Synthesis and Macrodomain Binding of Gln‐carba‐ADPr‐peptide

Mono‐ADP‐ribosylation is a dynamic post‐translational modification (PTM) with important roles in cell signalling. This modification occurs on a wide variety of amino acids, and one of the canonical modification sites within proteins is the side chain of glutamic acid. Given the transient nature of this modification (acylal linkage) and the high sensitivity of ADP‐ribosylated glutamic acid, stabilized isosteres are required for structural and biochemical studies. Here, we report the synthesis of a mimic of ADP‐ribosylated peptide derived from histone H2B that contains carba‐ADP‐ribosylated glutamine as a potential mimic for Glu‐ADPr. We synthesized a cyclopentitol‐ribofuranosyl derivative of 5′‐phosphoribosylated Fmoc‐glutamine and used this in the solid‐phase synthesis of the carba‐ADPr‐peptide mimicking the ADP‐ribosylated N‐terminal tail of histone H2B. Binding studies with isothermal calorimetry demonstrate that the macrodomains of human MacroD2 and TARG1 bind to carba‐ADPr‐peptide in the same way as ADPr‐peptides containing the native ADP‐riboside moiety connected to the side chain of glutamine in the same peptide sequence.


Introduction
Adenosine diphosphate ribosylation (ADP-ribosylation) is a post-translational modification (PTM), in which NAD + is used to transfer ADP-ribose onto an internal amino acid at an automodification domain, or onto an amino acid of another target protein. [1]Mono-ADP-ribosylation (MARylation) and the succeeding poly-ADP-ribosylation (PARylation) are both modifications catalyzed by enzymes belonging to the poly(ADP-ribose) polymerases (ARTD or PARP) family. [2]In total, the ARTD family constitutes 17 proteins that play essential roles in cell differentiation, chromatin regulation and genome integrity maintenance, and various prokaryotic ARTDs also act as toxins. [3]In humans, ARTD1 (PARP1), ARTD2 (PARP2) and ARTD 3 (PARP-3) detect and initiate single-and double-strand DNA break repair mechanisms, including base-and nucleotide excision repair. [4]hese roles in DNA damage repair have highlighted ARTDs as proteins of biomedical importance, most notably in relation to cancer and inflammatory disorders.Most ARTDs are transferases that can only transfer a single mono-ADP-ribose (MAR) onto their target proteins. [5]ARTD1 (PARP1), ARTD2 (PARP2), and tankyrases all catalyse PARylation. [6]Both PARylation and MARylation of proteins are reversed through the action of various hydrolases, such as, PARG, MacroD1, MacroD2 and TARG1, completing the ADP-ribosylation cycle. [7]he relation between ADP-ribosylation of acceptor amino acids in the target proteins and the corresponding biochemical function is poorly understood.Historically glutamate and aspartate have been considered to be the primary acceptor amino acids for ADP-ribosylation, but more recently lysine, arginine, serine, cysteine and histidine have been reported as ADPr acceptors as well. [8]Furthermore, non-covalent complexes between poly-ADP-ribose and histones have also been demonstrated. [9]Because of the chemical sensitivity of anomeric esters (or, more precisely, acylal linkages), data on ADPribosylation of histone proteins at glutamate or aspartate residues is conflicting. As part of a program to synthesize such molecular tools, we developed the synthesis of various ADP-ribosylated peptides, [10c,11] among which a part of mono-ADP-ribosylated human histone H2B. [12]10a] Although the anomeric glutamate 1, depicted in Figure 1 appears to be only a minor component (less than 5 %) of this regioisomeric mixture under the physiological conditions, [10a] it is a biologically important structure due to the fact that the α-configured 1 is the immediate product of the ADP-ribosylation of glutamic acid side chain by transferases from ARTD-family (PARPs) and the substrate for glycohydrolases responsible for the reversal of the ADP-ribosylation of Glu.That is why stable isosteres of Glu-ADPr are of interest for the studies on ADP-ribosylation of proteins.10c] Determination of the binding capacity with macrodomain-containing hydrolases showed that conjugate 2 binds to MacroD2 TM , while no binding was detected with TARG1 D125A , point mutants of MacroD2 and TARG1 that have impaired hydrolase activity, while ADPr and MARylated ATRD10 binding is retained by these engineered mutant proteins.In this paper, we report the synthesis and evaluation of peptide 3 (Figure 2) featuring a carba-ribofuranose moiety.Carba-ribofuranose as a ribose mimic has been applied in the past both successfully [13] and unsuccessfully [14] in various studies on ADP-ribosylation and it is interesting to explore whether this mimic can be used to study proteins that bind mono-ADP-ribosylated proteins as anomerisation is excluded.The amide bond as in Gln-ADPr peptide 2 is maintained in its carba-derivative 3, as migration of the original ester function cannot be excluded.The affinity of conjugate 3 toward human MacroD2 and TARG1 using isothermal titration calorimetry was determined to evaluate the suitability of Glncarba-ADPr to serve as a faithful isostere of Glu-ADPr.

Results and Discussion
10c] Retrosynthetically, this leads to building block 4, in which a Fmoc protected glutamic acid is functionalized with a protected and phosphorylated carba-riboside (Figure 2).Disconnecting building block 4 at the C 1 -glutamylamide and O 5 -phosphodiester residues reveals precursor 1-amino-carbaribofuraside 5. Using the method of Parry et al intermediate 5 is accessible through the photochemical addition of methanol onto the β-position of cyclopentenone 6.Although Parry et al. reported the synthesis of 6, their route entailed several low yielding steps with an overall yield of 10 % after four steps. [15]n alternative route to 6, described by Borcherdinger et al., [16] could not be reproduced with the same complications that Mariën et al experienced. [17]Therefore, the newly designed route of synthesis toward cyclopentenone 6, as shown in Scheme 1 was undertaken first.Fischer glycosylation of D-ribose with methanol and subsequent isopropylidation of the C 2 -C 3 -diol in the obtained crude methyl β-D-ribofuranoside using 2,2-dimethoxy propane and a catalytic amount of camphor sulfonic acid, provided 8 in 69 % yield over 2 steps.Iodination of the primary alcohol in 8 with iodine, triphenylphosphine and 1H-imidazole at 100 °C in toluene/MeCN, smootly produced ribofuranoside 9.In a two step one-pot procedure iodide 9 was converted into diene 11.Hence, lithium-iodine exchange at À 78 °C converted acetal 9 into aldehyde 10, that was immediately treated with vinylmagnesium bromide, to produce R/S-alcohol 11 in 90 % yield over two steps.Since the newly introduced chiral center is lost in a later stage of the synthesis, no efforts were undertaken separate the epimers of 11,.Ring-closing metathesis of diene 11 using 1 st generation Grubbs catalyst in DCM provided R/Scyclopentenoid 12 in 82 % yield.It is of interest to note that the 1 st generation Grubbs catalyst performed here equally well to the more expensive 2 nd generation variant if DCM was rendered oxygen-free.In addition, the catalyst loading could be reduced to 0.6 mol% on a 100 mmol scale.Final oxidation of R/S-cyclopentenoid 12 with excess manganese dioxide afforded target cyclopentenone 6.
With cyclopentanone 6 available, the synthesis of glutamine carba-riboside 4 was undertaken (Scheme 2).The photochemical addition of methanol onto the β-position of cyclopentenone 6 provided carba-riboside 13 in 61 % yield.Protection of the alcohol in 13 with TBSÀ Cl in the presence of imidazole gave fully protected 14.Condensation of ketone 14 with methoxylamine proceeded quantitatively to give O-methyloxime 15 as a mixture of isomers.The necessary stereoselective reduction of the imine turned out to be difficult and various conditions (BH 3 , LiAlH 4 , NaBH 3 CN, Zn/H + ) were explored, of which LiAlH 4 in refluxing THF proved to be the most effective, providing carbaribosamine 5 in 30 % yield and complete stereoselectivity.A hydride was delivered stereoselectively from the less-hindered convex β-face of compound 15, resulting in the exclusive 5 formation as the α-epimer.Significantly higher yield (81 %) was obtained when the 5-OH was protected with a trityl instead of a TBS.PyBOP-mediated condensation of amine 5 with commercially available Fmoc-Glu(OBn)-OH gave carba-ribosylated gluta- Selective phosphorylation of primary alcohol in triol 17 with ditert-butyl-N,N-diisopropyl-phosphoramidite and 1H-tetrazole, followed by in situ oxidation, afforded phosphotriester 18 in 50 % yield.Finally, hydrogenolysis of the benzyl ester provided target building block 4 that was applied in the forthcoming SPPS.
With carba-ribosylated glutamine building block 4 in hand, the solid-phase peptide synthesis (SPPS) of stabilized mono-ADP-ribosylated H2B conjugate 3 was undertaken (Scheme 3).The SPPS campaign comprised the use of the commercially available Fmoc-Lys(TFA)-OH, Fmoc-Pro-OH, Fmoc-Ala-OH and Fmoc-Ser(Trt)-OH amino acids and Tentagel resin functionalized with Fmoc-glycine via a 4-hydroxymethylbenzoic acid (HMBA) linker and standard Fmoc-based SPPS coupling methodology.After the construction of intermediate immobilized peptide Fmoc-Pro-Ala-Lys-Ser-Ala-Pro-Ala-Pro-Lys-Lys-Gly-HMBA the trityl group on Ser was replaced by an acetyl to allow simultaneous base mediated global deprotection and release from solid support at the end of the synthesis.Incorporation of carba-ribosylated glutamine with building block 4 proceeded uneventfully, as determined by LC-MS analysis of an aliquot of the peptide obtained from resin-bound conjugate 20.Installation of the ADP-pyrophosphate moiety starts with the removal of the tert-butyl groups from the phophotriester in 20 with 10 % TFA in DCM.As gauged by on-resin 31 P NMR spectroscopy performed as described [18] of 21 this deblocking reaction proceeded cleanly.On-resin phosphorylation of 21 with known adenosine phosphoramidite 22, under the agency of 5-ethylthiotetrazole (ETT), followed by CSO-mediated oxidation of the phosphate-phosphite intermediate, produced the protected resin-bound carba-ADP-ribosylated peptide.Final deprotection consisted of DBU-assisted elimination of the 2-cyanoethanol group and subsequent global deacylation and concomitant release from solid-support through ammonolysis.The crude 3 was precipitated and HPLC purified, to provide mono-carba-ADP-ribosylated H2B peptide 3 in 9 % (8.2 mg, 0.6 μmol) overall yield.
With carba-ADP-ribosylated H2B peptide 3 in hand, we established its affinity toward human MacroD2 and TARG1 using the same approach as used earlier for the binding capacity of its ADP-ribosylated counterpart 2 (Figure 1).In order to distinguish between the ability of a macrodomain module to recognize ADP-ribosylated peptides compared to their potential activity on the bound ligands as ADP-ribosyl-hydrolases, we used engineered point mutants of the two macrodomain proteins MacroD2 and TARG1 bearing specific point mutants.Thus, to circumvent possible complications from ADP-ribosylhydrolase activity, we used the MacroD2 mutant MacroD2 G100E/I189R/Y190N (referred to as MacroD2 TM ), as well as the TARG1 mutant protein TARG1 D125A as relevant point mutants.The catalytic activity of these mutants is impaired, while ADPr and MARylated ARTD10 binding is retained. [19]heme 3. The solid-phase peptide synthesis of mono-carba-ADP-ribosylated H2B conjugate 3. Assembly of core peptide 20:

The MacroD2 Protein Selectively Recognizes Mono-Carba-ADP-r H2B Peptide
Using isothermal titration calorimetry, we were able to assess the affinity of two distinct macrodomain modules for the synthesized conjugate 3. The peptide bound MacroD2TM mutant reliably with an affinity of 3.3 μM (Figure 3).In contrast, the TARG D125A mutant did not show any binding enthalpy for this peptide (Figure 3).These in vitro data indicate that conjugate 3 is a ligand of MacroD2, but not of TARG1.In order to ensure that the TARG1 D125 protein was natively folded and that the lack of binding to conjugate 3 was not due to protein misfolding, we used the TARG D125A protein after the ITC run conjugate 3 had been performed in a second ITC run, but now with monomeric ADP-ribose as a ligand.These results clearly showed that ADP-ribose binds the TARG1 D125 protein well, as expected (Figures 3S-5S, Supporting Information).Thus, conjugate 3 is selectively recognized by MacroD2 but not by TARG1, showing that distinct macrodomain proteins can exhibit high selectivity for ADP-ribosylated peptides.10c]

Conclusions
This paper describes the solid phase peptide synthesis of mono-ADP-ribosylated N-terminal H2B conjugate 3, in which a new bioisostere of ADP-ribosylated glutamic acid is incorporated.This bioisostere is characterized by the presence of carbariboside instead of the naturally occurring riboside and an amide substituting the native ester.For the SPPS of Gln-carba-ADPr peptide 3 a new building block (4) was developed.The application of this building block in SPPS and the on-resin introduction of the ADP-pyrophosphate proceeded successfully.The selected protective group pattern in the immobilized construct enabled simultaneous base-mediated global deprotection and release from resin, providing the target mono-ADP-C-ribosylated H2B conjugate 3 in multi-milligram amounts.The binding of the Gln-carba-ADPr-peptide to the ADPr-binding proteins MacroD2 and TARG1 closely mimics that of the ADPrpeptide containing the native ribose moiety.

Figure 3 .
Figure 3.The binding affinity of the MacroD2 and TARG1 macrodomain for carba-ADP-ribosylated histone H2B peptide 3. Representative isothermal titration calorimetry (ITC) profile for the titration of 3 into a solution containing either MacroD2 TM or TARG1 D125A (upper panel) and data derived from the interaction measurements between the macrodomains and the carba-ADPr peptide 3. Data are mean (n = 3) � SEM (values in brackets).