Purine disaccharides 2'-O-β-D-ribofuranosyladenosine-5''-phosphate (Arp, 1a) and 2'-O-β-D-ribofuranosylguanosine-5''-phosphate (Grp, 1b) were isolated for the first time from yeast methionine tRNA (tRNA^Met) and several plant sources [8, 10, 19]. This type of nucleoside is located at position 64 of tRNA^Met, in its T-loop [8]. The position of the additional phosphate group was determined by periodate oxidation of disaccharide nucleotides with their further β-elimination. It was also shown that an additional ribofuranose moiety is attached to a nucleoside via the β-glycoside bond. The formation and role of Arp in tRNA is still obscure [8]. It is formed by post synthetic transfer of phosphoribosyl moiety from 5-phosphoribosyl-1-pyrophosphate (PRPP) to Ado-64 catalyzed by ribosyl transferase enzyme (Rit). X-ray analysis disclosed the 5-phosphoribosyl moiety occupies a strictly defined place in a small groove on tRNA tertiary surface. The additional phosphate group forms a hydrogen bond with 2-NH₂ of neighboring Guo-63. Selective removal of 5-phosphoribosyl moiety led to tRNA beginning to function as an elongator, but less efficiently compared to a native tRNA^Met. Thus, it can be proposed that such a modification plays a definite role in the discrimination of initiation-elongation of protein biosynthesis, preventing the formation of tRNA complex with elongation factor EF-1a and GTP [8]. Indeed, we know a little about Arp and Grp activity as low-molecular bioregulators. Therefore, we made some attempts to test their activity on some protein targets (see part about therapeutic application). The first synthesis of Arp and Grp was elaborated by Mikhailov and colleagues [20–23] starting from 3',5'-O-(tetraisopropyldisiloxane-1,3-diyl)ribonucleosides with 1,2,3-tri-O-benzoyl-5-O-phenoxyacetyl-α/β-D-ribofuranose in the presence of tin tetrachloride (SnCl₄) in dry dichloroethane (DCE) at 0°С to form disaccharide products in 60–68% yields respectively. Further phenoxyacetyl deblocking with 0.1 М K₂CO₃ in MeOH and phosphorylation with bis[2-(4-nitrophenyl)ethyl]phosphate and further total deblocking afforded the desired products with 14% and 18% total yields for 1a and 1b respectively (Figure 1).

The corresponding disaccharide nucleosides (10a-b) can be synthesized in analogous conditions by simple condensation of 3',5'-O-(tetraisopropyldisiloxane-1,3-diyl)ribonucleosides with slight excess of peracylated ribofuranose. The reaction was performed in mild conditions (0°С, DCE, 7–16 h) in 74–82% yield, though at ambient temperature the reaction proceeded faster (30 min) but with lower yields (40–77%) [8].

PAR is a complex branched biopolymer, represented in Figure 2.

Native PAR molecules consist of ca. 200 monomeric units, in which linear segments alternate with branching motifs [10]. Enzymatic hydrolysis of the pyrophosphate bonds by snake venom phosphodiesterase (not shown in Figure 2) led to the formation of a mixture of disaccharide and trisaccharide nucleotides, subsequent treatment of which with alkaline phosphatase (Figure 2, pathway e) afforded disaccharide nucleoside (12) and trisaccharide nucleoside (13) respectively [24, 25]. Their structures were determined by NMR spectroscopy. Compound 13 represents the core motif of PAR branching. The common feature of compounds 12 and 13 is the presence of α(1→2) glycosidic bonds (Figure 2). The structure of PAR has some similarity to the structure of nucleic acids, DNA and RNA because antibodies raised against PAR can recognize RNA and DNA [7, 26]. Although PAR exists in an extended conformation in low salt buffers, it may also undergo some structural changes in high salt buffers, showing cooperative hyperchromic helix-coil transitions for a long molecule [26]. Three-dimensional structural information available by investigations of isotopically labeled PAR with nuclear Overhauser enhancement spectroscopy (NOESY) has revealed adenine base location in anti-conformation and thus heterocyclic moieties in PAR are accessible both for inter- or intramolecular interactions, that is in agreement with possibilities of helix-coil transitions, established by CD-spectroscopy [27]. It is not therefore surprising that PAR functions in cooperation with DNA and RNA structures and respectively associated proteins [7]. Thus, PAR interacts directly or indirectly with a series of regulatory and repair proteins and enzymes and modulates their functions [7, 28–31]. Covalent modification of proteins proceeds either via the formation of O-glycoside bond (Glu/Asp, Ser/Thr residues) or via the formation of N-glycoside bond (Lys/Arg, Asn/Gln residues) with short ADPR moiety being an anchor for PAR building up [32]. Covalent modifications of serine, acetyllysine, phosphoserine and covalent modification of DNA and RNA at guanine base were also described [33]. On molecular level functional role of PAR is based preferably on ionic interactions with target proteins and binding with nucleoside parts of ADPR-ADPR junction or terminal ADPR fragment. Charged lysine- and arginine-rich motifs are essential for ionic interactions of histones with PAR and thereby afford injured DNA unwinding for repair [34].

Biosynthesis and cleavage of PAR is catalyzed by several enzymes of various classes (Figure 2). In cellular nuclei, PAR is synthesized by PAR polymerases (PARPs) recruiting nicotinamide adenine dinucleotide (NAD⁺) as a substrate (Figure 2, pathway a). Nowadays, PARP family numbers about 17 enzymes. PARP-1 (EC 2.4.2.30) is responsible for the synthesis of 90% of PAR in cells and can be considered as the most widespread PARylating enzyme in higher eukaryotes. The first step associated with PAR biosynthesis is a covalent attachment of the first ADPR moiety to an acceptor amino acid. Further biosynthesis of PAR consists of elongation of PAR chain by joining of additional ADPR monomers; and generation of branching points [7, 35–39]. In fact, enzymes catalyzing these two steps strongly differ in structure and function. PARP-1 cannot operate with small threads of PAR, while the other representatives of the PARP family can transfer only short ADPR fragments. This leads to enormous complexity of PARP functions in cells. PAR is cleaved by three enzymes, viz., (ADP-ribosyl) protein lyase (ADPRL; Figure 2, pathway b), PAR glycohydrolase (PARG; Figure 2, pathway c), and NUDIX hydrolase, cleaving pyrophosphate linkages [38, 40–43]. O-Glycosidic bond between adenosine and ribofuranose moieties in PAR-polymer is split by PARG. ADPRL catalyzes the cleavage of O-glycosidic bond between ADPR residue and proteins [38]. Thus, at least 20 enzymes are involved in the biosynthesis and degradation of PAR polymer.

Intramolecular cyclization is another bioregulatory role of NAD⁺ which affords a series of cyclic ADPR (cADPR), the general function of which being immune regulation that is tightly associated with cytoplasmic Ca²⁺ release and entry [53–58]. Toll/interleukin-1 receptor (TIR) is a family of ancient immune modules present in all domains of life and possessing a NAD⁺ cyclization catalytic activity (Figure 6).

SARM1 belongs to the most evolutionally conserved type of TIR-receptor, expressed also in human cells. TIR/SARM undertake NAD(P)⁺ degradation and cyclization [58–63]. At a glance, on molecular level, NAD⁺-cyclization is hypothesized to take part in the development of innate immunity by affection on cytoplasmic Ca²⁺ release and entry. 1N-Ribofuranosyl-ADP or cADPR (35), the most widespread regulator, is a cyclic dinucleotide that is produced from NAD⁺ by intermolecular transfer of ribosyl moiety from nicotinamide of NAD⁺ to position 1 of adenine. In humans and mammalians, this transfer is catalyzed not by TIR, but by ADP-ribosyl cyclases CD38 and CD157 (Figure 6, red arrow) [56, 64, 65]. Firstly, discovered in sea urchins, cADPR then appeared to participate in wider biological roles, being a universal Ca²⁺ mobilizing second messenger, especially receptor stimulation in immune cells. Inside T-cell cADPR is involved in sustained Ca²⁺ release and also in Ca²⁺ entry. It mobilizes intracellular Ca²⁺ that in turn controls a diverse range of highly regulated cellular processes [64]. The physiological receptor for cADPR is the ryanodine receptor (RyR), while Ca²⁺ entry seems to be quite intricate and includes a series of receptor proteins with various functions [64]. Chemical lability of N1-linkage of cADPR to hydrolysis can introduce distortions into experimental data interpretation and make its chemical synthesis a complicated problem. A replacement of cADPR with its stable analogs can partially unravel mechanistic aspects of cADPR interaction with receptors, study cADPR/Ca²⁺ signaling pathways, and develop new medications. An increased basicity of 1N-substituted purine can even provide cADPR molecule to self-catalyze its own depletion [66]. Changing of purine ring properties can be applied to strengthen N1-linkage of cADPR analogs. Therefore, this can be achieved by the introduction of specific electronoacceptor substituents into the purine ring in the surroundings of 1N-gycosidic bond. An alternate synthetic protocol which can be considered is a replacement of easily eliminating ribose moiety with an alkyl analog to generate a more chemically stable 1N-alkyl bond. NAD⁺-cyclizing enzymes from molluscs were non-specific to chemically modified NAD⁺(NADP⁺)-analogs and utilized them to afford plenty of cADPR analogs (Figure 7). The nature of substituents both in purine or carbohydrate rings appeared to affect agonistic and antagonistic effects [56, 64].

8-Bromo-N1-cycloinosinediphosphoribose (8-Br-N1-cIDPR, 45) evoked Ca²⁺ signaling in the human T-lymphoma cell line Jurkat and in primary rat T-lymphocytes [64]. Ca²⁺ signaling induced by 8-Br-N1-cIDPR, consisted of Ca²⁺ release and Ca²⁺ entry, while 8-Br-cADPR (8-bromo-cADPR) manifested antagonistic effects on Ca²⁺ release. A replacement of N (northern)-ribose with an acyclic fragment or addition of phosphate groups at S (southern)-ribose fragment dramatically decreased the receptor stimulating activity of cADPR analogs [56, 64]. The use of 2'-deoxy-cADPR analogs (modification at S-ribose) strengthened their activity on sea urchin receptors but not in human T-cells [56]. In general, the structure-activity relationship of cADPR-induced Ca²⁺ release revealed by cADPR analogs is still evolving and presents a complex picture depending on the structure of compounds and the biological origin of receptors. Chemical synthesis of cADPR analogs is associated with lower regioselectivity compared to the enzymatic method due to the possibility of the formation of several conformers. These conformers can differently bind to receptors. It was shown by computational studies [65] the possibility of the two conformers for cADPR with various torsion angles χ O4'-C1'-N9-C4 in “southern” and “northern” ribose moiety (χ_S = 46.8°, χ_N = 9.4° for natural conformer, χ_S = –6.2°, χ_N = 57.6° for unnatural conformer, the two ones being syn/syn). Fortunately, the energetic barrier between the two conformers was large enough to presumably form natural conformation. L-Ribose, the mirror image of D-ribose, at the N1-position should exhibit the opposite lowest energy conformation to its D-ribose counterpart [65]. Therefore, according to NMR, only one conformer was formed during intramolecular cyclization of 1N-(5-O-phospho-D-ribofuranosyl)-AMP under Hata reaction conditions (nucleoside 5'-phosphate-5-thiophosphate-I₂/pyridine/molecular sieves) (Figure 7). Stereoisomerism of 1N-ribofuranose can determine specific rotation of molecular counterparts, which was demonstrated by the replacement of S-ribose with acyclic fragment, giving more conformational freedom to the obtained molecules [65]. The hydrogen bond between the N-ribose 2''-hydroxyl group and the hypoxanthine 6-oxo group managed conformer formation. By replacement of the N-D-ribose with L-ribose sugar, the cyclization of the linear precursor gave a cyclized product with opposite specific rotation and different biological activity [66]. “L-cIDPR” (50) was a less potent inhibitor of CD38-mediated cADPR hydrolysis than D-isomer and could be hydrolyzed by high concentrations of CD38. 8-Br-L-cIDPR was a potent inhibitor of CD38 acting in micromolar concentrations but was not utilized by the enzyme suggesting a non-competitive mechanism of inhibition. A range of disaccharide nucleosides and nucleotide regulators of immune response in plants have been synthesized at various times, with the most tedious one proposed for disaccharide 5',5''-dinucleotide (34). The multistep procedure consists of 15 steps of successive protection/deprotection steps with 5-O-phosphorylation near the end of the synthetic rote [67, 68]. A proposed protocol for the synthesis of 36 and 37 may include intermolecular phosphate condensation under Hata conditions. Structurally related compounds pRib-AMP/ADP (2'-(5''-phosphoribosyl)-5'-adenosine mono-/di-phosphate) (38–41), which trigger immune signaling in plants [67, 68] are metabolic precursors of various disaccharide nucleosides alternatively to NAD⁺ and may be formed from 2'/3'-cADPRs by depletion of the later. N1-Phosphoribosyl-AMP (34) (Figure 6), structurally similar to cADPR, is formed by an unusual reaction that joins ATP to PRPP to afford metabolic precursors in the biosynthetic pathway of histidine in plants [3]. Its structural similarity to cADPR, therefore, may indicate a relationship between various pathways of amino acid biosynthesis and cADPR utilization.

There are convincing medicinal proofs, that PAR is a participant in the pathogenesis of a range of diseases, i.e. cancer, diabetes, ischemia, and inflammations [69–75]. PAR-metabolizing enzymes are potential levers of PAR biosynthesis, with two major enzymes being PARP-1 and PARG. PARP-1 is the most studied enzyme among the PARP family. Being a key protein regulating base excision repair (BER) and DNA single-strand breaks (SSB) [39], PARP-1 can be considered as the prospective enzymatic target for the design of new anticancer and anti-inflammatory drugs [76–80]. It represents a six-domain construction: automodification domain (AMD) with 22 kilodaltons, three DNA-binding domains (Zn1, Zn2, and Zn3) with 42 kilodaltons total weight and catalytic subunit with a molecular mass of 54 kilodaltons (cat-PARP) [39, 80]. The use of alkylating drugs in the chemotherapy of tumors injures DNA both in healthy and cancer cells with further DNA repair in both cases. The use of PARP-1 inhibitors in combination with alkylating agents is expected to decrease the concentration of the latter and, hence, decrease the toxicity of chemotherapy. Breast cancer-associated genes BRCA1 and BRCA2 play an important role in the repair of double-strand breaks (DSB) and suppress tumor growth [11, 39, 69, 79]. The loss of one gene function (PARP-1 or BRCA1/2) is not a dramatic event, but the loss of both genes (PARP-1 and BRCA1/2) is lethal for a cell. In BRCA1/2 deficient cells, PARP-1 is upregulated and therefore in these cases, PARP-1 inhibitors can be used as single agents for the treatment of BRCA-deficient tumors [11, 39, 69, 79]. Among the PARP family PARP-2 comprises 60% similarity with PARP-1 in amino acid composition, being the most structurally related to PARP-1. Despite PARP-1 and PARP-2 differ by domain composition, relating to interactions with proteins, a structural similarity of catalytic centers can provide overlapping of their functions and their partial compensation in the case of arrest of one of them, which was evidenced by experimental procedures [81, 82]. For example, a simultaneous elimination of genes encoding PARP-1 and PARP-2 leads to the death of an organism in the embryonal stage [83]. The use of PARP inhibitors can strengthen the action of DNA-injuring drugs on cancerous cells and mitigate treatment toxicity due to the decrease in the concentration of chemotherapeutic compounds, used in combination with PARP-1/PARP-2 inhibitors [84]. Natural nucleosides and nucleotides manifest low PARP-1 inhibition activity, but disaccharide nucleosides seem to be an advantageous class of PARP-1 inhibitors due to their cell penetrability and low cytotoxicity [85]. A possible molecular mechanism of their action can be associated with mimicking of PAR disaccharide core structure. Therefore, a series of the structurally related 2'-O-α-D-arabinofuranosyl- and ribofuranosyladenosine analogs was obtained by various research groups as inhibitors of PARP-1 enzyme (Table 1) [85, 86].

Several PARP-1 inhibitors active in the micromolar range of concentrations were found among purine and pyrimidine disaccharide nucleosides. The hit-compounds were the most structurally related to the PAR monomeric unit and contained an additional methyl or isopropyl moiety at position N⁶ of purine. They manifested half maximal concentration (IC₅₀) values at 93–158 µM (see compounds 52, 54) [86].

The introduction of arabinofuranosyl moiety (see compounds 53, 55) drastically lowered PARP-1 inhibition [86]. The replacement of purine residue with pyrimidine one and modifications in the structure of disaccharide scaffold increased PARP-1 inhibition several times. Compounds bearing ribofuranose moiety at 2'-O-hydroxyl group manifested similar activity with purine derivatives, while 3'-O-β-D-ribofuranosyl deoxythymidine appeared to be several times more active [85]. The addition of phosphate groups reduced the inhibitory effect. The introduction of a bulky iodine substituent into position 5 of 3'-O-β-D-ribofuranosyl deoxynucleoside instead of methyl group had the most pronounced effect on PARP-1 inhibition. At the same time, PARP-1 appeared to be sensitive to N-glycoside bond configuration as α-thymidine derivative with drastically lowered activity. The most intriguing feature of pyrimidine disaccharide nucleoside PARP-1 inhibitors was their ability to accumulate in cell nuclei and inhibit PARP-1 inside them, as was shown by our experiments on SKOV-3 cancerous cell line [85]. Low toxicity of disaccharide compounds is probably explained by an absence of enzymes hydrolyzing O-glycosidic bonds inside cancer cells. It was also shown that the dialdehyde derivative of 3'-O-β-D-ribofuranosyldeoxythymidine manifested high PARP-1 inhibition potency but was rather toxic leading to cell agglutination (PARP-2 inhibition of the obtained compounds is now under study) [87].

A molecular mechanism of pyrimidine nucleoside action can be associated with hydrogen bond formation of pyrimidine carboxamide moiety with backbonds of serine (Ser-904) and glycine (Gly-863) in PARP-1 active site. This type of molecular interaction is similar to a variety of PARP-1 inhibitors used in medicinal practice and bearing the carboxamide pharmacophore group. The most important structural motifs in PARP-1 inhibitors are carboxamide fragments and aromatic moieties [77, 86–88]. Therefore, the structure of heterocyclic base is crucial for PARP-1 inhibition as a natural nucleoside thymidine (Thd) possesses PARP-1 inhibition activity. PARP-2 inhibition activity was recently described for some 5-substituted uracil analogs [84]. According to molecular docking data, their mechanism of action is highly possible due to hydrogen bond formation between pyrimidine carboxamide moiety and backbonds of serine (Ser-470) and glycine (Gly-429) in PARP-2 catalytic site.

It was previously shown in a series of investigations, that depending on its length and weight PAR can associate with various protein targets [15, 16, 28]. PARP-1 overactivation during exposure of cells to a toxic stimulus produces toxic levels of PAR [89]. Once being located in cytosol, PAR molecule triggers cell death by parthanatos mechanism (derived from the Greek Θάνατος, “death”). It is a programmed cell death, which is distinct from such processes as necrosis and apoptosis. Parthanatos is involved in the progression of a range of metabolic disorders such as Alzheimer’s/Parkinson’s disease, stroke, heart attack, and diabetes. As the name implies, PAR is the most electronegative natural polymer (in pyrophosphate pK₁ < 2.0, pK₂ = 2.64) [90] with two negative charges per monomeric unit. Other biopolymers, such as DNA and RNA, contain only one phosphate group per monomeric unit and have only one negative charge at neutral pH. PAR structure has much in common with natural polysaccharides heparin and heparan sulfate as the presence of sulfate and carboxyl groups defines the substantial negative charge of these biopolymers under neutral conditions. Heparin is the mostly charged polysaccharide with –2.7 on the disaccharide unit (–1.35 on the monosaccharide unit). Therefore, charged biopolymers may mimic ionic interactions with PAR-binding proteins, thus having the potency to affect biochemical processes associated with PAR and modulate some cellular functions (Figure 8).

Inhibition of PARG may be essential for the study of PAR dynamics and metabolism. The known PARG inhibitors can be both of monomeric or polymeric nature, such as ADPR-analogs ADP-(hydroxymethyl)pyrrolidinediol (ADP-HPD) and poly(etheno-ADPR) respectively [37, 91]. PARP-1 also manifests wide substrate specificity to the nicotinamide moiety of NAD⁺. Thus, modified NAD⁺ analogs may be utilized by the enzyme as efficiently as NAD⁺. Substitution of nicotinamide moiety with compatible aromatic or structurally related groups can be used for colorimetric detection of PARPs and other NAD⁺ consuming polymerases [92]. Chain terminators, which are NAD⁺ analogs lacking one or both hydroxyl groups at ribose of adenosine moiety in NAD⁺, are biochemically utilized by PARP-1 and provide an additional chemical tool for protein labeling and study of PAR’s mechanisms of action [93]. It was found, that 2-hydroxyl group of adenosine moiety is essential for PAR formation. Nevertheless, NAD⁺ derivatives without 2'- or 3'-hydroxyl group at adenosine moiety can be incorporated into PAR in the presence of NAD⁺ [93]. Therefore, chemically modified NAD⁺ can be used as substrates for enzymatic synthesis of chemically modified biopolymers, mimicking PAR functions.

All the described data suggest that ADPR and NAD⁺ may be a part of a delicate biochemical mechanism regulating transcription and translation in cells. ADP-ribosylation in host-virus conflicts triggers various routes of cellular resistance to infection. PARP-10-mediated modification of RNA ends could have an immune function as its expression inhibits Venezuelan equine encephalitis virus (VEEV) translation and participates in modulating inflammation. It was shown that VEEV and SARS-CoV macrodomains encoded within the non-structural proteins can remove ADP-ribosylation from PARP-10-modified RNA, counteracting PARP-10 activity [94]. It is hypothesized that the PARP-10 RNA recognition motif (RRM) specifically binds and ADP-ribosylates viral RNA, distinguishing it from a host RNA. It was shown in recent studies that SARS-CoV Mac1 reverses the PARP-14 (ART14)-catalyzed transfer of ADPR on signaling proteins thus abolishing immune response. Mac1 is a domain of a larger viral protein known as nonstructural protein (NSP3) and belongs to the family of macrodomains, involved in the post-translational modification (PTM) ADPR. Thus, ADPR represents the simplest disaccharide form of PAR and a platform for the design and synthesis of antivirals. Highly efficient disaccharide inhibitors of the coronaviral Mac1 domain have been synthesized for the first time by Dmitri V. Filippov research group using the phosphate-phosphoramidite coupling strategy (Figure 9) [95].

Using this synthetic approach, the nanomolar-range efficient compounds were obtained by combination of two pharmacophore groups based on remdesivir 5'-monophosphate (see compound 66) and modified carbohydrate 5'-phosphate. The proposed combinatorial approach allows the synthesis of libraries of novel compounds and therefore may have applications in antiviral therapy due to its potency to overcome drug resistance associated with mutations in the viral genome. Alternate ADPR synthetic modifications with propargyl group can be further functionalized by biotin and fluorescent dye and possibly applied in isolation and detection of viral and other ADPR-interacting proteins [51].