New Chemical Probe Targeting Bacterial NAD Kinase

Nicotinamide adenine dinucleotide (NAD) kinases are essential and ubiquitous enzymes involved in the tight regulation of NAD/nicotinamide adenine dinucleotide phosphate (NADP) levels in many metabolic pathways. Consequently, they represent promising therapeutic targets in cancer and antibacterial treatments. We previously reported diadenosine derivatives as NAD kinase inhibitors with bactericidal activities on Staphylococcus aureus. Among them, one compound (namely NKI1) was found effective in vivo in a mouse infection model. With the aim to gain detailed knowledge about the selectivity and mechanism of action of this lead compound, we planned to develop a chemical probe that could be used in affinity-based chemoproteomic approaches. Here, we describe the first functionalized chemical probe targeting a bacterial NAD kinase. Aminoalkyl functional groups were introduced on NKI1 for further covalent coupling to an activated SepharoseTM matrix. Inhibitory properties of functionalized NKI1 derivatives together with X-ray characterization of their complexes with the NAD kinase led to identify candidate compounds that are amenable to covalent coupling to a matrix.


Introduction
Nicotinamide adenine dinucleotide (NAD) kinases (NADK) are ubiquitous enzymes, which catalyze the phosphorylation of NAD to nicotinamide adenine dinucleotide phosphate (NADP), which is subsequently reduced to NADPH [1][2][3]. Since it is the only known enzyme producing NADP de novo, NAD kinase plays a crucial role in controlling the intracellular balance of NAD(H) and NADP(H) in many cellular metabolic pathways [4,5]. While the NADK enzymatic activity has been known for decades, their genes were identified more recently, leading to the discovery of orthologs in nearly all living organisms. The essentiality of the NADK gene has been experimentally validated in several bacteria, including Escherichia coli [6], Salmonella enterica, Bacillus subtilis [7,8], Mycobacterium tuberculosis [9] and Staphylococcus aureus [10][11][12]. Moreover, it was shown that human NAD kinase displays kinetic and structural features that differ considerably from that of prokaryotes [13,14]. Therefore, NADK represents a promising target for the development of novel antibiotics with an original mode of action.
Previously we solved the crystal structures of NADK from Listeria monocytogenes (LmNADK1) in the free state and bound to NAD and NADP, and described the first non-natural inhibitor of this enzyme [15]. Following a fragment-based approach, we identified a series of diadenosine derivatives with low micromolar inhibitory potencies against recombinant LmNADK1 and S. aureus NADK [16,17]. We subsequently discovered the first NAD kinase inhibitor (namely NKI1, Figure 1) active in mice infected with S. aureus, including antibiotic-resistant strains [18]. To get more information about the selectivity of our lead compound and its mechanism of action, we aimed at immobilizing NKI1 derivatives on a Sepharose TM matrix that could be further used in chemical proteomics [19]. Affinity chromatography is one of the most powerful chromatographic methods to analyze and purify proteins from complex mixtures or crude extracts. Various immobilized-nucleos(t)ides and cofactors are now commercially available. However specific affinity matrices may need to be developed by coupling the suitable ligand onto the matrix. The attachment chemistry and the spacer between the ligand and the matrix are important factors to be considered [20].
Here, we describe the synthesis of a series of functionalized diadenosine derivatives based on the chemical structure of NKI1 [18]. Short aminoalkyl chains were introduced at each flanking region or middle region of the diadenosine motif ( Figure 2) that could be used for coupling with a NHS-activated matrix. NADK binding properties of functionalized NKI1 were analyzed by X-ray crystallography and by measuring NADK activity, allowing selection of the candidate ligands for further covalent attachment to a matrix.

Chemical Synthesis of Functionalized Diadenosine Derivatives
The key step for synthesizing derivatives 2-5 is a Sonogashira cross-coupling reaction between a bromide (6 or 7) and an alkyne (8 or 9) (Figure 3). The appropriate building blocks were easily prepared according to classical methodology. The synthetic route to the N6-functionalized NKI1 derivatives 2 and 3 is depicted in Scheme 1. An aminoalkyl side chain (2 or 4 carbon) was introduced at the 6-position of adenosine by S N Ar from the 6-chloropurine derivative 10. Selective 5'-O-propargylation of 11a and 11b in the presence of propargyl bromide and NaH in DMF afforded the building blocks 8a and 8b, respectively, which were involved in Sonogashira cross-coupling reaction with bromide 6 to give 12a and 12b. Finally, removal of all protecting groups provided the desired compounds 2 and 3. Scheme 1. Reagents and conditions: (a) BocNH(CH 2 ) n NH 2 , DME, NEt 3 , 80 • C, 66% (n = 2) or 91% (n = 4); (b) NaH, propargylbromide, THF, 4 • C, 60% or 68%; (c) 6, Pd(PPh 3 ) 4  The synthesis of compound 4 was achieved by reaction of 4-(N-Boc-amino)butyric acid (Boc-GABA-OH) with the 5'-amino derivative 13 in the presence of PyBOP and DIEA in DMF, followed by C8 bromination of the resulting amide 14 to give the key intermediate 7 (Scheme 2). Sonogashira coupling of bromide 7 and alkyne 8c [16], followed by acidic treatment of the coupling product 15 gave the desired derivative 4 (Scheme 2). Finally, introduction of an aminoalkyl side chain at the middle position of the ligand was achieved from 5'-amino derivative 13 via the Ns strategy [21]. Nosylation of 13, followed by alkylation of the N-sulfonamide 16 and removal of Ns group afforded the N-monosubstituted amine 17 (Scheme 3). Alkylation of the secondary amine using propargyl bromide afforded the key N-alkylated intermediate 9 in good overall yield. Sonogashira reaction between 9 and 6 led to the coupling product 19. Finally, two-step removal of protecting groups gave the desired derivative 5.

X-ray Structures of LmNADK1 Bound to Compounds 2-5
A prerequisite to use a chemical probe is that it retains its binding affinity for its target. The binding properties of the four synthesized derivatives 2-5 were analyzed by X-ray crystallography. We checked the binding properties of these new di-adenosine derivatives by determining their complex with crystallized LmNADK1. The soaking procedure described previously for inhibitor 1 [18] was used to obtain crystals of the target in complex with compounds 2-5. The X-ray structures were solved by molecular replacement using PDB6RGC [18] as a starting template and refined to 2.2-2.4 Å resolution. The resulting structures showed the same overall mode of binding to LmNADK1, the ligands occupying both sub-sites A and N of the NAD binding pocket (Figure 4). While compound 3 failed to yield complex with the crystallized target, compound 2 soaking led to a complex with the target. The structure ( Figure 4B) showed a clear change in the orientation of the adenine moiety located in the subsite N (syn vs anti conformation), while that of the adenine moiety in the subsite A was maintained. The purine base flipping positioned the amino end out of the target, while the imidazole ring of the adenine still stacked on Tyr163. However, the N6 spacer arm appears to prevent the usual orientation of the ligand in the subsite N due to van der Waals clashes with neighboring residues, especially Asp150 which is normally hydrogen bonded to the N6 atom. The binding modes of compounds 4 and 5 were very similar to that of NKI1, the spacer arm having little impact on the orientation of the ligands. When the 4-carbon aminoalkyl chain is grafted at the 5'-end of the di-adenosine derivative (compound 4), no particular interaction of this added spacer arm with any residues from the protein could be observed ( Figure 4C) and the amino functional group appeared readily amenable to derivatization. Similarly, when the spacer arm was introduced between the two adenosine residues via the N-propargyl atom (compound 5), the aminoalkyl chain pointed outside the NAD binding site ( Figure 4D). In all three cases, the spacer arms appeared flexible in the crystal structures (weak electron densities; see Figure 4B-D) suggesting they are perfectly amenable to column grafting.

Effects of Compounds 2-5 on LmNADK1 Activity
Another important parameter to select the chemical probe is that it retains its binding properties. The potency of ligands 2-5 as inhibitors of recombinant LmNADK1 in vitro were determined using the coupling assay described previously [15].
The introduction of an aminoalkyl chain at position 6 of the adenosine residue located in the N subsite (compounds 2 and 3) led to sharp decrease in inhibitory potency ( Figure 5). This finding is consistent with our previous work describing the affinity drop induced by the introduction of a N6-cyclopropylamino group on the adenosine residue located in the subsite N [18]. Introduction of the spacer arm at the 5'-end of the diadenosine derivative (compound 4) resulted in a slightly reduced inhibitory potency as compared to that of inhibitor 1 (NKI1). In contrast, when the spacer arm was attached in the middle part of the ligand, between the two adenosine residues via the N propargyl atom, inhibitory potency of compound 5 was slightly increased.

Discussion
By taking advantage of the substrate recognition characteristics of NADK [15,16], we designed four ligand derivatives based on the structure of our lead NADK inhibitor NKI1 [18]. Successful synthesis of the four diadenosine derivatives bearing short aminoalkyl chains to be coupled to NHS-activated Sepharose TM matrix allowed investigation of their specific properties towards NADK. NADK from L. monocytogenes was selected as representative target for this study, as the recombinant enzyme is easily produced and more amenable to subsequent crystallization trials than NADK from S. aureus. Additionally, we showed that SaNADK shares 57% of sequence identity with its listerial counterpart [14,18]. While compound 3 failed to bind to LmNADK, the other three derivatives yielded complexes that diffracted with 2.2-2.4 Å resolution. Compounds 4 and 5 in complex with LmNADK1 showed a binding mode highly similar to that of NKI1. In contrast, compound 2 showed a different orientation which could explain its increased inhibition constant. In line with their binding properties, compounds 4 and 5 had inhibition potentials in the range of NKI1 Ki. Compound 5 had the highest apparent affinity for the purified enzyme of all derivatives tested in our study.
Our functionalized diadenosine derivative 5, and in a lesser extent compound 4, have the required features to develop a specific affinity chromatography column, which could be applied to bacterial NADKs. Ultimately, this chemical probe will help to decipher the selectivity characteristics of our lead compound.

General Information
Commercially available reagents and solvents, unless otherwise stated, were used without purification. Anhydrous reactions were carried out under an argon atmosphere. Analytical thin-layer chromatography (TLC) was performed on TLC plates pre-coated with silica gel 60 F 254 . Compounds were visualized with UV light (254 nm) and by spraying with a mixture of ethanol/anisaldehyde/sulfuric acid/acetic acid (90/5/4/1), followed by heating. Reactions were also monitored using an HPLC system (Agilent (Agilent Technologies, France, Les Ulis) 1100 equipped with a C18 reverse phase column) coupled to a mass spectrometer (ESI source). Flash chromatography was performed with silica gel 60 (230-400 mesh). HPLC purification was carried out on an Agilent system (1100 Series) equipped with a diode array detector using a C18 reverse phase column (Kromasil, 5 µm, 100 Å, 150 × 10 mm) and a linear gradient of acetonitrile in 10 mM triethylammonium acetate (TEAA) buffer over 15 or 20 min at a flow rate of 4 mL/min. 1  NMR spectra of prepared compounds in the Supplementary Materials). High-resolution mass spectra (HRMS) were recorded with a Q-Tof Micro mass spectrometer under electrospray ionization (ESI) using 0.1% formic acid in acetonitrile/water in positive ion mode. The purity of all tested compounds was greater than 97% (HPLC analysis). Retention time (t R ) and gradient are specified.
After stirring overnight at 50 • C, alkylation was incomplete and tert-butyl-3-bromopropylcarbamate (0.23 g, 0.98 mmol) was further added. After heating at 50 • C for 24 h, thiophenol (0.36 mL, 3.26 mmol) was added and stirring was maintained for 18 h at room temperature. Volatiles were then removed under reduced pressure and the residue purified by flash column chromatography (85 g SiO 2 70-230 mesh, 5 to 11% methanol in dichloromethane to yield compound 17 (0.61 g, 79%) as a white foam. 1   To a solution of 17 (0.60 g, 1.3 mmol) in DMF (13 mL), were added DIEA (1.35 mL, 7.7 mmol) and propargyl bromide (80% in toluene, 0.79 mL, 7.10 mmol). After stirring for 2 h at room temperature, the reaction was diluted with ethyl acetate (100 mL) and washed with water (2 x 50 mL). Organic layer was dried over Na 2 SO 4 , filtered and concentrated under reduced pressure. The residue was purified