Stability and cell permeability of sulfonyl fluorides in the design of Lys-covalent antagonists of protein-protein interactions

Abstract

Recently we reported on aryl-fluorosulfates as possible stable and effective electrophiles for the design of lysine covalent, cell permeable antagonists of protein-protein interactions (PPIs). Here we revisit the use of aryl-sulfonyl fluorides as Lys-targeting moieties, incorporating these electrophiles in XIAP (X-linked inhibitor of apoptosis protein) targeting agents. We evaluated stability in buffer and reactivity with Lys311 of XIAP of various aryl-sulfonyl fluorides using biochemical and biophysical approaches, including displacement assays, mass spectrometry, SDS gel electrophoresis, and denaturation thermal shift measurements. To assess whether these modified electrophilic “warheads” can also react with Tyr, we repeated these evaluations with a Lys311Tyr XIAP mutant. Using a direct cellular assay, we could demonstrate that selected agents are cell permeable and interact covalently with their intended target in cell. These results suggest that certain substituted aryl-sulfonyl fluorides can be useful Lys- or Tyr-targeting electrophiles for the design of covalent pharmacological tools or even future therapeutics targeting protein-protein interactions.

Graphical Abstract

We report that with proper substitutions, aryl-sulfonyl fluorides can be efficiently incorporated in protein-protein interactions antagonists to obtain potent and selective Lys or Tyr covalent agents. We report that such agents possess a proper balance between reactivity and stability and are cell permeable, hence can be used as pharmacological tools or perhaps even future novel covalent therapeutics.

Introduction

Recently we compared the reactivity of aryl-sulfonyl fluorides ^[1] and aryl-fluorosulfates ^[2] electrophilic “warheads” when inserted in binding peptides targeting proteins of the Inhibitors of Apoptosis Proteins family (IAPs). ^[3] These recent studies revealed that while benzamide-sulfonyl fluorides can react readily with lysine, or tyrosine, they are unstable in aqueous buffer at physiological pH values. On the contrary, we found that properly placed aryl-fluorosulfates could also react readily with lysine targets and were very stable in buffer and in plasma. Indeed, unlike for aryl-fluorosulfates containing agents that target XIAP,^[2] we could not clearly detect the formation of a stable adduct in cell with covalent agents containing a benzamide-sulfonyl fluoride electrophiles, presumably due to their instability.^[1] Here, we explored if further substitutions of aryl-sulfonyl fluorides placed at the X position of agents NMe-Ala-X-Pro-Phe-CONH₂ could result in compounds with increased aqueous stability and retained reactivity for their intended target (residue Lys311 of the BIR3 domain of XIAP). To extend these findings to Tyr residues, we also studied covalent adduct formation of the agents to a Lys311Tyr XIAP mutant. Subsequently, after selection of the most promising sulfonyl fluorides at the X position, we replaced the C-terminal Phe residue with a 4F,1-amino indane to increase cell permeability, and assessed the properties of the resulting agents in a direct cellular assay. Hence, we could demonstrate that selected agents can work as cell permeable, stable and effective irreversible agents targeting Lys and Tyr residues, further expanding the target space for the druggable proteome to protein-protein interactions (PPIs) that present such residues in proximity to their binding sites.

Results

Relative stability and in vitro activity of XIAP targeting agents containing sulfonyl-fluorides.

To qualitatively evaluate the stability and subsequent reactivity of a variety of aryl-sulfonyl fluorides, a library of agents that are synthetically accessible was prepared based on previous studies from our laboratory.^[1–2] We demonstrated that incorporation of benzamide- or benzylamine-sulfonyl fluorides in position X in the general tetra peptide of sequence AXPFCONH₂ (Table 1) resulted in agents that can covalently interact with Lys311 in the BIR3 domain of XIAP (Figure 1A) or Tyr311 in a Lys311Tyr BIR3 mutant (Figure 1B). The agents were synthesized and purified as we reported recently.^[1]

Table 1.

X group and data relative to test agents of structure NMe-Ala-X-ProPhe-CONH₂.

ID	X group	Aqueous stability [a]	DELFIA IC50 [nM][b]	ΔTm [°C] [c]
				wt-BIR3 (Lys311)	mut-B\R3 (Lys311Tyr)
1		>99% >99%	357 ± 9	4.7 ± 0.1 5.4 ± 0.1	3.0 ± 0.2 2.7 ± 0.5
2		54% 0%	15 ± 2	30.1 ± 0.3 29.9 ± 0.4	25.8 ± 0.2 24.7 ± 0.1
3		57% 0%	47± 3	5.1 ± 0.1 5.2 ± 0.3	24.8 ± 0.1 23.4 ± 0.2
4		39% 0%	173 ± 6	6.7 ± 0.2 6.1 ± 0.1	26.6 ± 0.2 25.2 ± 0.2
5		58% 0%	12 ± 1	28.5 ± 0.4 28.4 ± 0.4	25.9 ± 0.1 24.7 ± 0.1
6		80% 15%	21 ± 3	30.4 ± 0.1 30.0 ± 0.2	26.4 ± 0.1 25.1 ± 0.1
7		94% 73%	122 ± 1	25.8 ± 0.4[d] 25.5 ± 0.2[d]	26.3 ± 0.2 25.0 ± 0.1
8		96% 46%	28 ± 3	26.8 ± 0.7 27.4 ± 0.3	26.0 ± 0.2 24.6 ± 0.1
9		90% 71%	261 ± 16	5.3 ± 0.4 26.0 ± 0.4c	24.6 ± 0.1 23.5 ± 0.2
10		92% 47%	47 ± 2	25.8 ± 0.2 26.0 ± 0.1	25.5 ± 0.2 24.3 ± 0.1
11		86% 39%	268 ± 5	5.5 ± 0.2 25.6 ± 0.5	25.2 ± 0.1 24.1 ± 0.1

^[a]Remaining agent, as determined by ¹H 1D NMR spectroscopy measurements, after 5h at room temperature (top value), or after 5h at 37 °C (bottom value).

^[b]IC₅₀ values were obtained by displacement dose-response curves measured after pre-incubating protein and test ligand for 2h at room temperature.

^[c]Thermal shift (ΔTm) data were obtained after pre-incubating protein and ligand for 30min or 2h at room temperature.

^[d]In this experiment, the peak corresponding to the noncovalent ΔTm is still present.

Figure 1. Biochemical and biophysical characterization of Lys/Tyr targeting agents.

A) Molecular model representing the covalent docked pose of compound 8 in complex with the BIR3 domain of XIAP. B) Molecular model representing the covalent docked pose of compound 7 in complex with the Lys311Tyr mutant BIR3 domain of XIAP. The covalent docked structures of the compounds were obtained with Sybyl X (Certara). C) Superposition of the aromatic regions of the 1D ¹H NMR spectra of the compound 1 (left) and compound 8 (right) collected in aqueous buffer (500 μM, T= 37°C, pH = 8, 50 mM phosphate buffer, 150 mM NaCl in D₂O) a time zero (blue) and after 5 h (red). D) DELFIA displacement curves relative to the BIR3 domain of XIAP for compound 1 (blue), and compound 8 (red). E) Denaturation thermal shift curves collected for the BIR3 domain of XIAP in absence (blue) and presence (red) of compound 8. A compound induced shift in denaturation ΔTm) of ~ 27 °C is observed. Similar data were as those reported panels C, D, and E were used obtained to populate the values listed in Table 1 and are reported in the supplementary materials.

The stability of the agents in aqueous buffer was subsequently evaluated by solution ¹H NMR spectroscopy. Briefly, the agents (500 μM) were dissolved in 100 mM PBS buffer and their chemical stability was assessed by measuring a series of 1D ¹H NMR spectra over time both at room temperature and at 37 °C. In particular, the signals of the phenyl ring carrying the sulfonyl fluoride (Figure 1C) were monitored and used to assess the percentage of intact compound over time. A summary of these stability data is reported as percentage of remaining compound after 5 hr at room temperature or at 37 °C, hence at a physiological temperature relevant for cellular assays and in vivo studies (Table 1). As control, we also prepared compound 1 that, lacking a reactive aryl-sulfonyl fluoride, remained very stable at both temperatures (Table 1, Figure 1C). These studies suggested that compounds 2-6 presented a limited stability in buffer (Table 1) hence are deemed too reactive to be used as chemical probes or pharmacological tools.

On the contrary, significantly increased stability can be observed with substituted benzamide-sulfonyl fluorides compounds 7,8,9 as well as in the benzyl-sulfonyl fluorides 10 and 11 (Table 1). Next, to determine which of these agents were still capable to interact covalently with Lys311 in the BIR3 domain of XIAP, we performed four additional sets of assays.

First, we used a Dissociation-Enhanced Lanthanide Fluorescence Immunoassay (DELFIA) displacement assay as we described previously ^[4] that is designed to quantify the ability of each tested agent to compete for the binding of a reference biotinylated AVPI peptide. While IC₅₀ values for covalent inhibitors are time dependent, hence not suitable to quantitative evaluations, we could qualitatively compare IC₅₀ values obtained from dose-response curves measured at a given pre-incubation time of test agent and protein target (Figure 1D; Table 1). For most agents, introduction of the aryl-sulfonyl fluoride resulted in increased affinity for the target compared to the non-reactive reference compound 1 (Table 1).

Second, and to further corroborate the qualitative rank ordering provided by IC₅₀ measurements, we measured the ability of each agent to induce a denaturation thermal shift (ΔTm) on the BIR3 domain of XIAP with and without pre-incubating the target with a given agent (Table 1, Figure 1E) at two different pre-incubation times. Typical of non-covalent targeting agents, compound 1 displayed a ΔTm ~ 5 °C (regardless of the pre-incubation time of either 30 min or 2 h, Table 1), while putative covalent compounds showed significantly larger shifts ΔTm of > 20 °C; Table 1).^[1b] However, some agents displayed a larger denaturation thermal shift only at the longer incubation time (namely agents 9 and 11), suggesting a slower reaction with the target. Compound 4, compatible with its lack of stability in buffer (Table 1), did not induce large ΔTm values nor displayed low nanomolar affinity in the DELFIA assay (Table 1). As mentioned above, to assess if the agents were also capable to react with Tyr, we prepared a Lys311Tyr mutant and evaluated the above (Table 1).

Finally, we could qualitatively monitor the kinetics of covalent adduct formation directly using SDS gel electrophoresis at various times (Figure 2A). Final confirmation of adduct formation was also provided by mass spectrometry data (Figures 2B) after pre-incubation of compounds and protein for 2 hr at room temperature. These data qualitatively confirmed that among the most stable and potent agents, namely compounds 8 and 10, inducing large ΔTm (ΔTm > 20 °C) and concomitantly an increased affinity in the DELFIA assay, formed a stable covalent adduct with the BIR3 domain of XIAP (Figures 2A and 2B). Similarly we found that compound 7 can covalently interact with the Tyr311Lys mutant (Figure 2C,D,E).

Figure 2. In vitro characterization of covalent binding to the BIR3 domain of XIAP for selected agents.

A) SDS-PAGE gel electrophoresis followed by Coomassie staining of the BIR3 domain of XIAP incubated for 2 h at 25 °C with compounds 8 and 10 with a protein-ligand ratio of 1:2. B) LC-MS spectra of the BIR3 domain of XIAP in the absence (left), in the presence of compound 8 (center), and in the presence of non-covalent control compound 1 (right). The mass of the BIR3 domain of XIAP is 13,106 Da, and the mass increases by 629 Da in the presence of compound 8, corresponding to the expected MW of the covalent adduct. Similar MS data is available for all agents tested. C) SDS-PAGE gel electrophoresis followed by Coomassie staining of the Lys311Tyr BIR3 mutant of XIAP incubated for 2 h at 25 °C with compound 7 with a protein-ligand ratio of 1:2. D) LC-MS spectra of the BIR3 Lys311Tyr mutant domain of XIAP in the absence (left), in the presence of compound 7 (center), and in the presence of compound 1 (right). The mass of the BIR3 Lys311Tyr mutant of XIAP is 13,141 Da, and the mass increases by 651 Da in the presence of compound 7, corresponding to the expected MW of the covalent adduct. The reaction is still incomplete after 2h at 25 °C. Similar MS data is available for all agents tested (reported as supplementary materials). E) Denaturation thermal shift curves collected for the Lys311Tyr BIR3 mutant of XIAP in absence (blue) and presence (red) of compound 7. A compound induced shift in denaturation (ΔTm) of ~ 27 °C is observed.

Selectivity and cell permeability of covalent XIAP antagonists containing sulfonyl fluorides.

Our previous studies suggested that the tetra peptide AVPF-CONH₂ is not cell permeable while substitutions in the P4 position replacing the C-terminal Phe residue dramatically increased cell permeability. We recently reported that a 4-fluoro 1-amino indane represented a suitable Phe mimetic that increased potency and cell penetration of XIAP targeting agents. Hence, we selected the sulfonyl-fluorides with the best compromise between aqueous stability and reactivity with the target from Table 1, namely compounds 8 and 10, and replaced the Phe in their structures with 4-fluoro 1-amino indane, resulting in agents 12 and 13, respectively (Table 2). The synthesis of these agents followed the same experimental procedures as we reported recently.^[1]

Table 2.

Chemical structures, denaturation thermal shifts, and DELFIA IC50 values for XIAP, cIAP1, or cIAP2 BIR3 domains. IC50 values represent dose-response curves obtained after pre-incubating protein and ligand for 2h at room temperature. Thermal shift (ΔTm) data were obtained pre-incubating protein and ligand for either 30min (top value) or 2h (bottom value) at room temperature. We had previously reported compound 12^.[1b] Values and standard errors represent the average of 4 independent measurements.

ID	Structure	Thermal shift data ATm [°C] XIAP-BIR3	DELFIA IC50 values [nM]
			XIAP-BIR3	cIAP1-BIR3	cIAP2-BIR3
LCL161		9.8 ± 0.3 10.8 ± 0.1	36 ± 5	16 ± 2	13 ± 1
12		26.7 ± 0.2 26.5 ± 0.2	24.0 ± 0.7	22 ± 4	58.5 ± 0.1
13		27.0 ± 0.2 27.2 ± 0.1	15.6 ± 0.8	141 ± 15	492 ± 26

For these agents, we measured XIAP BIR3 denaturation thermal shifts (Δ Tm) as well as DELFIA displacement assays against the BIR3 domains of XIAP, cIAP1, and cIAP2 to assess selectivity (Figure 3A, Table 2) at a given incubation time. As a reference, we also tested the non-covalent, pan-IAP antagonist and clinical candidate LCL161 (Table 2). under the same experimental assay conditions. As evident from Table 2, and clinical candidate LCL161 (Table 2) under the same incubating protein and ligand for either 30min (top value) or 2h (bottom value) experimental assay conditions. As evident from Table 2, LCL161 displayed ΔTm values of ~ 10 °C compatible with a potent non-covalent agent, independent of the incubation time. On the contrary, both compounds 12 and 13 showed a dramatically increased stabilization with ΔTm values > 20 oC more typical of potent covalent agents. DELFIA data suggested that while LCL161 is more potent against cIAP1/2 compared to XIAP, agents 12 and 13 are more potent towards XIAP in agreement to their covalent nature targeting Lys311 of its BIR3 domain. Indeed, in cIAP1 and cIAP2 a Glu residue occupies the equivalent position where XIAP presents Lys311.^[1] Compound 13 was particularly intriguing as it was significantly more selective for XIAP compared to cIAP1/2, making it a unique pharmacological tool for future dissections of the role of XIAP in tumorigenesis and other cellular processes.

Figure 3.

Selectivity for compounds listed in Table 2. A) DELFIA displacement curves relative to the BIR3 domain of pan-IAP compound LCL161 against the BIR3 domains of XIAP (red), cIAP1 (blue), or cIAP2 (green). B) as in A) but with compound 12; C) as in A but with compound 13.

While IC₅₀ values at a fixed incubation times have been used to qualitatively rank order agents, together with timedepended SDS gel electrophoresis, and denaturation thermal shift measurements, for compounds 12 and 13 we also performed kinetics Kinetic analysis using the DELFIA assay measured at various concentration of agents and at different times (supplementary material). We found for compound 12 Ki = 0.07 ± 0.03 μM, k_inact = 0.87 min⁻¹, while for compound 13 we found Ki = 0.085 ± 0.003 μM, k_inact = 1.55 min⁻¹, indicating that the agents react rapidly and efficiently with the target, as predicted by the qualitative array of assays.

Next, to assess cell permeability of these compounds, and to verify that the covalent agents can react with the target covalently in cell, we have obtained a cell line that is stably transfected with HA-BIR3 of XIAP.^[5] Hence, exposure of this cell line to our agents, followed by western blot analyses of cell lysates using an anti-HA antibody, was used to monitor whether the test agents could enter the cell, and target covalently the BIR3 domain of XIAP. Covalent binding could be directly observed by a significant shift in molecular weight (band shift) similar to what can be observed in vitro.

However, as we reported recently, the unbound BIR3 domain of XIAP is fairly unstable at 37 °C, while the domain gets stabilized by ligand binding, as also corroborated by our reported denaturation thermal shift data (Tables 1, 2). Accordingly, LCL161 stabilized the BIR3 domain in cell as it is appreciable by a clear increase in band intensity in the western blot compared to untreated control cells (Figure 4A). Similarly, compounds 12 and 13 stabilized HA-BIR3 in the same experiment, resulting in increasingly more visible bands even compared to LCL161. Albeit more speculative and qualitatively, our agents appear to induce an appreciable gel shift, both consistent with cell permeability of the agent and covalent adduct formation in cell (Figure 4A). Consistent with kinetics data, compound 13 gave the most intense band being the most potent in targeting the BIR3 domain of XIAP. The data also suggest that the agents possess sufficient selectivity and do not react non-specifically with the components of cell culture media or cell membrane proteins. We used 10 μM based on previous experience with similar assays where the target is overexpressed. Collectively, these data suggest that it is possible to derive pharmacologically viable covalent PPI antagonists by direct incorporation of certain aryl-sulfonyl fluorides into binding peptides. Compound 13 represents a stable, covalent and selective XIAP BIR3 agent that could be further deployed to dissect the therapeutic potential of XIAP inhibition in oncology^[3c] and potentially in other indications. ^[6]

Figure 4. Cell permeability and target engagement in cell of selected aryl-sulfonyl fluorides.

Western blot analysis of HA-BIR3 of XIAP-expressing HEK293T cells after exposure to 10 μM of each of the indicated compounds. Band intensity is qualitatively proportional to compounds’ permeability and induction of protein stability in cell. Covalent adduct formation is clearly appreciable by a band shift due to the slightly increased molecular weight of covalent adducts (see Figure 2 for example) compared to the native protein. This is observed with both compounds 12 and 13.

Discussion and Conclusions

Targeted covalent therapeutics acting on surface Cys residues with mild electrophiles such as acrylamides have recently been successfully developed against several protein kinases ^[7] and more recently also against the KRAS G12C mutant ^[8]. The balanced reactivity of these electrophiles made it possible to derive potent, irreversible and Cys-covalent agents targeting these enzymes leading to several FDA approved covalent drugs, and as many clinical candidates that are currently being investigated. Because of the success of such agents in the clinic, increased drug discovery efforts have focused on the target space that contains at the least a druggable Cys residue (also known as the Cysteinome).^[7o–q] A logical expansion of this success includes the development of suitable electrophiles that can likewise react with more frequently occurring amino acids such as Lys, or Tyr.^{[1, 9]} However, because these residues are intrinsically less reactive than Cys, especially when surface exposed, increasingly more reactive electrophiles are needed. Stability in buffer, in cell, and ultimately in vivo of such electrophiles will determine if these could be used as chemical probes, pharmacological tools, or therapeutics. Several manuscripts are emerging that report on targeting of Lys residues in active sites of proteins by introduction of appropriately placed electrophiles on existing ligands.^{[9b, 9c, 10]} Our own recent studies ^[1–2] and others ^[11] focused on targeting Lys, Tyr, or His surface residues located at protein-protein interfaces, suggesting that in principle the target space of covalent antagonists of PPIs could be expanded to include these residues. Indeed, recent reports are proposing a Lysinome or a Tyrosinome as an ensemble of targets that present targetable Lys or Tyr residues in proximity of their binding sites in enzymes. ^[12] Our recent application included a side by side investigation of the merits and pitfalls of aryl-sulfonyl fluorides and aryl-fluorosulfates as possible electrophiles to target Lys, Tyr, and His in protein-protein interactions.^[1b] We found that when properly juxtaposed to a binding site Lys residue, aryl-fluorosulfates approach the proper balance of stability, reactivity and cell permeability, similar to what is observed with Cys and acrylamides. Hence, we envisioned that aryl-fluorosufates can be used as pharmacological tools or therapeutics, even when targeting protein-protein interactions. ^{[1b, 13]} On the contrary, we found that benzamide-sulfonyl fluorides were fairly unstable in aqueous buffer and in plasma, in agreement with previous in vitro studies, where the general chemical reactivity of these agents was more accurately determined and correlated with the electron withdrawing (Hammet σ constants) properties of the substituents.^[14] Hence, we tried to address whether substitutions on aryl-sulfonyl fluorides can result in agents with more balanced stability and reactivity to surface Lys, or Tyr residues. As a model system for these determinations we used the BIR3 domain of XIAP. Using NMR and enthalpy-based screening approaches,^[4] we previously identified Lys311 as a possible target for covalent agents (Figure 1A), and compound 2 was synthesized and tested.^[1] However, we found that the stability of this agent in aqueous buffer was limited as confirmed by our current studies using NMR (Table 1, Figure 1C). Hence, to explore whether substituted aryl-sulfonyl fluorides could afford a more balanced stability and reactivity with Lys residues, we first prepared a small library of derivatives (Table 1), and evaluated their stability in buffer as well as their reactivity with Lys311 (or Tyr311 using a Lys311Tyr mutant) of the BIR3 domain of XIAP. We adopted a variety of biophysical and biochemical assays to make such assessments. In particular, we found that measuring the denaturation thermal shift at various incubation times represented a simple and effective method to characterize covalent binding. First, covalent agents presented much larger ΔTm values even when compared to ΔTm induced by very potent reversible ligands. For example, the potent reversible pan-IAP antagonist and clinical candidate LCL161 (Table 2) displayed a ΔTm value for XIAP BIR3 of ~ 10 °C, independent of the incubation time (Table 1). On the contrary, covalent agents displayed ΔTm values > 20 °C. In addition, rapidly reacting agents displayed ΔTm values > 20 °C already after a short pre-incubation time (30 min), while slowly reacting agents required longer incubation (2 h) to induce a large ΔTm. Hence, a simple analysis of ΔTm measurements at two incubation times provided a robust readout of the possible reactivity of the agents in vitro with the target. Likewise, DELFIA displacement assays could substantiate these findings, with covalent agents approaching the low nanomolar IC₅₀ values, while agents that did not react effectively and/or that reacted with water before they could react with the target displayed triple digit nanomolar affinities, due to the reversible/non-covalent binding of the resulting agent (see reference non-covalent compound 1, for example). These observations matched nicely with more direct methods such as SDS gel electrophoresis and MS measurements (Figure 2). Based on these data, compounds 8 and 10 emerged as possibly possessing the best compromise between aqueous stability and Lys311 reactivity. Subsequently we introduced a 4-F 1-amino indane in lieu of the terminal Phe in these agents given that previous studies in our laboratory identified a 4-fluoro-Phe in position P4 that later was optimized into a 4-fluoro-1-amino indane^{[1, 15]} to increase drug-likeness, potency, and cell permeability. These modifications led to agents 12 and 13 (Table 2).

Note that the relative aqueous stability of the agents in Table 1 was not entirely surprising given the recent reactivity studies of aryl-sulfonyl fluorides for isolated Lys or Tyr have been recently reported. ^[13] However, not all geometries of stable compounds such as agents 7 and 9, presenting the sulfonyl fluoride in meta with respect to the carboxy-amide, and further stabilized by a methyl or a methoxyl, respectively, were compatible with reacting to Lys311 of XIAP, hence only compounds 12 and 13 were further considered. Of note is that compounds 7 and 9 reacted efficiently with the Lys311Tyr mutant (Table 1), suggesting that their “warheads” could be used efficiently for other targets to react not only with Lys but also with Tyr residues.

Hence, to assess the reactivity of compounds 12 and 13 we measured time dependent denaturation thermal shifts against the BIR3 domain of XIAP (Table 2). These data confirmed that the agents can react rapidly with the target with ΔTm values > 20 °C after only 30 min incubation (Table 2).

Several agents based on the tetrapeptide of sequence Ala-Val-Pro-Phe (AVPF) that interact with various members of the IAP family, including XIAP, cIAP1, and cIAP2,^[16] have been developed as potential pharmacological agents, ^[17] and even clinical candidates such as LCL161. ^[18] These agents however, are pan-IAP antagonists. Because cIAP1 and cIAP2 present a Glu residue in a position equivalent to Lys311 in XIAP (Figure 3), unlike LCL161, compounds 12 and 13 resulted more potent against XIAP compared to cIAP1 and cIAP2 (Table 2). Moreover, compound 13 was significantly more selective for XIAP compared to cIAP1/2, (Figure 3, Table 2).

Next, to evaluate whether compounds 12 and 13 were cell permeable and interacted covalently with their target in cell, we used HEK293 cells transfected with an HA-tagged BIR3 construct. We previously reported that the isolated BIR3 domain of XIAP, in its unbound form, was particularly unstable at 37 °C as also revealed by our thermal shift studies and our previous studies by NMR.^[1] Hence, we developed a simple assay based on western blot analysis using an anti-HA antibody of HEK293 cell lysates that express HA-BIR3. After transfection with HA-BIR3 XIAP plasmid and treatment with DMSO or test agents for 6 hours, only a fainted band for HA can be visible, unless the construct gets stabilized by potent, cell permeable BIR3 ligands, such as LCL161 (Figure 4A). While a sizable band was observed by exposure to LCL161, a much more intense band could be observed when exposing cells to covalent agents 12 and 13. Most importantly, these bands also appeared significantly shifted (much like we observed in vitro using SDS gel electrophoresis, Figure 2A) as a result of the covalent adduct formation and increased MW of the construct (Figures 2B, 4A). These data unequivocally suggest that compounds 12 and 13 are approaching the stability and cell permeability necessary for their use as pharmacological tools. Compound 13 is particularly interesting as it is uniquely covalent and selective for XIAP with limited activity against cIAP1 and cIAP2. Most ligands available to date are either pan-active or interact preferentially with cIAP1/2, making determinations of the possible therapeutic potential of XIAP difficult to gather using pharmacological inhibition.

With the success of covalent drugs targeting Cys residues via acrylamides-based Michael acceptors, our recent studies suggested that equally effective strategies targeting Lys or Tyr residues could be deployed by proper juxtaposition of aryl-fluorosulfates in small molecules or peptide mimetics.^[2] Here, expanding on our previous work, we further identified properly substituted aryl-fluoro sulfates (in particular those in agents 7–11) that could be potentially used as pharmacological tools in targeting Lys or Tyr in antagonists of PPIs as well as in small molecules enzyme inhibitors.^[19] Hence, together with our previous similar observations with aryl-fluoro sulfates, ^[2] the studies significantly widen the covalent druggable proteome from the Cysteinome ^[7o–q] to other more frequently occurring residues such as Lys, or Tyr. ^{[1b, 9a, 20]} In targeting PPIs, in particular, we envision that aryl-fluorosulfates, ^[2] and/or the substituted aryl-sulfonyl fluorides as reported here (agents 7–11), could be relatively easily incorporated into current drug discovery strategies, including fragment-, structure-, and/or NMR-based approaches, ^{[15, 17t, 21]} phage display, or DNA encoded libraries, ^{[12b, 22]} aimed at deriving, potent, selective, and cell permeable agents that can form the basis for further lead optimizations and drug development studies.

Experimental Section

General Chemistry

Solvent and reagents were commercially obtained and used without further purification. NMR spectra were recorded on Bruker Avance 600 MHz. High resolution mass spectral data were acquired on an Agilent LC-TOF instrument. RP-HPLC purifications were performed on a JASCO preparative system equipped with a PDA detector and a fraction collector controlled by a ChromNAV system (JASCO) on a Luna C18 10μ 10 × 250mm (Phenomenex) to > 95% purity. LCL161 was obtained from MedChem Express. Fmoc-amino acids and Rink amide resin were purchased from Chem-Impex and Novabiochem, while the different sulfonyl-fluorides were purchased from Enamine or Sigma-Aldrich. The BAL resin was acquired from Creosalus. Reductive amination was used to introduce the terminal amine on the BAL resin, using 3 eq. of the desired amine and 3 eq. of sodium triacetoxyborohydride overnight. After this reaction, the first amino acid introduced was left to react for 2h.

Peptides were synthesized by using standard solid phase Fmoc peptide synthesis protocols. For each coupling reaction 3 eq. of Fmoc-AA, 3 eq. of HATU and 5 eq. of DIPEA in 1 ml of DMF were used. The coupling reaction ran for 50 min at room temperature, followed by 3 washes with DMF. The Kaiser test was employed to monitor reaction completion. Fmoc deprotection was performed in two steps by treating the resinbound peptide with 20% 4-methylpiperidine in DMF for 5 min then 15 min at room temperature. Purity of tested compounds was assessed by HPLC using an Atlantis T3 3μm 4.6×150mm column (H₂O/CH₃CN gradient from 5% to 100% in 45min). All compounds have a purity >95%.

Compound 12:

4-((((S)-3-((S)-2-(((R)-4-fluoro-2,3-dihydro-1Hinden-1-yl)carbamoyl)pyrrolidin-1-yl)-2-((S)-2(methylamino)propanamido)-3oxopropyl)amino)methyl)benzene-1-sulfonyl fluoride. The synthesis and purification of compound 12 was described in our previous publication.^[1a]

Compound 13:

4-(((S)-3-((S)-2-(((R)-4-fluoro-2,3-dihydro-1Hinden-1-yl)carbamoyl)pyrrolidin-1-yl)-2-((S)-2(methylamino)propanamido)-3-oxopropyl)carbamoyl)-2methoxybenzene-1-sulfonyl fluoride. BAL resin was used as solid-phase support (0.05 mmol scale), and the previously described coupling conditions were used to obtain the peptidic part of the agent. The ivDde protecting group was removed using 4% N₂H₂ (3×5ml, 5 min each). The generated amine was reacted with 3 eq. of 4-(fluorosulfonyl)-3-methoxybenzoic acid, 3 eq. of HATU and 5 eq. of DIPEA in 1 ml of DMF. The reaction was left shaking overnight. After cleavage, the crude was purified by preparative RP-HPLC using a Luna C18 column (Phenomenex) and water/acetonitrile gradient (5% to 100%) containing 0.1% TFA, obtaining a white powder (7.8 mg, 55.4%). 1D ¹H NMR is reported as supplementary Figure S5. HRMS: calcd 635.2225 (M); obs 636.4896 (M+H)+.

Protein expression and purification

The BIR3 domain of human XIAP (residues 253–347) was expressed transforming a pET-15b vector, containing the XIAP-BIR3 cDNA and an N-terminal His tag, into E. coli BL21Gold(DE3) pLysS cells. The transformed cells were then grown in LB medium at 37°C in presence of 100 mg/mL of ampicillin. A total of 1 mM IPTG was added into the growing cells when the optical density (OD₆₀₀) reached a value of 0.6–0.7. After the induction, the cells continued to grow at 20°C overnight. The overexpressed protein was purified using Immobilized metal ion affinity chromatography (IMAC), followed by a buffer exchange in 25 mM Tris at pH 8, 150 mM NaCl, 50 mM Zn(Ac)₂, and 1 mM of DTT. The pET-15b vector containing the XIAP-BIR3 cDNA (residues 253–347) was mutated on the Lysine 311 to Tyrosine by GenScript (Piscataway, NJ). The transformation, expression, ad purification of XIAP-BIR3 Lys311Tyr were performed as previously described.^[1b] Recombinant BIR3 domains of cIAP1 (residues 258–363), and cIAP2 (residues 244–349) both with an N-terminal His tag were obtained from Reaction Biology Corp. (Malvern, PA).

Dissociation-Enhanced Lanthanide Fluorescence Immunoassay (DELFIA).

Each well of the 96-well streptavidin-coated plates (PerkinElmer) was incubated with 100 μL of 100 nM AVPI-Biotin (AVPIAQKSEK-Biotin) for 1 h. Plates were then washed 3 times with the DELFIA wash solution (PerkinElmer) to remove the unbound AVPI-Biotin. Once washed, each well of the plates was incubated for 2 h with a solution containing 89 μL of Eu-N1labeled anti-6xHis antibody (PerkinElmer) and a 11 μL-mixture of the protein and the test compounds. The final antibody concentrations used for XIAP-BIR3 and cIAP1-BIR3 were 22.2 ng/well and 29.7 ng/well for cIAP2-BIR3. The final protein concentrations were 30 nM for XIAP-BIR3 and cIAP1-BIR3 and 15 nM for cIAP2-BIR3. Plates were again washed 3 times and incubated with 200 μL of the DELFIA enhancement solution (PerkinElmer) for 10 min prior to taking fluorescence measurement with the VICTOR X5 microplate reader (PerkinElmer) using the excitation and emission wavelengths of 340 and 615 nm, respectively. DELFIA assay buffer (PerkinElmer) was used to prepare all the protein, peptide and antibody solutions and the incubations were done at room temperature. Samples were normalized to 1% DMSO and reported as % inhibition. GraphPad Prism version 7 was used to calculate IC₅₀ values with the SE values determining from replicate measurements. Kinetics measurements were obtained by measuring the percent inhibition at various concentrations (0, 24nM, 192nM, 155nM, 1250nM, 10000 nM) and a different times (0, 2min, 5min, 10min, 20min, 40min) to obtain k_obs(min⁻¹). Ki and k_inact values where then obtained by plotting k_obs as function of the ligand concentration (supplementary materials).

Gel electrophoresis

To detect gel shifts, a mixture of 10 μM of each test protein (wtBIR3 or Lys311Tyr BIR3 mutant of XIAP) was incubated with 100 μM of test compounds in a buffer containing 25 mM Tris pH 8, 150 mM NaCl, and 50 μM zinc acetate either at room temperature or 37 °C. Samples were taken after various incubation times and loaded onto the NuPAGE 12% Bis-Tris Protein Gels (Life Technologies) and electrophoresed using MES running buffer (Life Technologies) at 200 V for 35 min. Gels were subsequently stained with SimplyBlue SafeStain (Life Technologies) according to the manufacturer’s protocol.

Immunoblot assay

The HA-XIAP-BIR3 plasmid was a gift from Dr. Colin Duckett (Addgene plasmid #25689). One million HEK293T17 cells (purchased from American Type Culture Collection, ATCC) were plated in 6-well plates and left to attach overnight. The following day, cells were transfected with 0.5 μg of the HA-XIAP-BIR3 plasmid using Lipofectamine 2000 (Thermo Fisher) in complete Dulbecco’s modified Eagle medium (DMEM) media supplemented with 10% FBS and 1% PenStrep (Invitrogen). 18 h post-transfection, the media was replaced with serum-free DMEM containing 10 μM of compounds [1% of dimethyl sulfoxide (DMSO)] and incubated for an additional 6 h. Finally, the cells were lysed with lysis buffer [20 mM Tris, pH 7.4, 120 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 1% IGEPAL, 5 mM ethylenediaminetetraacetic acid (EDTA)] supplemented with EDTA-free protease inhibitor cocktail and PhosSTOP (SigmaAldrich) for 10 min on cold ice. Lysates were centrifuged and supernatants were collected. The protein content was quantified, and the samples were prepared using a NuPAGE antioxidant and LDS sample buffer (Thermo Fisher) and heated for 10 min at 70 °C. Each sample containing 10 μg of proteins were loaded into a 12% NuPAGE bis-tris precast gels and transferred to PVDF membranes. The membranes were blocked with 5% milk in TBS and 0.1% Tween (TBST) and incubated with anti-HA (Santa Cruz Biotechnology, Y-11, sc-805) overnight at 4 °C. Next day, the membrane was washed with TBST and incubated with goat anti-rabbit HRP secondary antibodies. The antigen-antibody complexes were visualized using a Clarity Western ECL kit (BIO-RAD). The membrane was stripped, and the western blot was repeated using a β-actin primary antibody (Santa Cruz Biotechnology, sc69879) to check for loading.

Denaturation thermal shift

Thermal shift assays for BIR3, or BIR3 Lys311Tyr, inhibitor complexes were obtained with a BioRad CFX Connect Real-Time PCR Detection System. Each data point was collected in triplicate. Incubation of BIR3 protein with inhibitor followed one of two parameters, either 30 min or 2 h at 25 °C. Protein/inhibitor complexes and 5000x SYPRO Orange dye (Sigma) were diluted using reaction buffer, 50 mM Tris pH 8.0, 150 mM NaCl, 50 μM Zinc acetate, to obtain final concentrations of 5 μM BIR3, 10 μM inhibitor, and 60x SYPRO Orange. Sample plates were heated from 10 °C to 95 °C with heating increments of 0.05 °C, over 30 min. Fluorescence intensity was measured within the excitation/emission ranges 470–505/540–700 nm.

Molecular modeling

Covalent docking of compounds in Figure 1, was obtained by non-covalent docking followed by manual covalent bond formation and energy minimization of the covalent adduct (SYBYL-X 2.1.1; Certara, Princeton, NJ; compound 2) using Protein Data Bank entry 3HL5 for XIAP. Likewise, single point mutation of Lys311 into Tyr311 was performed using SYBYL-X 2.1.1 (Certara, Princeton, NJ) and subsequently the model of the complex was prepared as described above.