Pyrrolo[2,3‐d]pyrimidine (7‐deazapurine) as a privileged scaffold in design of antitumor and antiviral nucleosides

Abstract 7‐Deazapurine (pyrrolo[2,3‐d]pyrimidine) nucleosides are important analogues of biogenic purine nucleosides with diverse biological activities. Replacement of the N7 atom with a carbon atom makes the five‐membered ring more electron rich and brings a possibility of attaching additional substituents at the C7 position. This often leads to derivatives with increased base‐pairing in DNA or RNA or better binding to enzymes. Several types of 7‐deazapurine nucleosides with potent cytostatic or cytotoxic effects have been identified. The most promising are 7‐hetaryl‐7‐deazaadenosines, which are activated in cancer cells by phosphorylation and get incorporated both to RNA (causing inhibition of proteosynthesis) and to DNA (causing DNA damage). Mechanism of action of other types of cytostatic nucleosides, 6‐hetaryl‐7‐deazapurine and thieno‐fused deazapurine ribonucleosides, is not yet known. Many 7‐deazaadenosine derivatives are potent inhibitors of adenosine kinases. Many types of sugar‐modified derivatives of 7‐deazapurine nucleosides are also strong antivirals. Most important are 2′‐C‐methylribo‐ or 2′‐C‐methyl‐2′‐fluororibonucleosides with anti‐HCV activities (several compounds underwent clinical trials). Some underexplored areas of potential interest are also outlined.

Some 7-deazapurine nucleosides even occur in nature-both as nucleosides and as components of nucleic acids.
Queuosine (1a) 1-3 and archaeosine (2) 4 are examples of 7-deazapurine nucleosides isolated from tRNA of prokaryotic and eukaryotic organisms and archaea, respectively. Recently, the presence of 2 ′ -deoxyqueuosine (1b) was also discovered in bacterial DNA 5 (Fig. 1). Several nucleoside antibiotics contain pyrrolo [2,3-d]pyrimidine nucleobase and their biological activities will be mentioned later in the text. The biosynthetic pathway to 7-deazapurines uses GTP as a precursor. More details about 7-deazapurine biosynthesis can be found in a detailed review. 6 Replacement of the N7 nitrogen by carbon changes the electronic properties of the five-membered ring, making it more electron-rich and thus (at least in theory) more prone to cation-or -interactions. Also, the presence of this additional C7 carbon gives a suitable position for attachment of substituents which (in DNA or RNA) point out to the major groove. It has been repeatedly shown that 7-substituted 7-deazapurine nucleoside triphosphates are good substrates for DNA or RNA polymerases and DNA containing 7-substituted 7-deazapurines forms stable duplexes (some 7-alkynyl-or 7-aryl-7-deazapurines nucleotides are even better substrates for polymerases than natural dATP or dGTP 7,8 and in DNA they stabilize duplexes 9 ). For applications in medicinal chemistry, the position C7 offers a possibility for functionalization and many 7-substituted 7-deazapurine nucleosides display important biological activities which are summarized and discussed in this review. It should be noted that a general review on syntheses and biological activities of pyrrolopyrimidines was published recently, 10 but it did not specifically cover nucleosides and completely missed many important classes of relevant compounds. The synthetic approaches to 7-deazapurine nucleosides are covered in the above-mentioned review 10 or in a recent book. 11 This review focuses on recent (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016) advances in medicinal chemistry of 7-deazapurine nucleosides and is based on SciFinder, Reaxys, and Web of Knowledge searches.

Natural compounds
Cytotoxic activities of naturally occurring 7-deazapurine nucleosides were discovered already in 1960s. Tubercidin (3), toyocamycin (4), and sangivamycin (5) were isolated from Streptomyces cultures. All of them show potent cytotoxicity against cancer cell lines [12][13][14][15] but despite their structural similarities, their modes of action are different (IC 50 values for compounds mentioned in this section are listed in Table 1). Tubercidin (3), toyocamycin (4), and sangivamycin F I G U R E 2 Structure-activity relationship among cytotoxic 7-deazapurine ribonucleosides (5) are phosphorylated by cellular kinases to their mono-, di-, and triphosphate forms and the resulting nucleotides are then incorporated into RNA and DNA which causes damage to nucleic acid functions. [16][17][18][19][20] Furthermore, tubercidin (3) impairs numerous cellular processes such as pre-mRNA processing and nuclear speckle formation, 21 mitochondrial respiration, purine synthesis, rRNA processing, and methylation of tRNA. 22 Tubercidin (3) is also a potent inhibitor of S-adenosylhomocysteine hydrolase. 23 Toyocamycin (4) showed inhibition of phosphatidylinositol kinase 24 , inhibitory effect on rRNA synthesis, and maturation. 25,26 In a more recent study, toyocamycin (4) was identified as a potent inhibitor of induced XBP1 (X-box binding protein 1) mRNA cleavage. IRE1 -XBP1 pathway is a key component of endoplasmic reticulum stress response. IRE1 -XBP1 inhibition by toyocamycin (4) leads to apoptosis in endoplasmic reticulum-stressed tumors and multiple myeloma. 27 On the other hand, cytotoxic effect of sangivamycin (5) is mainly caused through potent and selective inhibition of protein kinase C. 28 Recently, the mechanism of action of sangivamycin (5) in primary effusion lymphoma cells was studied showing that sangivamycin (5) acts through inhibition of Erk and Akt signaling in these cells. 15 Sangivamycin (5) is also able to bind heat-shock protein 70 (HSP70) (K D = 3.3 M) which is a molecular chaperone with a proposed importance in oncology. 29 Despite the attention paid to the naturally occurring 7-deazapurine nucleosides, none of the compounds proceeded to clinical use.
7-Iodotubercidin (6a) acts mainly as a potent inhibitor of adenosine kinase. 37 For example, it showed promising submicromolar cytotoxicity in canine osteosarcoma cell lines which show dysregulation of kinase activity. 38 7-Iodotubercidin (6a) is also a p53 activator, and generates DNA damage and G2 cell cycle arrest through its incorporation into DNA. 39 On the contrary, the cytotoxic effect of 7-bromotubercidin (6b) is caused by incorporation of its triphosphate form into cellular RNA and inhibition of RNA synthesis. 22 In our laboratory, we developed a new group of potent nucleoside cytostatics by the attachment of heterocycles to C-7 position of 7-deazaadenosine 12 (Fig. 3). It was shown that derivatives substituted with five-membered  40 Derivatives with five-membered rings showed mild antileukemic activity and sixmembered ring derivatives were inactive, however IC 50 values are not given in this study. Electronic properties of the substituents are also important for the activity. The presence of an electron-donating p-methoxy group on a phenyl substituent made compound 7d active in submicromolar concentrations. Direct connection of thiophene and 7-deazapurine moieties in 7a (known as AB61) is also crucial for the cytotoxic activity, as its analogue 8 with the thienyl group attached via a sulfur atom showed a substantial drop of activity (IC 50 > 20 M). 41 The mechanism of action of AB61 (7a) was studied in detail 42 (Fig. 3). AB61 (7a) is efficiently phosphorylated in cancer cell lines but not in fibroblasts which is the reason it shows selectivity toward cancer cells. The resulting ribonucleoside triphosphate (NTP) is incorporated both into cellular RNA and DNA where it causes block of translation and DNA damage, respectively. Interestingly, although tubercidin triphosphate was incorporated into RNA under the same experimental conditions, subsequent RNA translation proceeded smoothly. AB61 trisphosphate is a substrate for mitochondrial DNA polymerase and therefore AB61 (7a) might interfere with mtDNA replication and mitochondrial functions similarly as described for other nucleoside analogs. 43 The in vivo studies of xenograft models confirmed the promising properties of AB61 (7a) for its further development as an anticancer agent.
Structure-based drug design was used in development of inhibitors of DOT1L methyltransferase. This enzyme is a protein methyltransferase that methylates histone H3 on lysine 79 (H3K79). In so-called sangivamycin-like molecules (SLMs), the carbamoyl group of sangivamycin (5) is replaced by other carboxylic-group-related substituents and/or additional substituents are attached to position 8 of the 7-deazapurine moiety ( Fig. 6). SLMs, including SLM6 (20) and ARC (21a), showed potent in vitro cytotoxicity against multiple myeloma cell lines. 53 The efficacy of SLM6 (20) was also confirmed in vivo. Its cytotoxic effect is caused by inhibition of cyclin dependent kinase 9 (CDK9). ARC (21a), 8-hydrazinosangivamycin, attracted attention for its promising cytotoxic activities against colorectal cancer, 54 melanoma, 55 and neuroblastoma cells. 56 Synergistic effect was observed in treatment of cancer cell lines with ARC (21a) and a pan-Bcl-2 inhibitor ABT-737 (for structure, see 57 ). 58 ARC (21a) is a general transcriptional inhibitor acting through inhibition of positive transcription elongation factor b (P-TEFb), a complex of CDK9/cyclin T1. 59 ARC (21a) induces p53-independent apoptosis in malignant cells but not in normal cells. It inhibits Akt signaling pathway 56 and decreases expression of antiapoptotic proteins such as Mcl-1. 54,55 Nevertheless, ARC (21a) was found inactive in xenograft models in vivo, presumably due to rapid serum clearance. 60 It was shown that the activity and mechanistic aspects of ARC (21a) are identical to those of sangivamycin (5) so that the lack of efficacy of ARC (21a) is in accord with previous failures of sangivamycin (5) in clinical trials. 60 Compound 21b, an 8-substituted toyocamycin analogue, was found to bind HSP70 more efficiently than toyocamycin (4) and sangivamycin (5) (K D = 2.8 M). As inhibition of HSP70 in cancer cell lines induces apoptosis, 61 derivative 21b has a potential to possess cytotoxic activity but the data are not yet reported. Xylocydine (22a), also called BMK-Y101, a sugar-modified analogue of 8-bromosangivamycin, is a selective inhibitor of cyclin dependent kinases (CDK1, CDK2, CDK7, and CDK9) 62,63 which demonstrated its antitumor potential in hepatocellular carcinoma both in vitro and in vivo, while being inactive against cervical, prostate, and hepatic carcinoma or lung adenocarcinoma. 64 Further studies with leukemic HL-60 cell line showed that xylocydine (22a) induces apoptosis in these cells by CDK1 and CDK4 inhibition and upregulation of protein p16 INK4a , which is a CDK inhibitor. 65 Isobutyryl ester prodrug of xylocydine, ibulocydine (22b), also causes inhibition of CDK7 and CDK9. Subsequent inhibition of RNA polymerase II phosphorylation leads to rapid down-regulation anti-apoptotic of proteins Mcl-1, survivin, and XIAP, thus inducing apoptosis in hepatocellular carcinoma cell lines. 66 Hepatocellular carcinoma cell lines can be sensitized to TRAIL (tumor necrosis factor-related apoptosis-induced ligand)-induced apoptosis by treatment with ibulocydine (22b) 67 that is also F I G U R E 7 Structures of cytotoxic 6-(het)aryl-7-deazapurine ribonucleosides capable of inhibiting growth of hepatocellular carcinoma in a mouse xenograft model. 66 Combination of radiotherapy and ibulocydine (22b) treatment showed promising results both in vitro (apoptotic cell death accompanied with activation of caspases, decrease in Bcl-2/Bax expression, loss of mitochondrial membrane potential, and release of cytochrome c into cytosol) and in vivo (reduced tumor volume in lung cancer xenografts in mice). 68 Replacement of an 8-bromine atom of xylocydine (22a) by p-tolyl, m-or p-methoxyphenyl, and m-or p-bromophenyl groups in compounds 23a led to loss of CDK1 and CDK2 inhibitory activity 69 and also JRS-15 (23b) with 8-biphenylyl group is devoid of any CDK inhibiton. 64 However, JRS-15 (23b) showed cytotoxicity in broader panel of cancer cell lines compared to xylocydine (22a), even though these are only moderate (micromolar). 64 Carbocyclic derivative 24 that shares 7-deazapurine substitution pattern of xylocydine (22a) possesses mild cytotoxic activity against ovarian cancer cell line PA-1. 70 It is also important to mention that 8-substituted 7-deazapurine are likely to adopt syn-conformation 69 and binding modes to their molecular target may substantially differ from those of 8-unsubstituted 7-deazapurine nucleosides where anticonformation is preferred. In conclusion, nucleobase modifications of 7-deazapurine moiety in positions 7 and 8 cannot only dramatically change the cytotoxic activities but it also affects mechanisms of action of particular nucleosides.
Mechanisms of action of cytotoxic 7-deazapurine nucleosides are typically quite complex and often include activation in cells to obtain corresponding nucleotides that can interfere e.g. with DNA and RNA synthesis. It is common that one compound targets more pathways in parallel, which makes it difficult to decide which of them is the most significant.
Some 7-deazapurine nucleosides target specific enzymes such as adenosine kinase, CDKs or protein kinase A and C.
It is difficult to predict the mechanism of action of novel analogues of 7-and 8-substituted 7-deazapurine nucleosides because many compounds from this group significantly differ in their modes of action although their structures are very similar. Even though target-based design was successfully applied in the development of DOT1L inhibitors, most of discoveries in development of cytotoxic 7-deazapurine nucleosides are based on serendipity. More focused studies of binding of cytotoxic nucleotides to cellular DNA and RNA polymerases are required in the future in order to move structure-activity studies into more target-related research.

Synthetic nucleosides with C6 substituents
Substitution of 6-amino group of 7-deazapurine nucleosides by carbon substituents brings further possibilities for structure-activity relationship studies among 7-deazapurine nucleosides. As the substitution at position 6 of 7deazapurine nucleosides affects, or in some cases even prevents, formation of stable Watson-Crick base pairs, their mechanisms of action may differ from those of 7-deazaadenine nucleosides. In our lab, we prepared a large series of 6-(het)aryl-7-deazapurine ribonucleosides 25 and found that they show different cytotoxic activities based on the nature of the substituent (Fig. 7). Derivatives substituted with a small five-membered heterocycles, e.g. furan (25a) or thiophene (25b), are nanomolar cytostatics against a broad range of cancer cell lines, while phenyl derivative 25c is mostly inactive. 71 In this case, substitution of the phenyl ring did not bring substantial improvement of activity. Further substitution by fluorine in position 7 in derivatives 26a-b did not affect the efficacy of the parent compounds, on the other hand 7-chloroderivatives 27a-b were significantly less active or inactive. The decrease of cytotoxic potency in 7chloroderivatives is in agreement with the fact that intracellular phosphorylation of 27a is much less efficient than that F I G U R E 8 Phosphate-prodrugs of 6-(het)aryl-7-deazapurine ribonucleosides F I G U R E 9 2-Modified and sugar-modified derivatives of 6-hetaryl-7-deazapurine nucleosides of unsubstituted derivative 25a, indicating intracellular phosphorylation may be limiting the activity. The mechanism of action of 6-hetaryl-7-deazapurine ribonucleosides has not been fully examined yet, however, rapid and powerful cellular RNA synthesis inhibition was observed after treatment with 25a-b and 26a.
Since the efficient phosphorylation seemed to be crucial for cytotoxic activities of 6-hetaryl-7-deazapurine ribonucleosides 25 and 26, phosphate prodrugs of these compounds were synthesized in efforts to further improve their cytotoxicity (Fig. 8). CycloSal pronucleotides of 6-hetaryl-7-deazapurine (28) and 6-hetaryl-7-fluoro-7-deazapurine (29) nucleosides however showed similar or slightly lower cytotoxic activities compared to parent nucleosides in most cancer cell lines. 72 Again, derivatives with small furyl and thienyl substituents 28a-b and 29a-b were more potent than derivatives with bulky substituents such as benzofuryl 28c which were devoid of any cytotoxic effect. Pro-Tide approach was also applied and phosphoramidate prodrugs based on 6-hetaryl-7-deazapurine nucleosides were prepared. 73 Nevertheless, nanomolar cytotoxic activities of the parent nucleosides dropped to micromolar in their ProTides, no matter if methyl (30a, 31a), ethyl (30b, 31b), or benzyl (30c, 31c) alanine esters were used. The decrease of activities was presumably caused by efflux of ProTides out of the cells.
Also, the ribonucleoside moiety was shown to be crucial for keeping the potency of 6-hetaryl-7-deazapurine nucleosides as all of the sugar-modified derivatives, such as 33a-33e and 34, were devoid of any cytotoxic activity. 49,75,76 ( Fig. 9). Structure-activity studies of 6-hetaryl-7-deazapurine nucleosides mentioned in this section and those of Replacement of the amino group of tubercidin (3) by a methyl group led to 6-methyl-7-deazapurine ribonucleoside (35a) (Fig. 10). Compound 35a is a potent cytotoxic agent with low submicromolar activities against a broad range of cancer cell lines. 44 Unfortunately, 6-methyl-7-deazapurine ribonucleoside (35a) is poorly selective and showed similar cytotoxic effect also against normal fibroblasts. Substitution of compound 35a in position 2 is not tolerated as the resulting 2-fluoro (35b), 2-chloro-(35c), 2-amino (35d), and 2-methyl (35e) derivatives are completely inactive. 74 6-Trifluoromethyl-7-deazapurine ribonucleoside (36a) possesses similar cytotoxic activities against cancer cell lines as the 6-methyl analogue 35a. 77 Also in this case no 2-modified derivatives 36b-h showed any cytotoxicity. 77 Another option for modification in position 6 is N6-alkylation of 7-deazaadenine (Fig. 10). N6-benzyl and N6nitrobenzyl derivatives of tubercidin (3), toyocamycin (4), and sangivamycin (5) were studied as inhibitors of nucleoside transporter 1 (hENT1) which is a protein that plays key role in nucleoside drug uptake. Among the compounds synthesized in the study, N6-nitrobenzyl derivative of sangivamycin 37 showed both efficient (nanomolar) inhibition of the hENT1 transporter and mild (micromolar) cytotoxicity against several cancer cell lines. 78 Conjugation of tubercidin (3) and cisplatin, a clinically used cytostatic, through N6-nitrogen of tubercidin furnished derivative 38. Although compound 38 was able to react with purine residues in a similar way as cisplatin, its cytotoxic activity against cancer cell lines was only weak (micromolar) and dropped by one order of magnitude compared to cisplatin. 79

Fused nucleosides
7-Deazapurine nucleobase can be further extended by annulation with other aromatic rings leading to fused (tricyclic) nucleosides (Fig. 11). Triciribine (40a), also known as TCN or API-2, was synthesized already in 1971. 85 Triciribine (40a) suffers from poor solubility and therefore its soluble 5 ′ -monophosphate prodrug 40b (TCN-P) is often used. Antineoplastic activities of triciribine (40a) and TCN-P (40b) both in vitro and in vivo were studied in 1980s and early 1990s and were reviewed previously. 86 Clinical trials were halted due to toxicity in high dosing of TCN-P (40b) and mild efficacy presumably due to its low bioavailability. However, recently it was shown that the bioavailability can be significantly improved by application of phosphoramidate prodrug approach. 87 Triciribine (40a) acts through selective inhibition of Akt kinase 88  Most of the benzo-fused 7-deazapurine nucleosides prepared so far lack any cytotoxic activity against cancer cell lines (Fig. 12). Derivatives 42 and 43a-b are the only examples of weakly active (micromolar) compounds. 93,94 On the other hand, thieno-fused nucleosides, which are isosteric to above-mentioned benzo-fused nucleosides, showed interesting cytotoxic effects 95 (Fig. 12). Substituents in position 6 of 7-deazapurine part of the nucleobase play an important role in structure-activity studies. While dimethylamino derivatives (44a, 45a) were inactive and amino derivatives (44b, 45b) possess only moderate (micromolar) cytotoxic activities, methyl (44c, 45c), methoxy (44d, 45d), and  95 In conclusion, even though the family of fused nucleosides is relatively small, it has already brought interesting compound hits and deserves further development as a source of cytotoxic nucleosides. Detailed study of mechanism of action of these compounds is particularly needed and is under way.

Inhibitors of mammalian adenosine kinases
Adenosine kinase (ADK) is an enzyme that catalyzes the conversion of adenosine to adenosine-5 ′ -O-monophosphate (AMP). Extracellular adenosine is a ligand of adenosine receptors which regulates heart rate, neurotransmitter release F I G U R E 1 3 Examples of 7-deazapurine nucleosides as inhibitors of mammalian ADKs in brain, and inflammatory response. Stimulation of adenosine receptors is connected with anticonvulsant, analgesic, and anti-inflammatory activity. Inhibition of ADK leads to increased adenosine levels and so ADK inhibitors possess similar effect as adenosine receptor agonists. Moreover it was shown that ADK inhibitors show less side effects in antiseizure activity assays and therefore can be considered as a promising class of anticonvulsants. 96 Data from Section 3 is summarized in Table 2. 5 ′ -Deoxy-5-iodotubercidin (46a) is a naturally occurring nucleoside isolated from a marine red alga (Fig. 13). 5 ′ -Deoxy-5-iodotubercidin (46a) is a potent inhibitor of mammalian adenosine kinases 37,97 (IC 50 = 9 nM against human ADK) and does not bind to adenosine receptors. 98 Also other 7-halogenated 7-deazaadenine nucleosides show ADK inhibitory activity. Although 5-iodotubercidin (6a) 37,97 and 5-bromotubercidin (6b) 22,97 are less potent ADK inhibitors than 5 ′ -deoxy-5-iodotubercidin (46a), they still possess submicromolar IC 50 values. Both 5 ′deoxy-5-iodotubercidin (46a) and 5-iodotubercidin (6a) showed anti-seizure activities in vivo, 96,97 however, these compounds are not suitable for clinical use due to behavioral effects (decreased locomotor activity, hypothermia, and muscle flaccidity) 98 and cytotoxicity, 39 respectively. In order to obtain more suitable clinical candidates, a large structure-activity study was performed. 97 It was shown that the presence of a halogen atom in position 7 is crucial for the ADK inhibitory activity. The most potent activities were observed in derivatives with amino, chloro, or methylsulfanyl group in position 6 and 5 ′ -amino group. For instance, derivative 46b showed subnanomolar human ADK inhibition (IC 50 = 0.6 nM). Compound 46b showed anti-seizure activity in an in vivo model. However, it seems that anti-seizure activities generally do not fully correlate with ADK inhibitory effects, presumably due to different pharmacological properties of the tested compouds. 97 ADK inhibitors are also promising analgetics. Carbocyclic 7-bromo-7-deazapurine nucleoside 47 is a subnanomolar inhibitor of ADK (IC 50 = 0.47 nM against human ADK) which showed antinociceptive activity in animal acute, inflammatory, and neuropathic pain models (Fig. 13). 99 Replacement of iodine atom in 5-iodotubercidin (6a) by ethynyl group led to drop of ADK inhibitory activity, however, resulting 7-ethynyl-7-deazaadenosine (11a) still showed submicromolar ADK inhibition. 100 On the other hand, tubercidin derivatives with larger hydrophobic groups in position 7, such as 7-phenyl-7-deazaadenosine (7c), do not inhibit human ADK. 100 Despite this fact, when 7-phenyl group is combined with phenyl group attached to N6 nitrogen atom, the potent ADK inhibition in these so called diaryl derivatives is restored 101 (Fig. 13). Further sugar in rat formalin paw pain model was similar to that of standard opioid analgesic morphine. 103 Recently, submicromolar ADK inhibition was also observed in carbocyclic diaryl nucleosides such as compound 50 (IC 50 = 0.088 M). 104 However, in vivo activities of this class of compounds have not been studied yet. Despite potent activities of diaryl nucleosides, the compounds generally suffer from poor water solubility. Replacement of N6-phenyl group by glycinamide substituent led to compound 51 which showed improved solubility and nanomolar ADK inhibitory effect at the same time (IC 50 = 3 nM). 105 In vivo animal pain models provided promising results as derivative 51 was shown to be more potent than morphine. 105 Unfortunately, prolonged administration of compound 51 is connected with lethal toxicity and therefore further development of this compound was discontinued. In spite of this fact human ADK inhibitors still represent a very promising group of compounds and deserve attention for their antinociceptive, anti-seizure and antiinflammatory activities.

Inhibitors of mycobacterial adenosine kinase
Many pathogens lack enzymes for de novo purine synthesis and use purine salvage pathway instead. In this pathway purine nucleotides are formed from nucleobases and phosphoribosyl pyrophosphate or purine nucleosides are phosphorylated by kinases, e.g. adenosine kinase. Targeting ADK of the pathogen can therefore lead to anti-microbial compounds. Mycobacterium tuberculosis expresses enzymes from both salvage pathway and de novo purine synthesis pathway; however, the interdependence and regulation of these processes remain unclear. 106 (Fig. 14). Also 6-hetaryl-7-deazapurine ribonucleosides were shown to selectively inhibit with Mtb-ADK in submicromolar concentrations. 107 Derivatives with small hetaryl groups such as 53a and 53b  107 In this case, the mode of antimycobacterial activity is presumably independent of Mtb-ADK and rather suggests a more general cytotoxic mechanism as compound 35a is highly cytotoxic also toward human cells. This is in agreement with the fact that its 2-substituted analogues 35b-e showed neither antimycobacterial activity nor cytotoxicity. 74 In conclusion, the most potent 7-deazapurine nucleoside inhibitors of Mtb-ADK are only weakly active in antimycobacterial assays. Poor cellular uptake could be one of the reasons for this fact, however, initial attempts to improve the cell penetration by lipophilic prodrugs failed. 107 Therefore, showing that standard prodrug approaches for penetration for eukaryotic cells may not be effective in mycobacteria and requires further studies.

Ribonucleosides
Hepatitis C virus (HCV) is an RNA virus of the family Flaviviridae. It uses RNA-dependent RNA polymerase (NS5B) for replication of its genetic information. Because human cells do not express any RNA-dependent RNA polymerase and HCV RNA polymerase (NS5B) is structurally different from human RNA and DNA polymerases, the viral enzyme became a target for new anti-HCV compounds development. Thanks to their close resemblance to natural substrates of the viral RNA polymerases, many 7-deazapurine nucleosides were studied as potential antiviral candidates not only for HCV but also other RNA viruses that will be mentioned later in the text. Structure-activity studies of nucleobasemodified analogues of naturally occurring 7-deazapurine antibiotics such as tubercidin (3) and toyocamycin (4) brought several interesting structures with anti-HCV activities (Fig. 15). Modifications of toyocamycin (4) in positions 6 and 7 led to nucleosides 56 and 57 that showed submicromolar activities in HCV replicon assays. EC 50 and CC 50 values for compounds mentioned in Section 4.1 are listed in Table 3. Compounds 56 and 57 are non-cytotoxic and therefore they both represent good lead structures for anti-HCV drug development. 108 Also other 6-and 7-modified-7-deazapurine ribonucleosides showed interesting in vitro anti-HCV activities. 44 7-Furyl derivatives 9a and 9c showed submicromolar anti-HCV activities in both HCV 2A and HCV 1B replicon assays. Both 9a and 9c were non-cytotoxic to the HCV replicon cells but 9a showed submicromolar cytotoxicity towards normal fibroblasts and therefore it cannot be considered as suitable for further development as an anti-HCV agent. Also 7-ethynyl derivative 11c possessed submicromolar anti-HCV activity and no cytotoxicity in HCV replicon assay but was cytotoxic to fibroblasts. 44 On the other TA B L E 3 Anti-HCV activities of 7-deazapurine nucleosides

Compound EC 50 [ M] (replicon) a CC 50 [ M] (replicon)
Ref.  Examples of sugar-modified anti-HCV 7-deazapurine nucleosides hand, 6-N,N-dimethylamino-7-deazapurine ribonucleoside (58) seems to be more suitable lead structure because it showed submicromolar activities and no cytotoxicity in HCV replicon assay and only weak cytotoxicity towards fibroblasts. Sangivamycin-like molecule ARC (21a) that was developed as an anti-cancer agent showed in vitro anti-HCV activity while being non-toxic to host cells. 109 It was found out that anti-HCV activity of ARC (21a) is not caused by transcription or translation inhibition but the molecular target of ARC (21a) remains unknown. Also another antineoplastic compound, triciribine (40a), showed moderate anti-HCV activity (EC 50 = 2 M) and low cytotoxicity. 110 Triciribine analogue 59 was more potent (EC 50 = 1 M) and remained non-cytotoxic. 110 Janus-type nucleosides 41ab showed moderate anti-HCV activities (EC 50 = 5.7 and 3 M, respectively) which was accompanied by cytotoxicity toward Vero, CEM, and PBM cell lines. 91 Anti-HCV screening of thieno-fused 7-deazapurine ribonucleosides showed that derivatives 44c,e and 45c-e possess submicromolar activities and do not show any cytotoxic effect against the replicon cells. 95 The anti-HCV potency of these compounds is similar to that of anti-HCV agent mericitabine. However, thieno-fused 7-deazapurine nucleosides are mostly cytotoxic to fibroblasts and only derivative 45c was devoid of this cytotoxic effect. In conclusion, despite interesting anti-HCV activities of some 7-deazapurine ribonucleosides, the anti-HCV effect is often accompanied by cytotoxicity to normal cells and therefore further modifications of the lead structures are necessary in order to obtain suitable anti-HCV drug candidates.

Sugar-modified nucleosides
Decreased cytotoxic activities of sugar-modified 7-deazapurine nucleosides compared to corresponding ribonucleosides have already been mentioned. Because anti-HCV activities of 7-deazapurine ribonucleosides are often accompanied by cytotoxicity, sugar modifications seem to be suitable for restriction of the cytotoxic effect. Among the sugar-modified nucleosides (such as arabinosides, 3 ′ -deoxyribonucleosides, and 2 ′ -O-methylribonucleosides) 2 ′ -Cmethylribonucleosides (more precisely their NTPs) are usually the best HCV RNA polymerase inhibitors 111 (Fig. 16).
7-Deaza-2 ′ -C-methyladenosine (60a), 112 also known as MK-0608, was thoroughly studied as an anti-HCV drug candidate. 111 It showed submicromolar activity against HCV in a replicon assay and is non-cytotoxic. MK-0608 (60a) is efficiently phosphorylated in cells and the corresponding NTP is a potent inhibitor of HCV NS5B RNA polymerase.
In fact, the NTP is incorporated into the growing RNA chain and acts as a chain terminator. 111 A single mutation in the NS5B RNA polymerase, S282T, leads to a resistance to MK-0608 (60a). 111 The anti-HCV activity of MK-0608 (60a) was confirmed in HCV-infected chimpanzees where significant reduction of viral load was observed but viral loads rebounded after dosing ended in all tested animals. 113 Combinational therapy with MK-0608 (60a) and an HCV NS3/4A protease inhibitor vaniprevir (MK-7009), however, resulted in sustained virological response of viral negativity 6 months after treatment in chimpanzees. 114 Combinational treatment with MK-0608 (60a) was also successful in HCV-infected human hepatocyte chimeric mice. The combination of MK-0608 (60a) and another HCV NS3/4A protease inhibitor telaprevir led to sustained virological response but only in high dose regimen or in triple combination F I G U R E 1 7 Examples of anti-HCV derivatives of 7-(hetaryl)-7-deaza-2 ′ -C-methyladenosines and 7deazaneplanocin A therapy with MK-0608 (60a), telaprevir, and interferon. 115 In spite of these promising results, clinical trials with MK-0608 (60a) have been halted for undisclosed reasons.
Despite the effort made in development of new HCV RNA polymerase inhibitors as anti-HCV agents, none of the 7-deazapurine nucleosides proceeded to clinical use. One of the reasons for this fact could be toxicities caused by off target effects as many of other nucleoside NS5B polymerase inhibitors showed to be substrates of human RNA polymerase II and human mitochondrial RNA polymerase. 127 Finding proper balance between NS5B polymerase inhibition and selectivity, efficient cellular phosphorylation and cell penetration as well as cytotoxicity remains difficult. Still many of the HCV RNA polymerase inhibitors could be useful in development of compounds active against related RNA viruses from Flaviviridae family thanks to structural similarities among viral RNA polymerases. Efforts made in this field will be mentioned in the following sections. Data from Section 4.2 is summarized in Table 4. 6-Methyl-7-deazapurine ribonucleoside (35a) showed potent antidengue activity both in a replicon assay (EC 50 = 0.877 M) and in an infectivity assay. 128 Unfortunately, cytotoxicities against host cells were not reported in this study, however, 6-methyl-7-deazapurine ribonucleoside (35a) was shown to be highly cytotoxic towards human fibroblasts 107 and therefore anti-dengue activity caused by unspecific cytotoxicity of the compound 35a cannot be ruled out. On the other hand, 7-deaza-2 ′ -C-methyladenosine (60a) was able to reduce viral loads in a mouse dengue fever viremia model but its mechanism of action against DENV has not been studied. 129 Replacement of the 2 ′ -C-methyl group by ethynyl group led to derivative NITD008 (72a). NITD008 (72a) showed submicromolar activity against DENV in vitro (EC 50 = 0.64 M) and was devoid of cytotoxicity against host cells (Fig. 18). 126 Further studies showed that NITD008 (72a) is phosphorylated in vivo to its NTP 130 which then acts as a chain inhibitor of RNA-dependent RNA polymerase of DENV. 126,131 NITD008 (72a) is orally available, reduced viral loads in a mouse model and protected infected animals from death but showed toxicity when treatment was longer than one week. 126 Compound 72b, a 7-fluoro analogue of NITD008, showed slightly increased antidengue activity in vitro compared to NITD008 (72a) (EC 50 = 0.42 M) but also increased cytotoxicity. 130 On the other hand, compounds with CN (72c) and CONH 2 (72d) substitutions in position 7 showed lower anti-dengue activities (EC 50 = 3.1 and 2.0 M, respectively). 130 Compound 72d suffers from poor oral bioavailability in mice and rats. 132 In order to improve its gastrointestinal absorption and cell penetration, its prodrug, compound 73, was prepared.
Derivative 73 showed both better anti-dengue activity in vitro (EC 50 = 0.54 -0.71 M, depending on DENV strain) and improved bioavailability than the parent nucleoside 72d. 132 Compound 73 efficiently inhibits viral RNA synthesis and was found to be more potent than NITD008 (72a) in an in vivo mouse model. Nevertheless, in vivo toxicity in higher doses precluded further development of compound 73 as an anti-dengue agent. Some benzo-fused 7deazapurine ribonucleosides also showed submicromolar activities against DENV but in case of compounds 42 94

Nucleosides with antiviral activities against other viruses
7-Deazapurine nucleosides were subjected to antiviral activity screening against other viruses than HCV and DENV.
The largest structure-activity relationship studies against both DNA and RNA viruses were mostly performed in the 1980s. 31,35,[133][134][135][136][137] Nevertheless, most of the derivatives that showed antiviral effects were also cytotoxic. Therefore, only the compounds with sufficient selectivity between antiviral activity and cytotoxicity will be mentioned here. Xylotubercidin (75) possesses potent activity against herpes simplex virus (HSV-1 and HSV-2) (EC 50 = 0.75 and 0.26 M, respectively) and showed in vivo anti-HSV effect in mice, however, the therapeutic window is rather narrow because at higher doses of xylotubercidin (75) its toxicity to treated animals became apparent 138 (Fig. 19).
Significant activity against hepatitis B virus (HBV) was observed for fluoroderivative 76 which is able to efficiently inhibit episomal viral replication (EC 50 = 2.5 M) and was well tolerated in mice. 139,140 On the other hand, compound 76 significantly affected metabolism and morphology of mitochondria and therefore it is likely to be hepatotoxic. 141 Ethynyl derivative of 7-deazaneplanocin A 77 showed submicromolar anti-HBV activity (EC 50 = 2.5 M) and no cytotoxicity (CC 50 > 300 M). 125 Although the mechanism of action of compound 77 has not been studied yet, this derivative represents an interesting lead structure as it was also active against lamivudine-and adefovir-associated Antiviral activities of 7-deaza-2 ′ -C-methyladenosine (MK-0608, 60a) against HCV and DENV have been already mentioned. However, MK-0608 (60a) was also found active against other RNA viruses such as human rhinovirus type C (HRV-C) 144 , two viruses from Flaviviridae family: tick-borne encephalitis virus (TBEV) 145 and recently also Zika virus (ZIKV). MK-0608 (60a) showed promising anti-ZIKV activity both in vitro 147,148 (EC 50 = 8.92 M) and in an in vivo mouse model. 148 5 ′ -Triphosphate form of MK-0608 (60a) was identified as an inhibitor of ZIKV-RNA-dependent RNA polymerase (IC 50 = 7.9 M). 149 As well, NITD008 (72a) is not only a potent anti-dengue compound but also was shown to inhibit proliferation of enterovirus 71 (EV71) (CPE 50  family. 151 Co-treatment with NITD008 (72a) and an anti-inflammatory drug vorinostat (SAHA), a histone deacetylase inhibitor, was successful in a mouse model of West Nile virus (WNV) infection. 152 Structurally similar prodrug 73 was also found to be active against many members of the Flaviviridae family, such as yellow fever virus and WNV. 132 Acyclic analogues of 7-deazapurine nucleosides and nucleotides were also screened for antiviral activities but generally these compounds showed only poor activities, especially when compared with the corresponding purine analogues. [153][154][155][156] Nevertheless, some acyclic analogues of 7-bromo-7-deazaadenosine (81a-b) [157][158][159] and thiosangivamycin (82a-b) [160][161][162][163] possess selective activity against human cytomegalovirus that is comparable or better than that of ganciclovir.
In conclusion, 7-deazapurine nucleosides have a potential to provide new structures for development of antivirals against both DNA and RNA viruses and representatives of this class of compounds should be covered in antiviral screenings of compound libraries. 7-Deazapurine nucleosides seem in particular promising as compounds with antiviral activities against RNA viruses from Flaviviridae family. nucleosides are yet under study. Many 7-deazaadenosine derivatives are inhibitors of either human or Mtb-ADK but their antimycobacterial activities were rather moderate. Many sugar-modified 7-deazapurine nucleosides exert antiviral activities. In particular, 2 ′ -C-methyl-ribonucleosides and 2 ′ -C-methyl-2 ′ -fluororibonucleosides showed promising activities against HCV and several underwent clinical trials, although none of them made it to FDA approval so far.

CONCLUSIONS
There is still a lot of potential in this interesting class of compounds to investigate. Most of the derivatives were derivatives of 7-deazaadenosine, which are easier to synthesize. Much less attention was paid to derivatives of 7deazaguanosine, 7-deazainosine, or 7-deazaisoguanine. Undoubtedly, there is also a lot of space in other types of triand perhaps even tetracyclic hetero-fused deazapurine nucleosides and in other fused heterocycles (e.g. analogues of triciribine), but their synthesis is very challenging. Combination of a modified deazapurine base with modified sugars is also an underexplored area with some potential, especially for antiviral activities against new emerging viruses. Little is known about the inhibition of protein kinases by these compounds, as well as about their interactions with adenosine receptors. 7-Deazapurine analogues of some nucleotide cofactors could be also of interest both as drug candidates and as tools in chemical biology. We hope that this review will ignite higher attention to this important class of molecules.