM 2BP in serum-supplemented media overnight) and HEK293T cells (> 50 μM 2BP-azide in serum-supplemented media for 1 hour) [32,51]. It was also reported that short-term treatment with 250 μM 2BP diminished hyperpolarization of the inner mitochondrial membrane leading to substantial ATP depletion in rat adipocytes [58]. Correspondingly, our data suggested that common laboratory cell lines, such as HEK293T, HeLa, HepG2, MDA-MB-231, and 3T3-L1 pre-adipocytes, show signs of cytotoxicity when treated for 6–24 hours with 2BP in serum-free media at concentrations of 20–80 μM.

The cytotoxic effects of 2BP are believed to be due to its promiscuity. In the 1990s, 2BP and its CoA adduct, 2BP-CoA, were reported to inhibit a series of membrane-bound enzymes, including fatty acid CoA ligases involved in fatty acid metabolism [59]. Further studies showed that while 2BP’s inhibition of DHHC seemed to be CoA-independent, in cells, 2BP could indeed be converted to 2BP-CoA, a more active species with broad reactivity across the proteome [56]. Additionally problematic for the study of S-acylation cycles in cells, 2BP was found to interrupt the S-depalmitoylation of GAP43 by inhibiting APT1 and APT2, two critical S-acylation erasers, in cellulo and in vitro [60]. Structural analysis suggests that 2BP non-covalently inhibits APT1[61], and the reported inhibitory effects of 2BP on APT1 and APT2 were consistent with data from our group [47]. Most notably, these results suggest that 2BP, the canonical inhibitor of DHHC proteins, does in fact inhibit both S-acylation and S-deacylation within its range of working concentrations. Apart from S-acylation, 2BP can also disturb other post translational modifications (PTMs). For example, 2BP was found to inhibit both N-myristoylation and S-palmitoylation of Fyn kinase in Jurkat T cells[48]. Therefore, the use of 2BP as a putative DHHC inhibitor should be considered with caution, particularly when the S-acylation events under interrogation are dynamically regulated or involve other PTMs.

N-cyanomethyl-N-myracrylamide (CMA): a newly-identified broad-spectrum DHHC inhibitor

In an effort to develop a broad-spectrum DHHC inhibitor that overcomes some of 2BP’s major drawbacks, we explored the use of acrylamide as a covalent warhead. The acrylamide, ideally provides increased specificity for Cys over the 2-bromo electrophile of 2BP [62,63]. Moreover, removing the carboxylate present in 2BP would preclude the ability of the inhibitor to be converted to CoA moieties in cells, again ideally enhancing selectivity and predictability of the responses.

N-cyanomethyl-N-myracrylamide (CMA), featuring a C₁₄-aliphatic chain and an N-cyanomethyl group, was identified as an inhibitor hit on zDHHC20 (Figure 2D). Further in vitro analysis on zDHHC20 showed CMA’s slight improvement over 2BP in terms of IC₅₀ value (1.35 ± 0.26 μM vs 5.33 ± 0.77 μM with 1-hour preincubation), dissociation constant with zDHHC20 (K_i = 5.8 μM vs 6.4 μM), and rate constant _t (k_inac = 2.08 min⁻¹ vs 1.35 min⁻¹). Our structure-activity relationship studies suggested that the N-cyanomethyl group is indispensable for CMA’s inhibition of zDHHC20, although the broader structural basis of activity is not yet clear.

In live cells, CMA decreases global S-acylation level at the proteome level as well as on specific substrates, including GobX, Myd88, and Ras when applied at 20 μM in serum-free media for 6 hours. Critically, CMA induced less cytotoxicity across a panel of cell lines (HEK293T, HeLa, HepG2, MBA-MB-231, and 3T3-L1 pre-adipocytes) at 20 μM in serum-free media – the working concentration defined in our lab. Additionally, CMA showed no inhibition on purified human APT1 and APT2 in vitro, even at 50 μM, while we found 2BP significantly inhibited both eraser enzymes at 12.5 μM, and completely abolished their activity at 50 μM.

Treatment with CMA reproduced key reported S-acylation-related phenomena observed with either treatment of 2BP or knockdown of DHHC proteins. Treatment of MDA-MB-231 cells with CMA reduced EGFR acylation, and inhibited downstream AKT phosphorylation [18]. Additionally, treatment of NIH-3T3 L1 cells blocked CD36 acylation and lipid uptake and droplet formation [17,64].

Mechanistic interrogation using zDHHC20 revealed that CMA and 2BP compete with each other’s alkyne-containing analogues, implying that CMA shares a similar mechanism of action with 2BP. CMA also displayed a similar DHHC labeling spectrum as 2BP; in cellulo labeling with compound 8 (Figure 2D), an alkyne-containing analogue of CMA, indicated that compound 8 labeled overexpressed DHHC proteins that were also labelled by 16C-BYA, the alkyne-containing analogue of 2BP, with the exception of zDHHC3 and zDHHC12. This disparity is likely due to the difference in DHHC selectivity between CMA and 2BP, for which the structural basis remains unknown. Nonetheless, our data collectively confirms CMA’s engagement of zDHHC 2, 4, 5, 6, 9, 11, 13, 14, 15, 16, 18, 20, 23, and 24. The technical challenges – low expression levels, unavailability of adequate antibodies, and difficulty in detection by LC-MS/MS – precluded our analysis of CMA engagement and comparison with 2BP on other DHHC proteins.

Although CMA offers many advantages over 2BP as mentioned earlier, the following should be noted about the usage of CMA in cells. Firstly, our data suggests that the inhibitory effects of CMA could be impeded by serum. Higher concentrations of CMA will likely be needed if cell culture experiments require the presence of serum. Secondly, in situ labeling showed variable sensitivity of DHHCs on compound 8 and therefore CMA, which was consistent with the global ABE blots. Thus, the optimal CMA concentration can be target-specific. Finally, our preliminary proteomics in HEK293T cells revealed over 271 protein targets could be labelled by compound 8 and therefore potentially by CMA as well. Off-targets (i.e., targets other than DHHC family members) generally included a large number of membrane proteins or lipid-binding proteins, like FABP5 and HACD3, and over 75% percent of off-targets could be S-acylated. Therefore, more orthogonal approaches and careful validation should be taken when studying CMA-induced reduction in S-acylation and the downstream consequences.

Future outlook

There is a clear disconnect between our understanding of the biological significance of DHHC proteins and development of corresponding tools. Although direct genetic manipulation of DHHCs is useful, the robustness of genetic approaches is sometimes hindered by the functional redundancy and within-family regulation between DHHC proteins. On the other hand, an advantage of chemical inhibitors lies in the rapid perturbation of DHHC activity, reducing confounds stemming from compensatory expression of functionally redundant proteins. Moreover, isoenzyme-selective and pan-active inhibitors can block either a specific or a suite of related proteins, respectively, thereby clearly revealing the effects of the DHHC-mediated S-acylation events. Finally, targeting DHHC proteins with chemical inhibitors allows direct evaluation of their potential as drug targets in cells and could serve as the basis for future drug development.

Moving forward, identification of selective, covalent inhibitors of the whole DHHC family, or subgroups of DHHCs, will likely require new molecular scaffolds. One approach is to further optimize CMA, for example, by replacing the acyl chain with a more drug-like moiety in order to reduce lipid off-targets. Another approach would be to continue to seek out novel covalent scaffolds from unbiased libraries. Recent proteomic profiling of the targets of a large electrophile library suggested that some molecules label some of the DHHCs [65]. Therefore, it would be of great interest to evaluate their effects on enzyme activity. Although there is no evidence of direct engagement, inhibitors of specific targets’ S-acylation, like ketoconazole or others discovered from non-covalent molecule library screening, could be examined with purified DHHCs and possibly improved via installation of a covalent warhead. To do so, confirmation of target engagement on DHHC proteins and additional structural and computational studies would greatly aid in predicting optimal warhead positioning.

Discovering selective peptide binders is another promising strategy to develop broad spectrum or individually selective DHHC inhibitors. Although no consensus DHHC substrate sequence has been identified, several studies have reported some short peptides that specifically bind certain DHHC proteins and thus inhibit the protein-protein interaction [37], which serves as strong motivation for mapping short peptide binders of either the entire DHHC family or a subset of DHHC proteins of interest.

Improved DHHC assays, especially cell-based assays, are also needed in this field. While a robust toolkit to interrogate the APT “erasers” of S-acylation in live cells now exists [66–69], methods for measuring DHHC activity in cells fully rely on tracking the S-acylation of protein substrates [41] to report on the results of complex regulation, including the dynamic regulation of multiple DHHCs and APTs, and not on DHHC activity per se. An in cellulo DHHC-specific detection system would facilitate both the screening and characterization of DHHC inhibitors, as well as the regulatory study of DHHCs.