**1. Introduction**

Poly(ADP-ribose)polymerases (PARP; EC 2.4.2.30) belong to a family of eukaryotic proteins with diverse cellular functions mainly related to DNA repair, maintenance of genomic stability and cell death. The PARP family members [1] perform the catalytic activity of transferring ADP-ribose from the beta-nicotinamide adenine dinucleotide molecule (NAD+) to the acceptors; however, only three of them PARP-1, PARP-2 and PARP-3 possess DNA-dependent (ADP-ribose)transferase activity [1,2]. PARP-1 and PARP-2 catalyze the synthesis of a long stretch of poly(ADP-ribose polymers, whereas PARP-3 is a mono(ADP-ribose)transferase [3]. Acceptors of ADP-ribose, which is synthesized by PARP-1–3, can be both proteins and DNA [4–6]. The catalytic domain (CAT) of PARP-1–3 consists of two subdomains, a helical domain (HD) and the (ADP-ribosyl)transferase (ART) domain. This ART domain contains the active site, and is highly conserved in all members of the PARP family [7]. HD is an autoinhibitory domain to the PARP catalytic activity, and is conserved among DNA damage-dependent PARPs including PARP-1, PARP-2 and PARP-3 [7]. Due to a high homology between the catalytic domains of PARP-1 and PARP-2, small molecule inhibitors of PARP-1 usually possess the inhibitory affect to PARP-2 [8].

PARP-3 has two distinct functions: (a) participation in double strand break (DSB) repair pathway(s) and (b) the regulation of mitotic progression [9]. PARP-3 inhibitors sensitize breast cancer cells to vinorelbine, a vinca alkaloid used for the treatment of metastatic breast cancer [10]. Thus, there is a need for selective inhibitors of PARP-3, both for probing the functions of PARP-3 as an enzyme and evaluating its potential as a therapeutic target.

PARPs have been used as pharmacological targets for the treatment of different types of tumors with defects in DNA repair using synthetic lethality. PARP inhibition is considered a promising cancer treatment strategy, and a number of PARP inhibitors are currently undergoing clinical trials; for a review, see [11]. Four PARPi, olaparib, rucaparib, niraparib and talazoparib, have already been approved in the U.S. or Europe, mainly for the treatment of BRCA-deficient cancer [12–14].

Nicotinamide adenine dinucleotide (Figure 1) plays a key role in such vital processes, as maintaining the integrity of the genome, energy supply, cell death and others [15]. NAD+-metabolizing enzymes are considered by researchers as targets for the treatment of a variety of human diseases, including cancer, multiple sclerosis, neurodegeneration and Huntington's disease [16]. Nicotinamide, which is one of the structure elements of NAD+ and one of the major products of ADP-ribozylation, inhibits PARP-1 [17]. A large number of existed so far enzyme inhibitors, including some approved by the United States Food and Drug Administration (FDA), have been developed on the basis of nicotinamide [18]. Binding features of nicotinamide analogs include the formation of hydrogens bonds between the lactam or the carboxamide group of the inhibitor with the Gly863 of backbone chain and Ser904 of the side chain, as well as pi-stacking interactions with Tyr907 [19–21]. In spite of a great amount of developed analogs of nicotinamide for PAPR-1–3 inhibition, there is an increasing interest in creating novel ones, as evidenced by the growing number of publications in scientific journals [22–24]. However, a pharmaceutical usage of existing PARPi is still limited for a number of reasons: the presence of side effects, the appearance of tumor resistance to PARPi and the different efficacy in combination with traditional chemotherapy [14,25]. Therefore, the development of new structural classes of compounds with improved properties, such as selectivity to different types of PARP enzymes, increased inhibitor efficacy, lessened toxicity and higher bioavailability, remains an urgent task.

**Figure 1.** The structure of the beta-oxidized nicotinamide adenine dinucleotide (NAD+)-molecule.

Natural and modified nucleosides and their phosphorylated derivatives are indispensable tools in searching new anticancer, antiviral and antibacterial drugs [26,27]. However, a small number of studies on nucleoside derivatives as PARPi have been reported in the literature, although some of them exhibit moderate PARP-1 inhibition activity [28]. It has been shown that some thymidine derivatives modified at the 5- and/or 50 -position inhibited PARP-1 more efficiently than 3-aminobenzamide, which is the first generation of the PARP-1 inhibitors and the closest structural analog of nicotinamide [29]. Among phosphorylated adenosines (AMP, ADP, ATP, 30 ,50 -cycloAMP, 30 ,50 -diphosphoadenosine) a 3 0 ,50 -diphosphoadenosine inhibits PARP-1 most effectively [30]. Ueda's group show that 5-substituted derivatives of uridine and uracil exhibited an inhibitory activity against PARP-1 too [31]. Additionally, a number of small molecules containing uracil derivatives could be seen as a class of potent PARP-1 inhibitors [32]. There are known PARP-1 inhibitors with a high activity constructed on the base of isoindolinone derivatives and their conjugates with adenosine joined by various aliphatic spacers [33]. This fact may be explained by the dual binding of both parts of the inhibitor, nicotinamide- and adenosine-mimicking. According to the literature data, compounds binding simultaneously into the nicotinamide and adenosine binding sites of the active center of PARP family enzymes could be more selective inhibitors [34].

The NAD+ molecule may be modified at adenine or nicotinamide heterocyclic bases, ribose moieties or the pyrophosphate chain. The investigation of the applicability of fluorescent NAD+ analogs modified at different positions of the adenine moiety has been performed [35]. Recently, fluorescent NAD+ derivatives have been used for real-time cellular imaging of protein poly(ADP-ribosyl)ation [36]. Clickable NAD+ analogs modified at C8 of adenine are promising tools to illuminate the ADP-ribosylated proteome and investigate the molecular mechanisms carried out by individual PARPs upon different cellular signals [37]. Moreover, the well-known non-hydrolyzable NAD+ analog of benzamide adenine dinucleotide (BAD) has been successfully applied to establish the mechanism of PARP-1 interaction with DNA substrate [38]. All these studies demonstrate the power of modified NAD+ derivatives as potential tools for goals of molecular and cellular biology.

Previously, some naturally occurring dinucleoside 50 ,50 -pyrophosphates were investigated as potential PARP-1 inhibitors [39]. Diadenosine 50 ,50 -tetraphosphate was shown to inhibit the ADP-ribosylation of histone H1 by PARP from the bovine thymus [40]. However, there are published significantly fewer modern investigations devoted to the study of dinucleoside pyrophosphates or NAD+ derivatives as inhibitors of enzymes using NAD+ as a substrate and not as a cofactor. These were found to inhibit NAD kinase [41], bacterial DNA ligases [42] and CD38 NAD glycohydrolase [43]. Recently, we have developed a number of NAD+ mimetics presented by conjugates of ADP and nicotinamide riboside analogs [44,45].

It was shown that some of those provided an inhibitory effect to the autopoly(ADP-ribosyl)ation of PARP-1; IC<sup>50</sup> of the most effective conjugate consisting of ADP and morpholino-glycine thymine nucleoside [46] was 41.5 ± 3.5 µM [45]. The combination of data from in silico models of PARP-1 with the NAD+-molecule [47] and the crystal structure of the CAT-∆HD domain of PARP-1 with the non-hydrolyzable NAD+ analog of BAD [38] suggests the requirement of conducting a rational searching for the details of mechanisms of action and specificity of new NAD+ analogs, being the conjugates of ADP and morpholino nucleosides.

Morpholino nucleosides are widely used for the synthesis of oligonucleotide mimetics [48]. However, there are only few examples of the utilization of morpholino nucleoside monomers or their phosphorylated derivatives for the goal of molecular biology and biochemistry [49,50]. In the present study we describe the synthesis, inhibition activity and the molecular modeling of the novel conjugates combining from ADP and morpholino nucleosides (morpholino nucleoside adenine dinucleotides, MorXppA), where morpholino nucleosides mimic the nicotinamide riboside fragment of the NAD+ molecule (Figure 2).

**Figure 2.** Morpholino nucleoside adenosine dinucleotides (MorXppA).

#### **2. Results and Discussion**

#### *2.1. Chemistry*

### 2.1.1. Synthesis of ADP Conjugates Containing Phosphoester (P–O) Bond

Recently we have proposed a versatile method for the synthesis of ADP conjugates functionalized at the terminal phosphate [44,45]. In that study, we developed an effective protocol for the coupling of two monoester phosphate derivatives under the action of the redox coupling pair triphenylphosphine/2,20 -dipyridyldisulfide (Ph3P/(PyS)2) in the presence of 1-methylimidazole (MeIm). We applied the same strategy for the synthesis of the sought-for pyrophosphates MorXppA **4**, where X represents an oxygen atom (Figure 2, Scheme 1).

**Scheme 1.** Synthesis of the conjugates of adenosine diphosphate (ADP) with 2 0 -hydroxymethylmorpholino nucleosides. Reagents and conditions: (a) NaIO<sup>4</sup> , EtOH/H2O, 15 min; (NH<sup>4</sup> )2B4O<sup>7</sup> ·4H2O, Et3N, 1.5 h; NaBH3CN, 40 min; trifluoroacetic acid (TFA), pH 3–4, 1 h; TrCl, Et3N, dimethylformamide (DMF), 3 h; yield 60%–70%; (b) POCl<sup>3</sup> , Py, –15 ◦C, 15 min; 1 M triethylammonium bicarbonate (TEAB), yield 75%–90%; (c) Ph3P/(PyS)<sup>2</sup> , MeIm, 1,3-dimethyl-2-imidazolidinone (DMI); n-Bu3NH<sup>+</sup> salt of AMP; conc. aq. NH<sup>3</sup> for compounds **4A**,**G**,**C**; 80% aq. AcOH (v/v); yield 70%–80%.

At the first step, we provided preparation of *N*-Tr-protected morpholino nucleosides. The standard protocol for nucleosides with natural heterocyclic bases was Summerton's procedure [51]. Recently we have published a detailed protocol for the synthesis of 5-iodopyrimidine morpholino nucleosides based on this method [52]. Continuing our research, we have succeeded in obtaining *N*-Tr-protected morpholino nucleosides **2** from acyl-*N*-base-protected (when necessary) ribonucleosides **1** without chromatographic purification in an overall yield of 60%–70%.

To obtain monophosphate **3U**, we treated protected morpholino nucleoside **2U** with 2 eq of POCl<sup>3</sup> in dry pyridine (Py) under cooling in ice bath according to [53]. Unfortunately, the product **3U** was obtained in a yield of less than 50% after reverse phase chromatography (RPC). We varied the excess of POCl<sup>3</sup> (2–4 eq), the temperature (0 –−15 ◦C) and reaction time (5–30 min) and found optimal conditions for the monophosphorylation of nucleosides **2** (Scheme 1). The monophoshates **3A,G,C,IU** were obtained without chromatographic purification in a yield of over 90%; and monophoshates **3U,T,BrU,ClU** in a yield of 75%–80% after RPC.

To prepare ADP conjugates **4**, we activated a phosphate group in derivatives **3** by Ph3P/(PyS)<sup>2</sup> in a presence of MeIm in dry 1,3-dimethyl-2-imidazolidinone (DMI) at room temperature for 15–20 min according to [44]. Further coupling of phosphoro(*N*-methyl)imidazolidate of morpholino nucleoside with AMP was carried out in situ in dry DMI for 1 h. After chromatographic purification and deblocation procedures, we obtained target conjugates **4** in an overall yield of 70%–80% (Scheme 1).

### 2.1.2. Synthesis of ADP Conjugates Containing Phosphoramide (P–N) Bond

For the synthesis of these conjugates we prepared 20 -aminomethylmorpholino nucleosides **7** (Scheme 2).

**Scheme 2.** Synthesis of the conjugates of ADP with 20 -aminomethylmorpholino nucleosides. Reagents and conditions: (a) Ph3P, Im, DCE, I<sup>2</sup> , 0 ◦C → rt, 5 h; (b) NaN<sup>3</sup> , DMF, 12 h; (c) H<sup>2</sup> , 10% Pd/C, MeOH; (d) Ph3P, CBr<sup>4</sup> , DMI; (e) Ph3P (2 eq), Py; conc. aq. NH<sup>3</sup> ; (f) ADP n-Bu3N salt, Ph3P, (PyS)<sup>2</sup> , MeIm, DMI; conc. aq. NH<sup>3</sup> for compounds **10A**,**G**,**C**; 80% aq. AcOH (*v*/*v*).

An azido group is a convenient source for amino functions. In the literature, several methods for obtaining 50 -azidoribonucleosides containing natural heterocyclic bases are known. Methods are based on conversion of the hydroxyl group to halogen [54–59], sulfonate [60–62] or phosphate ester group [63] under anhydrous conditions. The further reaction with lithium or sodium azide leads to the formation of 50 -azidoribosyl derivatives in an overall yield of 14%–92%. Similar approaches were proposed for the synthesis of 20 -aminomethylmorpholino nucleosides containing natural heterocyclic bases. One of them is based on the one-pot synthesis of 20 -azidomethylmorpholino nucleosides under the treatment of compounds **2A,G,C,U,T** with Ph3P/CBrCl3/NaN<sup>3</sup> in dry dimethylformamide (DMF). After reduction of the N3-group and a full deblocation procedure, 2<sup>0</sup> -aminomethylmorpholino nucleosides were obtained in an overall yield of 39%–64% [46,64].

Another procedure is based on obtaining 20 -sulfonylmethylmorpholino nucleosides by the treatment of compounds **2A,G,C,T** with mesyl chloride in dry pyridine and a further substitution of mesyl with an azide anion. After the reduction of the N3-group 2<sup>0</sup> -aminomethylmorpholino nucleosides were obtained in an overall yield of 34%–45% [65].

A literature search for the synthesis of 50 -azidonucleosides containing 5-halosubstituted pyrimidines provided only a few reports. Most of suggested methods are based on the ribosylation of modified heterocyclic bases with 50 -azidosugar derivatives [66–68]. In 1976, Prusoff et al. reported a synthesis of the 50 -azido-5-halouridine by substitution of the sulfonate group of 50 -tosyl-5-halouridine with the azide one in a yield of 33%–73% [69]. The attempts to reduce the N3-group of 5 0 -azido-5-chlorouridine under the action of H2/PtO<sup>2</sup> resulted in the dehalogenation of 5-chlorouridine. 5 0 -Amino-5-halouridine was obtained by the halogenation of 50 -aminouridine with Hg(OAc)<sup>2</sup> and *N*-bromosuccinimide (NBS) or I2. Later, in a search for obtaining 5<sup>0</sup> -aminoarabinonucleosides containing 5-bromo- or 5-iodouracil, Prusoff et al. found that the halogenation of 50 -aminoarabinouridine by NBS or *N*-iodosuccinimide (NIS) did not lead to the formation of the target compounds [70]. In addition, attempts to replace a sulfonate group of 50 -tosyl-5-iodoarabinouridine with the azido one under the action of lithium azide failed. The only successful approach was the halogenation of 5 0 -azidoarabinouridine by NBS or NIS in a yield of 94% (for the Br-derivative) and 52% (for the I-derivative), and then the reduction of this azido function in the presence of Ph3P in Py in a yield of 56% (for the Br-derivative) and 41% (for the I-derivative). The authors noted that the reduction of the azido group by NaBH<sup>4</sup> or H2/Pd/C led to dehalogenation of the 5-bromo- or 5-iodouracil derivatives.

We did not find any data concerning the synthesis of 20 -aminomethylmorpholino nucleosides **7IU, BrU, ClU** containing 5-halosubstituited uracil or any modified heterocyclic base in the literature. Thus, at first, we tried to synthesize the azido derivative **6BrU** following the one-pot procedure using Ph3P/CBrCl3/NaN<sup>3</sup> in dry DMF according to the published method [46]. However, our attempts failed. Under using DMI instead of DMF and changing the order of the reagent's addition [58], we obtained azido derivative **6BrU** in a yield of 24% (Scheme 2). Since the one-pot method did not provide a satisfactory yield of the target compound, we applied a "step-by-step" approach and tested the method on the compound **2U**. To our surprise, after treatment of the compound **2U** with Ph3P/CBrCl<sup>3</sup> or Ph3P/CBr<sup>4</sup> in dry MeCN or DMI we observed a migration of the Tr-group from the 4<sup>0</sup> -*N* position to 20 -OH-methyl one giving *O*-tritylated morpholino nucleoside **8U**. Then, we tried to convert the hydroxyl group of morpholino nucleosides **2** into an iodine function and applied the I2/Ph3P/Im reagent [71]. The reaction of 20 -hydroxymethylmorpholino nucleosides **2** with this reagent in a variety of solvents (DMI, *N*-methyl-2-pyrrolidone, DCE, DCM, MeCN, THF) leads to a quantitative conversion of the compounds **2** into the compounds **5** in 5 h according to HPLC analysis. Further treatment of the reaction mixture with aq. sol. of NaHSO<sup>3</sup> and NaHCO<sup>3</sup> and chromatographic purification results in iodine derivatives **5** in a yield of 85%–93%. When the reaction mixture was treated with aq. sol. of NaHCO<sup>3</sup> only a partial detritylation of the target products were observed. Reaction of the hydroxyl group substitution with iodine also proceeds when pyridine was used instead of imidazole in the reagent mixture I2/Ph3P/Im [55,71,72]. However, when the reaction of **2T** with I2/Ph3P/Py was carried out in DCE, we observed the formation of two products **5T** and **8T** in a ratio of 1:2.

The reaction of the compounds **5** with sodium azide in DMF afforded morpholino nucleosides **6** in a quantitative yield. We tried to reduce the azido to an amino group by Pd/C-catalyzed hydrogenation and by Ph3P/Py treatment. However, in a case of treatment of the compound **6U** with Ph3P in dry Py we observed the reduction of the azido group and the migration of the trityl moiety to the 20 -aminomethyl position of the morpholine ring resulted in the compound **9U**. The successful reduction of azido derivatives **6** was carried out by Pd/C-catalyzed hydrogenation. In this way, 20 -aminomethylmorpholino nucleosides **7** were obtained in a nearly quantitative yield by "step-by-step" approach. It should be noted that under reducing the 20 -azidmethylmorpholino nucleosides containing 5-halo-substituited uracil **6IU, BrU, ClU** by both Pd/C-catalyzed hydrogenation and Ph3P/pyridine, we did not observe any dehalogenation of the heterocyclic base [52].

The synthesis of ADP conjugates **10** containing the phosphoramide (P–N) bond was carried out by Ph3P/(PyS)2/MeIm activation of the β-phosphate of ADP and further coupling with the 2 0 -aminomethylmorpholino nucleosides **7** (Scheme 2). Previously, the approach consisting in the Ph3P/(PyS)2/MeIm activation of terminal phosphate in oligonucleotides to obtain phosphoroamide derivatives was widely used [73]. After chromatographic purification and a deblocation procedure ADP conjugates **10** were obtained in an overall yield of 70%–75%.
