This Article Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Bennett, M. R. Articles by Schwartz, S. M. Search for Related Content PubMed PubMed Citation Articles by Bennett, M. R. Articles by Schwartz, S. M. Pubmed/NCBI databases Medline Plus Health Information Angioplasty Vascular Diseases

(Circulation. 1995;92:1981-1993.)
© 1995 American Heart Association, Inc.

Articles

Antisense Therapy for Angioplasty Restenosis

Some Critical Considerations

Martin R. Bennett, MD, PhD; Stephen M. Schwartz, MD, PhD

From the Department of Pathology, University of Washington, Seattle.

Correspondence to Dr M.R. Bennett, Unit of Cardiovascular Medicine, University of Cambridge School of Clinical Medicine, Department of Medicine, Level 5, Addenbrooke's Hospital, Hills Rd, Cambridge CB2 2QQ, UK. E-mail mrb@mole.bio.cam.ac.uk.

	Introduction

Top Introduction Pathogenesis of Angioplasty... Antisense: A Focus on... Promise of the Antisense... References

 
Percutaneous transluminal angioplasty is now a well-established and frequently performed procedure that has an initial success rate in reestablishing arterial patency of >95%.^{1 2} Although good symptomatic improvement occurs in the majority of cases, the procedure is complicated by restenosis in >30% of cases.^{3 4} Despite the apparent success of several therapeutic modalities in animal models, attempts to use pharmacological therapy to prevent restenosis in the clinical setting have not been successful. This failure has prompted research into alternative forms of intervention, including the use of antisense oligonucleotides therapeutically targeted to genes believed to be critical for the pathogenesis of restenosis.

The rationale for the use of antisense oligonucleotides to prevent restenosis is twofold. First, the prevailing view is that restenosis is the end result of a reactive proliferation of cells of the vessel wall after angioplasty. Thus, it follows that an agent that suppresses cell proliferation may suppress restenosis. Second, antisense agents have been used extensively to analyze genetic events associated with cell proliferation and the cell cycle (review in References 5 and 6). When any cell replicates, there is a characteristic sequential activation of a cascade of genes.^{7 8} This cascade of gene activation is also seen as cells are induced to proliferate after arterial injury.^{9 10 11 12} Because antisense agents can suppress the expression of genes associated with cell replication, the use of these agents to block cell proliferation after angioplasty is an attractive concept. Several studies have attested to the efficacy of antisense oligonucleotides directed at cell-cycle proteins in preventing neointimal formation after injury in animal models. The success of these animal studies has spawned widespread interest and enthusiasm for the use of antisense agents to prevent human restenosis. With that in mind, we review the critical issues that may determine whether or not an antisense strategy is likely to be successful in preventing human restenosis. The issues to be considered are as follows. (1) Is replication a critical step in restenosis? (2) Are antisense agents truly specific for their putative targets? (3) What factors determine the specificity and efficacy of an antisense agent? (4) Are side effects likely to be manifested as problems in clinical toxicology?

	Pathogenesis of Angioplasty Restenosis

Top Introduction Pathogenesis of Angioplasty... Antisense: A Focus on... Promise of the Antisense... References

 
Before discussing therapy, we need to agree on what we mean by restenosis. When an artery is dilated by angioplasty there is an "initial gain" in lumen size. Restenosis can best be defined as a "loss of gain,"¹³ that is, a late return of the vessel lumen to a size approaching that upon initial dilatation. It is important to note that this definition describes restenosis in terms of lumen size but does not attempt to identify the mechanisms involved in the changes in lumen size at angioplasty or in restenosis. Indeed, the mechanism involved in increasing the lumen size at angioplasty, the "acute gain," is still not entirely clear. Some of the most illustrative findings on the effects of angioplasty on the vessel wall have come from recent ultrasound studies that suggest that only a small amount of actual plaque mass is lost from the lesion site. Rather, most of the acute gain appears to be due to fracture and compression of the plaque, with fractures of the internal elastic lamina and overstretch of the vessel.^{14 15 16}

Next, it is essential to define "success" in the context of restenosis. After angioplasty, about 70% of patients have a persistently dilated vessel that approaches the desired lumen size. We can call this persistent dilatation "success" (Fig 1). In contrast, we can define loss of gain as "failure" and divide it into early and late forms. Early failure occurs when the lumen is occluded by a thrombus or by rapid recoil of the stretched vessel.^{13 17} These processes occur within hours or at most within a few days after angioplasty and are generally considered not to be part of restenosis. However, it is very important to remember that subclinical early recoil, combined with the mechanisms discussed below, may also be an important contributor to what appears to be late loss of gain ("Late Failure A" in Fig 1).

View larger version (44K): [in this window] [in a new window]   Figure 1. Flow chart shows possible outcomes after angioplasty and mechanisms responsible for restenosis. The events leading to both early and late failure after angioplasty are described in the text. It is important to note that an increase in neointimal mass may not necessarily cause restenosis if sufficient remodeling of the vessel occurs, and remodeling also may result in restenosis with minimal increase in neointimal mass. Inset shows desired lumen size.

The two possible mechanisms for late loss of gain are also shown in Fig 1. The first mechanism to consider is remodeling. Remodeling is a normal process that vessels use to maintain an appropriate lumen size or caliber, particularly in response to changes in blood flow.¹⁸ In early atherosclerosis, there is dilatation of the affected vessel.¹⁹ This initial dilatation may be analogous to remodeling seen after physiological changes in blood flow. However, when atherosclerosis becomes severe, the lumen size appears to be reset to an inappropriate caliber.¹⁹ Therefore, in a real sense, the goal of angioplasty is to prevent the vessel from healing and restoring itself to the inappropriate caliber after dilatation. Nonatherosclerotic vessels can remodel sufficiently to accommodate extensive amounts of intima.¹⁹ Thus, we might achieve success despite intimal hyperplasia if the vessel were somehow able to restore itself to a normal caliber by remodeling. Equally, remodeling itself after angioplasty may cause restenosis without increasing vessel wall mass.^{20 21} Unfortunately, virtually nothing is known about the molecular mechanisms involved in remodeling. As a result, to the extent that remodeling is critical to restenosis, we lack clear pharmacological strategies.

The second mechanism to consider is neointimal formation. When vessels are injured by any of a variety of processes, they respond by forming a new layer of intima.²² This mass of neointima can narrow the lumen. This would be an especially important mechanism if it were superimposed on an acute but less than critical extent of elastic recoil, as already noted. The extent to which neointimal formation will narrow the lumen is, of course, dependent on remodeling ("Late Failure B" in Fig 1). If remodeling restores the vessel wall to its preangioplasty dimensions, restenosis could occur without any intimal hyperplasia ("Late Failure C" in Fig 1). On the other hand, some degree of intimal hyperplasia may be tolerated if remodeling permits some compensatory dilatation ("Success" in Fig 1).

Neointimal formation is the result of cell migration from the intima or media, followed by cell proliferation and connective tissue formation. Although each process may contribute to neointimal mass, cell proliferation is the major source of smooth muscle cell accumulation in the neointima in the most often studied animal model, that of balloon injury to the rat carotid artery.²² Cell proliferation is a dramatic event in the injured rat vessel, in which three generations of replication occur within 2 weeks and can more than double the mass of the vessel.²² At present, however, we do not have evidence for a similarly dramatic proliferative event in the response of the human atherosclerotic vessel to angioplasty. The literature contains two contradictory papers. Strauss et al²³ reported extremely high levels of staining for markers of proliferation in atherectomy tissue from both primary and restenotic lesions. However, the values for proliferation in this study are suspect because high values were seen even months after angioplasty and because relatively few specimens were studied. Furthermore, values in primary atherosclerotic lesions were much higher than those described by others using autopsy or surgical excision tissue.^{24 25} In contrast, an extensive study by O'Brien et al²⁶ found only low values of proliferation even early after angioplasty. As pointed out by the latter authors, the possibility remains that replication in the clinical setting is too transient or at too low a level to be detected by random atherectomies; replication might also occur in layers of the vessel wall deeper than are usually sampled. It is also important to realize that the injured wall may produce new extracellular mass via mechanisms that are independent of proliferation. Collagen, elastin, and proteoglycans may all contribute to the loss of gain by forming a mass that occludes the lumen. Finally, a decrease in lumen caliber could occur by retraction or contraction of healing tissue in the wound. This latter mechanism would produce restenosis even if there were no actual increase in tissue mass.

In summary, restenosis is defined as a late loss of gain occurring weeks or months after angioplasty. The potential contributing processes include remodeling, healing of the injured vessel, smooth muscle cell proliferation, smooth muscle cell migration, and formation of new extracellular matrix. Any component of these interrelated processes is a theoretical target for intervention, and pharmacological approaches to each have been proposed (reviewed in References 27 and 28).

	Antisense: A Focus on Proliferation

Top Introduction Pathogenesis of Angioplasty... Antisense: A Focus on... Promise of the Antisense... References

 
Despite the complex possibilities just discussed, antisense approaches to restenosis have thus far focused only on smooth muscle cell proliferation. The focus of antisense strategies on proliferation reflects several issues. First, as we will discuss below, cell proliferation has several key steps. Blocking any one of these steps is sufficient to halt the entire cascade leading to DNA synthesis. In contrast, the other processes described are either poorly defined at a mechanistic level (remodeling) or lack any single molecular target that could stop the entire process (migration and matrix synthesis). Second, antisense approaches are most effective when the mRNA being targeted is of low abundance (see below). Many mRNAs required for replication are of low abundance and therefore make good targets. Finally, replication may be an early, transient event that follows angioplasty. This proposal is based on the absence of replication in human atherectomy specimens at late time points²⁶ and on results from animal studies using both antisense and conventional drugs in which blockade of replication early in the response to injury had surprisingly long-term effects on neointimal mass, which were seen weeks later. All these factors make replication easier to target than the other, more poorly defined, components of the response. However, it should be borne in mind that if proliferation is only a small part of restenosis after angioplasty in humans, as suggested by some studies,²⁶ then an antisense approach that targets proliferation alone may not work in humans.

Principles of the Antisense Approach
To understand how antisense agents suppress proliferation in animal models of restenosis or eventually how they may work in humans, it is necessary to analyze how antisense oligonucleotides themselves block gene expression. The synthesis of cellular proteins (whether structural components, enzymes, receptors, or proteins involved in cellular proliferation) occurs via a coordinated sequence of molecular events (Fig 2). The antisense approach to inhibiting gene expression is to block any one of the following processes: uncoiling of DNA, transcription of DNA, export of RNA, DNA splicing, RNA stability, or RNA translation. A large number of antisense approaches have been developed, including the use of antisense oligonucleotides, antisense mRNA (antigenes), and autocatalytic ribozymes (RNA molecules with enzymatic activity) and the insertion of a section of DNA to form a triple helix. The complexities of each approach are beyond the scope of this article (see References 5, 6, and 29 through 31 for reviews); rather, this review focuses on the use of antisense oligodeoxynucleotides, the most common antisense agent in use and the most extensively studied in vascular smooth muscle cells in vitro and in restenosis models in vivo.

View larger version (29K): [in this window] [in a new window]   Figure 2. Schematic representation of the events leading to expression of cellular proteins. DNA of the gene to be expressed is uncoiled, and a complementary mRNA copy is transcribed. This mRNA is processed, which involves splicing out of introns (areas of mRNA that are not translated), methylation at the 5' end, and polyadenylation at the 3' end, and exported from the nucleus. The mRNA is then translated into protein via binding and movement of the ribosome complex along the mRNA molecule.

How Antisense Oligodeoxynucleotides Work
Many studies have used the properties of antisense oligonucleotides to inhibit gene expression in cultured cells, and their use has also been extended to whole organisms (reviewed in References 6, 32, and 33). These studies have shown that antisense oligodeoxynucleotides targeted to cellular or viral RNA sequences can reduce target gene mRNA and/or protein product³⁴ and exert biological effects, manifested usually as a suppression of cell proliferation or differentiation. Theoretically, therefore, antisense oligonucleotides are extremely useful agents for targeting any gene sequence in vitro or in vivo.

Antisense nucleotides are short (usually <30 bp) complementary DNA or RNA sequences that will hybridize to a specific mRNA forming a hybrid duplex. Although the precise mechanisms by which antisense oligonucleotides reduce target mRNA and protein levels within the cell are imperfectly understood, two main mechanisms have been postulated. First, oligonucleotides have been suggested to exert steric interference to ribosome binding and translation or to splice excision. Evidence for steric interference comes from studies in which antisense to the 5' cap of the mRNA has been found to be most effective in inhibiting rabbit ß-globin protein synthesis (reviewed in Reference 32); the 5' cap is where a number of initiation factors bind for ribosome assembly, unwinding of DNA, and ribosome translocation along the mRNA.³⁵ Second, it has been postulated that much of the effect of antisense oligonucleotides is due to induction of cleavage of mRNA by the nuclease RNase H, which is widely present in mammalian cells and specifically recognizes DNA-RNA duplexes.³⁶ In the presence of RNase H, oligonucleotides directed to various parts of the coding and upstream sequences of a mouse globin mRNA were shown to be equally effective at inhibiting translation. Furthermore, when the enzyme was blocked by addition of a competitor DNA-RNA hybrid (poly dT/rA), there was a marked reduction in efficacy of globin mRNA degradation.³⁷ Most of the effect of chemically modified oligonucleotides (see below), eg, phosphorothioates, may be explained by this mechanism, and sequences directed downstream of the initiation codon usually fail to inhibit translation unless the hybrid is cleaved by RNase H. Antisense oligonucleotides can also enter the nucleus, where they may inhibit splicing,^{38 39} preventing the processing of pre-mRNA or mRNA, or block transport of the mRNA out of the nucleus. Introduction of antisense oligonucleotides thus results in reduction in specific mRNA and protein levels if mediated by RNase H or reduction in specific protein levels if mediated by steric interference.

Advantages of Antisense Approaches
There are many ways that cell proliferation can be inhibited pharmacologically, so what are the advantages of using antisense approaches compared with conventional inhibitors?

The first major advantage of synthetic antisense oligonucleotides, at least in theory, is the potential for design of agents with target specificity. The hybridization of base sequences between nucleic acids is very specific; only the complementary base (C-G, A-T) should be bound. Because the mRNAs of related proteins often have areas lacking significant homology, this specificity of base pairing means that an antisense sequence of bases should target only a single mRNA, without affecting the mRNAs of other genes. As evidence of this specificity, studies have shown that mRNAs can discriminate between oligonucleotides that differ by one or two bases.^{40 41 42} In the latter two studies, changes in a c-myc antisense sequence of only two bases resulted in almost complete loss of activity. The ability of antisense oligonucleotides to discriminate between mRNA sequences that differ by only a few bases has also been demonstrated and can be used to selectively target the mRNA of a mutated gene. In a study by Saison et al,⁴³ oligodeoxynucleotides directed against a point mutation in the Ha-ras gene could selectively inhibit expression of the mutant gene but not the normal gene. This specificity of binding is greater than can be achieved with most conventional pharmacological inhibitors, which frequently act on a variety of proteins with different binding affinities. However, although the specificity of binding of oligonucleotides is attractive in concept, it is not always achieved in practice (see below).

The second advantage is that antisense oligonucleotides targeted to specific mRNAs are much easier to design and synthesize than any previous class of drugs. The structures of oligonucleotides are relatively simple, consisting only of possible polymers of the four base pairs. Since the target sequence is known, rational drug design against a target is theoretically obvious without screening thousands of products as occurs with pharmacological agents. Also, antisense drugs have the potential for permanently altering the target tissue. Constitutive expression of antisense RNA in the target tissue can be achieved by inserting DNA into the host chromosome. In practice, this is usually not done with oligonucleotides but rather with full-length antisense mRNA sequences (antigenes). Finally, if delivered properly, the effects of polynucleotide-based drugs should be highly localized. Nucleotides are taken up into cells and are trapped in the intracellular compartment.⁴⁴ Any polynucleotide that remains outside the cell or undergoes exocytosis is likely to be rapidly degraded by serum nucleases (see below). This mechanism may help to restrict local delivery of antisense oligonucleotides to the site of delivery. Indeed, site specificity is a very important consideration if one is to target replication in restenosis. Potent antiproliferatives used in cancer chemotherapy almost always have systemic side effects that would be unacceptable in drugs used against restenosis. The side effects of these agents can be reduced by engineering antiproliferative drugs with target site specificity. Such drugs work well for tissues with hormone-sensitive proliferation pathways, such as breast or prostate. However, site-specific inhibitors of proliferation cannot, as yet, be used in the vessel wall because no smooth muscle–specific pathway has been shown to exist.

Targets for Antisense Agents Directed Against Proliferation
The obvious question arises of what the most appropriate target is for an antisense approach to inhibit replication in the vessel wall. Cell proliferation involves the complex interactions of mitogen binding to receptors, intracellular signal transduction pathways, and changes in the expression of specific genes. Mitogens may affect several intracellular signal transduction pathways, and pathways are branched, connected, interdependent, and in some cases redundant. This redundancy makes it unlikely that blockade of a single receptor or signaling pathway will be sufficient to suppress proliferation. In contrast to the many possible receptors and signal cascades, translation of genes concerned with proliferation is the requisite final common path into which all signal transduction pathways involved in replication converge. A large number of gene products are newly synthesized during the cell cycle and have been shown to be critical to cell-cycle progression (see Reference 45 for review). These gene products include enzymes involved in DNA and nucleotide synthesis (eg, thymidine kinase and DNA polymerases), DNA binding proteins and transcription factors (eg, c-myc, c-fos,c-jun, c-myb), and cell-cycle regulators (eg, cdc-2, cdk-2, and the cyclins). On theoretical grounds, these gene products may be the most effective targets to inhibit proliferation in smooth muscle cells. Since few conventional pharmacological inhibitors of these gene products exist, the use of antisense agents directed against growth-regulatory or cell-cycle genes remains attractive.

General Considerations for Use of Antisense Agents
Three criteria must be met for antisense agents to be useful experimentally and therapeutically. First, the antisense agent should be stable in vivo, both intracellularly and extracellularly. Second, the antisense agent must be capable of entering cells and binding to the target sequence with relatively high affinity, at concentrations that do not exert significant toxicity to the cell. Third, hybridization to the target sequence should induce suppression of gene expression of the target, and to no other nucleic acid sequences, or to intracellular proteins or lipids. Based on these criteria, a number of physicochemical characteristics of the oligonucleotide are considered when a sequence is selected for use as an antisense agent. In particular, the optimal length, target gene sequence, stability and uptake of the oligonucleotide, and nonspecific effects due to the agent all must be addressed.

Length of oligonucleotide. An oligonucleotide used for study should be long enough to be unique to the target mRNA but no so long that it binds to multiple mRNA species nonspecifically. Based on the complexity of the human genome, with approximately 3 to 4 million bases, it has been calculated that the shortest sequence required for recognition of a unique sequence is 12 to 15 bases.⁴⁶ In practice, most studies have used oligonucleotides of 15 to 30 bases. Increased length of oligonucleotide should improve binding and thus hybrid stability. However, this advantage is offset by an increase in the potential for binding to nontargeted sequences (see below), and longer oligonucleotides may also have variant uptake characteristics.

Target sequence. A number of theoretical considerations help in the choice of target sequence for antisense oligonucleotides within a specific mRNA. As most antisense-mRNA interaction is proposed to occur within the cytoplasm, areas of the mRNA with little secondary structure should offer attractive targets. This frequently means sequences directed around the initiation codon of the mRNA. For interactions involving nuclear mRNA, splice sites involved in mRNA processing and export have also been found to be effective. Other sites that have been found to be particularly effective are related to the 5' cap; the 5' cap is where a number of initiation factors bind for ribosome assembly, unwinding of DNA, and ribosome translocation along the mRNA (Fig 2).³⁵ Despite these considerations, however, a few base-pair shift in target sequence can profoundly affect the ability of an oligonucleotide to inhibit gene expression. In addition, sequences directed at different parts of the same mRNA have widely differing activities (eg, see References 47 and 48). Although the secondary structure of the mRNA may be partly responsible for differences in hybridization, the full explanation of this phenomenon is unknown. This makes design of oligodeoxynucleotide sequences an informed guess at best, and many sequences are usually tested before sequences are chosen that exert maximal suppression of target gene expression.⁴⁷

Uptake and stability. A further problem of antisense delivery into cells or tissues relates to uptake and stability of sequences. In cell culture, oligonucleotides are usually microinjected into cells or added to the culture medium, whereupon they are taken up into cells. Microinjection is feasible only for small numbers of cells, and therefore most studies in cultured cells use direct addition to the culture medium. However, the exact mechanism of oligonucleotide entry into cells by use of this method is unclear. Oligonucleotides are typically 15 to 30 bp long with molecular weights from 4500 to 9000 D. Oligonucleotides are also polyanions and cannot passively diffuse across cell membranes. Uptake depends on length of oligonucleotide, overall charge and hydrophilicity/lipophilicity (which in turn depend upon chemical modifications of the oligonucleotide; see below), and concentration of oligonucleotide. Uptake is also an energy-requiring process that is maximal at 37°C.⁴⁹ Studies using fluorescent acridine–labeled oligonucleotides have suggested that uptake of unmodified sequences is by a mechanism consistent with receptor-mediated endocytosis, and two surface proteins (34 and 80 kD) have been identified that may mediate the process.^{50 51} However, this route of uptake has yet to be conclusively proven, and it is also likely that the predominant method of uptake differs among modified oligonucleotides.⁴⁹ For instance, it has been shown that the 80-kD protein binds phosphodiester and phosphorothioate oligonucleotides but not methylphosphonates.^{50 52} Movement of oligonucleotides across cell membranes is also not a one-way process. Oligonucleotide exocytosis has been demonstrated in a number of cell types,⁴⁹ being temperature dependent, maximal at 37°C.

Uptake of both phosphodiesters and modified oligonucleotides is generally an inefficient process,⁵³ but it can be enhanced by complexing the oligonucleotide sequence with liposomes^{54 55 56 57 58} and/or by using a virus transport system such as that used by the hemagglutinating virus of Japan (HVJ).⁵⁹ Use of liposomes masks the negative charge present on many types of oligonucleotides, particularly unmodified and phosphorothioate-modified sequences, and may thus allow diffusion across the cell membrane. The HVJ-liposome system also bypasses receptor-mediated endocytosis; the HVJ-liposome complex fuses directly to the plasma membrane at neutral pH and can release DNA contained in the core of the complex into the cell. Addition of a nonhistone nuclear protein to the complex apparently results in translocation of the DNA sequence to the nucleus. DNA delivered in this way apparently shows a 10-fold higher incorporation into cells in culture than DNA-liposomes alone.⁶⁰

Although evidence indicating that antisense oligonucleotides suppress target mRNA levels suggests that oligonucleotides do enter the cell, direct evidence of cellular uptake comes from studies in which antisense oligonucleotide–RNA duplexes were directly demonstrated within the cell by S1 nuclease analysis.^{39 41 61} Many studies have also used radiolabeled or fluorescence-labeled oligonucleotides to monitor uptake and distribution of oligonucleotide within cells (Fig 3). In most cells, uptake is first demonstrated in a granular pattern consistent with internalization into endocytotic vesicles.⁵⁷ At later time points, and particularly if complexed with liposomes, nuclear staining of antisense sequences becomes evident, indicating transport to the nucleus.⁵⁷ Although in general uptake of oligonucleotides is poor, a recent study in human vascular smooth muscle cells demonstrated oligonucleotide uptake at 1 hour with persistence of full-length oligonucleotides within cells up to 16 hours.⁶² Thus, although the precise mechanism of entry has not been ascertained, the fact that the sequences do enter the cell has been established.

View larger version (71K): [in this window] [in a new window]   Figure 3. Photomicrographs show uptake of oligonucleotides into vascular smooth muscle cells. A 15-bp fluoresceinated phosphorothioate oligonucleotide to c-myc was added to the culture medium of rat vascular smooth muscle cells. A, At 1 hour, surface staining and some cytoplasmic staining are evident. B, At 2 hours, cytoplasmic staining in a granular pattern consistent with endocytosis is evident. C, At 16 hours, maximal staining occurred. D, High-power photomicrograph shows granular cytoplasmic staining with no evidence of nuclear transport. Scale bar=10 µm.

Delivery of antisense oligonucleotides in vivo to the arterial wall has been achieved by two methods: direct transfection and HVJ-liposome–mediated uptake. For example, a single application of phosphorothioate-modified oligonucleotides in a gel matrix to the adventitial surface of a rat carotid artery after injury can suppress target mRNA levels.^{42 48 63} However, intraluminal instillation with interruption of blood flow also appears to be effective in both the rat carotid artery and the pig coronary artery,^{59 60 64} and intravascular delivery can be enhanced by complexing the oligonucleotide with HVJ-liposome. With this latter system, significant uptake of oligonucleotide in the arterial wall can be observed after only 10 minutes, and oligonucleotides show persistence in the arterial wall up to 2 weeks after administration.⁶⁰

In addition to generally poor uptake, the instability of oligonucleotides has been a significant problem in their use in vitro and their potential use in vivo. Oligonucleotides are very sensitive to degradation by exogenous and endogenous nucleases (phosphodiesterases).⁶⁵ These enzymes are widespread, with significant activities being demonstrable in serum,^{66 67} and the presence of nucleases has previously precluded the use of unmodified oligonucleotides in studies of whole cells (but not all studies; see Reference 41). To improve stability against nucleolytic phosphodiesterases, the phosphate backbone of the oligonucleotide has been chemically modified in a variety of different forms (Fig 4).^{67 68} Compared with the unmodified phosphodiester linkage, chemical modifications such as phosphorothioate and phosphoroamidate bonding have improved nuclease resistance by up to 10-fold, thereby reducing the concentration at which a biological effect can be observed.^{69 70} These modifications, particularly the methylphosphonate form, can also increase cellular uptake significantly by removal of the net negative charge from the compound. Despite some reduction in the ability of modified agents to hybridize to the target sequence^{71 72} and increased nonselective inhibition of translation,⁷³ modified oligonucleotides in general and phosphorothioates in particular are widely considered to be the most promising agents for therapeutic use.⁷⁴

View larger version (7K): [in this window] [in a new window]   Figure 4. Chemical bonds show internucleoside phosphate linkages.

Despite the use of modified oligonucleotides, the inhibition of gene expression using antisense oligonucleotides is an inefficient process. Studies using unmodified oligonucleotides have required concentrations of 50 to 100 µmol/L to inhibit gene expression by 90%,^{75 76} representing very high molar ratios of oligonucleotide to mRNA. Indeed, in culture, genes that are highly expressed or amplified are inhibited poorly or not at all by antisense oligonucleotides.^{42 77} Studies of restenosis in the rat carotid artery injury model have used concentrations of phosphorothioated oligonucleotides of up to 40 µmol/L delivered periadventitially to achieve an effect,^{42 48 63} although concentrations as low as 3 to 15 µmol/L have been effective when delivered intraluminally.^{59 60} Although newer techniques such as oligonucleotide binding to the polypurine tracts of DNA to form a triple helix may be more effective inhibitors of gene expression,^{30 78 79} the inefficiency of conventional antisense oligonucleotides further limits the targets presently suitable for this mode of therapy.

Specificity of Action of Antisense Oligonucleotides
A major consideration in the design and use of an antisense oligonucleotide is its specificity for the target mRNA. The ability of an antisense molecule to act on its intended target might be called its "specificity." However, the concept of nucleotide specificity has a special meaning when applied to nucleotide hybridization. Because hybridization is specific for purine-pyrimidine pairings, only two combinations dictate hybridization, A-T and G-C. The strength of hybridization, or "stringency," is determined by the numbers of matches in a length of polynucleotide. Specificity of antisense action thus depends on two components: the uniqueness of the target sequence and the stringency of hybridization. As mentioned above, theoretical calculations suggest that the shortest continuous sequence required for uniqueness of an oligonucleotide is 12 to 15 bases.^{46 80} In practice, however, the length of oligonucleotide needed for hybridization to a specific mRNA sequence is unknown, and hybridization does not require a perfect match along the whole length of the oligonucleotide. It has been conclusively demonstrated that oligonucleotides with mismatched bases can still hybridize and induce target mRNA degradation.⁸¹ Thus, antisense oligonucleotides may inhibit expression of nontargeted genes in an unexpected and unpredictable fashion. More stable hybrids will form with longer oligonucleotides, but the longer the oligonucleotide the greater the chance of hybridization to nontargeted mRNAs by short sections of consecutive nucleotides. Indeed, the length of oligonucleotide-mRNA duplex required to mediate RNase H–induced cleavage may be quite short, such as 4 mer in vitro⁸² or 6 to 10 mer in oocytes.^{83 84} Therefore, within a 15- to 20-bp oligonucleotide (the most common size used) there may be more than one sequence that can mediate mRNA degradation. These sequences may be found in any number of nontargeted mRNAs, and cleavage at secondary sites within the target mRNA that do not have complementarity has also been reported from both unmodified and phosphorothioate oligonucleotides.^{85 86 87} The implication of this is that any antisense oligonucleotide introduced into cells will actually induce the degradation of a number of nontargeted mRNA species.⁸¹ This is especially true of longer sequences and has been used as an argument for restricting antisense oligonucleotides to 15 to 20 bases in length.

Other Biological Effects of Oligonucleotides
Another problem concerning specificity of action of oligonucleotides is that of possible non–sequence-specific pharmacological effects of large amounts of duplex or interactions of oligonucleotide sequences with cellular proteins. Double-stranded RNA (dsRNA) has been shown to induce interferon synthesis in a number of cell types⁸⁸ and also to activate the proteins of two interferon-inducible genes independently of interferon.⁸⁹ These proteins are the enzymes 2'5' oligoadenylate synthetase (2'5' AS) and p68 protein kinase. 2'5' AS can activate an endonuclease RNase L, which degrades transfer RNA, while P68 protein kinase phosphorylates the -subunit of the eukaryotic initiation factor eIF-2, leading to failure to initiate mRNA translation and inhibition of protein synthesis (see Reference 90 for review). dsRNA also activates adenylate cyclase, leading to a rise in intracellular cAMP levels.⁹¹ Thus, there are at least three ways that dsRNA can inhibit proliferation without hybridization to mRNA of a gene involved in proliferation. The induction of these pathways is irrespective of the target sequence and may therefore be responsible for a biological effect observed when using antisense oligonucleotides. Although a similar effect has not yet been demonstrated by DNA-RNA duplexes, interferon production has been shown to occur when oligonucleotide palindromes of six or more bases are used. Potentially, therefore, these nonspecific effects can occur with oligonucleotide binding to any expressed mRNA sequence. Furthermore, nonspecific effects on cell morphology and proliferation have been demonstrated with oligonucleotides that contain a stretch of 4G residues,^{92 1} and antiproliferative effects of a synthetic dsRNA poly(I.C) have been demonstrated in human endothelial cells by induction of interleukin-1.⁹³ Other mechanisms of nonspecific effects of antisense agents have also been demonstrated, including cleavage of nontarget mRNA at high concentrations of antisense oligonucleotides⁸⁵ and inhibition of specific enzymes associated with replication in a process not involving hybridization.⁹⁴

Another potential source of a non–antisense inhibition of biological processes relates to the fact that oligonucleotides can bind to cellular proteins in a sequence-specific manner. The binding of oligonucleotides, designated aptamers, to proteins can alter that protein's biological activity.^{95 96} Suppression of protein activity has also been documented in vivo, indicating that the oligonucleotide can be used pharmacologically without any specific antisense action being implicated.⁹⁷ In fact, the binding of oligonucleotides to specific proteins might actually provide a much more specific inhibition of a target protein than antisense oligonucleotide binding to the equivalent mRNA. Although the binding between protein and oligonucleotide is dependent on the sequence of oligonucleotide, it is not yet possible to predict which proteins will bind with which oligonucleotide sequences. Thus, sequence-specific binding to a protein may result in inhibition of an expression of a nontargeted gene when antisense oligonucleotides are used. Oligonucleotides will also bind proteins in a non–sequence-specific manner, including CD4,⁹⁸ protein kinase C-ß1,⁵² and albumin. Although the binding constants of oligonucleotides to proteins via non–sequence-specific interactions are usually significantly lower than those of a natural ligand, this is not always the case. Thus, introduction of an oligonucleotide into a cell may affect the function of a wide range of proteins, which may be responsible for the biological effect observed.

In summary, the occurrence of nonspecific effects of antisense oligonucleotides may explain a widely observed phenomenon, namely that antisense, sense, and random sequences can sometimes exert similar biological effects on cells or viruses.

Control Measures Used in Antisense Experiments and Targets
Since the specificity of antisense oligonucleotides for their target mRNA must remain in doubt, adequate control measures must be included in experiments using antisense agents together with the assessment of adequate end points before effects are attributed solely to the action of the antisense oligonucleotide.

The most obvious control measure is the demonstration that the antisense oligonucleotide has actually inhibited its target. This, of course, is ultimately the protein product and not the mRNA, particularly when the effect of the antisense sequence is not mediated by RNase H. It is quite disturbing to note that there have been a number of reports of "effective" antisense experiments that either failed to document loss of the protein or showed only a minimal diminution. In such cases, one must at least suspect that the biological effect observed may be due to the antisense binding to an unknown target. In addition to the antisense sequence itself, a combination of other control sequences has been used to assess nonspecific effects. In most studies, the sense sequence is used, together with mismatched or completely scrambled sequences. Demonstration of an effect with the antisense oligonucleotide and the lack of a similar effect with the sense, mismatched, or scrambled sequences are taken as evidence that the antisense is working to specifically inhibit the target gene product. However, neither sense, mismatched, nor scrambled sequences have an mRNA target within the cell; therefore, biological effects due to the presence of a hybrid DNA-RNA duplex may not be reproduced. In contrast, targeting of mRNA species of genes that possess biological actions unrelated to the gene of interest in an effective control measure for this possibility.

The most convincing controls are those that use overexpression of the target mRNA or more directly add back the protein to counter the effects of the antisense oligonucleotide. For overexpression, the target mRNA is injected into the cell or overexpressed from a transfected plasmid sequence. If the effects of the antisense can be reversed, then this is good evidence that the antisense oligonucleotide has acted by specific hybridization to its target mRNA (although it is still theoretically possible that RNA overexpression may block the biological effects of antisense oligonucleotides by a non–antisense mechanism). Suppression of antisense action by prior incubation with sense sequences is also a useful, but less conclusive, control.^{47 62} In addition, good evidence of specific hybridization is provided by an observation that multiple oligonucleotides to the same target mRNA induce a similar biological effect, because it is unlikely that oligonucleotides of very different sequences would have similar secondary targets. Thus, it is recommended that a combination of control sequences be used to circumvent some of the problems of non-antisense–mediated suppression of biological processes. It should be remembered, however, that the greater the number of control sequences used, the greater the chances that some nonspecific biological effects will occur.

The overall limitations of the antisense approach also govern selection of targets for this approach. Because it is difficult to achieve complete suppression of target protein levels with conventional antisense agents, the target should be of relatively low abundance, or incomplete suppression of target gene suppression should have a biological effect. On kinetic grounds, frequently translated mRNAs that produce stable proteins are the most efficacious targets for antisense agents because protein levels will be more sensitive to changes in mRNA.⁹⁹ If a rapid onset of action is required, an mRNA that is rapidly degraded and slowly translated and a protein that is rapidly degraded appear to be the most effective targets.⁹⁹ Alternatively, multiple antisense sequences against the same mRNA^{72 100} or multiple targets affecting the same biological process^{59 60} can be used to increase the desired effect. Another difficulty relates to maintenance of suppression of gene expression. A target that requires a single reduction in protein levels to irreversibly affect the biological process is more likely to be effective than one requiring constant suppression. The sequence of the target mRNA should be known; experiments using human sequences for animal studies and vice versa without adequate controls are particularly unsatisfactory. The ideal properties of a target mRNA sequence and possible control oligonucleotide sequences are outlined in Tables 1 and 2.

View this table: [in this window] [in a new window]   Table 1. Properties of an `Ideal' Target mRNA for Antisense Oligonucleotides

View this table: [in this window] [in a new window]   Table 2. Recommended Controls for Demonstration of a `Specific' Effect of an Antisense Oligonucleotide

Toxicity of Antisense Agents
The issue of oligonucleotide toxicity is an important one, particularly when antisense oligonucleotides are proposed as therapeutic agents. Although many of the physicochemical properties of antisense agents (eg, specificity, hybridization, selectivity, biological actions, and optimal length) are known, information on the pharmacology of antisense agents is sparse. In general, oligonucleotides are well tolerated in vitro at relatively high concentrations (up to 100 µmol/L) by a wide range of cells. However, the cytotoxicity of oligonucleotides is dependent upon cell type, chemical class of oligonucleotide modification, oligonucleotide length, and time of exposure. For instance, it has been shown that 50 µmol/L of a 28-bp oligo is not toxic to HeLa S3 cells¹⁰¹ but exerts marked toxicity when injected into Xenopus oocytes at a 100-fold lower concentration.⁷⁰ In general, the longer the oligonucleotide and the longer the exposure, the greater the toxicity; in some studies almost all compounds have caused a reduction in cell viability after 4 days' exposure.⁴⁹ In particular, both methylphosphonate and phosphorothioate oligonucleotides may bind intracellular components as full-length oligonucleotides and as breakdown products and have been shown to interfere with a variety of intracellular processes such as translation and protein synthesis.^{70 102} Furthermore, phosphorothioate oligonucleotides can also markedly inhibit human DNA polymerases and RNase H¹⁰³ ; the latter effect actually stabilizes target mRNA against RNase H–mediated breakdown. The effects of oligonucleotides on these enzymes are dependent on oligonucleotide length and concentration. However, the fact that phosphorothioate oligonucleotides can both activate and inhibit RNase H implies that there may be only a narrow therapeutic index of these oligonucleotides as antisense agents and that above a specific concentration range the nonspecific inhibition of RNase H may reduce any sequence-specific effects of the oligonucleotide. Toxicity of oligonucleotides is also not limited to modified forms. Unmodified oligonucleotides can be toxic by means of intracellular degradation into nucleotides, since even small changes in intracellular pools of free nucleotides have been shown to alter DNA synthesis.¹⁰⁴

Data relating to in vivo toxicity of oligonucleotides are more scanty. After intravenous or intraperitoneal delivery of a phosphorothioate oligonucleotide, there is rapid redistribution (t_1/2, 10 to 60 minutes). Excretion is predominantly via the urine,^{105 106 107} with a long elimination time (t_1/2, 20 to 40 hours). This indicates that dosing can be infrequent and still maintain an effective, therapeutic tissue concentration of oligonucleotide.¹⁰⁷ However, significant accumulation of intact oligonucleotides occurs in many tissues, such as the heart, stomach, and intestine.¹⁰⁶ Accumulation in nontargeted organs may ultimately govern overall toxicity because antisense oligonucleotides, particularly high concentrations of those with chemical modifications, exhibit nonspecific toxicity in cultured cells, manifested as a suppression of protein synthesis.¹⁰⁸ The widespread distribution of systemic oligonucleotides highlights further problems relating to toxicity. Because many genes used as targets for antisense agents (eg, proto-oncogene products) are not expressed only in the target tissue, suppression of gene expression in rapidly dividing tissue, such as bone marrow and intestine, also needs to be addressed. Another important consideration in using modified oligonucleotides relates to the incorporation of chemically modified bases into cellular DNA after oligo breakdown. Although not formally demonstrated, the potential for both mutagenesis and interference with normal DNA repair exists with these agents. Furthermore, while such effects may not be evident in cell culture systems, they are crucial to suitability of therapy in whole organisms.

Whereas many considerations regarding toxicity are important regardless of the route of delivery, local administration of antisense oligonucleotides can limit the effects on distance tissues. Local administration to the eye and brain and local perfusion of neoplastic tissue have resulted in reports of suppression of gene products in these locations.^{109 110 111 112} Another approach to localizing antisense actions is that of complexing oligodeoxynucleotides to cell-specific receptors¹¹³ ; eg, a complex of oligonucleotide-asialoglycoprotein can specifically direct much of a systemically delivered antisense agent to the liver.¹¹⁴ The use of such targeted vehicles awaits identification of specific receptors of each tissue of interest, but the approach offers great promise.

	Promise of the Antisense Approach

Top Introduction Pathogenesis of Angioplasty... Antisense: A Focus on... Promise of the Antisense... References

 
Despite theoretical and practical limitations, antisense oligonucleotides have successfully been used to partially inhibit vascular smooth muscle cell proliferation in vitro.^{42 62 63 115 116 117 118} Suppression of cell proliferation persists for several days, even after short (2 hours) exposure to antisense agents.¹¹⁹ Administration of antisense agents in these studies was associated with reduced levels of the target mRNA and in some cases of the target protein. The complementary (sense) sequence, scrambled sequences of the same bases contained in the antisense sequence, or mismatched controls exerted no biological effect; neither did these control sequences affect target mRNA or protein. Similarly, no effect was observed when cells were incubated with antisense oligonucleotides directed at genes unrelated to proliferation, such as -actin or GAPDH, or when the target mRNA was overexpressed by an introduced gene construct.⁴² Although the controls in each study were not exhaustive, taken together the evidence from all the studies suggests that the antisense oligonucleotides used may suppress vascular smooth muscle cell proliferation via an antisense mechanism.

The most exciting development in the use of antisense agents to suppress smooth muscle cell proliferation has been the demonstration that antisense oligonucleotides can suppress neointimal formation in rat arteries after injury with a balloon catheter.^{42 48 59 60 63} These studies targeted the proto-oncogenes c-myb and c-myc, the proliferating cell nuclear antigen (PCNA), and the cell cycle–dependent proteins cdc-2 kinase and cdk-2 kinase. Application of the antisense sequence resulted in suppression of neointimal formation after injury to the rat carotid artery (Fig 5),⁴² which in one study was associated with suppression of smooth muscle cell proliferation and a reduction in DNA content of the arterial wall.⁵⁹ Although suppression of neointimal formation was usually incomplete, addition of a second antisense oligonucleotide to a target gene also involved in proliferation resulted in an enhanced effect. Thus, the effect of antisense cdc-2 kinase could be increased by cotreatment with antisense PCNA or cdk-2 kinase.^{59 60} Importantly, suppression of neointimal formation appears to be localized to the antisense-treated regions^{42 60 63} and may be maintained for up to 8 weeks after a single application.⁵⁹

View larger version (134K): [in this window] [in a new window]   Figure 5. Photomicrographs show the effect of adventitial administration of antisense oligonucleotides to c-myc on neointimal formation after 2 weeks in the rat carotid artery.42 A, At 2 weeks, a prominent neointima is present in vessels treated with sense c-myc oligonucleotide. B, In contrast, neointima is reduced in vessels treated with antisense c-myc oligonucleotides, in some cases to very low levels.

It is somewhat surprising that in a process in which proliferation occurs relatively continuously, a single application of an agent may be sufficient to affect long-term luminal patency. This implies that suppression of an early proliferative event may be responsible for the effect of the antisense agent. This is similar to the effect observed after a short application of some pharmacological agents, in which application for only 2 to 4 days could affect arterial lumen size at 2 weeks after injury, suggesting that cells are committed early to replicate or not at all.^{11 120} However, application of oligonucleotides has been demonstrated to be associated with suppression of the specific mRNA species in the arterial wall both early, within a few hours,⁴² and late, at 2 weeks.⁶³ Suppression of target protein activity has also been shown to be maintained for up to 2 weeks.⁴⁸ Because the targets of two of these studies, the c-myc and c-myb proto-oncogenes, are expressed not only in cells induced to enter the cell cycle but also continuously in proliferating cells, some of the effect of antisense oligonucleotides may be directed at cells already in the cell cycle. The presence of antisense sequences detectable in the arterial wall up to 72 hours after injury supports this notion. A further possibility relates to an observed effect on the migration of smooth muscle cells in vitro. Antisense oligonucleotides directed against c-myc have been shown to inhibit smooth muscle cell migration in culture at concentrations below those that inhibit proliferation.¹¹⁷ Because migration has been shown to be an important component of the response to injury, at least in animal models,¹²¹ an additional effect on migration may be responsible for the potency of the antisense response.

Although most studies have been performed in the rat, catheter-based delivery of antisense c-myc oligonucleotides also suppresses neointimal formation after injury to the pig coronary artery.⁶⁴ Furthermore, the efficacy of antisense oligonucleotides on the proliferation of human vascular smooth muscle cells in vitro has been well documented.^{62 115} Thus, if cell proliferation is a prominent component of neointimal formation in humans, then it seems likely that the application of antisense oligonucleotides will suppress smooth muscle cell proliferation after arterial injury. Whether suppression of cell proliferation will lead to a reduction in the rate of human restenosis is harder to predict.

Antisense strategies have thus far concentrated solely on proliferation in arteries after injury. As already stated, however, doubt still exists as to whether proliferation is the key event in restenosis (see References 26 and 122). For example, anti–platelet-derived growth factor antibodies fail to inhibit proliferation but do prevent the formation of neointima in the rat carotid artery model^{123 124} by blocking migration. Moreover, formation of neointima itself may be less important in restenosis than the still poorly defined mechanisms involved in remodeling.^{20 21} Remodeling may involve many different processes, including cell replication, cell death, cell migration, and matrix degradation and synthesis. The obvious concern is that remodeling may not be as amenable to the targeting of a single critical molecule as the cell cycle, in which a number of molecules are required for cell replication.

Summary
The high affinity of even relatively short sequences of DNA for their target mRNA suggests that antisense agents represent an ideal method of suppressing specific gene products both in vitro and in vivo. In experiments performed thus far, an effect on the target mRNA in cultured vascular cells and in the vessel wall can be documented. The in vitro activity, toxicity, and pharmacokinetic data of antisense oligonucleotides are encouraging, and the in vivo animal experiments demonstrating suppression of neointimal formation are very promising. If animal trials presently under way show continued suppression not only of intimal formation but also of loss of lumen caliber after a single application, then effective delivery of antisense oligonucleotides is a realistic possibility.

Nevertheless, some words of caution regarding the use of antisense oligonucleotides are warranted. Potential nonspecific effects of antisense oligonucleotides should be carefully considered in studies in which antisense agents are used to define biological functions of specific genes. In particular, demonstration that the target mRNA has been suppressed does not prove that other sequences within the mRNA pool have not also been suppressed. Critical control measures include adding back the target mRNA or protein and demonstrating similar biological effects with antisense sequences, which also suppress target gene expression directed at different regions of the target mRNA.

At the clinical level, the systemic effects of antisense oligonucleotides, the dosage required, the timing of administration compared with mechanical intervention, and the toxicity of breakdown products all need to be established. In addition, the most appropriate targets for antisense use in restenosis remain largely obscure. Indiscriminate suppression of cell-cycle genes or proto-oncogenes may be as acutely toxic as current anticancer chemotherapy if the site delivery is not completely localized. Furthermore, much of the clinical evidence suggests that restenosis is a chronic process, continuing to develop weeks to months after the procedure. If this is the case, then the current approaches that rely on a transient, local application of an antisense agent may fail. If, however, a target gene is identified that is specific to vascular tissue, then repeated administration of an antisense agent may be tolerated via a systemic route. This approach has proved successful in targeting mutated genes with little suppression of closely related genes and with minimal systemic toxicity.^{109 125} An alternative approach is to transfect the target tissue with a gene that makes it susceptible to systemic delivery of a drug that is not normally toxic to mammalian cells. Such an approach has recently been demonstrated in studies using the herpesvirus thymidine kinase gene and the drug ganciclovir.¹²⁶

Finally, therapeutic success will depend to a great extent on whether our current models of restenosis are correct. As already noted, there is controversy about the role of proliferation in the clinical setting. Equally important is the fact that we lack information regarding the best time to deliver an antisense agent directed against proliferation. Thus, a lack of success in early clinical trials might lead to the false conclusion that this cannot be a successful approach.

	Acknowledgments

 
Dr Bennett is supported by a British Heart Foundation Clinician Scientist Fellowship, and this work was also supported by NIH SCOR Arteriosclerosis 47151.

	Footnotes

 
¹ Indeed, 4G-containing oligonucleotides have been shown to nonspecifically inhibit replication of smooth muscle cells (Villa A et al. Effects of antisense c-myb oligonucleotides on vascular smooth muscle cell proliferation and response to vessel wall injury. Circ Res. 1995;76:505-513; Burgess T et al. The antiproliferative activity of c-myb and c-myc antisense oligonucleotides in smooth-muscle cells is caused by a nonantisense mechanism. Proc Natl Acad Sci U S A. 1995;92:4051-4055.).

Received January 12, 1995; revision received March 8, 1995; accepted March 10, 1995.

	References

Top Introduction Pathogenesis of Angioplasty... Antisense: A Focus on... Promise of the Antisense... References

 

1. Greuntzig A, King SI, Schlumpf M, Siegenthaler W. Long-term follow-up after percutaneous transluminal coronary angioplasty: the early Zurich experience. N Engl J Med. 1987;316:1127-1132. [Abstract]
   
2. Parisi AF, Folland ED, Hartigan P. A comparison of angioplasty with medical therapy in the treatment of single-vessel coronary artery disease. N Engl J Med. 1992;326:10-16. [Abstract]
   
3. McBride W, Lange R, Hillis L. Restenosis after successful coronary angioplasty: pathophysiology and prevention. N Engl J Med. 1988;318:1734-1737. [Medline] [Order article via Infotrieve]
   
4. Califf RM, Fortin DF, Frid DJ, Harlan WR, Ohman EM, Bengtson JR, Nelson CL, Tcheng JE, Mark DB, Stack RS. Restenosis after coronary angioplasty: an overview. J Am Coll Cardiol. 1991;17(suppl B):2B-13B.
   
5. Cohen JS. Oligonucleotides as therapeutic agents. Pharmacol Ther.1991;52:211-225.
   
6. Colman A. Antisense strategies in cell and developmental biology. J Cell Sci. 1990;97:399-409. [Free Full Text]
   
7. Kindy MS, Sonenshein GE. Regulation of oncogene expression in cultured aortic smooth muscle cells: post-transcriptional control of c-myc mRNA. J Biol Chem. 1986;261:12865-12868. [Abstract/Free Full Text]
   
8. Gadeau AP, Campan M, Desgranges C. Induction of cell cycle-dependent genes during cell cycle progression of arterial smooth muscle cells in culture. J Cell Physiol. 1991;146:356-361. [Medline] [Order article via Infotrieve]
   
9. Miano JM, Tota RR, Vlasic N, Danishefsky KJ, Stemerman MB. Early protooncogene expression in rat aortic smooth muscle cells following endothelial removal. Am J Pathol. 1990;137:761-765. [Abstract]
   
10. Bauters C, de Groote P, Adamantidis M, Delcayre C, Hamon M, Lablanche JM, Bertrand ME, Dupuis B, Swynghedauw B. Proto-oncogene expression in rabbit aorta after wall injury: first marker of the cellular process leading to restenosis after angioplasty? Eur Heart J. 1992;13:556-559. [Abstract/Free Full Text]
    
11. Majesky MW, Schwartz SM, Clowes MM, Clowes AW. Heparin regulates smooth muscle S phase entry in the injured rat carotid artery. Circ Res. 1987;61:296-300. [Abstract/Free Full Text]
    
12. Majesky MW, Reidy MA, Bowen-Pope D, Hart CE, Wilcox JN, Schwartz SM. PDGF ligand and receptor gene expression during repair of arterial injury. J Cell Biol. 1990;111:2149-2158. [Abstract/Free Full Text]
    
13. Miller MJ, Kuntz RE, Friedrich SP, Leidig GA, Fishman RF, Schnitt, SJ, Baim DS, Safian RD. Frequency and consequences of intimal hyperplasia in specimens retrieved by directional atherectomy of native primary coronary artery stenoses and subsequent restenoses. Am J Cardiol. 1993;71:652-658. [Medline] [Order article via Infotrieve]
    
14. Losordo DW, Rosenfield K, Pieczek A, Baker K, Harding M, Isner JM. How does angioplasty work? Serial analysis of human iliac arteries using intravascular ultrasound. Circulation. 1992;86:1845-1858. [Abstract/Free Full Text]
    
15. Potkin BN, Keren G, Mintz GS, Douek PC, Pichard AD, Satler LF, Kent KM, Leon MB. Arterial responses to balloon coronary angioplasty: an intravascular ultrasound study. J Am Coll Cardiol. 1992;20:942-951. [Abstract]
    
16. The SH, Gussenhoven EJ, Zhong Y, Li W, van Egmond F, Pieterman H, van Urk H, Gerritsen GP, Borst C, Wilson RA. Effect of balloon angioplasty on femoral artery evaluated with intravascular ultrasound imaging. Circulation. 1992;86:483-493. [Abstract/Free Full Text]
    
17. Chesebro JH, Lam JY, Badimon L, Fuster V. Restenosis after arterial angioplasty: a hemorheologic response to injury. Am J Cardiol. 1987;60:10B-16B. [Medline] [Order article via Infotrieve]
    
18. Folkow B. The Benjamin W. Zweifach Award lecture: functional and structural `autoregulation'—some personal considerations concerning the century-old development of these microvascular concepts. Microvasc Res. 1989;37:242-255. [Medline] [Order article via Infotrieve]
    
19. Glagov S, Weisenberg E, Zarins CK, Stankunavicius R, Kolettis GJ. Compensatory enlargement of human atherosclerotic coronary arteries. N Engl J Med. 1987;316:1371-1375. [Abstract]
    
20. Kakuta T, Currier JW, Haudenschild CC, Ryan TJ, Faxon DP. Differences in compensatory vessel enlargement, not intimal formation, account for restenosis after angioplasty in the hypercholesterolemic rabbit model. Circulation. 1994;89:2809-2815. [Abstract/Free Full Text]
    
21. Post MJ, Borst C, Kuntz RE. The relative importance of arterial remodeling compared with intimal hyperplasia in lumen renarrowing after balloon angioplasty: a study in the normal rabbit and the hypercholesterolemic Yucatan micropig. Circulation. 1994;89:2816-2821. [Abstract/Free Full Text]
    
22. Schwartz SM, Reidy MA, Clowes A. Kinetics of atherosclerosis: a stem cell model. Ann N Y Acad Sci. 1985;454:292-304. [Medline] [Order article via Infotrieve]
    
23. Strauss BH, Umans VA, van Suylen RJ, de Feyter PJ, Marco J, Robertson GC, Renkin J, Heyndrickx G, Vuzevski VD, Bosman FT. Directional atherectomy for treatment of restenosis within coronary stents: clinical, angiographic and histologic results. J Am Coll Cardiol. 1992;20:1465-1473. [Abstract]
    
24. Gordon D, Reidy MA, Benditt EP, Schwartz SM. Cell proliferation in human coronary arteries. Proc Natl Acad Sci U S A. 1990;87:4600-4604. [Abstract/Free Full Text]
    
25. Spagnoli LG, Villaschi S, Neri L, Palmieri G, Taurino M, Faraglia V, Fiorani P. Autoradiographic studies of the smooth muscle cells in human arteries. Paroi Arterielle. 1981;7:107-112. [Medline] [Order article via Infotrieve]
    
26. O'Brien ER, Alpers CE, Stewart DK, Ferguson M, Tran N, Gordon D, Benditt EP, Hinohara T, Simpson JB, Schwartz SM. Proliferation in primary and restenotic coronary atherectomy tissue: implications for antiproliferative therapy. Cir Res. 1993;73:223-231. [Abstract/Free Full Text]
    
27. Herrman JP, Hermans WR, Vos J, Serruys PW. Pharmacological approaches to the prevention of restenosis following angioplasty: the search for the Holy Grail? part 1. Drugs. 1993;46:18-52. [Medline] [Order article via Infotrieve]
    
28. Herrman JP, Hermans WR, Vos J, Serruys PW. Pharmacological approaches to the prevention of restenosis following angioplasty: the search for the Holy Grail? part 2. Drugs. 1993;46:249-262.
    
29. Calabretta B, Skorski T, Zon G. Antisense oligonucleotides. Semin Cancer Biol. 1992;3:391-398. [Medline] [Order article via Infotrieve]
    
30. Hélene C. The anti-gene strategy: control of gene expression by triplex-forming oligonucleotides. Anticancer Drug Des. 1991;6:569-584. [Medline] [Order article via Infotrieve]
    
31. Neckers LM, Rosolen A, Whitesell L. Antisense inhibition of gene expression: a tool for studying the role of NMYC in the growth and differentiation of neuroectoderm-derived cells. J Immunother. 1992;12:162-166.
    
32. Goodchild J. Inhibition of gene expression by oligonucleotides. In: Cohen J, ed. Oligonucleotides: Antisense Inhibitors of Gene Expression. London, UK: Macmillan Press; 1989:53-77.
    
33. Murray J, Crockett N. Antisense techniques: an overview. In: Murray J, ed. Antisense DNA and RNA. New York, NY: Wiley-Liss Inc; 1992:1-49.
    
34. Heikkila R, Schwab G, Wickstrom E, Loke SL, Pluznik DH, Watt R, Neckers LM. A c-myc antisense oligodeoxynucleotide inhibits entry into S phase but not progress from G0 to G1. Nature. 1987;328:445-449. [Medline] [Order article via Infotrieve]
    
35. Kozak M. Influences of mRNA secondary structure on initiation by eucaryotic ribosomes. Proc Natl Acad Sci U S A. 1986;83:2850-2854. [Abstract/Free Full Text]
    
36. Wagner R, Nishikura K. Cell cycle expression of RNA duplex unwinding activity in cells. Mol Cell Biol. 1988;8:770-777. [Abstract/Free Full Text]
    
37. Walder R, Walder J. Role of RNAase H in hybrid-arrested translation by antisense oligodeoxynucleotides. Proc Natl Acad Sci U S A. 1988;85:5011-5015. [Abstract/Free Full Text]
    
38. McManaway ME, Neckers LM, Loke SL, al Nasser AA, Redner RL, Shiramizu BT, Goldschmidts WL, Huber BE, Bhatia K, Magrath IT. Tumor-specific inhibition of lymphoma growth by an antisense oligodeoxynucleotide. Lancet. 1990;335:808-811. [Medline] [Order article via Infotrieve]
    
39. Daum T, Engels JW, Mag M, Muth J, Lucking S, Schroder HC, Matthes E, Muller WE. Antisense oligodeoxynucleotide: inhibitor of splicing of mRNA of human immunodeficiency virus. Intervirology. 1992;33:65-75. [Medline] [Order article via Infotrieve]
    
40. Wang A, Creasy A, Lardner M, Lin L, Strickler J, Van Arsdell J, Yamamoto R, Mark D. Molecular cloning of the complementary DNA for human tumor necrosis factor. Science. 1985;228:149-154. [Abstract/Free Full Text]