CHAPTER 1: History and Development of Nucleotide Analogues in Nucleic Acids Drugs

Forty years ago, Zamecnik and Stephenson proposed the therapeutic use of antisense oligonucleotides on the basis of their finding that Rous sarcoma virus (RSV) replication could be inhibited by a synthetic oligonucleotide complimentary to the RSV genome.¹  This concept opened up a new approach to drug discovery, namely an oligonucleotide binding sequence-specifically via Watson–Crick base-pairing to a complementary target RNA.

Since then, continuous progress has been made towards realizing the potential of this novel scientific approach and this has led recently to the approval of five antisense drugs. While the underlying concept of antisense is very simple, a rigorous understanding of the chemistry of nucleic acids had to be developed for its use in humans. In this chapter we discuss the history of this chemistry of oligonucleotides in antisense and the lessons learned from preclinical studies and clinical trials that have guided the development in conferring drug-like properties.

In parallel to the development of antisense (see Chapters 2–4), the application of synthetic oligonucleotides as therapeutic agents has evolved into broad applications involving multiple modalities. These applications include ribozymes (see Chapter 18), small interfering RNA (siRNA, see Chapters 10, 11 and 12), microRNA (see Chapter 8), aptamers (see Chapters 15 and 16), non-coding RNA (see Chapter 9), splicing modulation (Chapter 6), targeting toxic repeats (Chapter 7), gene editing (Chapter 17), and immune modulations (see Chapters 5, 13 and 14).

The common feature of these applications is that drug candidates are composed of natural nucleosides or nucleoside analogues linked via phosphodiester or modified linkages.

In 1976, RSV was the only purified virus for which a sufficient quantity was available for potential sequencing. Maxam and Gilbert sequenced this RNA virus and noted that both ends of the linear viral genome bore the same primary sequence and were in the same polarity. It occurred to Zamecnik that the new piece of DNA synthesized by reverse transcription at the 5′-end of this retrovirus might circularize and hybridize with the 3′-end. Thus he considered the possibility of inhibiting viral replication by adding a piece of synthetic DNA to the replication system to block the circularization step by hybridizing specifically with the 3′-end of the viral RNA in a competitive way.

This experiment led to startling observations, including the inhibition of new virus particles and the prevention of transformation of chick fibroblasts into sarcoma cells. In a cell-free system, translation of the Rous sarcoma viral message was also dramatically impaired. These observations were the first to show proof of the antisense concept.^1,2 

Not much further progress was made in the field up to 1985, primarily for three reasons. First, there was still widespread disbelief that oligonucleotides could enter eukaryotic cells. Second, there was very little DNA (or RNA) genomic sequence available for targeting by antisense, and third, efficient automated methodologies to synthesize oligonucleotides in sufficient quantities were only just beginning to become established.

Although the principle of solid-phase oligonucleotide synthesis was first introduced by Letsinger and Mahadevan in 1965,³  development of more efficient methods of oligodeoxynucleotide (ODN) synthesis on solid support took place from 1975 in the Gait laboratory by the phosphodiester chemistry⁴  and from 1979 by the phosphotriester method in the Itakura laboratory⁵  and the Gait laboratory.^6,7  These methods were superseded by the outstanding phosphoramidite chemistry of Caruthers and colleagues,⁸  which was automated by Applied Biosystems and other companies. This transformed the ability of non-chemists to obtain ODNs for biological purposes.

In the mid-1980s the discovery of human immunodeficiency virus 1 (HIV-1) and the availability of its RNA sequence led Zamecnik to employ an antisense approach in attempts to inhibit HIV replication.²  Antisense ODNs with phosphodiester linkages (PO-ASO, Figure 1.1) complementary to various regions of HIV-1 mRNA were synthesized using automated synthesizers. In HIV-1-infected cells, these PO-ASOs inhibited HIV-1 replication and suppressed expression of HIV-1 related markers.⁹  Such experiments were carried out using primary human cells and cellular uptake of ODNs was not a limiting factor. In these studies a control PO-ASO showed minimal inhibition of HIV-1 replication, providing evidence of sequence-specific antisense activity. These studies re-established the potential application of antisense as a therapeutic approach.

It was realized that the use of PO-ASOs would have limited therapeutic application, since these ASOs would be degraded rapidly in biological fluids. Soon, the focus of research shifted into identifying novel analogues of oligonucleotides that would have increased stability against nucleases and maintain the sequence-specific hybridization for use in antisense studies. The mechanisms of ASO activities (RNase H and steric blocking) are reviewed in Chapter 2. The characteristics required for a good ASO oligonucleotide are summarized in Figure 1.2.

The ASO therapeutics field took a major step forward in the mid 1980s through the chemical synthesis of phosphorothioate (PS) ODNs,¹⁰  based on the pioneering work of Eckstein. In these analogues a simple sulfur atom replaces an oxygen atom (Figure 1.1). PS-linked ODNs are much more resistant to nuclease degradation than phosphodiesters and thus cellular activities were found to be much higher. However, mixed diastereomeric PS-ODNs, accessible readily by automated synthesis, have lower binding affinity to target RNA as compared with PO-ODN. Synthetic methodologies were optimized to synthesize milligram quantities of PS-ODNs for use in experiments as ASOs.

Studies with PS-ASO targeted to various regions of HIV-1 mRNA as well as non-complementary analogs including homo-oligomers, were conducted in HIV-1-infected cell-based assays and showed potent dose-dependent inhibition of viral replication and antiviral activity.^11,12  The antiviral activity was related to the base composition of the analogues, and longer PS oligonucleotides were more effective than shorter ones. These studies also established that in primary human cells, cellular uptake of PS-ASO was efficient and that no carrier was required. Furthermore, PS-ASOs also showed very potent and durable inhibition of HIV-1 replication in chronically HIV-1-infected cells.^13–15 

On the basis of the promise of these results, antisense technology gained the attention of the broader scientific community and PS-ASOs became the choice as first-generation ASOs. PS-ODNs thus quickly became the primary choice for therapeutics development by newly formed biotechnology companies, such as Gilead Sciences, Isis Pharmaceuticals (now Ionis), Hybridon (now Idera Pharmaceuticals) and several others. Over the next few years, hundreds of reports appeared in the literature on the use of PS-ASOs targeting various viruses,^16–18  oncogenes,^19,20  kinases^19,21  and other targets.²²  Soon it was realized that cellular uptake in transformed cells in culture was poor and lipid-based formulations were needed for efficient uptake and antisense activity. Also it was noted that the duplex of a PS-ASO with a target RNA elicited RNase H activity,^14,15  thereby allowing PS-ASOs to cleave RNA strands and thus to inhibit translation. This was more efficient than by using a steric blocking mechanism.^23,24  However the efficiency of RNase H cleavage of RNA by a PS-ASO was poor as compared with a PO-ASO.^14,15 

The first pharmacokinetic and tissue disposition study of a systemically delivered PS-ASO occurred in mice and showed that the plasma half-life was very short and that there was a broad tissue distribution.²⁵  The highest concentrations of the administered PS-ASO were observed in the liver and kidneys, with the lowest concentrations in the brain. Delivered PS-ASO remained stable in tissues for days and was excreted primarily in urine in degraded form. Degradation was primarily from the 3′-end, and modifications of the 3′-end increased the stability further.^26,27  The binding of PS-ASO to serum proteins played a major role in plasma half-life and tissue disposition and was affected by the presence of secondary structures.^28,29 

Studies with multiple PS-ASOs of varying sequence composition and length were conducted in animal models of viral diseases and cancer.^30,31  In these studies PS-ASOs exerted very potent activity. However it soon become clear that in some cases the control PS-ASOs employed also had some activity, leading to questions on what a good control would be for PS-ASO and/or if a PS-ASO had off-target activity.³²  In one study a PS-ASO targeted to human papillomavirus inhibited papillomavirus-induced growth of implanted human foreskin in a mouse xenograft model. However, it also showed activity in a lethal mouse cytomegalovirus (CMV) model, in which the PS-ASO was not expected to have antisense activity.³³  Detailed studies in immune-compromised mice led to an understanding that the antiviral activity of PS-ASO was largely due to immune activation of the host.

The impact of immune-stimulatory properties of PS-ASOs also became evident during the non-clinical safety evaluations of multiple candidates in support of investigational supporting studies. In mice and rats, repeated systemic administration of PS-ASO candidates caused general inflammation, splenomegaly, thrombocytopenia, elevation of transaminases and histological changes in multiple organs.^34,35  In addition, in primates, bolus administration of the first PS-ASO (GEM91) led to severe hemodynamic changes, which subsequently were found to be due to activation of the alternative complement pathway.³⁶  Complement activation thus became recognized as a second off-target effect of PS-ASOs. The alternative pathway complement activation cascade was attributed to the poly-anionic nature of PS-ASOs and largely found to be a plasma concentration threshold effect. As such, it could be mitigated by subcutaneous administration or by slow intravenous infusion to keep a low plasma concentration. Based on this observation, the US Food and Drug Administration (FDA) implemented guidelines on dosing of PS-ASOs and recommended use of non-human primates as the non-rodent species for non-clinical safety studies.³⁷  Meanwhile, early development programs with PS-ASOs continued to proliferate.

Over 25 PS-ASO drug candidates targeted to viral RNA, oncogenes and cellular targets had advanced to human trials.³⁸  Routes of administration included intravitreal, intravenous infusion or subcutaneous.^39–41  In humans the plasma half-life and excretion of PS-ASOs was similar to that which was observed in pre-clinical models.^42,43  Clinical development of most of the PS-ASO drug candidates were discontinued, either due to lack of activity or a poor therapeutic index.⁴⁴  Experience with a specific PS-ASO drug candidate, GEM91, in humans was very informative. Subcutaneous administration of GEM91 caused flu-like symptoms, swelling of the draining lymph nodes, prolongation of activated partial thromboplastin time (aPTT) and more importantly, rather than suppressing HIV-1, it increased HIV-1 RNA levels in blood.⁴⁵  Importantly, administration of the same dose by the intravenous route had minimal effects on these parameters. This was puzzling at that time, but much later it became clear that PS-ASO containing the unmethylated CpG motif were activating the immune responses by binding to Toll-like receptor 9 (TLR9), an innate immune receptor present in immune cells that recognizes DNA containing CpG dinucleotide motifs (see Chapter 14). Treatment with many other PS-ASO drug candidates also had shown flu-like symptoms and injection-site reactions. It is important to note that most of the PS-ASO drug candidates that had been advanced to human studies contained a CpG motif.⁴⁶  For example, the PS-ASO Fomivirsen, targeted to CMV and delivered intravitreally, had been approved but is no longer marketed. The mechanism of action of Fomivirsen has been questioned.

Collectively, the development of PS-ASOs from preclinical to clinical studies has provided very important insights into the properties of PS-ASOs.^47–49  These could be classified as a class effect, including an affinity to target RNA, stability towards nucleases, serum protein binding and poly-anionic-related side effects, such as complement activation and prolongation of aPTT. The immunostimulatory effects of PS-ASOs have been shown to be sequence-dependent and associated with the presence of CpG motifs.⁵⁰ 

Much debated was whether some of the issues observed with PS-ASOs could be due to the presence of a mixture of diastereoisomers in the synthetic PS-ASOs. A sulfur substitution for a non-bridged oxygen on phosphorus results in both Rp and Sp diastereoisomers leading to 2¹⁹ isomers in a 20-mer (Figure 1.1). It was well known, based on the pioneering work of Stec and colleagues, that polymerases and nucleases interact and exert action on PS-ODN in a diasterioisomeric-selective fashion.⁵¹  To obtain large amounts of stereo-enriched PS-ODN, nucleoside bicyclic oxazaphospholidium as synthons were introduced and used for synthesis of various configurations of stereo-enriched PS-ASOs, including all Rp, all Sp, or segments containing Sp–Rp–Sp or Rp–Sp–Rp.⁵²  As expected, binding affinity to RNA was higher for PS-ASO containing Rp linkages followed by Sp–Rp–Sp, equal to Rp–Sp–Rp, more than stereo-random and Sp isomers. Stability towards nucleases was in the reverse order where Sp > Sp–Rp–Sp > stereo-random > Rp. RNase H activation was preferred by Rp isomers followed by Rp–Sp–Rp > Sp isomers.⁵²  In cell culture studies, stereo-enriched PS-ASOs targeted to mouse double minute 2 homolog (mdm2) of different configurations showed very similar antisense activity compared with synthetic stereo-random PS-ASOs.⁵²  In an independent follow-up study stereo-enriched PS-ASOs targeted to scavenger receptor class B type 1 (SR-B1) showed similar antisense activity to stereo-random PS-ASOs.⁵³  However, in a recent study, stereo-pure PS-ASOs were shown to have improved activity in cell culture and in in vivo studies.⁵⁴  It is likely that the sequence composition of ASO and the particular placement of stereo-specific linkages at specific positions have an effect on the antisense activity.

The most extensively studied non-anionic oligonucleotide analogues are those containing alkylphosphotriester^55,56  methylphosphonate⁵⁷  and phosphoramidate linkages⁵⁸  (Figure 1.3). These internucleotide linkages consist of a mixture of distereoisomers. Ts'o, Miller and colleagues had carried out extensive studies with methylphosphonate oligodeoxynucleotides (MP-ODNs) and had shown that MP-ODNs are stable under physiological pH, resistant to nucleolytic degradation and had a lower binding affinity to target RNA compared with PO-ODN.⁵⁷  In earlier studies, short MP-ODN had been employed as antisense in a cell-free system. To evaluate MP-ODNs as antisense, solid-phase synthetic methodology using methylphosphonamidites as synthons was developed^57,59  to obtain milligram quantities of longer length MP-ODN. MP-ASO of 15- to 20-mers targeted to HIV-1 showed dose-dependent inhibition of HIV-1 replication, longer MP-ASOs were more active and the effective concentration ranged from 30 to 150 micromolar.⁵⁹  This could have been due to a lower affinity to target RNA and a lack of RNase H activation. Such ASOs failed to activate RNase H and inhibited translation by blocking the ribosome machinery, which is referred to as ‘translation arrest’. In addition, it was noted that MP-ASOs of longer length had poor solubility under physiological conditions. In vivo delivery and disposition of MP-ASOs were limited due to their non-ionic nature. No MP-ASO candidates were advanced to human trials.

The anti-cancer agent GRN163 is a 13-mer ASO containing all-phosphoramidate internucleotide linkages, that inhibits telomerase⁵⁸  and did make it to a clinical trial, but did not receive regulatory approval as an antisense drug. Studies with an ASO containing phosphoramidate linkages (PN-ASO) targeted to HIV-1 had also been carried out in cell-based assays and showed similar results to those observed with MP-ASOs, and no further studies were conducted.¹¹ 

Phosphorodiamidate morpholino oligonucleotides (PMO) are charge-neutral oligonucleotides in which bases are attached to a morpholino ring, linked through a phosphorodiamidate group (Figure 1.3).⁶⁰  Replacement of anionic phosphates by uncharged phosphorodiamidate groups eliminates ionization in the physiological pH range. PMO-ASO inhibits translation by ‘steric block’ (see Chapter 2) since a duplex of PMO and RNA is not a substrate for RNase H⁶¹ (see Chapter 2). Cellular uptake of PMO is not very efficient and requires very high doses for in vivo delivery. There have been reports in which PMO ASOs have been evaluated against Ebola, Marburg and Chikungunya viruses, for example.⁶²  One of the PMO-ASO candidates designed to induce alternative splicing of mutant dystrophin, eteplirsen, has shown an acceptable safety profile and restoration of dystrophin in Duchenne muscular dystrophy (DMD) patients receiving 50 mg kg⁻¹ dose weekly and recently has been approved for clinical use (see Chapter 6). The chemistries used in exon-skipping activities have been reviewed.⁶³  More specifically, conjugation of cell-penetrating peptides to PMOs has allowed for significant improvements in delivery and increased exon-skipping activity in in vivo models in muscles as well as in heart.⁶⁴ 

The major use of PMO has been in research to study the role of specific mRNA transcripts in embryos. Several studies have been conducted in which PMO-ASOs have been injected into eggs, embryos of zebrafish, Xenopus and sea urchins to elucidate gene function.⁶⁵ 

Peptide nucleic acids (PNA) are charge-neutral synthetic mimics of DNA or RNA containing N-2 aminoethylglycine units linked by amide bonds as repeating units in place of a sugar–phosphate backbone (Figure 1.3).⁶⁶  PNA provided two key properties for use as antisense, a significant increase in affinity to target RNA, due to the absence of electrostatic repulsion, and resistance to degradation by nucleases or proteases. Since the duplex of PNA with RNA does not activate RNase H, PNA-ASO would act by ‘translation arrest’. PNA-ASOs have been broadly studied as anti-cancer,^67,68  anti-viral^69,70  and anti-bacterial agents^71,72  as well as inhibitors of micro RNAs.⁷³  However, once again very high doses are needed in in vivo applications, due to poor cellular uptake, unfavorable pharmacokinetics and the need for efficient delivery vehicles. No PNA-ASO agents have been advanced to clinical development.

One of the approaches pursued to improve the affinity of ASO to target RNA was to employ modified heterocyclic bases. There were a number of studies carried out with the incorporation of heterocyclic modification that provided an additional hydrogen bond to its complementary base pair to improve binding affinity. These included modifications in the pyrimidine moiety, at positions C-2, C-4, C-5 and C-6. A large number of these were incorporated in ASOs. However, the results observed were not exciting. Incorporation of modified purines resulted in lowering of the binding affinity of antisense. In a detailed study, various heterocyclic bases including C-5 propynyl, 5-methyl cytosine (5-MeC), phenoxazine and G-clamp were compared for antisense activity and these showed that G-clamp had potent dose-dependent antisense activity.⁷⁴  However, these modified ASOs containing G-clamps were found to be highly toxic in in vivo studies. Out of hundreds of heterocyclic bases that have been studied in ASOs, only a few are currently being employed. The most used modified nucleoside in ASO is 5-methyl-2′-deoxycytidine (5-MedC).⁷⁵ 

One of the key properties of ASOs is a strong affinity to target RNA, and it is well known that the affinity of the duplex of RNA–RNA is stronger than the duplex of DNA–RNA. Based on these criteria, the use of synthetic oligoribonucleotides (ORN) in antisense has been studied (Figure 1.4). To provide nucleolytic stability, phosphorothioate ORNs (PS-ORN-ASO) were synthesized. Evaluations of PS-ORN-ASOs showed an increased affinity to target RNA.⁷⁶  However they had lower potency as compared with the corresponding deoxynucleotide PS-ASOs, probably due to reduced nucleolytic stability and the lack of RNase H activation by the RNA–RNA duplex.

To achieve further increases in the affinity of ASO to target RNA and an increase in nucleolytic stability, the use of 2′-O-methylribonucleosides⁷⁷  was evaluated. These are naturally occurring nucleosides, and, importantly, phosphoramidites of 2′-O-methylribonucleosides had become available commercially. Studies carried out with 2′-O-methyloligoribonucleotide phosphorothioates (2′-OMe PS-ASO), showed nucleolytic stability as compared with PS-ORN and higher affinity to target RNA.⁷⁸  However, they also showed less antisense activity compared with PS-ASOs, indicating that RNase H activation was a key parameter for antisense.^23,24  Since then, studies have been conducted with many additional 2′-O-alkyl analogs and the results have been similar. These studies led to the identification of 2′-O-methoxyethylribonucleosides (2′-MOE),⁷⁹  which are now used widely in ASO studies (see Chapters 2–4).

Additional studies with PS-ORN and 2′-OMe PS-ASO provided very good insights into the nature of PS linkages in RNA and DNA. As discussed above, PS-ASOs had shown dose-dependent activation of complement and prolongation of aPTT and strong binding to serum proteins. These activities were believed to be due to the poly-anionic nature of the phosphorothioate backbone. Studies with PS-ORN and 2′-OMe-PS-ASO showed significantly reduced effects on these parameters, indicating that the poly-anionic nature of the PS backbone in PS-ORN and PS 2′-OMe had different characteristics.⁸⁰  This insight was helpful in designing gapmer antisense (see Chapters 2–4).

Lack of RNase H activation and increased affinity to target RNA also provided the opportunity for the application of 2′-O-alkyl modified PS-RNA for steric-blocking applications, such as splice switching^81,82  (see Chapter 6) and microRNA inhibition (see Chapter 8). Clinical development of 2′-O-alkyl ASOs targeting splice sites have yielded mixed results. Drisapersen, a 2′-OMe PS-RNA, failed to show clinical benefits in patients with DMD and also caused significant adverse events. In contrast in the last two years, an 18-mer 2′-MOE phosphorothioate (2′-MOE PS-ASO; nusinersen) that redirects the splicing of the SMN-2 gene has been approved for the treatment of spinal muscular atrophy (see Chapter 6). It is important to note that the safety issues observed with the use of drisapersen include thrombocytopenia and renal toxicity. This could be largely due to repeated administration of 2′-OMe PS-ASO, which is very stable to nucleolytic degradation. Drisapersen is retained and accumulated in tissues after repeated administration due to its long half-life. Nusinersen, a 2′-MOE PS-ASO, is also very stable to nucleolytic degradation, but it is administered intrathecally at lower doses and less frequently, which minimizes the effects of tissue accumulation.

Reducing the conformational flexibility of nucleotides has been shown to increase their binding affinity to complementary RNA. These are collectively known as bridged nucleic acids (BNA). Locked nucleic acids (LNA) (Figure 1.4) link the 2′-oxygen and 4′-carbon of ribose and have shown a significant increase in affinity to target RNA.⁸³  All LNA oligomers of octamers or longer tend to aggregate, thereby limiting their therapeutic utility. Therefore they are generally used as mixmers with 2′-deoxy nucleotides or 2′OMe nucleotides (see Chapters 6 and 8). Other bridged nucleic acids which have been studied include a methylated analog of LNA, known as ‘constrained Ethyl’ (cET)⁸⁴  (see Chapter 3), tricyclo-DNA (tcDNA)⁸⁵  (Figure 1.4) and 2′-O,4′-C-ethylene linked nucleic acid (ENA)⁸⁶  (not shown). These bridged nucleic acids have also shown very strong affinity to target RNA and increased nucleolytic stability, but they are not substrates for RNase H. These modifications are being employed in ASOs mostly for splicing modulation and gapmer approaches. Encouraging results have been reported with ENA, tcDNA or combinations of 2′-OMe and ENA in preclinical models of DMD. Increased affinity has allowed for the use of shorter ASOs. In a recent study an ASO, DS-5141 containing segments of 2′-OMe PS-RNA and ENA has shown promising activity in an mdx mouse model and a phase 1/2 clinical trial in DMD has been initiated in Japan.⁸⁷ 

Based on the experience with PS-ASOs and other modified ASOs, including MP-ASOs and 2′-OMe PS-ASO's, it became clear that each of these modifications carry specific desirable properties. However, each of them lack one or more of the required characteristics for antisense (Figure 1.2). For example, desirable properties of PS-ASOs include activation of RNase H, whereas the characteristics of MP-ASOs only include nucleolytic stability and a non-anionic backbone. For 2′-OMe PS-ASO characteristics include increased nucleolytic stability, increased affinity to target RNA and reduced poly-anionic characteristics and immune-stimulatory effects. This led to a concept that an optimal ASO could be designed in which segments of PS-DNA and MP-ASO or 2′-OMe PS-RNA could be appropriately placed. For example, a PS-DNA segment could be placed in the middle flanked by segments of 2′-OMe PS-RNA at both the 3′- and 5′-ends^23,88  (see Chapter 2). Studies have also been carried out in which a segment of 2′-OMe PS-RNA has been placed in the middle and segments of PS-DNA are placed at both the 3′- and 5′-ends.

Early studies were carried out with ASOs containing segments of PS-ASO and MP-ASO, which showed that RNase H-mediated excision of the target could be directed to specific sites.¹⁴  However, these ASOs had lower affinity to target RNA. ASOs containing segments of PS-ASO and 2′-OMe PS-ASO had many of the desired characteristics, including high affinity to target RNA, increased nucleolytic stability, reduction in poly-anionic characteristic and immune-stimulatory activity and activation of RNase H.^23,80,88  Over the years these types of ASO constructs have been referred to as mixed backbone-oligos, hybrids or, more recently, as gapmers. In these studies 2′-MOE nucleotides⁸⁴  have shown promising results and have been widely employed in gapmer ASOs (Chapters 2–4).

Gapmer ASOs of varying sequence and composition, length and configuration have been studied for hundreds of RNA targets in cell culture and in animal models and have yielded very potent antisense activity. Pharmacokinetic studies in mice and primates have shown very similar plasma half-life and tissue disposition as PS-ASO, with significantly increased in vivo stability, resulting in longer persistence in tissues of the administered gapmer ASO.^89,90  This increased in vivo stability of gapmer ASO should have allowed for less frequent dosing in preclinical and clinical studies. However, this has not yet been achieved. This indicates that gapmer ASO retained in tissues is not bio-available for antisense activity. Gapmer ASOs also showed reduced inflammatory responses⁹¹  and improved safety parameters as compared with PS-ASO (discussed in detail in Chapter 20). The increased nucleolytic stability of gapmer ASO has also allowed for potential oral delivery.^28,92  Clinical experience with gapmers is described in Chapters 3 and 4. Unfortunately, the list of discontinuations of clinical development of gapmer ASO drug candidates has also been growing, largely either due to lack of efficacy or to safety signals that limit the therapeutic index.

Incorporation of LNA into gapmer ASOs has proved disappointing and the clinical development of a number of drug candidates have been discontinued due to safety signals. Incorporation of cEt in gapmer ASO has shown improvements in potency in preclinical studies and the clinical development of gapmer ASO drug candidates is in progress (see Chapter 4).

It is important to consider whether safety signals with gapmer ASOs (discussed in Chapter 20) are due to specific sequence and motifs and/or extensive tissue build-up. It was postulated that increased nucleolytic stability and tissue retention of gapmer ASO may result in less frequent dosing to achieve therapeutic benefit. However, this has not been the case, and in a majority of the studies, gapmer ASOs are being injected on a weekly regimen. It is evident from pre-clinical studies in mice and primates that after repeated injections significant amounts of administered gapmer ASO build up in certain tissues.⁹³  What are the fates of gapmer ASOs and their degradation products in tissues? Do they interact with local innate immune receptors? Does the accumulated gapmer ASO with very high affinity to RNA, and in some cases use of shorter ASO sequences, result in excessive activation of RNase H by binding to unrelated RNA targets? Insights into these questions will further provide guidance on improving the design of gapmer ASOs to improve therapeutic index.

The results of administration of PS-ASOs and gapmer ASOs have shown short plasma half-lives, wide deposition in multiple organs and elimination primarily via urinary excretion (Chapter 20). Serum protein binding is very important for both of these ASO types. ASOs of a non-ionic nature, like MP and PMO, do not bind to serum proteins, have a very short plasma half-life, are not distributed to many organs and are rapidly eliminated by urinary excretion. A number of chemical conjugates, including cholesterol,⁹⁴  lipids,⁹⁵  peptides,⁹⁶  antibodies⁹⁷  and sugars,⁹⁸  have been studied for targeted in vivo delivery. The formulation of ASOs with lipids led to preferential disposition to liver, but was also associated with immune activation. In the last few years, use of N-acetylgalactosamine (GalNAc), a ligand for asialoglycoprotein receptor expressed on hepatocytes has been widely studied (see Chapters 4 and 11). Conjugation of GalNAc has led to increased liver uptake of gapmer ASOs, resulting in improved efficacy and allowing use of lower doses. It is to be noted that a significant fraction of conjugated gapmer ASOs are also present in kidneys. Emerging clinical data with GalNAc gapmer ASO conjugate targeted to ApoCIII has been encouraging and long-term exposure data will guide further development. At the same time, it is important to note that clinical development of two GalNAc ASO conjugates, RG101 and RG125, has been discontinued due to clinical or pre-clinical safety signals.

Cell-penetrating peptide conjugates of PMO ASOs have shown enhanced uptake in muscles⁹⁹  and, in some cases, heart muscles.^64,100  There is now a significant effort to find specific receptor-mediated targeting into other tissue types.

As discussed above and in Chapter 14, the first indication of immune stimulatory activity of a PS-ASO in humans was observed in 1993–1994. It was not until the year 2000, when TLR9 was discovered, that this provided the basis for the observed immune activation. Since then, tremendous progress has been made in the discovery and the role of the family of innate immune receptors. Many of these receptors, including Toll-like-receptors (TLR) 3, 7, 8 and 9, RIG-1, Inflammasome and STING, are activated upon recognition of patterns of foreign nucleic acids of viral and bacterial origin and, under certain conditions, also with endogenous nucleic acids (see Chapters 5, 13 and 14).

Under normal conditions, activation of these receptors leads to a defense mechanism. Under pathological conditions, unintended activation of these receptors could lead to autoimmune inflammatory disorders. In brief, TLR9 is a receptor for bacterial DNA containing an unmethylated CpG motif, and upon recognition leads to the initiation of a signaling cascade resulting in activation of nuclear factor kappa B and induction of T helper 1 (Th1)-type cytokines. Many of the PS-ASOs advanced to clinical development contained a CpG motif and had shown immune-stimulatory properties, thereby leading to the questioning of the intended mechanism of action (see Chapter 14). To mitigate immune-stimulatory activity of PS-ASOs or gapmer ASOs containing CpG motifs, substitution of C in the CpG motif by 5-MedC has been employed. As has been reported recently, the presence of 5-MedC in the CpG motif is indeed not immune-stimulatory, but acts as an antagonist of TLR9,¹⁰¹  so caution should be exercised. Such has been a recent case with a PS-ASO targeting small and mothers against decapentaplegic homolog 7 (SMAD-7), in which CpG motifs were substituted by 5′-MedC in dCpG motifs and which failed in a phase 3 trial.¹⁰²  These types of ASOs can act as antagonists to TLR9. Similarly gapmer ASOs, depending on the sequence composition and position of 2′-O-alkyl substitutions, could act as antagonists of TLR7, 8 and 9 and interfere with the intended mechanism of action (see Chapter 5). Nucleic acids that either activate or inhibit innate immune receptors have shown very broad therapeutic applications (see Chapters 5 and 14).

It is important to give due considerations to innate immune receptors when selecting ASO sequences and modifications to be employed (Figure 1.2). In addition, the role of these innate immune receptors with respect to accumulated gapmer ASOs and their metabolites in tissues has yet to be elucidated. Does this interaction lead to local signaling cascades and thereby inflammation and safety signals? In a preliminary study, an oligonucleotide which had shown inflammatory signals and histopathological changes in C57 black 6 mice, showed significantly reduced effects in TLR7 and 9 double-knockout mice (Agrawal, unpublished data).

Tremendous progress has been made in the last 40 years. The field of nucleic acid-based therapeutics has evolved to become a major drug-discovery platform. In parallel to continued developments in the antisense field (Chapters 2–4, 6 and 7), there has been tremendous progress made in several related fields. Understanding of the role of RNA in biology has led to identifications of miRNAs (Chapter 8), long non-coding RNAs (Chapter 9), small nucleolar RNA (snoRNA) etc. and their interplay in gene regulation, thereby expanding the pool of potential targets for antisense. New mechanisms of gene regulation have also been uncovered, including the harnessing of ribozymes (Chapter 18), siRNA (Chapters 10–12) and CRISPR (Chapter 17). Furthermore, the discovery of innate immune receptors for nucleic acids has helped in providing answers to many questions in the field of nucleic acid therapeutics and has resulted in a new therapeutic platform (Chapters 5, 13 and 14). In addition, aptamers have been developed as therapeutics (Chapters 15 and 16).

To date, a number of nucleic acid-based drugs have been approved. These include an aptamer, Macugen, and five antisense oligonucleotides, fomivirsen, mipomersen, eteplirsen, nusinersen and inotersen. A siRNA drug candidate, patisiran, has been approved recently. Clinical developments of a number of other ASO and siRNA candidates are in progress. This indicates that the potential of antisense and other nucleic acid-based therapeutics is being realized. This has required development of large-scale synthesis techniques (Chapter 19).

However challenges remain. Out of the over 70 ASO drug candidates that have been advanced to clinical development, the list of ASO drug candidates being discontinued also continues to grow. There could be number of reasons for discontinuations, such as an unvalidated gene target, ASO sequence composition and motifs, level of knockdown achieved not being sufficient for therapy, potency of the ASO, off-target activity and, most importantly, therapeutic-index-limiting treatment (Chapter 20).

Since 2001, the focus has been on the development of gapmer ASOs. Gapmer ASOs provide optimal properties for antisense, both in cell-based assays as well as in vivo models and have yielded positive phase 3 results. However, treatment has also been associated with safety signals. There could be many reasons for this, including sequence composition, off-target activity, nucleotide motifs, or even secondary structures, or interactions with innate immune receptors. A common feature in all of these gapmer ASOs is increased in vivo stability, resulting in accumulation and retention in many organs. The retained gapmer ASO and its metabolites are not functionally active, since repeated dosing is needed for sustained target RNA or protein knockdown.

The focus is shifting towards further improvements in the potency of gapmer ASOs by incorporation of novel nucleotide analogues (e.g. LNA, cEt and others). Improved affinity of these new gapmer ASOs has allowed for the use of shorter gapmer ASOs (Chapters 3 and 4). Studies carried out with a few of these gapmer ASO candidates have shown increased potency, resulting in the use of reduced dose and dosing schedule in vivo models. However, without addressing the in vivo stability issues repeated dosing would still lead to tissue retention and accumulation.

Throughout the studies with gapmer ASOs, it has been postulated that their increased potency is also related to the increased nucleolytic stability. A recent study provided a new insight into nucleolytic stability and its effects on antisense potency.^103,104  A comparative study of a 19-mer PS-ASO, 19-mer PS-ASOs linked together via their 3′-ends, or via their 5′-ends, or via a normal 3′–5′ linkage, in cell culture and an in vivo model, showed that 5′–5′ linked PS-ASO had more potent antisense activity than the other three PS-ASOs in cell-based assays and in the in vivo model. If nucleolytic stability is a key factor in exerting the antisense activity, a PS-ASO with 3′–3′ linkage, with the highest nucleolytic stability, would have been the most active as antisense. Furthermore a 5′–5′ linked PS-ASO showed more potent activity than a gapmer ASO. This indicates that nucleolytic stability of ASO may not be as important a factor as previously thought, but more important is how the duplex of ASO and RNA engages RNase H in excision of the target RNA. Increased antisense activity of the 5′–5′ linked PS-ASO was associated with excision of the target RNA in the center of the ASO, as compared with other ASOs, in which excision products were at the 5′-end. This insight provides further direction on the design of optimal ASOs.

One of the key emerging issues in the broad applicability of ASOs is how to address safety signals. This perhaps may be achieved by improving the potency of the ASOs, avoiding tissue build-up and by minimizing interactions with innate immune receptors. Applications of novel delivery vehicles could be of equal importance. There is more work to be done. With the proof-of-concept of antisense in humans established, it is not too distant to expect the application of ASOs for the treatment of genetic disorders, a true approach to precision medicine.

S.A. is indebted to my co-author and co-editor, Mike Gait for his mentorship. While working with him, I learned about aspects of nucleic acid chemistry, but more importantly, how to approach research to seek answers. My journey in the antisense field started in the lab of the late Paul Zamecnik, known as the ‘father of antisense’. He was a great teacher, a friend, and together we started the journey of Hybridon, now Idera Pharmaceuticals. I am also grateful to all of my colleagues and collaborators whose names appear in the references cited.