Published OnlineFirst February 13, 2019; DOI: 10.1158/0008-5472.CAN-18-2474 Cancer Review Research Deoxyribozymes as Catalytic Nanotherapeutic Agents Levon M. Khachigian Abstract RNA-cleaving deoxyribozymes (DNAzymes) are synthet- and nanosponges, and the emerging role of adaptive ic single-stranded DNA-based catalytic molecules that can immunity underlying DNAzyme inhibition of cancer be engineered to bind to and cleave target mRNA at growth. DNAzymes represent a promising new class of predetermined sites. These have been used as therapeutic nucleic acid–based therapeutics in cancer. This article dis- agents in a range of preclinical cancer models and have cusses mechanistic and therapeutic insights brought about entered clinical trials in Europe, China, and Australia. This by DNAzyme use as nanotools and reagents in a range of review surveys regulatory insights into mechanisms of basic science, experimental therapeutic and clinical appli- disease brought about by use of catalytic DNA in vitro and cations. Current limitations and future perspectives are also in vivo, including recent uses as nanosensors, nanoflowers, discussed. DNAzyme Catalysts: Mechanistic and transfected with commercial delivery agents or electroporated Design Considerations into cells. DNAzyme use in experimental animal models can be hampered by delivery issues, especially in regard to systemic DNAzymes are synthetic single-stranded enzymatic DNA mole- administration (10). This has motivated local delivery meth- cules that bind to their target mRNA via Watson–Crick base odologies in animals, such as intracardiac, intratumoral, and pairing and cleave a specific interbase junction in the mRNA by intraarticular injection, or tissue immersion. Novel biodegrad- a deesterification reaction (1–3). This involves metal-assisted able template-based DNAzymes have recently been developed 0 deprotonation of 2 -hydroxyl in the RNA, producing RNA frag- that facilitate cancer cell recognition and internalization (11). 0 0 0 ments terminating in a 2 ,3 -cyclic phosphate and a 5 -hydroxyl That said, oligonucleotides may not necessarily require a spe- (Fig. 1; ref. 4). DNAzymes were originally developed as catalysts cific delivery vehicle for endosomal or lysosomal seques- in vitro and then used as gene-silencing agents within cells where tration (12), and indeed there are numerous examples of endogenous mRNA is targeted. Breaker and Joyce reported the first DNAzyme efficacy in experimental models without use of any DNAzyme in late 1994, cleaving a ribonucleotide linkage by way carrier. In experimental tumors, this may involve an enhanced 2þ of a transesterification reaction requiring Pb (5). Studies have permeability and retention (EPR) effect (13). DNAzymes have since demonstrated that DNAzymes also use other divalent ions been used in the clinic with or without a specifictransfection 2þ 2þ 2þ as cofactors, including Mg ,Zn , and Ca (6). DNAzyme agent as discussed below. Second, the half-life of the target RNA action is not confined to RNA cleavage; these agents can also would ideally be short and background mRNA expression of 0 catalyze bond formation, such as ligation between the 3 -hydroxyl the target low. Cleavage destabilizes RNA and accelerates its 0 and 5 -triphosphate terminus in RNA (7, 8). The first high- degradation, meaning less protein is produced. In our own resolution (2.8 Å) crystal structure of a DNAzyme in its post- work, we have generated DNAzymes targeting immediate-early catalytic state complexed with RNA was recently reported with genes that are poorly expressed in normal tissues and have one such DNAzyme. Ponce-Salvatierra and colleagues found that relatively short half-lives. For example, c-jun mRNA (targeted the 44 nucleotide RNA-ligating DNAzyme 9DB1 forms a double by the DNAzyme Dz13; ref. 14) rapidly decays with a half-life pseudoknot in complex with a 15-nucleotide RNA strand (9). of 20 minutes (15). The half-life of early growth response-1 This review focuses mainly on RNA-cleaving DNAzymes. (Egr-1) mRNA (targeted by ED5; ref. 16) is even shorter (17). There are several considerations when designing DNAzymes These genes are expressed at low or undetectable levels in as endogenous RNA-cleavage–based inhibitors. First, there uninjured, growth quiescent arteries (14, 16, 18). Third, should be sufficient intracellular accumulation of the oligonu- although target site specificity is prescribed by the choice of cleotide and access to mRNA. Typically in vitro, DNAzymes are bases in the binding arms, DNAzymes, like other antisense- based targeting strategies, can have "off-target" effects, arising from partial complementarity of DNAzyme arm sequences Vascular Biology and Translational Research, School of Medical Sciences, with an unintended target or Toll-like receptors that recognize Faculty of Medicine, University of New South Wales, Sydney, Australia. CpG-motif–containing DNA (TLR-9; ref. 19). This can be Corresponding Author: Levon M. Khachigian, University of New South Wales, addressed experimentally using DNAzymes as controls bearing High Street, Gate 9, Sydney, NSW 2052, Australia. Phone: 61-2-9385-2537; single-nucleotide mutations (e.g., G6>C6) in the 15 nucleotide Fax: 61-2-9385-1797; E-mail: [email protected] 10–23 catalytic domain (i.e., 50-GGC TAG6 CTA CAA CGA-30; doi: 10.1158/0008-5472.CAN-18-2474 refs.1,4),renderingtheDNAzymeunabletocleavebut Ó2019 American Association for Cancer Research. remaining identical to the test DNAzyme in all other www.aacrjournals.org OF1 Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst February 13, 2019; DOI: 10.1158/0008-5472.CAN-18-2474 Khachigian R Y O 2’ O O P O 3’ OOH O O - 5’ Figure 1. O HO Mechanism of action by the Y RNA-cleaving 10–23 DNAzyme. The 10–23 DNAzyme (depicted here with R O þ OOH O 9 9 nucleotide arms) hybridizes O N P OOH N with its target RNA through O N N O N – O Watson Crick base pairing and N – N N N N N catalyzes cleavage between an N N 3’ N R N N unpaired purine (R) and a paired N N N N N N N pyrimidine (Y) through an internal N N G N 5’ N N A phosphoester transfer reaction N G 0 N G whereby the 2 -hydroxyl attacks the 5’ N N C N adjacent phosphodiester bond. This 3’ N C T results in one fragment ending with a 10-23 20,30-cyclic phosphate and the other A DNAzyme A ending with a 50-hydroxyl. Other A G sugar phosphates in the DNAzyme or C C RNA are not shown for simplicity. A T R, A or G (purine); Y,CorU (pyrimidine); N, G or A or T or C. © 2018 American Association for Cancer Research respects (20). Another way is to test for TLR9/NF-kBactivation rat carotid arteries, a finding later confirmed by Liu and col- alongside reference oligonucleotides (21). Such "off-target" leagues who used ED5 to inhibit Egr-1 and TGFb expression, effects should not be confused with DNAzymes targeting resulting in increased levels of nitric oxide synthase and NO master regulators, such as transcription factors, where and preservation of endothelial cell function (31). Wang and that DNAzyme would also be expected to affect other genes colleagues further found ED5 reduced levels of Egr-1 and reliant on the targeted factor. Fourth, DNAzyme stability platelet-derived growth factor (PDGF)-BB, cdk4, cyclin D1, against nucleolytic attack is typically achieved by structural monocyte chemotactic factor (MCP)-1, and intercellular adhe- modifications such as inverted bases, phosphorothioate or sion molecule (ICAM)-1 (32). In a rat autologous vein graft phosphoramidate linkages, locked nucleic acids, and 20-meth- model, Liu and colleagues used essentially the same DNAzyme oxy substitutions. These modifications can be incorporated (EDRz) to suppress Egr-1 expression, SMC proliferation, and during oligonucleotide manufacture. Fifth, although there are intimal thickening (33). EDRz also reduced formation of potentially hundreds of target sites in a typical mRNA, con- abdominal aortic aneurysm in rats, suppressing Egr-1, matrix ventional sequence selection for a DNAzyme is time-consum- metalloproteinase (MMP)-2, and MMP-9 levels (34). Egr-1 ing, labor-intensive, and cumbersome (22). Investigators have DNAzymes also inhibited intimal thickening in a rat model used predictions on RNA folding, low free energy, and second- of carotid ligation following adventitial delivery (35), reduced ary structure (23, 24), but this does not ensure DNAzyme myocardial infarct size in the area at risk in a rat model of efficacy (14). Finally, it should be noted that DNAzyme cata- myocardial ischemia–reperfusion injury (36), in-stent resteno- lytic efficacy also depends on the length of binding arms even sis in a porcine model of stenting (37), and improved for a single target site (6, 21, 25). left ventricular systolic function in porcine models of myocar- dial ischemia–reperfusion injury and intracoronary deliv- DNAzymes as Versatile Experimental ery (38, 39). Dickinson and colleagues further demonstrated the effects of Egr-1 suppression on pulmonary vascular remo- Drugs in Noncancerous Settings deling in an experimental flow-associated pulmonary arterial Since the discovery of the 10–23 DNAzyme by Santoro and hypertension rat model. Intravenous administration of ED5 (in Joyce reported in 1997, so named from the 23rd clone from the DOTAP) reduced pulmonary vascular Egr-1 expression, vascu- 10th round of an in vitro selection procedure (1, 4), DNAzymes lar remodeling and neointima formation, and reduced expres- have been used as experimental drugs in a diverse range of models. sion of PDGF-B, TGFb, IL6, and p53 (40). ED5 also inhibited New monovalent cation-dependent RNA-cleaving DNAzymes, pulmonary vascular resistance, right ventricular (RV) systolic such as Ag10c (26, 27), Ce13d (28), EtNa (29), and NaA43 (30), pressure, and RV hypertrophy. Nakamura and colleagues deliv- with catalytic domains that differ from 10–23 have also emerged, ered ED5 into interstitial fibroblasts in vivo by electroporation and it will be interesting to see how these progress within via the ureter that reduced interstitial fibrosis in obstructed preclinical systems.