Cell-Based Proteome Profiling Using an Affinity-Based Probe (AfBP) Derived from 3-Deazaneplanocin A (DzNep)
Abstract: 3-Deazaneplanocin A (DzNep), a global histone methylation inhibitor, has attracted significant inter- est in epigenetic research in recent years. The molecular mechanism of action and the cellular off-targets of DzNep, however, are still not well-un- derstood. Our aim was to develop novel DzNep-derived small-molecule probes suitable to be used in live mam- malian cells for identification of poten- tial cellular targets of DzNep under physiologically relevant settings. In the current study, we have successfully de- signed, synthesized, and tested one such probe, called DZ-1. DZ-1 is a ‘clickable’ affinity-based probe (AfBP) derived from DzNep with minimal structural modifications. The probe was found to be highly cell-per- meable, and possessed similar anti- apoptotic activities as DzNep in MCF- 7 mammalian cells. Two additional con- trol probes were made as negative la- beling/pull-down probes in order to minimize false identification of back- ground proteins due to unavoidable, in- trinsic nonspecific photo-crosslinking reactions. All three probes were subse- quently used for in-situ proteome profiling in live mammalian cells, followed by large-scale pull-down/LC-MS/ MS analysis for identification of poten- tial cellular protein targets that might interact with DzNep in native cellular environments. Our LC-MS/MS results revealed some highly enriched proteins that had not been reported as potential DzNep targets. These proteins might constitute unknown cellular off-targets of DzNep. Though further validation experiments are needed in order to un- equivocally confirm these off-targets, our findings shed new light on the future use of DzNep as a validated chemical probe for epigenetic research and as a potential drug candidate for cancer therapy.
Introduction
Post-translational modification of histones is an essential event for the regulation of chromatin structure and gene ex- pression in eukaryotes. 3-Deazaneplanocin A (DzNep, Figure 1) is a carbocyclic analog of adenosine that was origi- nally developed as a potent inhibitor of S-adenosylhomocys- teine hydrolase (SAHH) for potential therapeutic applications.[1] In recent years, there has been renewed interest in the application of DzNep in epigenetic research, primarily due to the discovery that DzNep could modulate histone methylation by disrupting the activity of histone-lysine N- methyltransferase EZH2.[2] The latter is a catalytic subunit of the polycomb repressive complex 2 (PRC2) that functions to silence tumor suppressor genes by the addition of three methyl groups to lysine 27 of histone 3, a modification lead- ing to chromatin condensation.[3] EZH2 is also known to associate with numerous cellular proteins, such as the embryonic ectoderm development protein, the VAV1 oncoprotein, and the X-linked nuclear protein (XNP).[4] It may also play a role in the hematopoietic and central nervous systems. EZH2 overexpression has been shown to have implications in tumorigenesis and to correlate with poor prognosis in sev- eral tumor types.[5] EZH2 inhibition by DzNep has been documented in cultured cancer cell types, including primary and acute myeloid leukemia (AML) cells.[6,7] For example, human AML cells treated with DzNep were found to have elevated expression levels of several cell-cycle regulators in- cluding p21, p27, and FBXO32, before undergoing cell cycle arrest and apoptosis.[6] These findings have led to the pursuit of using DzNep and similar compounds as potential thera- peutic agents in the treatment of cancer.[7] Notwithstanding the intensive efforts on DzNep in epigenetic research, exist- ing biochemical and cellular data on DzNep and its potential biological targets are at best inconclusive.[8] As an ade- nosyl nucleoside analog, DzNep might be expected to bind to a variety of other cellular proteins that possess intrinsic binding affinity toward AMP/ADP/ATP, S-Adenosyl me- thionine (SAM), NADH/NADPH, and other nucleotide an- alogs, in addition to its putative targets SAHH and EZH2. Many such proteins are present in mammalian cells, includ- ing kinases, methyltransferases, ATPases, oxidoreductases, and others. It is therefore imperative to decipher the precise molecular targets of DzNep and to elucidate its mechanism of action. Strategies capable of large-scale proteome-wide analysis and identification of potential cellular targets of DzNep under physiological settings have, however, not been reported thus far.
Figure 1. (A) Overall workflow of the cell-based proteome profiling approach followed by large-scale pull-down/LC-MS/MS for identification of poten- tial cellular targets of DzNep using affinity-based probes (AfBPs) shown in panel B. (B) Structure of DzNep, the DzNep-derived probe DZ-1, and the negative control probes NP-1 and NP-2. The two click reporters, rhodamine-N3 and biotin-N3, used in the current study are also shown.
Several chemical profiling strategies for large-scale, pro- teome-wide protein target identification exist, but most of them use affinity beads immobilized with the compound of interest (e.g., the probe) to capture its interacting proteins from cellular lysates.[9,10] In more recent examples, protein– drug interactions were studied under native cellular settings by adding the drug of interest directly to cells and initiate binding to its intended cellular targets. Upon cell lysis, im- mobilized probes were then added to the cell lysates as “baits”. By measuring the amount of a protein captured on beads by quantitative mass spectrometry, with and without drug treatment, this method, when compared to previous methods,[9] is a step forward in the right direction and ena- bles the in-situ profiling of protein–drug interactions indi- rectly.[10] Nevertheless, the strong reliance on bead-bound probes and/or cellular lysates to capture protein–drug inter- actions inevitably leads to many ‘‘false positives’’ and ‘‘missed targets’’ in all these strategies, as genuine protein–drug interactions (defined hereafter as those that occur in living cells under physiologically relevant conditions) are governed by a variety of other factors besides the protein/li- gands themselves, including protein compartmentalization and localized protein concentrations. Our long-term re- search aim has been to develop chemical strategies capable of interrogating protein–drug interactions in native cellular environments (i.e., in live cells, not cell lysates).[11,12] This calls for the design of small-molecule probes that are cell- permeable, minimally disrupt protein–drug interactions, and effectively capture interacting cellular targets in situ (Fig- ure 1 A).[12] Previously, we had successfully carried out sever- al large-scale liquid chromatography–tandem mass spec- trometry (LC-MS/MS) experiments on small-molecule drugs (e.g., Orlistat and Dasatinib) by using such design principle- s.[12a,c] From these studies, we have firmly established that such a cell-based proteome profiling approach is indeed ef- fective in the rapid identification of potential cellular targets of biologicallly relevant small molecules. A key element to the success of this approach is the design of suitable cell-per- meable, small-molecule probes that are modified versions of the target compounds but are capable of recapitulating most if not all the genuine protein–interacting events under phys- iological settings (Figure 1 B). As part of our on-going inter- est in DzNep and related compounds, we have undertaken the challenge of developing DzNep-derived probes that might be useful for identification of potential cellular targets (both on- and off-targets) of DzNep. Herein, we report the successful design, synthesis, and biological testing of DZ- 1 (Figure 1 B&Scheme 1), a ‘clickable’ affinity-based probe (AfBP) derived from DzNep with minimal structural modi- fications. The probe was found to be highly cell-permeable and to possess similar anti-apoptotic activities as DzNep in MCF-7 mammalian cells. We have successfully carried out in-situ proteome profiling of DZ-1 in MCF-7 cells, followed by large-scale pull-down LC-MS/MS analysis for the iden- tification of potential cellular protein targets of DzNep. Our LC-MS/MS results revealed some highly enriched proteins that were previously unknown DzNep targets. These might constitute potential cellular off-targets of DzNep.
Scheme 1. Synthesis of probes DZ-1, NP-1, and NP-2 used in this study. DIAD=diisopropyl azodicarboxylate, DMAP= 4-dimethylaminopyridine, TBAF=tetra-n-butylammonium fluoride, TBAI=tetrabutylammonium iodide, TMSN3 =trimethylsilyl azide.
Results and Discussion
Overall Workflow of Cell-Based Proteome Profiling
As shown in Figure 1 A, in our cell-based proteome profiling approach, a cell-permeable DzNep-derived probe (e.g., DZ- 1) bearing a photoreactive group and an alkyne handle was used to convert non-covalent protein–drug interactions into covalent linkage in situ. Upon cell lysis, these protein–probe cross-linked complexes would be subsequently conjugated to an azide-containing reporter tag (e.g., rhodamine-N3 or biotin-N3) by the well-established bioorthogonal click chemistry.[13] Further in-gel fluorescence scanning and/or pull-down (PD)/LC-MS/MS analysis would enable large- scale identification of potential DzNep-binding cellular pro- teins. This approach could alleviate many of the aforemen- tioned problems encountered in the ‘traditional’ PD ap- proaches,[9,10] in that it would enable genuine noncovalent protein–probe intereactions to be faithfully captured in situ prior to downstream in-vitro cell lysis, protein enrichment, and LC-MS/MS analysis for target identification.
Design and Synthesis of DZ-1
The probe DZ-1 contained several critical elements of design needed for a successful cell-based proteome profiling experiment. First, the probe preserved most of the core structure in DzNep, namely the adenine and the cyclopen- tendiol moieties, thus retaining most if not all of the pro- tein-binding ability and apoptotic activities when compared with wild-type DzNep (see below). Second, a photo-reactive group that allows conversion of non-covalent protein–probe interaction in situ to covalent linkage upon UV irradiation was introduced into the probe, and this was accomplished through an aryl azide installed at the C2 position of the ade- nine group, which retained the essential structural features in adenine and, at the same time, delivered an aryl azide that can be converted into a highly reactive nitrene upon UV irradiation (254 nm). Finally, an aryl alkyne moiety was installed at a position remote from the core pharmacophore. Based on known structure–activity relationships of DzNep analogs and their apoptotic activities (data not shown),[1b] this position was not expected to interfere with protein– probe binding. The alkyne would serve as a convenient click reporter moiety for subsequent detection, purification, and isolation of probe-labeled proteins. For the purpose of re- ducing false-positive hits, which are intrinsic and in most cases unavoidable in protein photo-crosslinking experi- ments,[14] two negative probes, NP-1 and NP-2 (Figure 1 B), were designed and used in all subsequent cell-based pro- teome profiling and PD/LC-MS/MS experiments.
The synthesis of DZ-1 is summarized in Scheme 1. It com- menced from the optically pure O-protected cyclopentendiol 1 which was obtained from d-ribose in 8 steps.[15] Mitsunobu coupling between compound 1 and 2-amino-6-chloropurine gave the single adduct 2 in excellent yield (95 %). Unfortu- nately, subsequent direct amine-to-azide conversion on the purine ring of 2 to deliver the corresponding azide 5a in a single step was unsuccessful when n-butyl nitrite in the presence of trimethylsilylazide or sodium azide was used. Thus, a longer sequence was subsequently employed to ach- ieve the desired chemical transformation through the forma- tion of intermediates 3 and 4. In this synthetic route, the ar- omatic amine 2 was initially converted into the correspond- ing iodide 3 by a diazotization–iodination protocol. This was followed by amination at the 4-chloro position of the purine ring to furnish 4-amino-2-iodoadenine (i.e., compound 4). Next, the iodo group in 4 was converted into an azido functionality under copper-catalyzed microwave conditions, thereby giving rise to 5a (61 % yield). The corresponding amine side-product 5b was also obtained in 11 % yield. The exo-amino functionality on the purine ring of 5a was then protected as the N,N-diBoc derivative 6. Subsequent depro- tection of the trityl group in 6 was achieved by treatment with p-toluene sulfonic acid to yield the cyclopentendiol 7 in 74 % yield. Dess–Martin periodinane oxidation of 7 gave the corresponding aldehyde 8 (quantitative yield), which was used in a subsequent Grignard reaction without further purifications. The Grignard reaction between the aldehyde 8 and 4-[(trimethylsilyl)ethynyl]phenyl magnesium bromide delivered compound 9 as an inseparable 10:1 diastereomeric mixture (56 % yield). Finally, deprotection of the TMS group on the alkyne with TBAF and treatment with HCl gave the desired target probe DZ-1 as a mixture of diaste- reomers (58 % yield over two steps).
The synthesis of the two negative probes, NP-1 and NP-2, is summarized in Scheme 1. Probe NP-1 was prepared in three steps from (4-aminophenyl)methanol. The first step in- volved the formation of azide 11 by diazotization–azidation of the aromatic amine starting material with sodium nitrite and sodium azide in acetic acid at 0 8C (83 % yield). The hy- droxy group in 11 was then converted into a bromo group to give 12 (77 % yield). The subsequent SN2 reaction between the alkoxide form of 12 and pent-4-yn-1-ol furnished NP- 1 in 37 % yield. The other negative probe, NP-2, was synthe- sized in a three-step route. First, a direct coupling reaction between the chloropurine and pent-4-yn-1-ol was successful- ly carried out under Mitsunobu reaction conditions (with tri- phenylphosphine and diethyl azodicarboxylate as coupling reagents) to afford 13 in 24 % yield. Next, direct amine-to- azide conversion from 13 into 14 in a single step was suc- cessful by using n-butyl nitrite and trimethylsilyl azide (TMS-N3) in acetonitrile at 15 8C (46 % yield). Finally, amination of the chloro moiety on 14 gave the expected product NP-2 in 39 % yield. All probes including their inter- mediates were fully characterized by NMR spectroscopy and mass spectrometry before being used in subsequent bio- logical and proteomic experiments.
Biological Evaluation of the Probes
We first assessed the cell permeability of the probes. DzNep was used as the positive reference compound. Many small- molecule probes developed thus far have bulky reporter groups such as a fluorophore or biotin directly attached to the compound of interest.[16] This makes downstream protein analysis and isolation straightforward, but the main problem is that the chemical structure of the resulting probe becomes significantly different from that of the original unmodified compound. Consequently, the biological properties, includ- ing cell permeability, cellular localization, and protein-bind- ing ability, of such probes might be significantly altered. Our design of DZ-1 took a “minimalist” approach to minimize structural changes to DzNep, and we thus expected DZ- 1 and DzNep to possess similar cellular properties including cell permeability. To confirm this expectation, we subjected all three probes (DZ-1, NP-1, and NP-2) and DzNep to a bi- directional permeability assay in MDCK cells and the re- sults are summarized in Figure 2 A. With apparent permea- bility (Papp) values of 18.57, 10.16, and 32.35 mms—1 for DZ- 1, NP-1, and NP-2, respectively, all three probes had excellent cell permeability that is comparable to that of DzNep, which gave a Papp value of 12.29 mm s—1 under identical assay conditions. These results indicate that our DzNep-derived probes are suitable for experiments in live cells.
Figure 2. (A) Results of the cell-permeability assay with the three probes and DzNep. (B) Dose-dependent inhibition of cell proliferation in MCF-7 and DLD1 cancer cells with DzNep and the three probes (DZ-1, NP-1, and NP-2) by using the XTT anti-proliferation assay. The data represent the average and standard deviation for two trials. (C) In-vitro and in-situ proteome labeling profiles of MCF-7 and DLD1 by the three probes at different probe con- centrations (5 and 20 mM). Only the in-gel fluorescence scanning profiles are shown. (D) Bioimaging of MCF-7 and DLD1 cells treated with DZ-1 (5 and 20 mM). Panels 1, 3, 5, and 7: Cells treated with the probe followed by click chemistry with rhodamine-N3; panels 2, 4, 6, and 8: Merged images of the images shown in panels 1, 3, 5, and 7 with Hoechst-stained nuclei (pseudocolored in blue). Scale bar, 10 mm.
DzNep is known to induce apoptotic cell death in a number of cancer cells, such as MCF-7 (breast cancer), but not in normal cells.[2a,5] In addition, a combination treat- ment of DzNep and suberoylanilide hydroxamic acid (SAHA), which is a well-known histone deacetylase inhibi- tor on DLD-1 cells (a colorectal cancer cell line) was found to effectively induce cell death (unpublished results).[1b,6] In order to determine whether DZ-1 and DzNep possess simi- lar cellular activities, we performed anti-proliferative assays against MCF-7 and DLD1 cell lines to compare their cell- killing activities (Figure 2 B). Following drug treatment for 72 h, both DzNep and DZ-1 showed good anti-proliferative activities against MCF-7 cells. The effect was dose-depen- dent and noticeable even at concentrations as low as 1 mM. Side-by-side comparison of the two compounds showed that DzNep consistently displayed a slightly higher cell-killing effect than DZ-1 at all tested concentrations, thus indicating that the structural modifications introduced into DZ-1 did cause marginal but tolerable changes in the probe’s cellular activities. The fact that neither DZ-1 nor DzNep showed any effect against DLD1 cells indicates that these com- pounds alone were insufficient to induce apoptotic cell death in this cell line. The two negative control probes, NP- 1 and NP-2, showed no apparent inhibitory activities against either MCF-7 or DLD1 cell growth. This is expected as nei- ther probe possesses the two obligatory structural features of DzNep. Our results thus indicate that DZ-1 likely pre- served most if not all of DzNep’s original biological and cel- lular activities, and therefore may be used for subsequent cell-based proteome profiling experiments for large-scale identification of potential cellular targets (on- and off-tar- gets) of DzNep.
In-Vitro and In-Situ Proteome Profiling
Next, we carried out endogenous proteome labeling, under both in-vitro (cell lysates) and in-situ (live cells) settings, using both MCF-7 and DLD1 cell lines. All three probes (DZ-1, NP-1, and NP-2) were run side-by-side to compare their proteome labeling profiles under two different probe concentrations (5 and 20 mM). Upon probe labeling, followed by rhodamine-N3 click chemistry, the resulting samples were separated by SDS-PAGE and visualized by in-gel fluores- cence scanning (Figure 2 C). Relatively weak fluorescence labeling profiles were observed for both negative probes (NP-1 and NP-2) at both low and high concentrations (e.g., 5 and 20 mM, respectively), thus indicating that neither probe had any significant number of specific cellular targets. On the other hand, DZ-1 showed weak labeling at 5 mM probe concentration, but at 20 mM probe concentration strongly la- beled bands, which likely corresponded to specific cellular targets of the probe, started to appear. This indicates that the effective cellular concentraction of DZ-1 was relatively high. The overall number and intensity of the fluorescent bands produced by DZ-1 labeling were consistently higher than those produced by the two negative probes (e.g., NP- 1 and NP-2) under the same probe concentrations in both MCF-7 and DLD1 cell lines. This result also indicates that more cellular proteins were targeted by DZ-1 in identical proteome environments. There were obvious differences in the in-vitro and in-situ proteome profiles of both MCF-7 and DLD1 produced by DZ-1, suggesting that the cellular targets of this probe might be different under different cel- lular settings. Similar results were observed in our previous cell-based proteome profiling studies.[12] This underscores the significance of our approach in its ability to label endog- enous protein targets in live cells in their native cellular en- vironments. There were clear differences in the labeling be- tween the two cell lines, under both in-vitro and in-situ set- tings, indicating that DzNep likely has different cellular tar- gets in MLC-7 and DLD1 cells. Both in-vitro and in-situ proteome profiling experiments were repeated on cells pre- treated with an excessive amount of DzNep (Figure S1 in the Supporting Information). The results showed that most DZ-1 labeled bands were effectively competed away, indi- cating that these labeled proteins were likely true cellular targets of DzNep.
Bioimaging
There has been enormous interest in the development of small-molecule-based imaging probes capable of reporting in-vivo protein–drug interactions as well as enzymatic activi- ties.[17–19] In some recent examples, several groups have suc- cessfully reported small-molecule probes that could be used to image different endogenous kinases in live cells.[12c,19] Given the excellent cell permeability of DZ-1, as well as its ability to closely imitate endogenous cellular activities of DzNep, we wondered if this probe could also serve as a useful imaging reagent to study the cellular uptake and lo- calization of DzNep in live mammalian cells. It should be noted that, since DzNep could likely bind to a variety of dif- ferent cellular proteins, DZ-1 would therefore not be very useful in bioimaging of specific protein targets. We carried out live-cell imaging experiments in MCF-7 and DLD1 cells using DZ-1. First, the cells were treated with DZ-1 (5 and 20 mM) for 2 h. No cell death was observed at the end of probe treatment. Subsequently, the cells were irradiated by UV light to initiate covalent protein–probe linkage, thereby “fixing” the probe to its cellular targets/locations. Cells were then fixed with formaldehyde, permeabilized by using Triton X-100, and treated with rhodamine-N3 following our previ- ously optimized click chemistry protocols.[12c] The same cells were further stained with Hoechst to visualize their nuclei. Finally, the cells were imaged by confocal fluorescence microscopy (Figure 2 D). In MCF-7 cells, strong fluorescence signals were observed throughout the whole cell excluding the nucleus at 20 mM probe concentration. When the probe concentration was decreased to 5 mM, there was a corre- sponding decrease in the overall fluorescence signals of the cells, with most of the fluorescence localized mainly on the edge of cell nuclei, indicating that, upon uptake into cells, DzNep likely was localized to these regions under physio- logical settings. In DLD1 cells, we only observed significant fluorescence signals at 20 mM probe concentration, which was in accordance with both the anti-proliferative results and the in-situ proteome labeling experiments (Figure 2 B and 2 C) that suggested that DzNep had fewer cellular tar- gets in this cell line. Of the two putative cellular targets of DzNep, SAHH is a cytosolic protein and EZH2 is localized to the nucleus of mammalian cells. The fact that our imaging results showed mostly cytosolic but not nuclear localization of the DZ-1 probe provides yet another line of evidence that DzNep likely has many other unknown cellular targets that reside outside of the cell nucleus. The endogenous ex- pression of EZH2 was either too low to be detected by the probe, or EZH2 may not be a true cellular target of DzNep (see below).
Large-Scale Pull-Down/LC-MS/MS Analysis
Finally, we performed large-scale pull-down/LC-MS/MS ex- periments on live MCF-7 cells in situ labeled with DZ-1, NP-1, and NP-2 to identify potential cellular targets of DzNep using previously published procedures, with some modifications.[12c] NP-1 and NP-2 were used as the negative control probes in order to subtract background protein la- beling originating from intrinsic non-specific labeling caused by photo-crosslinking experiments.[14] Protein extracts from the labeled cells were enriched (following click-chemistry conjugation with biotin-N3) by avidin-agarose beads, sepa- rated by SDS-PAGE (see Figure S2 in the Supporting Infor- mation), subjected to in-gel trypsin digestion, and identified by LC-MS/MS analysis. The complete LCMS results of all three probes are shown in the Supporting Information, with selected results summarized in Figure 3. As in the case with most large-scale LC-MS/MS experiments, a large number of proteins hits were identified in our results (> 1000 different proteins). Among them, many were inevitably highly abun- dant, nonspecific proteins that were labeled indiscriminately in live cells due to their “sticky” nature and high levels. As a result, they were labeled by all three probes. These pro- teins were automatically eliminated as false positives. Of the remaining proteins, we focused on those that met the follow- ing two criteria: 1) proteins with an exponentially modified protein abundance index (emPAI) value of greater than 0.1, and 2) proteins with a protein score of greater than 30. The emPAI offers an approximate, label-free, relative quantita- tion of proteins in our pull-down samples based on the pro- tein concentration. In total, there were 41 such proteins present and they were deemed likely true binders of DZ-1. When we further grouped these proteins based on their functions (Figure 3 A), we found 6 methyltransferases, 7 kin- ases, 8 ATPases, and 5 phosphatases, most of which possess a nucleotide-binding pocket in their active sites and there- fore likely have an intrinsic binding affinity toward DzNep as well. Some of the most interesting proteins hits are also summarized in the table shown in Figure 3 B. Both MAPK and PKA, which are well-known protein kinases, as well as histone arginine methyltransferase and phosphatase methyl- transferase, which like EZH2 are methyltransferases that utilize S-adenosylhomocysteine as a cofactor, are well repre- sented in this list. While our list of potential cellular protein targets of DzNep shown in Figure 3 should be further vali- dated by other independent experiments and more detailed biological studies, the results themselves clearly support the assumption we set out to confirm at the start of the current study, that is, compounds such as DzNep, given their struc- tural homology to adenosine, are likely binders to a variety of cellular proteins. Interestingly, neither SAHH nor EZH2 was positively identified from our LCMS results (see the Supporting Information). This again indicates that either both proteins were not endogenously expressed to a signifi- cant extent to be effectively labeled by DZ-1 in our experi- ments or that they are not true cellular targets of DzNep.
Figure 3. (A) The number and percentage of each protein subgroup enriched from the MCF-7 cell line. (B) Representative proteins hits identified from MCF-7 cells. “1” signals detection only in positive PD/LC-MS/MS (see the Supporting Information for details).
Conclusions
We have successfully developed a “clickable”, cell-permea- ble probe, DZ-1, suitable for cell-based proteome profiling and target identification of DzNep in live MCF-7 cells. The probe possessed similar cell permeability and cellular apop- totic activities as the wild-type DzNep. Through a cell bio- imaging experiment we confirmed that DZ-1 might be useful for the study of cellular uptake and cellular localiza- tion of DzNep. From our in-situ large-scale pull-down/LC- MS/MS results, we have successfully obtained a list of poten- tial cellular targets (on- and off-targets) of DzNep. Among the list were several kinases, phosphatases, ATPases, and methyltransferases. Although further studies will be needed in order to validate some of these newly identified, potential DzNep targets, our current findings show that DzNep likely has many unknown cellular targets and that both existing putative targets, EZH2 and SAHH, might not be true tar- gets of DzNep under native cellular settings. Our findings should have important implications in the consideration of using DzNep as a validated chemical probe for epigenetic research (especially against EZH2) and as a potential drug candidate for cancer therapy. Finally, our strategy should be generally useful for the off-target identification of other suitable drugs and/or candidates.
Experimental Section
General Information
All chemicals were purchased from Aldrich or Alfa Aesar and used as received, unless specified otherwise. Dried solvents (CH2Cl2, DMF, THF, acetonitrile, Et2O, and toluene) were drawn from a Glass Contour sol- vent dispensing system. All reactions requiring anhydrous conditions were carried out under argon atmosphere using oven-dried glassware. The reaction progress was monitored by thin-layer chromatography (TLC) on pre-coated silica plates (Merck 60 F254, 0.25 mm), and spots were visualized by UV light or phosphomolybdic acid stain. Flash column chromatography was carried out using Merck 60 F254, 0.040–0.063 mm silica gel. 1H and 13C NMR spectra were recorded on a Bruker 400 MHz instrument equipped with a CryoProbe. Chemical shifts for pro- tons are reported in parts per million (ppm) that were referenced to the NMR solvent (CDCl3: 7.26 ppm; [D6]DMSO: 2.50 ppm; CD3OD: 3.31 ppm). Chemical shifts for carbon are given in ppm that were refer- enced to the solvent (CDCl3: 77.0 ppm; [D6]DMSO: 39.5 ppm; CD3OD: 49.0 ppm). Data are presented as follows: chemical shift, multiplicity (br =broad, s=singlet, d =doublet, t=triplet, q =quartet, m=multiplet), integration and coupling constants (J) in Hertz (Hz). Electron impact mass spectra (EIMS) (low and high resolution) were measured using a Finnigan MAT95XP double-focusing mass spectrometer. Electrospray ionization (ESI) mass spectra were recorded using a Waters Quattro Mi- croTM API instrument for low-resolution and an Agilent 6210 Time-of- Flight LC/MS instrument for high-resolution mass spectrometry. In-gel fluorescence scanning of SDS-PAGE gels was carried out with a Typhoon 9410 fluorescence gel scanner (Amersham Biosciences), and where applicable, bands were quantified using ImageQuant 3.3 (Molecular Dynam- ics). Tris(2-carboxyethyl) phosphine (TCEP) and tris[(1-benzyl-1H-1,2,3- triazol-4-yl)methyl]amine (“ligand”) were purchased from Sigma–Al- drich. Hoechst 33342 was purchased from Invitrogen. DzNep (SML0305) was purchased from Sigma–Aldrich. MCF-7 mammalian cells were
grown in Dulbecco’s modified Eagle’s medium (DMEM, Invitrogen,Carlsbad, CA) containing 10 % heat-inactivated fetal bovine serum (FBS; Invitrogen), 100 UmL—1 penicillin, and 100 mg mL—1 streptomycin (Thermo Scientific, Rockford, IL), and maintained in a humidified 37 8C incubator with 5 % CO2. Rhodamine-N3 and biotin-N3 were synthesized as previously reported.[12]
Cellular Imaging in MCF-7 Cells
MCF-7 cells were seeded in glass bottom dishes (Mattek) and grown until 70–80 % confluence.[12c] Cells were then treated with 0.5 mL of DMEM containing the probe at different concentrations. After 2 hat 37 8C, the medium was removed, and cells were gently washed twice with PBS, followed by UV irradiation (254 nm) for 5 min. The cells were sub- sequently fixed for 15 min at room temperature with 3.7 % formaldehyde in PBS, washed twice with cold PBS, and permeabilized with 0.1 % Triton X-100 in PBS for 10 min. Cells were then blocked with 2 % BSA in PBS for 30 min, washed twice with PBS, and subsequently treated with a fresh- ly prepared premixed click chemistry reaction solution in a volume of 100 mL (final concentrations of reagents: 1 mM CuSO4, 1 mM TCEP, 100 mM TBTA, and 10 mM rhodamine-N3 in PBS) for 2 h at room tempera- ture under vigorous shaking. Cells were washed with PBS at least three times. The same cells were then incubated in PBS containing 0.25 mg mL—1 of Hoechst for 15 min at room temperature to stain nuclear DNA, and washed with PBS for 5 min with gentle agitation and a final wash with deionized water (1–2 min with gentle agitation) before mounting. Cellular imaging was done using a Leica TCS SP5X confocal micro- scope system equipped with Leica HCX PL APO 63 ×/1.20 W CORR CS, 405 nm diode laser, white laser (470–670 nm, with 1 nm increments, with eight channels AOTF for simultaneous control of eight laser lines, each excitation wavelength provides 1.5 mV), and a photomultiplier tube (PMT) detector ranging from 410 to 700 nm for steady-state fluorescence. Images were processed with Leica Application Suite Advanced Fluores- cence (LAS AF).
Pull-Down (PD) and LC-MS/MS Analysis
The general in-situ proteome labeling and pull-down procedures were based on previously reported procedures,[12] with the following optimiza- tions. Briefly, the probe (10 mM) was directly added to live MCF-7 cells, followed by incubation for 5 h at 37 8C under 5 % CO2. DMSO never ex- ceeded 1 % in the final solution. The medium was aspirated, and cells were washed twice gently with PBS to remove excess probe, followed by UV irradiation (254 nm) for 5 min on ice. The cells were then trypsinized and pelleted by centrifugation. Subsequently, the cell pellets were resuspended in PBS (50 mL), homogenized by sonication, and diluted to 1 mg mL—1 with PBS. The labeled lysates were then subjected to click re- action with biotin-N3, and all subsequent experiments were carried out as previously described.[12c] Control labeling/PD experiments using the nega- tive probe (NP-1 or NP-2 at 10 mM final concentration) was carried out concurrently. After PD, all protein samples were separated on 10 % SDS- PAGE gels, followed by coomassie or silver staining. Trypsin digestion was performed as previously described.[12] Digested peptides were ex- tracted from the gel with 50 % acetonitrile and 1 % formic acid. The pep- tides were separated and analyzed on a Shimadzu UFLC system (Shi- madzu, Japan): A mobile phase A (0.1 % formic acid in H2O) and mobile phase B (0.1 % formic acid in acetonitrile) were used to establish a 60 min gradient comprising 45 min of 5–35 % B, 8 min of 35–50 % B, and 2 min of 80 % B, followed by re-equilibrating at 5 % B for 5 min. Peptides were then analyzed on LTQ-FT mass spectrometer with an Advance CaptiveSpray Source (Michrom Bio Resources) at an electrospray potential of 1.5 kV. A gas flow of 2 Lmin—1, ion transfer tube tempera- ture of 180 8C, and collision gas pressure of 0.85 mTorr were used. The LTQ-FT was set to perform data acquisition in the positive-ion mode as previously described,[23] except that the m/z range of 350–1600 was used in the full MS scan. The raw data were converted to mgf format. The da-
tabase search was performed with an in-house Mascot server (version 2.2.07, Matrix Science) with MS tolerance of 10 ppm and MS/MS toler- ance of 0.8 Da. Two missed cleavage sites of trypsin were allowed. Carba- midomethylation (C) was set as a fixed modification, and oxidation (M) and phosphorylation (S, T, and Y) were set as variable modifications. LC-MS/MS results obtained from the above experiments (in situ, and for negative probes) were processed as shown below, and results are sum- marized in the Supporting Information. As in the case of most large-scale LCMS experiments,[12] a large number of proteins were identified from each LCMS run, many of which were “sticky” and/or highly abundant proteins. These were automatically removed. “False” hits that appeared in negative control pull-down/LCMS experiments were further eliminat- ed. Detailed LC-MS/MS results are presented in the Supporting Information.