From a Concept on a Chalkboard to Gapmer Antisense

Abstract

Gapmer antisense oligonucleotides emerged from our efforts to improve the stability, affinity, and potency of early phosphodiester and phosphorothioate (PS) antisense oligonucleotides. Working with Paul Zamecnik during the late 1980s and early 1990s, I witnessed the transition from a simple steric-block model of antisense action to the recognition of RNase H as the central driver of potency. Our systematic exploration of backbone and sugar modifications revealed the limitations of single-chemistry approaches and led to the development of hybrid antisense constructs that combined deoxyribonucleotide with 2′-modified ribonucleotide segments, all PS-modified. Such designs later became known as gapmers and provided enhanced stability, improved affinity, and efficient RNase H activation. Although our initial efforts with phosphorothioate oligonucleotides faltered in the clinic, continued refinement of the hybrid designs ultimately enabled the development of multiple approved drugs. Here I describe how a concept sketched on a chalkboard evolved into a widely adopted therapeutic platform.

Introduction

Gapmer antisense oligonucleotides emerged from efforts to improve the stability, affinity and potency of early phosphodiester and phosphorothioate (PS) antisense oligonucleotides. They contain a central region of DNA flanked on both the 5′ and 3′ ends by modified RNA “wings”. When hybridized to complementary RNA sequences within mammalian cells, the DNA region forms DNA/RNA duplexes with the target. Such DNA/RNA duplexes are the preferred substrate of endogenous RNase H1 which cleaves the RNA strand. The gapmer is then released and can engage in multiple rounds of target binding, resulting in the efficient knockdown of the intended RNA. The flanking modified RNA wings enhance affinity since RNA/RNA hybrids adopt the more stable A-helix. Importantly, gapmers function in both the nucleus and cytoplasm, enabling knockdown of RNA species beyond mRNA. Chemical modifications of the backbone, ribose rings and bases can further enhance stability, potency and efficacy. Continued refinement of the gapmer design over the last 25 years ultimately enabled the development of multiple approved, commercially successful drugs.

Although the initial observations that oligonucleotides (ODNs) could inhibit viral replication and RNA translation date back to two papers published by Paul Zamecnik in 1978, the implications of these reports only became clear in the late 1980s and early 1990s. In those papers, Paul introduced the term “hybridon” to describe an oligonucleotide acting through competitive hybridization. As part of Paul’s laboratory from 1987 to 1991, and later as a founding scientist at Hybridon (renamed Idera Pharmaceuticals in 2005), the company he established, I had an up-close view of how our understanding evolved under his direction. Here, I will recount only the work I directly participated in or observed in the lead up to successful therapeutic gapmer designs, as a comprehensive account of the field at that time is beyond the scope of this personal perspective.

Drawing the Background

I joined Paul’s group at the Worcester Foundation for Experimental Biology (WFEB) in Shrewsbury, Massachusetts, USA, a small town about 35 miles west of Boston in January 1987. The day before my arrival, heavy snowfall had blanketed the town in more than two feet of brilliant white snow. The bright sunshine reflecting off the snow was a striking contrast to the dreary freezing rain I had just left behind in the UK, where I had spent the previous two years working with Mike Gait at the MRC’s Laboratory of Molecular Biology (LMB), Cambridge, UK, on solid-phase oligonucleotide synthesis methods.

In my first meeting with Paul, I did a double-take: he was wearing his sweater inside out under his lab coat, with his reading glasses perched precariously on the end of his nose. Much later, his wife Mary explained that he wore his sweaters inside out while cooking breakfast to avoid stains and then sometimes forgot to turn them the right way around afterward. Paul then showed me around the laboratories, a cluster of rooms with biohoods and chemical hoods. His group consisted of John Goodchild and Paul’s wife Mary.

One of the most remarkable features of the lab was a bookshelf stacked with Paul’s lab notebooks dating back to 1932, freely available to everyone. Over the following months, I would take out lab notebooks from the 1940s and 1950s to read his handwritten experimental designs and results. It was extraordinary to see, in his own hand, the work that had uncovered the components and mechanisms of protein synthesis, including tRNA. Looking through those pages felt like stepping directly into the history of molecular biology, and it gave me a profound sense of continuity. Paul had also recently published evidence of “cellular oligonucleotides. . .that···may play a regulatory role in intracellular metabolism” later recognized as microRNAs.

Our early meetings focused on ongoing projects in the lab and on Paul’s collaboration with Robert Gallo’s group at the National Cancer Institute, on antisense inhibition of the recently discovered Human Immunodeficiency Virus (HIV). Much of the work in Paul’s lab had used phosphodiester (PO) antisense oligonucleotides to show inhibition of HIV, but despite confirming antisense action, therapeutic application was limited as PO oligonucleotides were rapidly degraded when injected in vivo. Paul concluded that more nuclease-resistant oligonucleotides were needed - ideally without sacrificing affinity - and that reducing the number of negative charges on the backbone might improve cellular uptake.

If at First You Do Not Succeed, Try, Try Again

Paul’s conviction that chemistry held the key to antisense, defined my role and work in his lab: addressing the limitations of PO oligonucleotides by exploring alternative backbone chemistries. Considering my experience in synthesizing oligonucleotides using phosphotriester and amidite chemistry from working in Mike’s lab, we decided that my first goal in Paul’s group was to make antisense oligonucleotides containing methylphosphonate (PM) linkages (Figure ). To produce these on our automated Biosearch 8600 synthesizer, we first had to develop the chemical synthesis of the corresponding nucleoside methylphosphonamidite monomers, since this had not been described at the time. This was also my first experience working with an automated synthesizer!

1.

Gapmer antisense oligonucleotide and chemistry. Gapmers consist of a PS ODN segment flanked by 2′-modified ORN sequences at both termini. The central PS ODN section provides resistance to exonucleases and forms a DNA/RNA heteroduplex upon hybridization with the target RNA. The surrounding 2′-modified ORN regions confer stability against both exo- and endonucleases, increase affinity for the target RNA, and modulate the polyanionic character of PS linkages. These ORN segments may contain PO, PS, or mixed backbones and widely used 2′-substitutions (R) such as 2’-O-methyl and 2’-O-methoxyethyl. In our 1992 patent application, the examples encompassed broader classes of 2′-substitutions, including 2’-alkyl, 2’-O-aryl, 2’-O-allyl, and 2’-O-amino ribonucleosides. PS backbone linkages in both ODN and ORN sequences may be present as random mixtures of the Rp or Sp diastereomers, incorporated in stereodefined form, or introduced as phosphorodithioates.

We initially made PO oligonucleotides with one or two PM linkages at each terminus. A single PM linkage did not improve nuclease stability, but incorporating two at each end significantly slowed degradation. We then synthesized a series of antisense candidates containing various combinations of PO and PM linkages, as well as fully PM-modified oligonucleotides. Increasing the number of PM linkages improved nuclease resistance, but reduced affinity and caused solubility problems. In cell culture studies, fully PM-modified oligonucleotides were more potent than pure PO oligonucleotides, whereas mixed PO/PM oligonucleotides were not, despite their enhanced nuclease stability. This suggested that even internal PO linkages remained vulnerable to cellular nucleases.

In parallel, we optimized the experimental conditions for making antisense oligonucleotides containing other phosphate-backbone modifications. Using H-phosphonate chemistry, we synthesized intermediates that could be converted to either phosphorothioate (PS) linkages with a sulfurizing reagent, or phosphoroamidates via amines during the last step of synthesis. We also prepared several phosphoramidate analogues, including morpholidate (PN), n-butyl amidate (PB), and piperazidate (PP) (Figure ). At the time, preparative HPLC columns were not yet readily available, so we purified the oligonucleotide analogues using thin-layer chromatography (TLC) followed by desalting on Sephadex columns. When Paul saw the TLC setup, he recounted how Kurt Randerath had developed it in his lab in the 1960s and pulled one of his old lab notebooks from the shelf to show me his handwritten notes about these experiments – I really enjoyed hearing Paul’s reminiscences about the foundational work done in his lab.

With this expanded panel of analogs in hand, we began a systematic comparison to understand how each modification shaped the behavior of the antisense candidates. We synthesized ODNs of differing lengths and sequences, including several homo-oligomers (e.g., A[15] and C[20]). Because some experiments required larger quantities of specific oligonucleotides, I set up an Omnifit assembly for manual synthesis on self-packed controlled-pore glass support columns, similar to the system I had used in Mike’s lab. The drawback of this setup was the time commitment: once synthesis of a 20-mer was initiated, it demanded more than 15 h of continuous hands-on work.

The PN-, PB-, and PP-modified oligonucleotides we produced in this way showed increased nuclease stability to varying degrees, but all were poorly soluble in PBS, similar to the PM modification. Melting temperature analysis also revealed that all three phosphoramidate analogues had even lower affinities than the PM oligonucleotides. In contrast, the PS analog was highly soluble in PBS, resistant to nucleases, and exhibited a melting temperature higher than that of PM, albeit slightly lower than the corresponding PO oligonucleotide.

When we tested the various backbone-modified antisense oligonucleotides in cell culture against HIV-1, all showed improved potency relative to the unmodified PO controls, with the PS by far the most effective. However, the PN, PB, and PP analogues caused marked cytotoxicity at higher concentrations, limiting their usefulness despite their increased nuclease stability.

The most surprising observation in these experiments was that even PS oligonucleotides with mismatched sequences or simple homo-oligomers such as A[15] or C[20] showed inhibitory activity when added to cells at the same time as HIV. Similar sequence-independent inhibition had also been reported by Samuel Broder’s lab. This was puzzling at the time, but subsequent experiments showed that the mismatched oligonucleotides lost inhibitory activity when duplexed, suggesting that the single-stranded form was required for this effect. We concluded that the sequence-independent inhibitory effect was due to the polyanionic nature of the PS backbone resulting in nonspecific binding to either viral or cellular surface proteins and subsequent interference with viral entry. In chronically infected cells, however, PS oligonucleotides showed sequence-dependent inhibitory activity.

These studies showed us that nuclease stability and affinity are critical, but solubility is important, too. Compared to PO, PS antisense nuclease stability and potency increased, but affinity decreased. For PM, PN, PP, and PB modified oligonucleotides, nuclease stability significantly increased, but potency and affinity decreased compared to PS ODNs.

We were uncertain whether these effects arose from intrinsic chemical properties or from differences in how efficiently the oligonucleotides entered cells. To investigate this, we worked together closely: Mary cultured the HeLa and 3T6 cells, and Paul treated them with the fluorescently conjugated modified oligonucleotide analogues I synthesized. The two of them always carried out their experiments together, growing the cells and examining them under the microscope. Working with Mary was always a pleasure. She was a warm and thoughtful presence in the lab, and often left a cookie on my table when she returned from lunch, a small gesture that always made me feel at home in the lab.

For PO ODNs, they observed only faint fluorescence in the cytoplasm and nucleus. In contrast, both PS and PM ODNs produced intense fluorescence, which could have been due to increased nuclease stability and persistence in the intracellular compartment, so these studies did not explain the differences in potency between them. I only recently realized that the paper presenting these results is the only one I coauthored with both Mary and Paul!

Yet even as we accumulated data on stability, affinity, solubility and uptake, a deeper question began to surface. How did antisense oligonucleotides exert their effects inside the cell? At the time, Paul had a simple diagram on the blackboard in his office, showing an antisense oligonucleotide hybridizing to its target RNA, blocking ribosome migration, and thus causing translation arrest. He had first proposed this mechanism in his 1978 papers, and I can still picture him standing there with a piece of chalk in his hand, explaining how he had arrived at this explanation with the clarity and directness that characterized his approach to science. What strikes me now is how that sketch captured our understanding of antisense action at the time, a model centered on physical obstruction of cellular processes rather than the RNaseH-based mechanism, as we know today. It was an elegant and straightforward explanation.

Our results clearly showed that turning this model into a practical antisense strategy would require the right chemistry. At that point, only the PS modification remained a viable option, so improving the affinity of PS modified oligonucleotides became the next logical step. One possibility was to combine PS backbone modifications with oligoribonucleotide (ORN) instead of oligodeoxyribonucleotide (ODNs) sugar rings in our antisense candidates. At the time, synthetic methodologies for ORNs were still developing, and several groups were working to optimize 2′-protecting groups, most prominently Kevin Ogilvie. We chose the H-phosphonate approach and used the t-butyldimethylsilyl group for 2′-protection, but these monomers were not commercially available, so we had to synthesize all four ourselves.

Jin Yan Tang from Marvin Caruther’s lab joined us, and together we synthesized PO ORNs ranging from 6-mers to 63-mers for a series of biochemical studies. Soon after, Valeri Metelev from Zoya Shabarova’s lab at Moscow State University and Konrad Misiura from Mike Gait’s lab also joined Paul’s lab and worked closely with me to expand the ORN and ODN program. The PS ORNs we synthesized exhibited higher affinity than PS ODNs while maintaining comparable nuclease stability, yet in HIV-1-infected cells, they showed reduced antisense activity.

During this time, Dan Brown from Cambridge, UK, also spent two years in the lab, after I requested Paul to invite him to discuss chemistry-related topics. Paul had known Dan since visiting Alexander Todd’s laboratory at Cambridge (UK) in the early 1950s, where he also became friends with Har Gobind Khorana, the very same Nobel laureate whose photograph in my schoolbook had inspired me to pursue chemistry. Mike Gait had worked with Khorana as a Post-Doc in the 1970s and John Goodchild had previously been Dan’s Post-Doc. While Dan was visiting, we held several meetings with him and Dr Khorana to review the data, discuss interpretations and consider the path forward. It was such a valuable learning experience to be in their company and to see that science need not be complex. Both Gobind and Dan encouraged me to write a book on oligonucleotides, anticipating that demand would grow as antisense research expanded. I eventually edited two volumes. Dan contributed the introductory chapter on the history of oligonucleotide synthesis and Gobind wrote a short foreword. He kindly asked me several times to call him Gobind, but out of respect I continued to address him as Dr. Khorana.

We also synthesized 2′-O-methyl modified ORNs with PS backbone linkages (Figure ), which demonstrated further increases in nuclease stability and affinity compared to PS ODN antisense, but these improvements did not translate into greater antisense potency. Taken together, our studies with these various chemical modifications left us without clear answers: 2′-O-methyl ORN PS antisense molecules were more stable and bound their targets with higher affinity, yet they were less potent than PS ODNs. Others soon reported similar findings, including Shibahara, reinforcing our conclusion that increased affinity alone was insufficient to enhance antisense activity. This challenged our assumptions about what governed antisense efficacy. It was at this point that Paul asked the question that cut to the heart of the problem: “Is antisense as simple as written on the chalkboard?”

Is There Another Mechanism at Play?

The simplicity of Paul’s chalkboard sketch no longer explained everything we were observing experimentally. The first indication that another mechanism might be involved was published in 1988, when Joe Walder reported that a PO ODN antisense duplexed with its target RNA was cleaved by RNase H in rabbit reticulocyte lysate. Similar RNase H activation by PS ODNs was soon reported by others while our studies were still in progress, although the extent of cleavage varied across assay systems. Because we were aiming for potential therapeutics, we chose to focus on human RNase H and therefore used nuclear extracts from HeLa cells in our assays. In collaboration with Thoru Pederson and Sandra Maynard, whose laboratories were located above Paul’s, we targeted a 514-nucleotide RNA using our various antisense analogs.

Several results emerged from these studies. First, we could not detect any target cleavage with PM, PP, and PN modified ODNs, suggesting that RNA/DNA duplexes formed by these analogues are not substrates for RNase H. Second, short 4- and 6-mer PO ODNs were unable to form stable duplexes in cell extracts and a 15-mer PO or PS ODNs were required for cleavage of the target RNA at the duplex site. However, cleavage was significantly reduced with PS ODNs, which we interpreted as a possible consequence of either the observed lower affinity or the replacement of one oxygen with a bulkier, more polarized sulfur atom. Another likely explanation was that PS-modified oligonucleotides contain a mixture of chiral Rp and Sp linkages and that RNase H is selective for one of these diastereomers, a reasonable assumption given the known stereoselectivity of many enzymes.

We had already shown that mixing chemical modifications could improve key antisense properties when we combined PS backbone modifications with 2′-O-methyl-modified sugar. Taking this further, we designed 15-mer mixed-phosphate ODNs, and PM ODNs with 2, 4, or 6 PO linkages in the center to test whether limited PO content could restore RNase H activity. This proved to be the case: PM-modified oligonucleotides that previously showed no RNase H cleavage demonstrated clear activity with 4 PO linkages and even stronger activity with 6. Taken together with the absence of duplex formation by short oligomers, these results indicated that although short PO sequences could not function by themselves, they could support RNase H cleavage when placed within an ODN that mediated duplex formation. An ODN with 6 PS linkages on the 3′-end and 9 PM modifications also served as an RNase H substrate, although its efficacy was reduced compared with a similar ODN containing a PO segment.

These studies provided us with insights into why PS ODNs exhibited potent antisense activity compared with the other analogues we had examined. Nuclease stability and affinity were important, but activation of RNase H and subsequent catalytic degradation of the target RNA was also a factor in potency. We then evaluated ODNs with various configurations of PS linkages flanked by PM linkages as antisense agents targeting HIV-1. These ODNs showed increased nuclease stability but reduced affinity relative to fully PS-modified ODNs, and in cell culture they did not demonstrate improved potency (Agrawal, unpublished data).

By this time, PS ODNs had become the choice as first-generation antisense and were being pursued by many other investigators as antisense agents, with a number of lead candidates in clinical development. We ourselves were developing a 25-mer PS ODN antisense targeting the gag region of HIV-1.

In the meantime, we continued our efforts to find more potent chemical modifications or combinations thereof, so we returned to the chalkboard to summarize our observations and determine the most promising direction. The notes on the chalkboard are replicated here in the top half of Table .

1.

Property	PO-ODN	PS-ODN	PN-ODN	PS-ORN	2′-O-subsituted PS-ORN	Gapmer
Internucleotide linkage	Ionic	Ionic	Nonionic	Ionic	Ionic	Ionic
Affinity to RNA	++++	+++	++	++++	+++++	+++++
Nuclease stability	-	+++	+++++	++	+++++	+++++
RNase H activation	+++++	+++	-	-	-	++++
Antisense activity	-	+++	++	+	++	++++
Post 1992 patent filing
CpG/immune activation	++	+++	-	+	-	-
Complement activation	-	++++	-	-	+	+
Prolongation of aPTT	-	++++	-		+	+
Protein binding	+	+++++	-	+++	+++	+++
Oral delivery	-	-	-	-	-	+

^aAntisense activity in this table refers specifically to RNase-H-dependent gapmer activity. Steric-blocking 2′-O-substituted PS oligoribonucleotides used for splice switching typically exhibit substantially higher activity than indicated here. ,

Try, Try, and Try Again, and at Some Point, You Will Succeed

Clearly, PS modifications on ODNs conferred both nuclease stability and RNase H activation, although their affinity could be improved. Fully PS-modified 2′-O-methyl ORNs provided greater affinity and increased nuclease stability, but lacked RNase H activation. We had tried to unify the positive characteristics of both modifications in one ORN antisense molecule, but our experiments combining PS and PM linkages now made it clear that a 4–6-mer PS or PO ODN segment was required for RNase H activity. Others had explored such hybrid oligonucleotides as a detection method for specific sequences or used combinations of PO ODN and 2′-O-methyl ORN without improvements in therapeutic antisense potency, , however, the combination of PS ODN with PS 2′-O-ORN segments was, at that time, unrecognized and unexplored.

We had also learned that, in vivo, PS ODNs are degraded primarily from the 3′ end, and that protecting the 3′ end therefore increases the amount of functional antisense in tissues. Thus, for our initial studies, we placed the PS 2′-O-methyl ORN segment on both the 3′ and 5′ ends of the PS ODN, only on the 3′ end, only on the 5′ end, or interspersed throughout.

All designs containing a segment of PS 2′-O-methyl ORN demonstrated increased affinity compared with pure PS ODN, and nuclease stability increased when the PS 2′-O-methyl segment was positioned at the 3′ end or at both ends. RNase H was activated by PS ODNs containing PS 2′-O-methyl ORN at both the 3′ and 5′ ends, as well as at the 5′ end alone. In cell culture studies, the construct with a central PS ODN segment flanked by PS 2′-O-methyl ORN at both the 3′ and 5′ ends showed the greatest improvement in antisense potency. (Figure )

2.

Gapmer antisense mechanism of action. A gapmer antisense oligonucleotide consists of a PS ODN segment (dark blue) flanked by 2′-modified ORN segments. Upon hybridization to the target RNA, the PS ODN segment forms a DNA/RNA heteroduplex that serves as a preferred substrate for endogenous RNase H, which cleaves only the RNA strand. After cleavage, RNase H releases the gapmer, allowing it to rebind additional target RNAs, and thereby sustaining a catalytic cycle. Depending on the length of the PS ODN segment and the kinetics of RNase H-mediated cleavage, the hybridized gapmer may also sterically block translation or initiation of translation.

Finally, we had created an antisense that was more potent than PS ODN!

When Paul heard the full account of how we had pieced together these seemingly disparate observations, he complimented me and shared a few stories about similar challenges from his early career. He told me that it was now time for me to begin publishing independently, so this became my last piece of work in his lab. Our friendship continued for many years until his passing, sustained by frequent visits, lunches and dinners full of scientific discussions, Paul’s characteristic one-liner remarks and his stories of how the science had evolved in the 1960s and 1970s. One particularly cherished memory is a reply to a paper I sent him in 2008 for publication in Proceedings of the National Academy of Sciences, just a year before his death, a letter that hangs framed in my office to this day. He began, “I regret having to send this communication,” and then recounted how, on the morning Fritz Lipmann learned he had won the Nobel Prize, a reporter mentioned that the Commonwealth Fund had just rejected one of his grant applications. In response, Lipmann turned to his secretary and dictated a telegram to the Fund: “You fools, I won the Nobel Prize.” Paul ended his note to me by adding “Anyway, I think your paper is an imaginative, superior one.” It was such a quintessential Paul moment - honest, wry and always encouraging.

But returning to the period when our hybrid antisense work first came together, the next step was clear. Rather than publishing a paper at that time, we filed a patent in July 1992 disclosing the antisense constructed from combinations of PS ODN and PS 2’-substituted ORN. This patent was assigned to WFEB (now UMass Chan Medical School) and exclusively licensed to Hybridon. We termed our novel design “hybrid antisense” as it united DNA and RNA within a single molecule. This configuration later became known as “gapmer”. The patent claims covered antisense constructs comprising PS ODN and PS 2′-substituted ORN segments, with broad coverage of 2′-modifications, including 2′-O-methyl and 2′-O-methoxyethyl. This filing resulted in several related patents. Our own observations remained in our lab notebooks until we published them in 1994 and 1996. In July 1993, nearly a year after our filing, Monia et al. published a paper describing similar concepts. In 2001, Isis Pharmaceuticals (now Ionis) entered into a licensing agreement with Hybridon (now Idera) to gain coexclusive rights to the gapmer antisense and other patents for use in creating antisense candidates and for sublicensing.

From the New Chalkboard Concept to the Clinic

A new model of antisense had emerged, but the challenges were far from resolved. Our hybrid antisense work drew little interest initially because attention remained fixed on the clinical development of PS ODN antisense. As we advanced our PS ODN candidate targeting HIV (GEM91) through preclinical studies and early trials, a series of issues became apparent (bottom half of Table .): inflammatory responses, , complement activation linked to the polyanionic backbone, pharmacokinetics and degradation profiles, , and immune effects that influenced both safety and activity. , Between the mid-1990s and early 2000s, several high-profile clinical studies using PS ODN antisense, including our own GEM91, were discontinued due to reporting similar observations, safety concerns, or lack of efficacy.

This was a dark period for the antisense field, as confidence in PS ODNs collapsed and many companies exited the field. In the aftermath, we refocused on our hybrid antisense designs, which were just beginning to be referred to as gapmers and soon became the preferred format for the second generation of antisense. Throughout the midnineties, we repeatedly returned to the bench to assess how the undesirable characteristics we had seen could be mitigated or improved in gapmer designs. − Our studies demonstrated how we could reduce inflammatory and immune responses, improve safety profiles, enhance in vivo stability, and lower complement activation. For example, incorporating PO linkages into the 2′-substituted ORN segment reduced protein binding and complement activation. We also investigated whether the increased in vivo persistence of gapmer antisense could enable both oral and rectal delivery.

Those experiments marked the beginning of a trajectory that would take decades to unfold. Steady refinement and clinical validation eventually carried the gapmer concept into therapeutic practice. As it turned out, Paul’s initial thoughts on how antisense works, as described in his 1978 papers and illustrated by the chalkboard sketch in his office, proved to be accurate for certain therapeutic applications. In the mid-1990s, Ryszard Kole contacted me to explore whether our PS-modified 2′-substituted ORNs could modulate splicing, and together we demonstrated the first proof-of-concept for splice modulation in intact cells. This work confirmed that physical obstruction of splice-recognition sites could redirect alternative splicing events, precisely as Paul had envisioned, and laid the foundation for later therapeutic applications such as Duchenne Muscular Dystrophy.

Today, an increasing number of approved drugs use either the gapmer or splice-switching oligonucleotide design. What began as a simple sketch on a chalkboard has, through persistence and iteration, become two widely adopted platforms that continue to shape modern antisense therapeutics. The principles established during this period - how chemical modifications, their precise placement, and overall oligonucleotide design influence the function of cellular proteins involved in the mechanism of action - also provided a solid foundation for the later development of small interfering RNA therapeutics. For readers interested in the more recent history of the antisense field, including the shift from purely medicinal chemistry optimization toward spatial designs to further improve efficacy, I recommend these citations.

The author declares no competing financial interest.