Making sense of antisense

HISTORY OF ANTISENSE TECHNOLOGY Knowledge of nucleic acid structure and function, including DNA hybridization and mRNA production, has opened the door to the concept of the antisense mechanism in disease therapeutics as well as a specific investigational tool in animal and in vitro cell studies. In 1978, Zamecnik and Stephenson (1) found that a 13 nucleotide long strand of oligodeoxyribonucleotide complementary to a sequence in the respiratory syncytial virus genome had potential as an antiviral agent. This concept was refined progressively during the late 1970s and 1980s by Ts’o et al (2), who experimented with phosphate backbone modifications to develop the stability necessary for antisense molecules to be highly active in vitro and in vivo because native nucleic acids are rapidly degraded by nucleases. In the mid to late 1980s, phosphorothioate backbone modification, investigated as initial evidence of in vitro activity against a spectrum of viral and mammalian targets, was developed by using several antisense compounds, including methylphosphonates and phosphorothioates. In the 1990s, in vivo demonstrations of efficacy in animal models and preliminary evidence of clinical activity in a spectrum of diseases have been obtained with phosphorothioate antisense molecules, and the first new drug application (NDA) for an antisense drug, a phosphorothioate deoxynucleotide administered intravitreally for cytomegalovirus (CMV) retinitis, has been filed. Despite these early successes with phosphorothioate deoxynucleotides, extensive experimentation with heterocycle, sugar and backbone modifications has been and is continuing in an effort to develop antisense drugs with increased binding affinity to the target message, increased nuclease resistance, improved pharmacokinetic properties, improved cellular and intracellular distribution, decreased or altered nonspecific binding to proteins and perhaps oral bioavailability. Over this same time period, synthetic methods and capacity have improved substantially, and economically feasible, commercial scale synthesis is now achievable.

the DNA template (3).This process permits transcription of one strand of DNA into a 'sense' strand of pre-mRNA.Pre-mRNA, by excision of introns and splicing, is further processed into mRNA.mRNA processing then continues outside of the nucleus within the cell cytoplasm.Here, mRNA translation or degradation results in either protein production or mRNA catabolism, respectively.
A typical message contains one thousand to several thousand bases, most of which comprise the coding regions or exons of the mRNA.The 'head' (5¢) end of the mRNA usually contains several hundred nucleotides that are not translated but appear to play an important role in a variety of events, including control of translation, ribosomal assembly, mRNA nuclear transport and message stability.The 'tail' (3¢) untranslated region (UTR) is the poly A tail and serves to orient the mRNA.The 3¢ UTR also appears to be important in regulating message metabolism and stability.Theoretically, any region of the message could be a fruitful target for an antisense drug.From a statistical perspective, an oligonucleotide of 11 to 15 bases is required to assure uniqueness to a specific human message, and 15 to 19 bases to assure uniqueness to a DNA target (4).From the pragmatic standpoint, it appears that, at least with present compounds, a minimum sequence of six to seven bases is required to provide adequate binding affinity to target and/or catalytic activity to provide reliable pharmacological activity.Increasing the number of bases increases the binding affinity and the likelihood of absolute specificity to the target message, but also increases the complexity and expense of synthesis, and perhaps nonspecific protein binding.With present compounds, approximately 15 to 25 bases seems to provide the optimal balance among competing factors, and all antisense drugs used in the clinic are on the order of 20 bases in length.The construction of antisense molecules is mainly empirical; however, computer programs are used to help in difficult areas, such as for palindromic sequences.
Antisense drugs, depending primarily on their chemical class, are thought to work by two different general mechanisms (Figure 2).First, they can bind to DNA or mRNA and physically inhibit any of the key steps in transcription or translation, respectively.This is the theoretically less favoured mechanism because it uses one molecule of antisense for each mRNA or DNA molecule, and requires a relatively high binding affinity for the target message.More efficient use of antisense is provided by exploitation of catalysis, whereby one antisense molecule, specifically bound to its target message and then released, acts as a catalyst for the degradation of many mRNA molecules by RNases such as RNase H.This latter concept has been the theoretically preferred mechanism of action, and one on which most current antisense activity is based.Catalytic mechanisms depend on recognition of the particular antisense chemistry by the enzyme in question and are, therefore, restricted to certain chemical classes of antisense molecules, depending on the catalytic enzyme involved.
Compared with traditional drugs, antisense molecules, containing approximately seven to 30 bases or nucleotides, are relatively large, with molecular weights between approximately 3000 and 12000 kDa.These molecules require parenteral delivery to achieve significant systemic bioavailability.Most antisense drugs in clinical studies are phosphorothioate oligodeoxynucleotides (ODNs), differing from

IN VITRO STUDIES OF ANTISENSE
The activities of different antisense compounds employing both catalytic and physical mechanisms have been demonstrated in vitro in a spectrum of cell types against a variety of viral and mammalian targets (7,8).Of these compounds, the phosphorothioates are the most studied and understood.Early investigations primarily involved compounds with phosphate backbone modifications, such as methylphosphonates, phosphorotriesters and phosphorothioates, but recent demonstrations have used heterocycle or sugar modifications.In addition, hybrid or chimeric molecules, using two or more chemical modifications, have been investigated (9,10).Uptake of oligonucleotide by a broad spectrum of cell types has been demonstrated in vitro and in vivo.In general, in the in vitro situation, unlike in vivo, lipid formulations have been required to optimize pharmacological activity.
In theory, an effective antisense drug can be developed against any known molecular target.All available antisense drugs target either mammalian or viral RNA.Determining the best 15 to 25 base sequence to target in a given message is largely an empirical exercise, and effective sequences have been found in virtually every region of individual messages from the 5¢ cap to the 3¢ UTR.In developing a drug candidate, typically some 30 to 50 separate sequences spaced along the entire length of the message, lacking in selfcomplementarity and, therefore, thought to lack secondary structure, are manufactured and screened for in vitro activity before selecting the optimal candidate for further study.

ANIMAL STUDIES
Successful in vivo studies of antisense against a broad spectrum of viral and mammalian molecular targets have been conducted, primarily in rodents (7).Cardiovascular targets have included a spectrum of molecules involved with smooth muscle proliferation (eg, c-myb, cdc-2 kinase, proliferating cell nuclear antigen, cyclin B1); cancer targets have included a spectrum of signalling pathway, transcription factor and antiapoptotic molecules (eg, N-myc, c-myb, protein kinase C-alpha [PKC-a], Ha-ras, C-raf kinase and bcl-2); inflammatory targets have included a spectrum of cell adhesion molecule, transcription factor and cytokine receptor molecules (eg, intercellular adhesion molecule-1 [ICAM-1], type I interleukin-1 receptor and nuclear factor-kappaB p65); and neurological targets have included a spectrum of neurotransmitters and receptors with intracerebral administration (eg, dopamine receptors, N-methyl-D-aspartate receptor, opioid receptors, hormone receptors [corticotropinreleasing hormone, progesterone], oxytocin and substance P receptor).In animals, as in vitro, the phosphorothioate class of antisense molecules, first synthesized in 1969 (11), are the most studied and best understood.Phosphorothioate molecules are typically 17 to 25 bases in length to exert efficacy in vivo; however, longer molecules have been studied (3).
The in vivo toxicity and pharmacokinetics of phosphorothioates directed against a variety of molecular targets have been studied extensively in animals -to doses greater than 500 mg/kg in mice and 100 mg/kg in monkeys (12).The toxicological profile of these compounds appears to be class specific and largely independent of base sequence and route of administration, and the pharmacokinetic profile is consistent for a given route of administration.Toxicity resulting from intended or exaggerated pharmacology (sequence dependent) has not been reported.This is probably due to multiple factors, including the high or absolute specificity of these molecules, the careful choice of molecular target, the presence of alternative pathways or molecules for essential functions in mammalian cells, downregulation rather than total ablation of target protein synthesis, and the choice of a viral-specific target in the applicable instances.Evidence of antigenicity has not been observed with repeated dosing for up to six months administered by the subcutaneous and intravenous route.
In rodents, phosphorothioates produce a nonspecific immune stimulation characterized by a mixed mononuclear infiltration in multiple organs and hepatotoxicity at very high exposures.Transient and modest decreases in platelet counts have also been observed at higher doses, and this appears to be due to a temporary sequestration effect.Marked immune stimulation appears to be unique to rodents and includes an apparent component of nonspecific B cell mitogenesis.In primates, toxicity has been primarily limited to transient prolongation of the intrinsic pathway of coagulation, as assessed by the activated partial thromboplastin time (APTT), and alternative pathway complement activation with dose regimens that produce peak plasma drug levels of greater than 50 µg/mL.Nonspecific protein binding appears to be responsible for these effects; inhibition of APTT appears to be mediated through the inhibition of intrinsic tenase (factors IXa, VIIIa, phospholipid and calcium) activity (13) and complement activation through interaction, either directly or indirectly, with factor H, an inhibitory protein of the alternative pathway (14).
Phosphorothioate ODNs undergo low avidity, high capacity protein binding in plasma, and this likely accounts for the very small percentage of drug excreted in urine.Although in animals (and in humans) these compounds are cleared from plasma rather rapidly (approximate half-life of 30 to 60 mins), they are distributed extravascularly and taken up by cells, with catabolic half-lives ranging from one to five days, depending on the tissue.These compounds are catabolized by normal nucleotide degradation pathways.Principal distribution is to liver, kidney and spleen, but uptake in vivo by a broad spectrum of cells and tissues has been demonstrated (5,15).

HUMAN STUDIES OF ANTISENSE
Twelve antisense compounds, all with uniform phosphorothioate backbones, are currently in clinical trials (Table 1).Six of these drugs are in development for cancer (targeting c-myb, bcl-2, protein kinase A [PKA], PKCa, C-raf kinase and Ha-ras, respectively), four target viruses (CMV, human immunodeficiency virus [HIV]), one targets c-myc for the prevention of smooth muscle proliferation and restenosis following balloon angioplasty, and one targets an inflammatory disease target, ICAM-1.Eight of these antisense drugs are uniform phosphorothioate ODRs and four are hybrids of deoxy and 2¢ substituted nucleotides.In the hybrid oligonucleotides, several of the nucleotides at the 5¢ and 3¢ ends have been modified to contain sugars with a 2¢ substitution (either O-methyl or O-methoxyethyl) to enhance nuclease resistance and even offer the potential for low level oral bioavailability.All of these drugs are being administered parenterally, with the exception of GEM92, targeting HIV.Preclinical studies have suggested the possibility of low level oral bioavailability for this compound, and this concept is being tested in a combined oral and intravenous phase 1 study in patients with acquired immunodeficiency syndrome (AIDS).
Antisense has been tried with varying degrees of success in viral diseases.ISIS 2105, directed against human papilloma virus 6 and 11, showed evidence of activity against venereal warts when injected intradermally in phase 2 studies, Can J Gastroenterol Vol 13 No 9 November 1999 but was discontinued from development because of the apparent need for frequent injections and the lack of a formulation that would maintain skin levels.GEM91, targeting the gag message of HIV, has been discontinued from phase 2 studies, in which inconsistent evidence of antiviral activity was demonstrated, in order to focus on development of GEM92, a hybrid deoxy/2¢ O-methyl phosphorothioate with the same molecular target.This compound offers the potential for low level but sufficient oral bioavailability for clinical activity (16); this concept is being tested in a combined oral and intravenous phase 1 study in AIDS patients.ISIS 2922 (fomivirsen) has completed phase 3 trials for CMV retinitis in AIDS patients, and an NDA has been filed.This antisense molecule targets the major immediate early region of the human CMV genome (Figure 3) (17).Studies of intravitreal administration of this drug with once weekly dosing for three weeks induction followed by maintenance doses every other week in patients with CMV retinitis who had failed approved therapy with foscamet and ganciclovir revealed good response rates, with some responses exceeding 50 weeks (18).In patients with previously untreated CMV retinitis, phase 3 results comparing observation with intravitreal treatment demonstrated a median time to progression of 71 days in the fomivirsen group versus 13 days for the deferred treatment group (P=0.0001)(19).Trials in patients with advanced disease revealed similar median times to progression with either a once weekly induction for three weeks followed by every other week maintenance or induction on days 1 and 15 followed by maintenance injections every four weeks; the median time to progression was 90 days.In addition, fomivirsen was well tolerated when coadministered with oral ganciclovir.ISIS Pharmaceuticals (Carlsbad, California) has begun a phase 1-2 study with a second generation chemistry oligonucleotide, ISIS 13312, a deoxy/2¢ O-methoxyethyl hybrid, targeting the same sequence in the CMV genome as fomivirsen.Although fomivirsen was well tolerated, it is expected that a second generation compound will demonstrate even better local tolerance, with the potential for even less frequent intravitreal dosing.Hybridon (Cambridge, Massachusetts) has also initiated phase 1-2 studies, both intravitreally and systemically, with a second generation (a deoxy/2¢ O-methyl hybrid) anti-CMV antisense drug, GEM132.

Figure 3) Antisense has been tried with varying degrees of success in viral diseases. Of these, ISIS 2922 (fomivirsen) has completed phase 3 trials for cytomegalovirus (CMV) retinitis in acquired immunodeficiency syndrome. This molecule targets the major immediate early region of the human CMV genome. This molecule is specific to the CMV virus and not a generic mechanism of action of virally directed antisense molecules
Most of the antisense compounds in clinical trials are being developed for oncological indications (20,21).ISIS 3521/ CGP64128A is an antisense inhibitor of PKC-a expression.PKC-a is a member of a multigene family of signal transduction proteins that participate in the regulation of information flow in and out of cells and modulate cellular responses to environmental stimuli.PKC-a has been implicated in the growth of a spectrum of solid tumours.ISIS 3521 has demonstrated activity in a range of human tumour xenografts in mice (22).PKC-a, in conjunction with the effects of ras-GTP, activates C-raf-kinase from its inactive form (Figure 4).In the two phase 1 trials of ISIS 3521, patients with a variety of cancers, including pancreas, colon, stomach, lung, breast and ovarian, received treatment in cycles of 21 days by continuous or three times weekly 2 h infusions followed by seven days of rest (23,24).ISIS 3521 demonstrated good tolerance with disease stabilization, partial or complete response of several tumours, including three ovarian, two lymphoma and one nonsmall cell lung cancer for up to 18 months.Phase 2 single agent and combination studies against a spectrum of solid tumours, including ovarian, prostate, breast, brain, colon and lung cancers and melanoma have been or are in the process of being initiated.
ISIS 5132/CGP69846A is complementary to a region of the 3¢ UTR of human C-raf kinase (Figure 4) (21).C-raf kinase and ras are both serine-threonine kinases that participate in the mitogen activated protein kinase signal transduction pathway.C-raf kinase is activated by ras, which itself is a GTP-regulated molecular switch that has been implicated in up to one-third of human cancers.C-raf-kinase participates in the transduction of signals from cell surface receptors for, among others, several growth factors and cytokines, and the T cell receptor.Signals are transmitted to downstream phosphorylators, or kinases, ultimately producing activation of transcription factors.In this way, C-raf-kinase participates in the regulation of cell proliferation and differentiation.ISIS 5132 has also completed two phase 1 studies in which the drug was administered as in the ISIS 3521 trials (25,26).Several tumours, including one colon, one ovarian, one pancreatic and two renal cell cancers, were stablized for up to 10 months.Phase 2 single and combined agent trials in a spectrum of solid tumours are also commencing with this compound.In addition, a phase 1 study with ISIS 2503, targeting Harvey-ras (Ha-ras) has begun.
Results of a phase 1 study with G3139, an anti-bcl-2 Can J Gastroenterol Vol 13 No 9 November 1999 749 antisense compound, in nine patients with recurrent non-Hodgkin's lymphoma, were recently reported (27).Overexpression of bc1-2 produces resistance to programmed cell death or apoptosis.G3139 was administered by continuous subcutaneous infusion for two weeks in increasing doses from 4.6 to 73.6 mg/m 2 of body surface area.The bcl-2 antisense was well tolerated other than causing local inflammation at the infusion site.In two patients, computed tomography scan revealed a reduction in tumour size (one minor and one complete response).The number of circulating lymphoma cells decreased in two patients, and serum lactate dehydrogenase levels fell in four patients.A c-myb-directed antisense drug, LR-3001, is in phase 1 studies in patients with chronic myeloid leukemia (CML) and acute myeloid leukemia (AML).c-myb is a transcription factor that appears to be preferentially expressed in hematopoetic cells.LR-3001 targets codons 2 through 9 of the human message.LR-3001 is being used in the ex vivo purging (24 to 72 h) of CD34+ marrow stem cells in CML patients undergoing autologous bone marrow transplantation.This compound is also being administered systemically by sevenday continuous intravenous infusion to patients with advanced CML and AML.In addition, a second generation (hybrid) antisense drug directed against PKA has begun phase 1 studies.The R1-a subunit of PKA is overexpressed in a variety of cancers and is associated with rapid cellular growth.

PHASE 1-2 PILOT STUDY OF ANTISENSE TO ICAM-1 IN CROHN'S DISEASE
We completed a double-blind, placebo controlled, randomized pilot study of antisense to ICAM-1 (ISIS 2302) in Crohn's disease (28).This protocol, designed as a phase 2 study, was conducted in 20 patients between the ages of 18 and 80 years with moderately active Crohn's disease (Crohn's disease activity index [CDAI] 200 to 350) despite stable background doses of steroids (prednisone 40 mg/day) with or without 5-acetylsalicylic drugs.Patients received 13 doses of antisense or placebo by 2 h intravenous infusion.Corticosteroid dosages were kept stable for the 26-day infusion period, and then were adjusted by the investigator according to blinded clinical judgement.Four patients each were assigned to the 0.5 and the 1 mg/kg dose groups, and the remaining 12 patients were assigned to the 2 mg/kg group.All patients were followed for six months.At the end of the treatment period, seven of 15 (47%) antisense-treated patients and one of five placebo-treated patients, a patient already in remission at baseline, were in remission by clinical criteria (CDAI less than 150).At the end of the study, five of the seven ISIS 2302-treated remitters were still in remission.In addition, a significant difference in corticosteroid usage between the antisense and placebo-treated patients was observed.No clinically significant adverse events were reported, and there was no effect on routine laboratory safety indexes.As expected, transient elevations of APTT, of approximately 3, 5 and 10 s at respective dosages of 0.5, 1 and 2 mg/kg, were seen.A pivotal quality trial in 300 steroid-dependent Crohn's patients is underway to assess definitively the safety and efficacy of this compound.

SUMMARY
Since the identification of the double-stranded DNA helix by Watson and Crick (29) in 1953, the gene has been a potential specific therapeutic target in human disease.Although DNA is an attractive target site for drugs, its use as a drug itself has been a relatively recent concept.Researchers are more familiar with the use of antisense gene regulation in bacteria.These organisms use promoter region-specific antisense RNA to regulate plasmid, copy number and membrane proteins (30).Antisense technology represents a novel in vitro and in vivo approach to cell biology, and, potentially, clinical therapeutics.The development of antisense has gone from the early methylphosphonate oligonucleotides to the current clinical studies using first and second generation phosphorothioate antisense to a wide variety of molecular targets.There are currently 12 antisense compounds in clinical development for a spectrum of human diseases, including malignancies, viral diseases and inflammatory diseases.Tolerability of these compounds has been excellent, and the first NDA with an antisense drug (fomivirsen for CMV retinitis) has been filed.In addition, suggestions of clinical effect in phase 1 to 2 studies have been reported in patients with a spectrum of malignancies using antisense to bcl-2, PKC-a and C-raf kinase, and in an inflammatory disease (Crohn's disease) with an antisense to ICAM-1.Antisense technology offers laboratory and clinical opportunities to study specific gene targets.Specificity of this group of molecules provides a unique and novel opportunity to study and treat human diseases.

AntisenseFigure 4 )
Figure 4) A number of antisense molecules are being developed for oncological indications.These include ISIS 3521, targeting protein kinase C-alpha (PKC-a) expression, ISIS 5132 targeting human C-raf kinase and G3139, targeting bcl-2 in non-Hodgkin's lymphoma.Potential molecular targets in this pathway are illustrated.The son of sevenless (SOS) is a ras nucleotide exchanger that is brought through the cell membrane by the GRAB-2 (GRB-2) adapter protein.DAG Diacyl glycerol; MAPK Mitogen activated protein K; MEK Map erk kinase