Profiling and Inhibiting Reversible Palmitoylation

Abstract

Protein palmitoylation describes the post-translational modification of cysteines by a thioester-linked long chain fatty acid. This modification is critical for membrane association, spatial organization, and the proper activity of hundreds of membrane-associated proteins. Palmitoylation is continuously remodeled, both by spontaneous hydrolysis and enzyme-mediated de-palmitoylation. Bioorthogonal pulse-chase labeling approaches have highlighted the role of protein thioesterases as key regulators of palmitoylation dynamics. Importantly, thioesterases are critical for regulating the spatial organization of key oncogenic proteins, such as Ras GTPases. New inhibitors, probes, and proteomics methods have put a spotlight on this emerging post-translational modification. These tools promise to advance our understanding the enzymatic regulation of dynamic palmitoylation, and present new opportunities for drug development.

Introduction

Protein palmitoylation was first reported just a few months before the classic discovery of tyrosine phosphorylation[1–2], yet more than 30 years later, the importance of palmitoylation is only now gaining significant attention as a widespread, dynamic post-translational modification. This is likely due to a historical lack of robust methods for sensitive analysis of this non-polar, non-antigenic modification. Until recently, the only method to study palmitoylation involved metabolic labeling with [³H]-palmitate, followed by lengthy exposure times ranging from days to weeks. Given the lack of straightforward methods, the dynamics and regulation of protein palmitoylation is largely unexplored.

Protein palmitoylation is clearly important in establishing the spatial localization of many well studied signaling complexes. Cellular transformation by oncogenic v-Hras (H-Ras^G12V) requires membrane anchoring[3–4], and mutation of a single palmitoylation site eliminates the protein’s oncogenic potential[3]. The rate of palmitate turnover on inactive GDP-bound H-Ras is accelerated > 15 times upon activation[5]. Similarly, activation of G-alpha-s accelerates palmitate turnover nearly 50-fold[6]. Similar findings have been observed for the synaptic scaffolding protein PSD-95, which is rapidly depalmitoylated following glutamate stimulation[7]. Based on these observations, dynamic palmitoylation may be a general regulatory mechanism controlling signal-dependent spatial localization.

The goal of this review is to present recent advances for the detection, annotation, and quantification of dynamic palmitoylation, as well as a discussion of the potential for thioesterase inhibitors to modulate key signaling pathways.

Non-radioactive detection of Palmitoylation

Two complementary methods have been developed in recent years for the non-radioactive detection, enrichment, and mass spectrometry-based annotation of palmitoylated proteins. The first method, termed acyl-biotin exchange, is useful for the static analysis of palmitoylated proteins in native tissues or cells[8–10]. In this method, lysates are first treated with methyl methanethiosulfonate (MMTS) or maleimide to block free thiols. Next, thioesters are hydrolyzed with hydroxylamine, which releasing the acyl chain and exposes new free thiols for disulfide capture[11]. One drawback to this approach is the enrichment of proteins with native thioesters, such as ubiquitin ligases and lipoamide-linked dehydrogenases. New modifications of this approach employ thiol resins for more simplified enrichment[12] (Figure 1A).

Figure 1.

Methods for palmitoylated protein enrichment. (A) Resin-assisted capture of hydroxylamine-sensitive cellular thioesters for static analysis of palmitoylation. After reduction and alkylation, lysates are treated with hydroxylamine to hydrolyze thioesters. Free thiols are captured by disulfide formation using thiopropyl sepharose resin. (B) Bioorthogonal enrichment of 17-ODYA metabolically labeled sites of palmitoylation. Biotin-azide is conjugated by click chemistry to 17-ODYA labeled proteins for streptavidin enrichment.

The second method uses metabolic labeling with the bioorthogonal fatty acid analogue 17-octadecynoic acid. The alkynyl fatty acid analogue is incorporated by the endogenous palmitoylation machinery into native sites palmitoylation. After lysis, labeled proteins are ligated to azide-linked reporter tags by click chemistry[13–14] (Figure 1B). Importantly, all reagents are commercially available and relatively inexpensive. The key advantages are a simplified workflow, high sensitivity, reduced non-specific labeling, and the ability to examine palmitoylation turnover dynamics by classic pulse-chase methods. Unlike ABE, this method only enriches native sites of long-chain fatty acid modification, and not other endogenous thioesters[9–10].

Both enrichment methods have been used to globally annotate palmitoylated proteins by mass spectrometry in a variety of organisms, tissues, and cell lines[9–10,13,15–17]. Altogether, more than 500 palmitoylated proteins have been annotated in mammalian cells. This list contains both integral and membrane-associated proteins, including channels, receptors, and scaffolding proteins. Based on these results, there are likely thousands of palmitoylated cysteine residues in the proteome[15], solidifying protein palmitoylation as pervasive as other widely studied polar post-translational modifications.

Quantitative Analysis of Palmitoylation

Ras is the prototypical palmitoylated protein, and has been used as a model to study the spatial organization, dynamics, and turnover of protein palmitoylation. Upon microinjection of fluorescent, palmitoylated N-Ras, the semi-synthetic protein rapidly distributes to all membranes, and enters a pathway of dynamic palmitoylation and de-palmitoylation[18–19]. N-Ras is quickly de-palmitoylated in the periphery, but re-palmitoylated at the golgi and recycled back to the plasma membrane through the secretory pathway. Complementary live-cell fluorescence imaging with transfected photo-convertible fluorescent protein fusions confirmed these observations. Based on these experiments, palmitoylation is hypothesized to stabilize the membrane attachment and increase the residency time of N-Ras at the plasma membrane[20]. This specific example demonstrates how dynamic palmitoylation can promote spatial organization and function of a key oncogenic signaling protein. Despite these promising observations, there is little evidence studying the dynamics of native proteins to support these findings.

Bioorthogonal metabolic pulse-chase labeling provides a direct approach to profile the global dynamics of palmitoylation on native proteins. Preliminary studies in Jurkat T-cells demonstrated an increase in the rate of Lck de-palmitoylation after pervanadate treatment[21]. Importantly, the rate of palmitoylation turnover was attenuated by treatment with the generic lipase inhibitor methyl arachidonyl fluorophosphonate (MAFP)[22]. This was the first demonstration on native proteins that unspecified serine hydrolases regulate the turnover of palmitoylation on native proteins.

To measure fractional changes in palmitoylation, SILAC quantitative proteomics methods[23] were applied in conjunction with metabolic 17-ODYA and enrichment. BW5147-derived mouse T-cell hybridoma cells were first passaged in isotopic media according to standard SILAC protocols[24]. First, a list of high confidence palmitoylated proteins was established by performing two parallel experiments. In the first experiment, light cells were treated with 17-ODYA for 8 hours, and the heavy cells with palmitic acid, or vice versa (Figure 2). In a second experimental group, heavy and light cells were both treated with 17-ODYA for 8 hours, and one cell lysate was incubated with hydroxylamine to cleave thioesters and release 17-ODYA. In both experimental groups, isotopic paired lysates were mixed at a 1:1 ratio, conjugated to biotinazide, enriched, and digested for high resolution mass spectrometry analysis. The relative enrichment ratio is calculated by comparing the precursor extracted ion chromatograph for each isotopic peptide pair (17-ODYA / Control). A ratio of 1 signifies no specific enrichment, but the majority of peptides were completely absent in the control sample yielding an infinitely high ratio. Approximately 400 proteins displayed specific enrichment in reciprocal labeling experiments in both experimental groups. Similar quantitative proteomics studies in mouse neuronal stem cells identified a partially overlapping set of approximately 300 palmitoylated proteins[17]. These studies highlight the value of reciprocal SILAC approaches for providing high confidence annotation of low abundance palmitoylated proteins.

Figure 2.

Quantitative proteomic analysis of palmitoylated proteins using SILAC. Cells are grown for several passages in media with light or heavy isotopic lysine and arginine. Next, SILAC pairs are labeled with either palmitic acid or 17-ODYA. After metabolic labeling, lysates from isotopic pairs are mixed and conjugated to biotin-azide by click chemistry for high resolution mass spectrometry identification and quantification. Dashed arrows represent replicate experiments switching the order of fatty acid / probe labeling, which is used for combined reciprocal analysis to reduce false positives.

Global dynamics of palmitoylation

The deacylation of Ras is accelerated by protein thioesterases[18–19], yet little is known about the commonality of this mode of regulation, and whether this is specific for small GTPases. Several enzymes have been ascribed protein thioesterase activity, including the lysosomal hydrolase palmitoyl protein thioesterase 1 (PPT1)[25], as well as the soluble lysophospholipases LYPLA1 (APT1)[26–27] and LYPLA2 (APT2)[28–29]. Given the possibility that multiple hydrolases contribute to dynamic palmitoylation, we sought to develop tools for defining the subset of palmitoylated proteins regulated by any combination of lipases.

Any candidate protein thioesterase is likely a member of the serine hydrolase enzyme family[30] with preference towards lipidated substrates. The palmitate analogue hexadecylfluorophosphonate (HDFP) covalently inhibits approximately 20 lipases in T-cell hybridoma cells (including APT1, APT2, and PPT1), without inhibiting serine peptidase, proteases, or small molecule hydrolases[24]. To explore the global dynamics of palmitoylation turnover, SILAC pairs of T-cell hybridoma cells were pulse-labeled with 17-ODYA, and two groups of experiments were performed, one to determine the palmitoylated proteins with rapid turnover, and a second to profile palmitoylated proteins stabilized by HDFP[24]. Using SILAC quantitative analysis, both experiments were performed without over-expression and analyzed globally on native palmitoylated proteins after bioorthogonal enrichment.

Pulse-chase proteomic analysis revealed several important findings (Figure 3). First, the majority of palmitoylated proteins did not change over the 4 hour chase time period, suggesting dynamic palmitoylation is not a general phenomenon. Secondly, there were several dozen palmitoylated proteins with both rapid turnover dynamics and stabilized by HDFP. This group includes many Ras family GTPases, G-alpha proteins (GNAS and GNA13), and membrane-associated guanylate kinase (MAGUK) proteins (MPP1 and MPP6), which all belong to enzyme families previously characterized as targets of dynamic palmitoylation turnover[31], as well as many other many other palmitoylated proteins implicated in cancer pathogenesis[32–33]. This study does not quantify specific sites of palmitoylation, which likely biases the results towards palmitoylated proteins with fewer, simultaneously regulated sites of palmitoylation. Also, longer pulse-labeling times are required for sufficient 17-ODYA labeling for mass spectrometry detection, making it more difficult to remove 17-ODYA from cellular lipid pools and thus reducing the chase efficiency. Therefore, these findings only highlight the most direct targets of thioesterases with the shortest half-lives. Such global experiments establish a role for protein thioesterases in the regulation of select palmitoylated proteins with established roles in cell growth and cancer.

Figure 3.

Pulse-chase proteomic analysis of dynamic palmitoylation. In experiment 1 (x-axis), SILAC cell pairs were labeled with 17-ODYA for 2 hours, then harvested or chased with 10-fold excess palmitic acid for 4 hours. This experiment highlights proteins with rapid de-palmitoylation or protein turnover. In experiment 2 (y-axis), SILAC cell pairs were labeled with 17-ODYA for 2 hours, and chased with excess palmitic acid with or without the lipase inhibitor HDFP. This experiment highlights palmitoylated proteins where probe labeling is stabilized by lipase inhibition. Blue lines are arbitrary thresholds. Red dots are enzymatically regulated proteins with rapid turnover kinetics. Green dots are proteins with enhanced degradation, as determined by unenriched SILAC proteomics experiments. Adapted with permission from [24]. Copyright (2012) Nature Publishing Group.

Protein thioesterases

Acyl-protein thioesterase 1 (LYPLA1) was first characterized as a lysophospholipase[27], but has several hundred-fold higher activity as a protein thioesterase[26]. Genetic deletion of the yeast homologue of LYPLA1 eliminates the in vitro de-palmitoylation activity from lysates, yet shows a very minor decrease in the turnover of palmitate on GNAS in vivo[34]. Furthermore, LYPLA1 mutant yeast have no defects in growth, mating, or deacylation of other palmitoylated proteins[34]. In mammalian synapses, LYPLA1 mRNA is silenced by the microRNA mi138[35]. After glutamate stimulation, microRNA suppression is relieved by RISC degradation, inducing local synaptic translation of LYPLA1 mRNA[36]. Suppression of LYPLA1 results in a partial reduction in synaptic volume, although there is no direct evidence this is due to the de-palmitoylation activity, and no synaptic protein substrates of LYPLA1 have been identified[35]. Overall, more genetic and biochemical evidence is needed to fully support the hypothesis that endogenous LYPLA1 is responsible for all enzymatic palmitoylation turnover.

Inhibitors of Dynamic Palmitoylation

Protein thioesterases regulate the dynamic palmitoylation turnover on Ras and other important proteins involved in the progression of cancer, making LYPLA1 an attractive candidate for inhibitor development. The over-the-counter weight loss drug tetrahydrolipstatin (orlistat) inhibits a diverse subset of lipases to prevent intestinal lipid uptake[37–38]. This natural product contains a reactive β-lactone that is targeted by the nucleophilic serine of a wide range of hydrolases[39]. Careful diversification led to the discovery of a potent LYPLA1-directed, stereo-specific β-lactone inhibitor, named Palmostatin B[40] (Table 1). This compound blocks the acylation cycle of microinjected or transfected Ras in cells, and inhibits the proliferation of myeloid cells transformed with oncogenic H-Ras or N-Ras, but not K-ras4B (which is not palmitoylated)[41]. Competitive activity-based protein profiling using an alkynyl Palmostatin derivative shows inhibition of LYPLA1, LYPLA2, and PPT1 by both Palmostatin B, and the higher potency second generation derivative Palmostatin M[42–43] (Table 1). In vitro, PPT1 has broad thioesterase activity towards many palmitoyl proteins, but is a lysosomal resident hydrolase[44] and unlikely to contribute to dynamic palmitoylation turnover of plasma membrane-bound proteins. Mutations in PPT1 lead to the human neurodegenerative lysosomal storage disease infantile neuronal ceroid lipofuscinosis[45]. Mass spectrometry analysis shows a wide array of other enriched cellular targets[42], highlighting the need for more selective tools to separate the functional roles of each candidate thioesterase.

Table 1.

LYPLA1 and LYPLA2 inhibitors.

Inhibitor	Structure	LYPLA1 Inhibition	LYPLA2 Inhibition	References
Palmostatin B		IC50 = 5.4 nM	IC50 = 37.7 nM	[40,42–43]
Palmostatin M		IC50 = 2.5 nM	IC50 = 19.6 nM	[42–43]
LYPLA1 Inhibitor 21		Ki = 300 nM	Ki >10 µM	[47]
LYPLA2 Inhibitor 1		Ki >10 µM	Ki = 230 nM	[47]

In order to develop selective inhibitors with more favorable drug-like properties, both LYPLA1 and LYPLA2 were screened in parallel against the NIH Molecular Library of 315,004 compounds using the robust competitive fluorophosphonate-rhodamine fluorescence polarization high-throughput assay (FluoPol-ABPP)[46–47]. Active site occupancy prevents FP-Rhodamine labeling of the enzyme, and hence reduces the fluorescence polarization signal[46]. This time-dependent competitive activity-based assay selects for active site inhibitors with potent binding and long residency times. Lead inhibitors for both LYPLA1 and LYPLA2 converged on a common piperizine amide chemotype[47]. In vitro competitive activity-based protein profiling (ABPP) of initial hits led to the identification of highly selective LYPLA1 and LYPLA2 inhibitors despite their significant homology (68%) (Table 1). Steady-state kinetic analysis reported K_i values of 230 nM (LYPLA2) and 300 nM (LYPLA1)[47]. No other T-cell hybridoma serine hydrolases were inhibited in competitive ABPP-SILAC experiments. Furthermore, these inhibitors maintain potency and selectivity in vivo in mice. These improved reversible probes promise to assign the individual functional roles of both LYPLA1 and LYPLA2, and accelerate the in vivo study of dynamic palmitoylation.

Conclusions

Emerging analytical tools and inhibitors have greatly accelerated our understanding the role of dynamic palmitoylation in the spatial organization of cellular signaling complexes critical for cell growth and membrane polarization. Non-radioactive, bioorthogonal labeling approaches have initiated a new focus on the uncharacterized dynamics of palmitoylation. Exploring the competitive role of palmitoylation with other cysteine modifications may begin to unravel the fundamental role of the palmitoylation cycle and the importance of specifically targeting a subset of palmitoylated proteins for dynamic turnover.

Highlights.

New methods allow direct enrichment of palmitoylated proteins for LC-MS annotation.

SILAC proteomics methods enhance the annotation of low abundance palmitoyl proteins.

Alkynyl fatty acids facilitate pulse-chase labeling for measuring palmitoylation dynamics.

Thioesterase inhibitors are promising new tools to dissect palmitoylation dynamics.