Dna Analogs for the Purpose of Gene Therapy

From DEPARTMENT OF LABORATORY MEDICINE Karolinska Institutet, Stockholm, Sweden

DNA ANALOGS FOR THE PURPOSE OF GENE THERAPY

Mathias G. Svahn

Stockholm 2007

Published and printed by Karolinska University Press Box 200, SE-171 77 Stockholm, Sweden © Mathias G. Svahn, 2007 ISBN 978-91-7357-290-3

to my wife and children ABSTRACT

This thesis describes DNA analogs and their potential use for gene therapeutic purposes, primarily by sequence specific strand-invasion of double-stranded DNA. Gene therapy is classically thought of as restoration of a defect gene. Correction of dysfunctional genes is not feasible with today’s technology, instead a new therapeutic gene is introduced.

Nature’s own gene delivery system, the virus, has gained its potency through evolution. Viruses can be modified, to instead of delivering its genome, be exploited as a vector to convey a therapeutic gene. Biotechnologically engineered viruses are expensive to produce, involving extensive monitoring for recombinant competent viruses, and if to become readministrated, the isotype has to be changed in order to prevent a specific immune response. Furthermore, humans have also developed defense mechanisms to prevent pathogens as viruses to fulfill their purpose. Non-viral gene delivery techniques are being developed to avoid viral associated complications. Bioplex (biological complex) is based on attachment of biologically active molecules, called functional entities (FE), on double-stranded DNA. Gene delivery is a complex process where FEs of different sources and origins are anticipated to guide the genetic material through the body, taken up by the correct cell-type and gain entrance to the nuclei. The FEs are anchored to DNA at specific positions by DNA analogs with superior binding affinity. Analogs presented here are peptide nucleic acid (PNA) and locked nucleic acid (LNA).

Paper I addresses the assembly of the Bioplex. Instead of having a FE directly conjugated to the anchor, multiple molecules could be added to each anchor by self- assembling of single-stranded oligonucleotides, presenting exponentially growing number of anchoring sites without compromising the structure of the DNA. There are circumstances when it is advantageous to detach the FE from the cargo. Paper II presents a possible strategy by insertion of an amino acid recognition sequence for the endosomal protease, cathepsin L. In this way, the FE is released when the pH drops and the enzyme is activated. The FE could be carbohydrates or peptides of endogenous or viral origin, or synthetic. Furthermore, combinations of FEs are needed for optimal effect. Literature offers many alternatives and to try all of them in combinations is manually undoable, even for small ligand libraries. Paper III utilizes robots for transfection and analyzation. Ligands in combinations of two are screened for combinatorial effect on cellular association of 6 different human cell-lines. Even though only 22 ligands are included, it equals 484 unique combinations, resulting in 10 000 transfections.

The field of oligonucleotide mediated gene therapy was founded with the discovery of antisense in the early 1970s. Paper IV is a study aiming to optimize our own oligonucleotide based construct called Zorro LNA. It inhibits transcription by directly blocking the polymerase machinery sterically by stably hybridizing to both strands of the helical DNA. Zorro has two arms of LNA oligonucleotides, each hybridizing to one strand of the DNA strands. They are interconnected by base paring, creating a Z-like structure within the double-stranded DNA.

LIST OF ABBREVIATIONS

BS Binding site CPP Cell-penetrating peptide dsDNA double-stranded DNA FE Functional entity HA2 Hemagglutinin peptide LNA Locked nucleic acid NLS Nuclear localization signal PNA Peptide nucleic acid RGD Tri-peptide arginine-glycine-aspartic acid RISC RNA-induced silencing complex RITS RNA-induced transcriptional RNAi RNA interference siRNA short interfering RNA ssDNA single-stranded DNA SUN Self-assembling UNiversal anchors WGA Wheat germ agglutinin

LIST OF PUBLICATIONS

I. Svahn M.G., Hasan M., Sigot V., Valle-Delgado J.J., Rutland M.W., Lundin K.E., Smith .C.I.E. Self-assembling supramolecular complexes by single- stranded extension from plasmid DNA. Oligonucleotides. 2007 Spring;17(1):80-94.

II. Svahn M.G., Salih G., Simonson E.O., Smith C.I.E., Branden L.J. Protease-induced release of functional peptides from bioplexes. J Control Release. 2004 Jul 23;98(1):169-77.

III. Svahn M.G., Rabe K.S., Barger G., Brandén L.J., Dumy P., Niemeyer C.M., Smith C.I.E. High-throughput identification of combinatorial ligands for DNA delivery in cell culture. Manuscript

IV. Svahn M.G., Ge R., Perálvarez A., Hugonin L., Pande V., Valle-Delgado J.J., Rutland M.W., Nilsson L., Gräslund A., Smith C.I.E. Optimizing anti-gene Zorro locked nucleic acid constructs. Manuscript

TABLE OF CONTENTS

1 Introduction ...... 11 1.1 Gene therapy...... 12 1.2 Non-viral gene therapy ...... 13 1.3 Bioplex technology ...... 14 1.4 Oligonucleotide mediated gene therapy...... 18 1.4.1 Antisense oligonucleotides ...... 19 1.4.2 RNA interference...... 20 1.4.3 Oligonucleotide inhibition of transcription...... 22 1.5 Nucleic acid derivates...... 24 1.5.1 Peptide nucleic acids ...... 26 1.5.2 Locked nucleic acid...... 28 1.6 Ligand-mediated gene therapy...... 30 1.6.1 Cell specificity and entry...... 30 1.6.2 Intracellular translocation ...... 31 1.6.3 Combinatorial ligands ...... 32 2 Aims of present study...... 35 3 Results, discussions and future plans...... 37 3.1 Paper I ...... 37 3.2 Paper II...... 39 3.3 Paper III...... 40 3.4 Paper IV...... 44 4 Populärvetenskaplig sammanfattning...... 47 5 Acknowledgements...... 49 6 References ...... 51

1 INTRODUCTION

Gene therapy is generally defined as the transfer of genetic material into cells in order to treat or alleviate symptoms of a disease. For hereditary diseases this means that a defective mutant allele is supplemented with a functional one in recessive disorders. Ideally, the defect allele would be repaired, necessary when dominant disorders are treated, but this is currently not feasible due to technical reasons. The concept is herein used to also cover genetically mediated therapy, as for example antisense therapy. The aim is then instead to regulate gene expression either at DNA or RNA level.

1.1 GENE THERAPY

In the clinical setting it was originally thought that the transfer of DNA would mainly serve as corrective treatment for inherited genetic diseases. The potential for treating a patient using gene transfer has been realized in an exponential fashion, with paralleling advances in molecular biology over the last four decades (1-4), but has also faced serious setbacks and unexpected complexity. While experimental gene therapy for cancer has become the most frequent application, also acquired diseases, such as infection by human immunodeficiency virus (HIV), and cardiovascular disease are being investigated (5-7). For tumors, multiple therapeutic approaches have been employed in order to supplement tumor suppressor genes, induce apoptotic mechanisms, kill tumor cells by the expression of suicide genes or by the transfer of replicative viruses. In the case of cardiovascular disease, the gene therapy is not directed against any known genetic defect, but instead is used as a drug delivery system in an attempt to correct the phenotype by promoting new blood vessel formation. These versatile applications result in a demand for safe and reliable vector systems.

The theory of building a modified virus to “transmit information” was initially proposed in 1968 (8) and today most of the clinical trials being conducted use viral vectors. The first attempt to accomplish gene therapy in humans was performed in 1990 by Anderson, Blaese and Culver (9). The then four-year-old Ashanthi DeSilva was treated for adenosine deaminase (ADA) deficiency, a rare genetic disease in which children are born with a severe immunodeficiency and are prone to recurrent, serious infections. The ADA gene was inserted into T cells by retroviral-mediated transduction ex vivo. Ashanthi is living a normal life but is still treated with polyethylene glycol- conjugated ADA enzyme, as a safety-net (10). Viruses are highly specialized organisms, which have developed numerous ways of invading the cells of the body. This makes them highly suitable as efficient gene transfer vectors. However, the generation of recombinant viruses is dependent on the use of packaging cell lines carrying viral genes in trans necessitating cumbersome and expensive analyses in order to exclude the existence of rare, recombinant, replicative viruses in batches for clinical use. A more common inherent caveat is the fact that viruses frequently mutate at high rates making the composition of viral supernatant heterogeneous. Virus particles are often empty and do not contain viral nucleic acid. Another limitation is the fact that viruses often are highly immunogenic, especially upon readministration.

Eleven patients with X-linked severe combined immunodeficiency (X-SCID) responded very well in what was believed to be the first significant clinical step forward for gene therapy (11,12). However, later reports have concluded that four of the patients developed leukemia as a consequence of the retroviral integration near to an oncogene, resulting in its upregulation (13,14). The New York Times published an article 27th of July 2007 entitled: “Patient in experimental gene therapy study dies, FDA says”. The patient was getting treatment for inflammatory arthritis and had earlier received the first injection. The patient became very ill after the second injection and died a few days later. Unfortunately this is not the first time that gene therapy treatment causes death (not formally investigated), though the first time using an adenoassociated virus (AAV) vector, which is considered “safe”.

For these reasons, the development of gene transfer systems depending on highly defined components with low immunogenicity is warranted.

1.2 NON-VIRAL GENE THERAPY

In 1973 successful uptake of “naked DNA” into cultured eukaryotic cells was reported (15), later followed by microinjection into nuclei of cultured cells (16) and the direct injection of reporter genes into mouse muscle in vivo (17). These seminal papers clearly demonstrated that purified nucleic acids could be efficiently delivered to live cells. Graham and van der Eb used calcium phosphate transfection to deliver adenoviral DNA (15). Using the words of others, there are only three problems encountered that limit gene therapy applications: delivery, delivery and delivery (18). There are other mechanical procedures than microinjections for gene delivery, for example gene gun, electroporation, magnetofection and microneedles, but they all have accessibility limitations, only reaching a few millimeters into the tissue (19). If instead of simply shooting the DNA into the nucleus of the cell, all evolutionary hurdles invented to protect the host are considered, the task becomes more delicate. Transfection (non-viral delivery) is a complex process involving several steps: adsorption of the transfection complex to the cellular surface, membrane translocation or endocytosis, escape from the endosome/lysosome, nuclear translocation with or without chromosomal integration and finally the effect of the nucleic acid, normally in the form of expression of a reporter/therapeutic gene.

Non-viral vectors are particularly suitable with respect to simplicity to use, ease of large-scale production and lack of specific immune response (20). However, they are inefficient as compared to viruses. Increasingly efficient methods have been developed and the prevailing transfection methods are either based on the addition of cationic lipids (21,22) or cationic polymers (23,24) to the nucleic acid. The physical and chemical characteristics of oligonucleotides have to be considered when delivering oligonucleotide-based therapies. The anionic phosphate backbone of DNA is attracted by cationic lipids or polymers and together they form micelles or electrostatic complexes. By these means, the DNA is protected against degradation and when correctly formulated, the size of the DNA complex can be fine-tuned and most reagents even condense the DNA. Size and stability of DNA particles have been considered to be critical parameters for in vivo gene delivery because of known physiological constrains, such as passing through capillary endothelium, diffusion within tissues and the cytosol of cells (25).

Cationic lipids (lipoplex) have furthermore the possibility to fuse with the cell membrane but the lipoplexes might also fuse with each other and form aggregates. Cationic polymers (polyplex) are positively charged and are attracted to the negatively charged cellular membrane. Also polyplexes have a tendency to aggregate under physiological conditions, first demonstrated for poly-lysine oligomers with DNA (26). Aggregates are efficient in vitro, since the macro-molecular complexes sediment on the cells. Important to note is that the behavior of transfection reagents in vitro differs markedly compared to when used in vivo. Without addition of ligands to these complex formulations, the uptake is unspecific and suboptimal.

The non-viral gene delivery technique we have developed is referred to as Bioplex (biological complex) in analogy with the nomenclature for other transfection reagents (27,28). The basis of this technology is to attach biologically active molecules to “naked”-DNA in order to enhance the specific cellular uptake of and eventually expression of the gene of interest.

1.3 BIOPLEX TECHNOLOGY

The Bioplex technology is a novel, module-based platform for increasing the cellular uptake of ”naked”-DNA in transfections. Typically the DNA is bacterially, expanded in the form of super-coiled plasmid, encoding a promoter and the gene of interest. Untranscribed regions of the plasmid are used as attachment points for biologically active molecules, called functional entities (FEs). Examples of FEs are cell receptor ligands; the tri-peptide RGD (29,30), which binds integrins for receptor mediated endocytosis, membrane disruptive peptides such as the influenza hemagglutinin peptide HA2 for endosomal escape (31) and nuclear localization signals (NLS) for active transportation through the nuclear pores by karyopherins into the nucleus (32-35).

The Bioplex idea originated from earlier attempts to equip “naked”-DNA with FEs (Figure 1). Initially, DNA and free NLS peptides were mixed to form NLS/nucleic acid complexes by charge interactions (36). Subsequently, methods were developed to enhance the peptide association with the nucleic acid. By using chemical linkage the location of the peptide binding is not sequence specific (37), rather generating a heterogeneous population of plasmids with varying number of covalently attached peptides. Sequence specificity was achieved by hybridizing and subsequently ligating a peptide-conjugated DNA oligonucleotide to the plasmid (38).

Figure 1: Schematic history of Bioplex. A. DNA- peptide charge interaction B. DNA-peptide chemical linking C. DNA-peptide ligation dependent D. DNA-peptide sequence specific hybridisation dependent. Brandén LJ et al. Methods Enzymol (2002;346:106-24)

Bioplex was first described using an NLS peptide (39). The DNA analog, peptide nucleic acid (PNA) was utilized as a sequence specific “genetic glue”. PNA has superior binding affinity to DNA and is able to strand-invade super-coiled plasmid (described more in detail in 1.5.1). Since PNA-peptide chimeras can be synthesized as normal peptides, it is possible to extend the PNA sequence with amino acid residues as continued peptide synthesis. This makes the integration of a peptide function, or property, into the PNA molecule possible. The PNA-NLS is an example of such a

14 construct, hence the NLS peptide is anchored to a specific position in the plasmid by PNA:DNA duplex formation (40). However, the kinetics and efficacy of strand- invasion by a single PNA-FE is rather inefficient. By positioning multiple binding sites (BSs) for the PNA anchors, the PNA has the ability to strand-invade double-stranded DNA (dsDNA) in a cooperative manner. If the BSs are positioned adjacently, interference will occur between the BSs and if the distance between the BSs is greater than 5 bp the cooperative effect is lost. Optimal positioning was found to be a separation of 2 bp between the BSs (41) (Figure 2A). In addition, the simian virus SV40, from which the truncated NLS peptide originates, presents multiple copies of the protein, containing the NLS peptide. Therefore, attaching multiple PNA-FEs per plasmid increases the biological activity, stability and kinetics of strand-invasion.

The binding of linear PNA anchors to plasmid DNA was found to be stable only under low ionic conditions and the target DNA had to be super-coiled, otherwise, the anchors were kicked out by the DNA base pairing. By hybridizing to both strands of the super-coiled plasmid with partially overlapping BSs, the binding of linear PNAs to super-coiled plasmids was stabilized at physiological conditions, albeit only for short periods of time. A set of oligonucleotides was first hybridized to the plasmid in order to open up for the PNA- FEs by displacing the strand with the BS, i.e. generating an open bulge of single- stranded DNA (ssDNA) (Figure 2B). The kinetics was increased and excess of PNA- FE needed was decreased. Additionally, it was hypothesized that the DNA might also be better protected against endonucleases, since the displaced strand is hybridized to the liberated D-loop (42-45).

A variant of PNA oligonucleotide is bisPNA, a clamp of PNA that binds the DNA strand both in Watson-Crick and Hoogsteen mode. One of the limitations of bisPNA is that the BS must be a homopurine sequence but the advantage is that bisPNA Figure 2: The anchor region of Bioplex. A. creates more stable complexes, better suited The anchors hybridize cooperatively when for in vivo applications. To achieve good the binding sites are separated with 2 bp. B. strand-invasion, a 10-1000-times molar Hybridization to the displaced DNA strand increases the stability. C. BisPNA generates excess of PNA per binding site is normally a more stable hybridization than linear PNA. required (46), and positively charged amino D. LNA is synthesized with extensions, sticky- acids are frequently used to facilitate strand- ends, which can be used for further invasion (47,48). The bisPNA per BS ratio hybridizations in a single-stranded manner. Functional entities are depicted as circles.

15 was decreased to 10 and still gave 90% hybridization, but only when a few positively charged lysines were added to the ends of the bisPNAs and the target plasmid had three or four adjacent binding sites, which permits cooperative binding (41).

In order to reduce the amount of PNA-FEs needed and the risk of unspecific binding further, opener oligonucleotides were also exploited in this setting to allow more efficient hybridization of bisPNAs (Figure 2C). With a displaced DNA strand, shorter incubation time, together with the need for less PNA, led to a reduction of nonspecific interactions, without compromising the stability of the bisPNA hybridization to the plasmid. Locked nucleic acid (LNA) is another DNA analog, which is closely related to RNA (described in detail in 1.5.2). The LNA is synthesized with conventional phosphodiester (or phosphorothioate) chemistry. In contrast to PNA, which is made with peptide chemistry, LNA synthesis is compatible with DNA and RNA, therefore allowing mixed bases in the oligonucleotide. In the search of more stable anchors and combinations of DNA analog oligonucleotides for optimization of the Bioplex, it was found beneficial to use LNA as opener oligonucleotides and bisPNA as anchors in respect to both stability and kinetics (49). When the oligonucleotides are designed to have 3 bp overlap, the plasmid can be incubated with all oligonucleotides simultaneously.

However, the number of FEs per plasmid might be a bottleneck of the Bioplex technology. Virus particles display hundreds of ligands. Having only one FE per anchor would then require many anchors that might destabilize the plasmid. Synthesizing multiple FEs per anchor would greatly increase the manufacturing cost. Instead, the LNA opener oligonucleotide could be extended with uncomplementary bases, with respect to the plasmid. These bases would then be used for hybridization of, for example, a second oligonucleotide with multiple anchor-binding sites for a third hybridization and so forth. All oligonucleotides, or for economical reasons, only the last oligonucleotide in the branching structure could bear a FE (Figure 2D). The oligonucleotide itself may well be the function, acting as an aptamer (50). An exponentially growing number of FEs could be presented on the plasmids by self- assembling hybridization of single-stranded oligonucleotides extending from the plasmid. Sequences were carefully chosen to only give the correct final product in a predefined fashion. FEs can be spaced as brush-like structures or exactly positioned, relative to each other, to give combinatorial effects. We call the technology of extending the opener oligonucleotides with a single-stranded region, Self-assembling UNiversal anchors (SUN). Bioplex constructs with the SUN technology only need a few BSs in the plasmid for the addition of large number of FEs. The Self-assembling reflects the easy to use of technology in terms of formulation. The sequences can easily be calculated not to cause unspecific cross hybridization between the participating oligonucleotides, but the extreme hybridization affinity of LNA:LNA and LNA:PNA has to be considered. All oligonucleotides can be incubated simultaneously, self- assembling into the desired structure. If large FEs are to be attached, the kinetics will however be slower. Limitations in diffusion suggest that when high molecular weight oligonucleotide-conjugates, are assembled as a Bioplex, they ought to be hybridized stepwise, starting with hybridization of the anchors to the plasmid followed by secondary oligonucleotides and so on. The SUN anchors are designed to be “UNiversal” in the sense that the anchor presents a sticky-end that could be used for

16 hybridizing any kind of secondary oligonucleotides that have the cognate recognition sequence. Accordingly, the same anchors can be used for adding any kind of FE to the plasmid.

Self-assembling supra-molecular complexes are of great interest for bottom-up research like nanotechnology. DNA is an inexpensive building block, which could be used for applications beyond the scope of this thesis, for example in surface chemistry, material science, nanomechanics, nanoelectronics and nanorobotics. The starting point is usually ssDNA, which is rather easily accessible for base pairing and duplex formation. When long stretches of dsDNA are desirable, serving either as genetic codes or electrical wires, bacterial expansion of plasmids is an inexpensive approach with scale-up properties. The SUN technology is presented in paper I, as a strategy of generating simple structures, such as junctions and DNA lattices. In paper III, the oligonucleotide part of the Bioplex is isolated for identification of optimal FEs and combinatorial FEs for specific and efficient delivery of nucleic acid. The Bioplex technology is applicable on oligonucleotide based gene therapeutics. The earlier denoted “secondary” oligonucleotide could be the gene drug itself. No such examples will be presented here.

1.4 OLIGONUCLEOTIDE MEDIATED GENE THERAPY

DNA was first described in scientific correspondence by Friedrich Miescher in 1869, later published in 1871 (51). However, DNA was first recognized as hereditary material in 1944 by Avery, MacLeod and McCarty (52). The most celebrated discovery was the structure of the DNA duplex by Watson and Crick in 1953 (53). In 1961, Nirenberg and Matthaei performed an experiment with extract from bacterial cells. Adding an artificial form of RNA, polyuridacylic acid, to the extract caused it to make an unnatural protein composed entirely of the amino acid phenylalanine (54). This showed that RNA controlled the production of specific types of protein even when no intact living cells were present. The experiment cracked the genetic code, but still left the investigators unaware of the number of bases coding for an amino acid or if all continuous bases were translated. In the same experimental setup, it was also shown that co-incubation of polyuridacylic acid and polyadenylic acid resulted in inhibition of translation. This is probably the first reported assay for post-transcriptional gene silencing, although with a synthetic form of mRNA.

Nirenberg and Matthaei did not propose any therapeutic application for the gene silencing, but only suggested that RNA has to be single-stranded in order to be translated. Their hypothesis was confirmed by Paterson et al. in 1977 in an assay called “hybrid arrested translation” (55). RNA from total cell extract was incubated with DNA oligonucleotides and translated in vitro. The mRNA sequences of the untranslated proteins were then found to reflect the hybridization of the added DNA oligonucleotides. Shortly thereafter, Zamecnik and Stephenson published, what is considered the proof-of-principle for antisense technology, blocking of both protein translation and replication of Rous sarcoma virus, in vitro (56,57). As a result, they presented a therapeutic application for post-transcriptional gene silencing for the first time, 12 years after gene therapy was proposed as a therapeutic application (58).

An antisense mechanism is defined as the process in which an antisense drug works after it hybridizes to a target RNA to form a Figure 3: Post-transcriptional gene silencing duplex. The hybrid prevents the RNA from by antisense. The antisense oligonucleotide functioning normally and from producing binds to the target mRNA, and blocks protein products by enzymatic digestion of translation. If the antisense oligonucleotide resembles a DNA:RNA duplex, it will be the mRNA or by steric hindrance of; recognized and cleaved by RNase H. splicing, nuclear-cytoplasmic transport or

18 translation. Conventional drugs on the other hand usually bind to proteins and thereby modulate their function. The major classes of antisense agents are in addition to antisense oligonucleotides and RNA interference, which are discussed here, ribozymes and DNAzymes.

1.4.1 Antisense oligonucleotides

There are two mechanisms of action for antisense oligonucleotides; one of which is that the hybridization of the oligonucleotide to the target mRNA inhibits translation by steric block. The hybrid formation will interfere with the ribosome’s access to the mRNA or its ability to proceed, thus stopping the translation. The second mechanism is a continuation of the first, where the DNA:RNA duplex is recognized by the ubiquitous ribonuclease RNase H. The catalytic RNase H is a sequence independent endonuclease that hydrolysis the RNA strand in RNA:DNA hybrids but will not degrade DNA or unhybridized RNA. The truncated mRNA is quickly degraded by exonucleases due to the exposed ends (Figure 3). The specificity of RNase H towards RNA:DNA heteroduplexes is a limitation for the degrading pathway, which is the more efficient, for the majority of antisense oligonucleotides composed of DNA analogs. Antisense activity occurs also in the nucleus. The hybridization to mRNA then blocks nuclear export (59).

If the antisense oligonucleotides are designed to target pre-mRNA, i.g. splice regions, it can cause alteration of the splicing by blocking its machinery (Figure 4). Kole and colleagues demonstrated the capability of inducing exon skipping in a thalassaemia mutation of the β-globin gene. This introduced a novel 5’ splice site in the second intron of β-globin, resulting in the activation of a cryptic 3’ splice site, inclusion of extra exonic sequence in the mature mRNA, and premature termination of the β-globin reading frame (60,61).

The system has been modified into an assay for evaluating the efficacy of cellular delivery and nuclear translocation of antisense oligonucleotides. The reporter gene luciferase is interrupted by the mutated Figure 4: Exon skipping induced by antisense second β-globin intron, which causes oligonucleotides. A. Normal splicing of the aberrant splicing of luciferase pre-mRNA, pre-mRNA. B. An antisense oligonucleotide interferes with the splicing machinery by preventing correct translation. However, blocking the 5’ splice site. No cryptic splice treatment with antisense oligonucleotides sites are activated in this schematic. induces correct splicing, restoring luciferase activity. Inhibition systems suffer from several drawbacks: low signal-to-noise ratio,

19 delayed response originating from the stability of protein translated from the targeted mRNA, and unknown contribution of cytotoxicity from a given strategy to global protein synthesis. This positive assay avoids these impediments and, furthermore, it verifies the location of the oligonucleotide(62,63).

1.4.2 RNA interference

The latest contribution to the antisense family is RNA interference (RNAi). The breakthrough came when Fire and Mello followed up the observation that, either sense- or antisense-orientated RNAs inhibited gene function, by injecting both the sense and the antisense strands into worms to elucidate if they had an additive effect. However, the effect was not additive but rather a 10-fold increase (64). Fire and Mello received the Nobel Prize for their discovery in 2006.

Retrospectively, the phenomenon had been identified in plants earlier but the mechanism was thought to be saturation of the investigated machinery rather than specific degradation of the mRNA. Double-stranded RNA was already at that time also a well- established method to silence gene expression in mammalians but in a non- specific manner. Translation was arrested due to interferon response, recognizing dsRNA longer than 30 base pairs as viral. Activated protein kinase R (PKR) stalls translation by phosphorylation of the translation initiation factors. Another mechanism is mediated by oligoadenylate synthetase, which activates RNase L induced digestion of dsRNA by (65-67). Both responses are sequence-unspecific and mask any RNAi. Nematodes lack this defense mechanism, only the RNAi effect was identified when the 742 bp dsRNA was administered. Base-paired 21 and 22 RNA

Figure 5: RNA interference. Dicer cleaves the dsRNA non-specifically in duplex of approximately 21 bp with 2 bp overhang in the 3’-end. One of the strands from the active siRNAs is loaded into the RNA- induced silencing complex (RISC). The target mRNA is cleaved by argounate (ago), a member of RISC, after hybridization to the siRNA antisense strand.

20 oligonucleotides with overhanging 3’-ends were shown to mediate efficient sequence- specific mRNA degradation in lysates prepared from Drosophila embryos by Tuschl et al. and they showed that this design also is active in mammalian cells (68,69).

RNAi plays an important biological role in regulating endogenous gene expression and protecting the genome against viral infections. It has also been shown to reduce instability of the genome in C. Elegans, caused by the accumulation of transposons and repetitive sequences, inherited from parental infections (70,71). The human genome consists of 3 billion base pairs and only about 1.2% codes for proteins. However, a much larger fraction of the genome is transcribed, resulting in 98% non-protein-coding RNAs; thus there are many potential RNAi molecules endogenously expressed (72,73). Depending on species, structure and origin of the dsRNA, the nomenclature and protein complex compositions differ, but the basic mechanism is similar for all. Herein, only exogenous RNAi will be discussed.

The ribonuclease protein Dicer binds and cleaves exogenous dsRNA molecules to produce double-stranded fragments of 20-25 base pairs with a few unpaired overhang bases on each end. The short double- stranded fragments, called short interfering RNAs (siRNAs), are bound by several proteins, collectively named RNA-induced silencing complex (RISC). The strands are separated by RISC and the strand with the more stable 5’-end is usually integrated into the active RISC complex. When the antisense strand, associated with RISC, has bound the target mRNA, the catalytic RISC protein, a member of the argonaute family, cleaves the target mRNA (74) (Figure 5). The cleavage occurs between the 10th and 11th base pair, in respect to the antisense strand(75).

Hence, the smallest active molecule for RNAi is siRNA, typically 21mers of dsRNA, with 2 nucleotides overhang at each 3’-end with stable 5’-end of the antisense strand.

Figure 6: Transcriptional gene silencing by siRNA. The RNA-induced transcriptional complex (RITS complex) separates the strands of the siRNA and guides it to its target site, usually a promoter, within the chromosomal DNA. The catalytic part of RITS, argounaute (ago), then causes histone and/or CpG methylation, which silences the promoter and the gene of interest.

RNAi, and in particular siRNA, was found to induce silencing also at a transcriptional level in mammalian cells 2003 (76). Transcriptional gene silencing in mammalian cells can methylate cytosine in CpG motifs (77) and induce chromatin methylations (78) at positions that are characteristic of facultative heterochromatin (79). The RNA-induced transcriptional (RITS) complex binds the antisense strand of siRNA, which sequence specifically identifies the target sequence in chromosomal DNA, usually a promoter region, where an argounate homolog is responsible for the methylation of the histones (80) (Figure 6). However, some of the nuclear activity of RNAi might be pre-mRNA targeting with subsequent cleavage by argonaute (81), i.e post-transcriptional gene silencing.

1.4.3 Oligonucleotide inhibition of transcription

The last 15 years of advances in the field of DNA analogs have provided oligonucleotides capable of strand-invasion and hybridization to dsDNA with high affinity. If the binding is strong enough to sustain the transcriptional machinery, transcriptional gene silencing is obtained without relying on protein complex assemblies or enzymatic reactions. In earlier studies, it has been shown that triplex- forming oligonucleotides and their derivatives are able to invade the major groove of oligopyrimidine/oligopurine regions in dsDNA and interfere with gene transcription initiation and elongation (82-84). However, this Hoogsteen-binding dependent invasion is considered slow and the nucleotide excision repair machinery is targeted to helical distortions and altered hydrogen-bonding pattern caused by Hoogsteen binding (85,86). The first, supposedly minor groove binding DNA analog reported to have transcriptional gene silencing capabilities was PNA (87-89). However, linear PNAs only work efficiently at very high PNA to target site ratios (90). Corey and colleagues have also more recently published data describing transcriptional silencing of chromosomal DNA using linear PNA conjugated with polycationic peptides as well as lipid mediated delivery of LNA oligonucleotides (91,92).

In our hands, neither of the DNA analogs could efficiently silence gene expression as linear oligonucleotides when targeting an inserted BS in-between the promoter and the gene of interest. Oligonucleotides up to 27 bases were designed to bind either the coding or the template strand. The refined PNA oligonucleotide design, bisPNA, hybridizes to duplex homopurine sequences of DNA with an extremely stable PNA:DNA:PNA triplex with a looped-out DNA strand(93-95). Moreover, it has recently been reported that further optimized PNA, which is extended outside the bisregion, widens the application for strand-invasion and inhibits transcription in vitro(96,97). However, in our previous studies, we have shown that unless extremely high ratios of PNA to DNA target sites were applied, bisPNA could not efficiently block RNA polymerase II or III-mediated transcription (98,99).

The exception was an LNA construct called Zorro LNA consisting of one oligonucleotide binding to the coding strand, whilst another LNA oligonucleotide binds to the opposite template strand. These two oligonucleotides are linked together by anti- parallel hybridization, creating a Z-like structure, which seems to strand-invade into super-coiled DNA and simultaneously hybridize to both strands (Figure 7). Zorro LNA induces sequence-specific inhibition of both RNA polymerase II and III-dependent

22 transcription (98,99). Noteworthy is that two adjacent target sequences are needed for potent inhibition and that, to date, only a single target site sequence has been evaluated. Figure 7: Blocking transcription by Zorro LNA. Two LNA rich oligonucleotides are The Zorro LNA has been prehybridized with simultaneously hybridized to the coding and the target plasmid, or cotransfected, the template strand in dsDNA. The requiring 10-fold and 150-fold Zorro LNA oligonucleotides are interconnected by base to target site ratio, respectively for ≥90% pairing. inhibition. The ability of Zorro LNA to inhibit target gene expression of prehybridized target plasmid was shown in two mouse model systems using either hydrodynamic delivery of the nucleic acids to mouse liver or intramuscular injection. The degree of inhibition of pol III-dependent transcription in mice was similar on days 1, 3 and 6 (98,99). As previously reported, the persistence seen in inhibition might in part also be due to loss of DNA or to promoter silencing (100). Paper IV in this thesis is addressing the design of the Zorro construct, focusing on the linker region, by which the two oligonucleotides are hybridized.

Figure 8: Overview of antisense technologies discussed herein. Post-transcriptional gene silencing is mainly found in the cytoplasm where the translation is blocked sterically by hybridization or cleaved by RNase H (DNA antisense oligonucleotide mediated) or by argonaute (ago, siRNA mediated). Splice alteration will not silence the gene but by changing the splicing of the pre-mRNA, a new mRNA will be processed. Transcriptional gene silencing in the nucleus is mediated by methylation of histones and CpG regions by RITS assisted siRNA. Novel DNA analogs have shown to be able to stop the polymerase machinery by steric block, for example Zorro LNA.

1.5 NUCLEIC ACID DERIVATES

One of the major challenges for oligonucleotide mediated gene therapy is the stability. Oligonucleotides in their natural form as phosphodiesters are rapidly hydrolyzed by nucleases in blood, extracellular fluid and cells. The half-life in plasma is only a few minutes for unmodified nucleotides but by chemical modifications, the stability can be increased significantly. Nucleic acid analogs and mimics are commonly modifications of native nucleic acids at the nucleobase, the sugar ring or the phosphodiester backbone. The DNA analogs are clustered corresponding to their introduction in antisense technology. Modifications of the nucleobases are omitted in this thesis, comprehensively reviewed by Herdewijn (101).

The first generation of DNA analogs is the relatively simple modification on the phosphodiester linkage by replacing one of the non-bridging oxygen molecules with sulfur, called phosphorothioate chemistry (Figure 9). The oxygen substitution is known to dramatically improve biologic stability with a half-time of an hour extended to 10 hours in human serum (102-104).

The modification was later found to have other side effects due to increased non-specific association with plasma proteins (105), primarily heparin and albumin. Antisense oligonucleotides are subject to rapid renal clearance, since they in general are rather short. The theoretical lower limit of an antisense oligonucleotide is 15 bases with respect to specificity (106), but they are typically 15-20 bases long, which is optimal for cellular uptake (107). A positive consequence of protein association is that the oligonucleotide is protected against renal clearance thanks to the increase in size of the complex (108). Their extended time in circulation promotes a more broad distribution to organs and tissues, with the highest accumulation in liver and kidney (109,110). Figure 9: Modification On the other hand, the association with plasma proteins has of the phosphate back- resulted in complement activation (108,111). Another bone. The substitution of a non-bridging oxygen shortcoming is the slight decrease in melting temperature, for a sulfur improves the about 0.5°C per nucleotide, partly compensated by an biological stability increased specificity compared to phosphodiester (112). dramatically. The second generation of modifications is focusing on the weaknesses of phosphorothioate. Therefore, the second generation has slightly increased melting temperature and are less toxic than its predecessor. The members of this class have 2’- O modifications such as; fluoro, pentyl and propyl, but the most widely used are 2’-O- methyl and 2’-O-metoxy-ethyl RNAs (Figure 10). A common feature is that they, in contrast to phosphorothioate modifications, do not have RNase H activity. This can be circumvented by usage of gapamers, i.e. an oligonucleotide that can have any kind of modifications in the ends but only nucleotides with natural sugar rings in the middle, which is the site of RNase H activity (113). This is a strategy found useful for a variety

24 of DNA analogs that lack RNase H sensitivity. However, in the case of splice alteration, RNase H activity is not desired

Figure 10: Second generation of analogs. RNA modifications, that increases the binding affinity through 2’-O-modifications. Though, these modification abolish RNase H recruitment.

The third generation is more diverse, presenting modifications of the sugar ring and/or the phosphodiester backbone. Conformational restriction has widely been used to improve melting temperature and nuclease resistance. The sugar ring has the freedom to switch between two configurations in DNA and RNA. Analogs have been developed to have totally constrained ribose ring in bicyclic nucleic acids (BNAs), which is implied by the name “locked” of the most well known variant, LNA. Other modifications have restrained ring structures, like tricyclo-DNA (tcDNA) and cyclohexene nucleic acid (CeNA). Morpholinos were developed as a cost-efficient analog utilizing less expensive ribonucleosides for oligonucleotide synthesis and avoiding expensive catalysts and post-coupling oxidation steps. In addition to the restrained morpholine ring, the anionic phosphodiester backbone is replaced with a non-ionic phosphoramidites, thus avoiding electrostatic repulsion towards its target. This feature is adopted by other techniques; an example PNA. The ribose phosphate backbone is replaced by non-charged polyamide linkage, making the oligonucleotides neutral and flexible. We will focus on the Danish analogs (PNA and LNA), which together with morpholinos (114) are the most promising third generation analogs with therapeutic properties (Figure 11).

Figure 11: The three most widely used, commercially available, third generation DNA analogs.

1.5.1 Peptide nucleic acids

PNA (44,115-118) is composed of repeated N-(2-aminoethyl)glycine units with the nucleobases attached through methylenecarbonyl linkers. The achiral backbone of PNA is as the name indicates similar to a peptide with N and C terminus. In comparison to phosphate backbone, PNA has poor water solubility. Their tendency to aggregate is length and purine to pyrimidine ratio-dependent (119). Cationic amino acids are usually added to improve the solubility, which furthermore increases the binding affinity to anionic oligonucleotides. The electrostatic attraction is of non-specific character, which possibly might cause off-target effects.

The thermal stability of PNA:DNA and PNA:RNA duplexes is higher compared to DNA:DNA and DNA:RNA duplexes (120,121). The stronger binding is largely attributed to the lack of charge repulsion between the neutral PNA and the negatively charged DNA or RNA strand. A direct consequence of the neutral backbone is also that PNA:PNA duplexes are insensitive to variations in salt concentration and PNA:DNA only exhibit minor dependence from 10-500 mM sodium chloride (122). This is in contrast to DNA:DNA duplexes, which are highly dependent on ion concentration. At low salt concentration, PNA can bind to a target sequence at temperatures where DNA:DNA hybridization is suboptimal.

Interestingly, a mismatch causes a greater disruption in a PNA:DNA duplex than a mismatch in a DNA:DNA duplex. The average cost of a single mismatch is estimated to be in the order of, or larger than, the gain of two matched base pairs, which explains the high degree of sequence specificity generally observed for PNA binding to a ssDNA target. The composition of neighboring base pairs has been found to be a primary determinant of mismatch stability, where flanking G-C pairs seem in general to stabilize the mismatch relative to flanking A-T base pairs. In this respect, mismatches in PNA:DNA duplexes exhibit behavior similar to that of DNA:DNA (123) and RNA:RNA (124) duplexes.

The structures adopted by duplexes of PNA-nucleic acid hybrids can be characterized as having Watson-Crick type base pairing and stacking patterns similar, though not identical, to those of DNA:DNA duplexes. The PNA:DNA duplex is somewhat unwound with a helical rise of 42 Å with 13 bp per turn (compared to DNA:DNA with 10.4 bp per turn), and a slightly larger helix diameter of 23 Å, with a major groove wider, and a minor groove narrower (125-127), than those of the DNA:DNA duplexes (128). PNA adapts the sugar puckering of its partner oligonucleotide to a great extent when hybridized to DNA or RNA, having a helical characteristic of both A and B- form.

PNA was initially considered to mimic DNA in the sense that DNA oligonucleotides can bind to dsDNA as triplex forming homopyrimidine oligonucleotides (129,130) (Figure 12A). The DNA duplex is instead invaded by a Watson-Crick hybridized oligonucleotide anti-parallel orientation plus the expected Hoogsteen hybridized oligonucleotide in parallel orientation (Figure 12B). The strand-invasion can be promoted by PNA-peptide conjugates or preferably by conjugating the Hoogsteen and Watson-Crick binding oligonucleotides by a flexible linker (Figure 12 C). The bisPNA

(or PNA clamp) interacts more rapidly with dsDNA relative to monomeric ones, but still exhibits excellent sequence specificity(48,93-95). The bisPNA and the target DNA strand forms a triplex with a new helical structure, called the P-form (131). This unique helix form is wide and has a large pitch (28 Å and 18 bp, respectively), with a base tilt similar to B-form helix. The bases are displaced from the helix axis even more than in A-form helix. Hydrogen bonds between the DNA backbone and the Hoogsteen PNA backbone explain the increased affinity and stability of these hybrids. Nucleobase modifications makes it feasible to even achieve double strand-invasion, utilizing pseudo- complementary bases (132) (Figure 12D). At least partial strand-invasion can be obtained in linear dsDNA but binding sites in negatively super-coiled plasmid can be saturated (133,134). An obvious disadvantage with bisPNA is that only pyrimidines Figure 12: PNA hybridization of are allowed in the sequence and since it is binding as dsDNA by: A. 1 PNA binding a clamp, it will be longer even though the binding Hoog-steen, B. 2 PNAs binding element can be reduced in comparison with linear Hoogsteen and Watson-Crick, C. 1 PNA binding both Hoogsteen and PNA. Further optimized PNA with a mixed bases Watson-Crick, D. 1 PNA binding extension outside the bis-region, increases the both DNA strands with Watson- probability to find suitable chromosomal target Crick base paring. sequences (96).

The present model for strand-invasion or strand displacement binding of PNA involves DNA base pair “breathing”, defined as the transient (lateral) opening of a base pair. The estimated base pair opening probabilities for A:T and G:C pairs are roughly 10-5-10-6 and 10-6-10-7, respectively (135,136), although sequence dependent (137). The exposure of the inaccessible bases, buried in the double helix, is crucial. This is at least partially, a rate-limiting step in the process. Rates of strand invasion are in analogy with rates of breathing and base stacking; temperature-, salt- and sequence-dependent (137,138).

In addition, PNAs are resistant to nuclease and protease degradation (139). PNA was designed with antisense in mind(140) but is a potent analog candidate for gene therapy, diagnostics, and molecular biology applications (141). Synthetic molecules, which bind sequence-specifically and with high affinity to dsDNA might provide a way for modulating gene expression. Thus, PNA has been shown to arrest transcription in cell culture through blockage of RNA polymerase elongation by DNA strand-invasion (96,97) and recently, Corey and colleagues showed, inhibition of chromosomal gene expression with PNA and for the first time also with LNA (91,92).

1.5.2 Locked nucleic acid

LNA (142-145) is a bicylic RNA analog restricted in a C3’-endo/N-type sugar conformation by an O2’ to C4’ methylene linkage (143,145). This lock reduces LNA’s flexibility and lowers the entropic penalty for hybridization, which results in increased melting temperatures to both RNA and DNA. Thus, LNA is less flexible than RNA and DNA, which have different degrees of the two possible states of the furanose ring. The ring can be either of S- or N-type (Figure 13A). RNA is to a high degree the N-type whereas LNA is locked in the N-type (Figure 13B). DNA is mainly S-type, which makes LNA more a RNA analog, forming slightly more stable hybrids with RNA than DNA. Incorporations of LNA monomers into DNA or RNA oligonucleotides change the configuration of the neighboring bases towards that of LNAs, i.e A-type helix with N-type sugar puckers (144,146). In other words, LNA monomers preorganize the oligonucleotide and the entropy penalty for hybridization is reduced.

Figure 13: The sugar rings have two possible configurations. A. DNA and RNA oligonucleotides have rather flexible structures that are able to adapt. B. The ring is locked into the C3’-endo (RNA) configuration by the O2’ to C4’ methylene linker.

LNA has been developed as a diagnostic tool for real-time PCR assays and single nucleotide polymorphism detection, were DNA probes have been potentiated by the substitution to LNA bases. It has been found that LNA pyrimidines contribute with more stability than purines, especially while preceded with an adenosine, but it is context dependent (146,147). LNA increases the melting temperature by either preorganization or improved stacking of the oligonucleotide, but not both simultaneously (146,147). Two factors are mainly responsible for the stability of the DNA double helix: base pairing between complementary strands and stacking between adjacent bases. The bases of the individual nucleotides are on the inside of the helix, stacked on top of each other like the steps of a spiral staircase. The ”steps” are stabilizing the double helix in addition to base pairing hydrogen bonds between T:A and C:G (Figure 14A). The GC:GC duplex has the greatest overlap and the largest stacking energy; GC:GC > CG:CG > GG:CC. Similar observation can be correlated among other duplexes, e.g. TA:TA, AA:TT and AT:AT duplexes(148).

The organization of the backbone and the sugar rings in a C3’-endo fashion favors a more efficient stacking. This is equivalent to the structural changes from B-DNA (C2’- endo conformation) to A-DNA (C3-endo conformation)(149) (Figure 14B). Base

stacking is a net phenomenon involving hydrophobic, electrostatic and dispersion components(150,151).

LNA has been used for antisense (152,153), siRNA (154), decoy (155,156), LNAzyme (157), alternate splicing (158) and aptamer (159) technology. These studies show LNA’s increased stability in vivo and also suggests that LNA has a low toxicity, although this is still debated (160). LNA is capable of strand-invasion of super- coiled plasmids (161), although in our experience, not as efficiently as PNA (116), and especially bisPNA (49,162). A direct comparison of LNA, PNA and DNA oligonucleotides shows the following relation between their melting temperatures at physiological conditions; LNA:LNA > LNA:PNA > PNA:PNA > LNA:DNA > PNA:DNA > DNA:DNA (163).

Figure 14: Helical stacking in A or B-form. Stacking occurs between adjacent nucleotides and adds to the stability of the molecular structure. A. illustrates how the aromatic rings in the bases of the nucleotides are positioned nearly perpendicular to the length of the DNA strands when a duplex is formed. Thus, the faces of the aromatic rings are arranged parallel to each other, allowing the bases to participate in aromatic interactions, creating a large net stabilizing energy of the helix. B. shows on the difference of stacking (top view) between the A and B-form helix, dominant in RNA and DNA, respectively.

A. In closing of this chapter it should be noted that combinations of modifications are standard-procedure today. Examples are LNA with phosphorothioate backbone and PNA with pseudoisocytosine bases for pH-independent Hoogsteen mediated hybridization (164). Other add-ons are intercalators that do not act in a sequence specific manner but further increases the melting temperature (165).

1.6 LIGAND-MEDIATED GENE THERAPY

The optimal gene therapeutic should be an efficient delivery of the intact nucleic acid to a specific cell type and no others, without stimulating a significant immune, inflammatory and cytotoxic response. After binding to the target cell, the nucleic acid should be internalized, escape the endosome, translocate intracellulary to the compartment of activity and finally potently express, repress or up-regulate the gene of interest (Figure 15).

The most concrete method to achieve cell specificity is to remove the patient’s cells and transfect ex vivo. The cells are then transplanted back into the patient. This approach is rather efficient and safe, depending on the vector used. However, it is expensive, might cause excessive suffering and be difficult to extract the target cells, depending on their nature. A more convenient approach would be systemic delivery of the gene therapeutic vehicle that possesses all the qualities for a safe, specific and efficient delivery in vivo. Nucleic acid has been formulated into complexes relying on electrostatic or hydrophobic forces to avoid degradation in the bloodstream and renal excretion but there are also examples of pegylation (conjugated with the inert polyethylene glycol). The complexation can also promote the cellular uptake in an unspecific manner. Cell targeting and efficient intracellular translocation is directed by receptor-ligand interactions. Ligands used for the purpose of gene therapy are usually of endogenous or viral origin and some examples will be given.

1.6.1 Cell specificity and entry

Carbohydrates and particularly N-acetylgalactosamine (GalNAc) recognizing asialoglycoprotein receptors (ASGPr) are abundantly and selectively expressed on hepatocytes. Hence this hepatic lectin has been well characterized and studied as a promising candidate target for drug and gene delivery into hepatocytes (166,167).

Short peptides can act as ligands and one of the shortest examples is the tripeptide RGD (arginine-glycine-aspartic acid) found on fibronectin (29). This peptide binds to integrins, cell surface receptors, frequently over-expressed in cancer cells. The tripeptide is more potent in cyclic conformation, cRGD (168). Truncation of proteins to find the smallest active peptide is a common method for identifying potential ligands. A field has emerged within gene therapy, focusing on oligonucleotide delivery by using short cationic peptide sequences in order to mediate cellular uptake. These peptides, collectively named cell-penetrating peptides (CPPs), were first identified while investigating the proteins involved in infection of mammalian cells by HIV. The TAT protein (Trans-Activator of Transcription) was found to penetrate cells, thereby being able of bystander effects (169,170). A positively charged peptide of 9 amino acids was found to be the minimum sequence required for translocation (171). Since these initial observations, other naturally occurring CPPs have been identified such as penetratin from the transcription factor antennapedia in drosophila (172,173). Furthermore, promising chimeric peptides like transportan have been synthesized (174), later optimized and named transportan 10 (175) but more simple designs of polyarginine has also been found potent (176).

Positive charge of CPPs is crucial. The general approach has been to conjugate non- charged DNA analog oligonucleotides directly to CPPs. The cellular uptake was found to be efficient but the therapeutic effect was typically lower than anticipated. As the name indicates, CPPs were originally thought to directly translocate the plasma membrane and gain access to the cytoplasm. This was later found to be a technical artifact and revised to make use of the endocytic pathway (177). Nevertheless, there are still few CPPs considered to be capable of entering via a non-endosomal pathway (178- 180). As a result of massive endosomal entrapment of peptides following endocytosis, modern CPPs are designed with the ambition of endosomal escape (181), e.g [R-Ahx- R]4 (182,183) and EB1 (184).

The entrapment of nucleic acid in endosomes seems to be a universal obstacle for all types of gene therapy. Co-administration with chloroquine, calcium or high concentration of sucrose has been used as endosomolytic reagents. Lipoplex and polyplex technology have efficient strategies for endosomal escape, destabilizing the endosomal membrane by pH-dependent disruption of the lipid bilayer or acting as proton sponges that causes osmotic collapse of the endosome, respectively (185,186). In experimental settings, large quantities of the lysomotropic reagent are normally utilized. Quantities of this magnitude could be toxic, leading to destabilization of cellular membranes and is unacceptable in the clinic.

1.6.2 Intracellular translocation

Plasmid DNA has been shown to remain at the site of microinjection, thus being too large to have diffusional freedom in the cytoplasm (187). Oligonucleotides on the other hand rapidly translocate into the nucleus upon cytoplasmic delivery (25). Consequently, oligonucleotide based gene therapy is potent without further designation, whereas plasmid based gene therapy needs to exploit the cytoplasmic infrastructure, at least in non-dividing cells.

The microtubule motor protein, cytoplasmic dynein, transports various cellular cargo by "walking" along microtubules towards the minus-end, i.e. towards the nucleus. Dynein is a protein complex involved in transportation of many different molecules. It binds for instance to a conserved amino acid recognition sequence found in a wide variety of proteins that potentially can be used for plasmid translocation (188).

All active and passive transport into and out of the nucleus occur through the nuclear pore complexes (NPCs) present in the nuclear envelope. The pores limit free diffusion of globular proteins over 40 kDa into the nucleus (189). The spatial segregation of DNA replication, RNA biogenesis in the nucleus and protein synthesis in the cytoplasm of eukaryotic cells are therefore using NLS, or nuclear export signals (NES) to enter or exit the nucleus, respectively. In spite of this, the first NLS peptide found is of viral origin, a truncated form of the protein found in SV40 (32,190). The basic paradigm for nuclear import is that the NLS cargo is bound by an import receptor in the cytoplasm, translocated through the NPC, and released in the nucleus. In addition to NLS, Lentivirus utilizes G-quartets, also called DNA-flaps. They bind to cellular proteins and form a large DNA-protein complex, the pre-integration complex (PIC) which is

31 translocated into the nucleus allowing integration into the host cell genome (191,192). However, a recent study states that DNA-flaps are only involved in uncoating of HIV-1 and lacks nuclear import abilities (192).

Finally, expression is the last step of controlling the gene delivery. Cell specific promoters offer the possibility to have gene expression only in the target cell even if other cells and tissues are transfected (193).

Figure 15: Schematic presentation of the main hurdles in gene therapy and systemic gene delivery. 1. The DNA should resist extracellular nucleases and inactivating plasma proteins. (Oligonucleotide mediated gene therapy has to avoid renal excretion). 2. Binding to the cell surface, preferably cell specific interaction. 3. Uptake is usually directed through receptor-mediated endocytosis. 4. The DNA is trapped in the endosome, 5. From which it must escape before hydrolysis occurs due to decreasing pH. 6. Large molecules have to be transported to the nucleus, using the infrastructure of the cell, whilst smaller molecules can rely on diffusion. 8. The nuclear pore complex requires active translocation of macromolecules. 9. Many gene delivery techniques involve packing of the DNA. To be accessible for the transcription machinery, it has to be unpacked. 10. Finally, the gene of interest should be efficiently expressed and not become methylated. Further specificity can be gained by choosing a cell specific promoter.

1.6.3 Combinatorial ligands

The oligonucleotides have been equipped with ligands originating from viruses, endogenous proteins or synthetic new designs, to promote cellular binding and uptake. However, several viruses have been shown to interact with more than one receptor for optimal internalization. For example, adenovirus binds to integrins and to the CAR protein (194,195) whereas HIV binds to both CD4 and chemokine receptors, such as CCR5 and CXCR4 (196). These interactions are crucial and it has been demonstrated

32 that blocking the CCR5 receptor by siRNA technology impairs viral infection (197,198). Chimeric peptides are becoming more a rule than an exception in CPP design (180,199). Paper IV presents a cost efficient approach to find suitable ligand combinations, not restricted to peptide but of any chemistry and origin. The novel concept is based on DNA-directed nanoassembly of supra-molecular constructs, which are capable of carrying cargos across cellular membranes. DNA-based nanofabrication allows for fabrication of spatially well-defined, molecular devices by taking advantage of the unique specificity of Watson-Crick base-pairing of designed synthetic nucleic acid moieties (200,201).

2 AIMS OF PRESENT STUDY

The aim of this thesis was to further develop Bioplex, a non-viral gene therapy technique, for attachment of biologically active molecules onto plasmid DNA by sequence specific strand-invasion generated by superior DNA analogs. The ambition was to potentiate the Bioplex technique as a result of; i) increased complex stability, ii) multiplicity of ligands per plasmid iii) combinations of ligands, ideally in a combinatorial manner. Furthermore, to exploit the Bioplex strategy for smaller gene therapeutic vectors, and to elucidate novel applications for DNA analogs.

Specific aims in the individual papers were:

I To simplify the assembly of multiple biologically active molecules per plasmid binding site through single-stranded hybridizations, created by strand-invading DNA analog oligonucleotides with single-stranded extensions. II To detach biologically active molecules from the complex in a controlled manner. A proof-of-concept of removing used ligands and an example of how to avoid receptor or protein entrapment. III To present a high-throughput screening assay, in search of ligands with combinatorial effects. The assay should be based on cellular uptake of DNA oligonucleotide constructs in complex with biotin-labeled ligands. IV To optimize the DNA analog oligonucleotide based transcription inhibitor, Zorro LNA. Elucidating the origin of its potency, focusing on the linker region.

3 RESULTS, DISCUSSIONS AND FUTURE PLANS

3.1 PAPER I

This study is an extrapolation of the findings and ideas obtained in the preparation of earlier publications(41,49), addressing the assembly and stability of Bioplex at physiological salt concentrations. We have earlier shown that PNA with covalently linked FEs can be utilized for anchoring functionality to naked DNA. When bisPNA strand invades a plasmid, it displaces the DNA strands. Lundin et al. demonstrated the possibility of using LNA oligonucleotides to hybridize to the displaced strand of the plasmid for both increased stability and binding kinetics (49). However, the number of FEs per plasmid might be a bottleneck. Virus particles display hundreds of ligands. Having only one FE per anchor would then require many anchors that might destabilize the plasmid. Synthesizing multiple FEs per anchor would greatly increase the manufacturing cost. Instead, the LNA anchor, which is the first oligonucleotide hybridization, a) could be extended Figure 16: Self-assembling with uncomplementary bases, with respect to the oligonucleotides, hybridized to plasmid, and these bases would be used for plasmid. a) LNA anchor oligonucleotide with extension. b) hybridization of an second oligonucleotide (b) with Secondary oligonucleotide with BS multiple anchor-binding sites for a third for the third oligonucleotide c). hybridization (c) and so forth (Figure 16).

An exponentially growing number of biologically active molecules could be presented on the plasmids by self-assembling hybridization of single-stranded oligonucleotides extending from the plasmid. Sequences were carefully designed to only give the correct final product in a predefined fashion.This study only presented up to three consecutive hybridizations but could be extrapolated. The SUN technology only needs a few anchor-binding domains in the plasmid for the addition of large number of FEs. The anchors are shown to be a Self-assembling and UNiversal tool and when the anchor is inserted into the plasmid, the extension can be used for hybridization of any kind of oligonucleotide. The oligonucleotide itself could be the function, acting as an aptamer. FEs can be spaced as brush-like structures or exactly positioned, relative to each other, to give combinatorial effects (Figure 17A).

Some molecules have to be combined with correct relative orientation for effect(202). Example presented in this study is, surface adhesion of plasmid via SUN attachment in an exact orientation for purification. If the process is reversible, this could be a method to purify hybridized plasmids or to concentrate DNA. Cleavable linkers can be included in the complex for controlled release (203). The single-stranded oligonucleotide extending from the plasmid has also the capability to hybridize with another plasmid, directly or indirectly. Even if dimers can be formed in the absence of anchors (204), anchors would simplify their formation. Plasmids with pyrimidine-purine tracts can form both inter- and intramolecular triplexes. Thus, Hoogsteen base pairing between plasmids can cause dimerzation but only at low pH and has limited stability at physiological conditions (205,206) (Figure 17B).

Self-assembling systems in nanotechnology require suitable building blocks and accessible sticky-ends for the generation of bottom-up constructed structures. Most structures are assembled by ssDNA oligonucleotides hybridized to form junctions. Rothemund recently published beautiful atomic force microscope (AFM) pictures of the western hemisphere in scale 1:2×1014 drawn by scaffolded DNA origami (207).

His starting material was the naturally occurring Figure 17: SUN technology applications. A. Plasmid with ssDNA (ssDNA) of the M13mp18 virus, which was combinations of multiple ligands. folded in different ways by synthetic ssDNA B. Dimerization of plasmid DNA. oligonucleotides, showing the relatively simple self- C. Nanoelectronic circuit with DNA wires. assembly of ssDNA. Others have used DNA oligonucleotides to generate larger building blocks

38 with different geometry and rigidity in order to form grids or supra-molecules in a “Lego-like” fashion (reviewed in (208,209)), for construction of, for example, nanomotors (210,211) and nanoelectric circuits (212-215) or DNA lattice for organizing molecules (216) as well as for single molecule detection (217). We believe that future applications of the SUN technology could primarily be as double-stranded backbones in simpler structures and lattices for nanoelectronics (Figure 17C) or single molecule detection, in addition to gene therapy.

3.2 PAPER II

The objective of this study was to investigate the use of sequence-specific cleavage for removal of biological functions from the Bioplex conjugate. By choosing a specific enzyme and introducing its target sequence within the peptide, it was shown that it is possible to engineer not only where, but also when, cleavage will occur during the transfection. Cathepsin L was chosen in order to obtain proof of principle on the premises of its ubiquitous expression, specific activity and intracellular location, being present in the late endosome/early lysosome. The assembled Bioplex was exposed to purified cathepsin L. Digestion occurred and resulted in loss of the functional entity although the kinetics were context dependent. The functional entity was found to be deleted in cell-culture in a manner compatible with endosome maturation and pH activation of lysosomal proteases, i.e. cathepsin L.

Cathepsin L is active at pH 5.5, when the endosome is converting into a lysosome and degradation is imminent. Thus, products might even be degraded at the time of cell harvesting, resulting in loss of material in vivo. Compatible with this possibility, the shift, which illustrates the loss of a peptide in vivo, was not as pronounced as in the in vitro studies. A more homogenous Bioplex product might be detectable if escape from the endosome, or degradation within the lysosome, are prevented. When using PNA409 with two different functional peptide entities, the transferrin peptide could be removed by cathepsin L in vitro with sustained NLS activity in vivo. No significant difference between digested and undigested PNA409 could be detected (Figure 18).

Figure 18: Protease-induced release of functional entity.

In this study, constructs with different functional entities were tested, but only one type of linker. To overcome accessibility problems with bulky or charged functional entities, the nature of linkers could be the solution. Length, hydrophobicity and flexibility are some of the variables of interest. When a larger degree of cleavage is crucial, alternatives to enzymes might be more appropriate. Enzymes are proteins of substantial size, whereas non-catalytic disruption can be obtained by considerably smaller

39 molecules. In reactions with chemical linkers, which are broken as pH drops, or through redox reactions, the reactive small molecules do not experience the steric hindrance of enzymes. Disulphide bonds are known to be reduced in the cytoplasmic milieu. Disruption of disulphide bonds has been used in non-viral gene therapy to free DNA from cationic lipids after endocytosis (218,219). Bioplexes could therefore be synthesized with a disulphide bridge located between the peptide and the functional entity, eliciting a gain in cleavage rate with a concomitant change in subcellular targeting. In this study, we chose an enzyme present in a particular compartment of all eukaryotic cells, whereas redox reactions would have a different spatial distribution.

The capacity to release functional moieties could be advantageous at many of these steps: i) If the receptor-ligand interaction is very strong, it might impair the release of genetic material from the complex, causing reduced nuclear translocation of transfected material. ii) Cleavage of the receptor ligand after uptake will reduce the bulkiness of the transfection complex, thereby potentially enhancing intracellular transport at multiple levels, including the final routing to the nucleus, resulting in increased transfection efficacy. iii) The released moiety could also serve as a functional unit, such as a lysosomal disruption peptide. As an example, the structural and functional information available for the HA2 peptide indicates that the peptide has to be untethered to allow it to form homo-trimeric complexes, which subsequently form octameric complexes that create a pore in the vesicle membrane (220). It is feasible that other types of functional moiety could also be activated in this fashion. However, the choice of cathepsin L for HA2 release might not be the optimal one. Thus, cathepsin L only becomes active as the endosome is converting into a lysosome where the degradation already has started. The HA2 peptide should ideally integrate into the membrane before the pH drops and its configuration changes. For this reason, its induced release might be too late, or at least suboptimal. Another potential limitation of cathepsin L would be its ubiquitous expression, preventing tissue specificity. iv) A special case is when multiple functional entities are linked consecutively onto the same PNA, and the functional peptide must be exposed at the terminus to be active.

Our findings demonstrate the feasibility of altering the composition of the Bioplex. This has implications for controlled plasmid-mediated gene delivery. By the choice of disrupting agents, alterations of the constructs can be directed to e.g. the cytoplasm or the cell surface. The strategy of specific molecular switches to induce loss of functional entity or exposing consecutive functions etc. could be used to obtain cell or tissue specificity. Regulation through different systems in vivo transfection and a release of functional moieties in cells expressing the correct protease, for which a target site is present in the Bioplex, is one of them. Receptor interaction with a defined receptor ligand or combination of ligands targeting a selected subset of cells, as well as inclusion of cell-specific promoters in the plasmid construct, are other examples of how specificity might be achieved.

3.3 PAPER III

Here, we turn to the ligands. Several viruses have been shown to interact with more than one receptor, for example HIV, which binds to both CD4 and chemokine

40 receptors, such as CCR5 and CXCR4, or adenovirus, which binds to integrins and to the CAR protein. These interactions are crucial and it has been demonstrated that blocking the CCR5 receptor by siRNA technology impairs viral (221). We have designed bioactive complexes with defined structures and charges. The oligonucleotide complex was designed as a 44 bp dsDNA oligonucleotide with one central streptavidin molecule and a second streptavidin at the terminus (Figure 19A). The complexes can be utilized for determination of cell-specific adhesion, uptake and intracellular localization dependent on the specific nature of each complex. Importantly, it was observed in studies for protocol optimization that the 44mer carrier oligonucleotide was internalized only when ligand-loaded streptavidin- DNA-conjugates (preferably containing wheat germ agglutinin, WGA) were hybridized to it. Thus, this observation clearly indicates that the entire construct crossed the cell membrane, verified on unfixed HUVEC cells by confocal microscopy. It should be emphasized that this result confirms our hypothesis, that DNA-based supra-molecular constructs posses sufficient stability in complex media and are thus well suited for combinatorial screening and interrogation of cellular systems. However, the final screening in this study was only analyzing cellular association of the complex after fixation, unable to discriminate between membrane bound, internalized or compartmentalized (177).

The experimental setup ensures the generation of defined complexes with ligands designated specifically to either “OligoA” or “OligoB”. An alternative strategy would be to have equimolar solutions of the ligand combinations and incubate with streptavidin alone. However, this would give a random distribution of all possible combinations, thus reducing the level of control that we have in the present set-up. There is a potential risk for ligand exchange on the streptavidin, i.e. larger biotinylated molecules are exchanged for smaller as a result of less steric hindrance and low diffusion rate. Blocking available binding pockets on streptavidin with free biotin would decrease these interactions. Mixing OligoA and OligoB immediately prior to adding the complex would reduce the risk of ligand exchange. The assembly of the 484 different constructs was accomplished in 3 hours and could be administered to Figure 19: Schematic the last of the 10 000 wells within 3 hours. Thus, ligand representation of the exchange should be evaluated. Here, the mass of the oligonucleotide complex. ligands spans from the RGD-peptide of less than 1 kDa A. Present complex B. An example of future to transferrin, which is 80 kDa. The majority of the complexes. ligands used have a mass of 2-5 kDa.

An alternative method to labeling the scaffold oligonucleotide “OligoC”, the oligonucleotides OligoA and OligoB, is to conjugate the scaffold with fluorescent streptavidin. Commercially available fluorescent streptavidin yields a stronger signal and has on average 5 fluorophores per molecule, thus giving a higher signal-to-noise ratio. However, it would also give rise to signal from incomplete complexes. An

41 oligonucleotide complex with fluorescent streptavidin would present a more direct comparison of the ligand combinations efficacy in transfecting negatively charged cargos relative to non-charged cargos, i.e. with or without the charged DNA oligonucleotides. The absolute intensities between the two complexes could be compared. The choice of streptavidin as connector between the ligand and the DNA is not optimal. NeutraAvidin for example, is a uncharged deglycosylated form of avidin and would be a better candidate, since it is less likely to have unspecific interaction with the cell membrane and thus cause higher variance (Figure 19B).

The ligand positioning was not anticipated to be of any significant importance contrary to the results. Switching position was thought to be an internal control to validate the sequence independence on cellular uptake. Instead the positional effects were far greater than the hybridization efficacy or concentration errors of the oligonucleotides can be accounted for. The effect was seen in both directions, i.e. in HUVEC, WGA has to be ligandA but in HEL299 cells, WGA should be ligandB and a carbohydrate should be ligandA. The reason for the remarkable differences for the HUVEC cells might have a more technical than biological explanation; OligoA and OligoC were prehybridized for convenience. This allowed the OligoA to hybridize before ligands were added to the conjugated streptavidin, but OligoB was first exposed for ligands that probably caused a steric hindrance effect on the subsequent hybridization to OligoC. This difference was most pronounced for HUVEC since these cells were very susceptible to WGA binding. The majority of the ligands are less likely to cause interference of the hybridization due to their relatively small sizes, compared to streptavidin, and only WGA and transferrin have molecular weights and sizes of normal proteins. If the streptavidins would be positioned in each end of the DNA complex, position would most likely not influence receptor interactions and switching positions of ligands would instead be included as internal controls.

The Z’ factor was developed as assay quality assessment for high-throughput screenings of larger magnitudes and has been used here as an optimization tool. An ideal method has Z’=1 and a value ≥0.5 indicating an excellent assay with statistical reliability (222).

An oligonucleotide receptor has been identified in Hep G2 cell line (223). This could explain the high background of the oligonucleotide complex and there might be a similar effect in HEK-293. Other striking differences between the cell lines are their shape and size. HUVEC has a flat morphology with large cytoplasm and suits the detection technique well. This is also true for HEL299 with the difference being that HEL299 has a more irregular shape and is larger. Even though the number of HEL299 cells per field of view is lower than for HUVECs, the cell-to-cell variance was low, suggesting HEL299 to be a good cell-type for this assay. However, the WGA ligand was not efficient in cellular binding to HEL299. The Z’ factor is therefore low for HEL299, but this might simply be due to the fact that WGA is not an adequate positive control in this case. HEK-293 and Hep G2 are rather small, spherical cells that tend to grow on top of each other. They both have relatively high level of auto-fluorescence in the spectrum used for imaging and a large nucleus to cytoplasm ratio, characteristics that are suboptimal for data acquisition with the present imaging methodology. Also SH-SY5Y and SK-HEP-1 grow in clusters but are otherwise good candidates. Two

42 liver-derived cell lines were included in this study, since we have special interests in hepatic uptake through the asialoglycoprotein receptor with carbohydrates as ligands. Hep G2 expresses the asialoglycoprotein receptor but SK-HEP-1 does not. Thus, the more cumbersome cell line, Hep G2 was included in the study with SK-HEP-1 as a negative control. Finally, the washing procedures have been developed for HUVECs, other less adherent cell lines could possibly detach to a higher degree than the HUVEC cells. Some wells had cell free areas but rarely with cell traces, stained by Alexa 647 Fluor conjugated WGA. After receptor-ligand interactions with basolateral receptors, cells are more vulnerable than normally and might even detach without exterior forces. The overall characteristics will reflect how well the explicit cell line will function in this particular assay, reflected in the Z’ factor.

A high-throughput approach is pivotal for finding functional, combinatorial ligands in an efficient manner. This study shows the strength of our technique for identification of ligand combinations as well as the need for automation and advanced imaging systems. Only 22 different ligands were tested in this pilot and the total number of transfections performed was in the order of 10 000. The Z’ factors obtained were low and show that each cell line has to be optimized separately in respect to controls used. HUVEC, which the protocols have been optimized for, has an acceptable Z’ factor. The foundation of this study is the automated robotic system with Liconics robotic incubator, two autonomous robotic arms equipped with for example 96-tip fixed pipette head, plate readers, plate washers and Velocity 11 plate sealer. Everything is integrated in the same hood (Figure 20). The cells are analyzed by an automated laser line-scanning confocal microscope system, IN Cell Analyzer 3000 (GE Healthcare, USA). Hence, there are no manual steps.

Figure 20: Photograph of the robotic platform used in paper III.

The oligonucleotide complex is to be attached directly to plasmid DNA in order to transfer the plasmid across the plasma membrane and/or to facilitate the intracellular translocation of the plasmid. Owing to the possibility to strand-invade dsDNA by the usage of high affinity DNA analogs, these anchoring oligonucleotides could be a part of, or hybridize to, the oligonucleotide-ligand complex (162). The potency of ligands is cargo-dependent. Most of the post-translational gene regulatory techniques are based on oligonucleotides between 20 and 100 bases in length. We postulate that the oligonucleotide-based complex presented here could be used to find common ligand combinations that can be directly transferred to the gene regulatory constructs, without

43 compromising neither ligand efficacy nor specificity. The methodology delineated in this work has the potential of having a significant impact not only on delivery of nucleic acids but also on drug delivery. The prospect of systemic delivery combined with highly specific targeting of cell-types will enable re-formulation of a plethora of existing drugs that today have serious side effects due to non-specific uptake.

Future studies aim to have a biological readout and not relying on fluorescence. The complex can be modified to have antisense ability or other kind of oligonucleotide mediated gene therapy effect (Figure 18B). Splice correction, is an appealing alternative with a positive readout, possible to measure directly in a plate reader. The advantage is a fast readout but the results would then only give quantitative values. Potential combinatorial ligand hits would then later be further evaluated based on compartmentalization in live cells using the IN Cell Analyzer. Refinements of the technology and by choosing a control ligand specific for each cell line/type will enable us to generate more precise data with less variability in the results. It is possible to draw conclusions from our pilot study in terms of positional effects and combinatorial effects unique for the cell-types tested. A future study would include comparison of cell-types and cell lines relevant for specific diseases such as a comparative study of a cancer cell and a non-transformed cell against a broad panel of control cell-types. Furthermore, by identifying a broad collection of cell receptor ligands collected both from normal cell signaling as well from ligands identified from pathogens, we intend to define a range of ligand combinations with a high degree of specificity for a defined cell-type. By implementing a drug formulation from an identified, unique and highly specific ligand combination, it would seem like a natural step to subsequently apply the formulation in vivo using an animal model relevant for the specificity of the identified ligand.

3.4 PAPER IV

In paper IV, the DNA analog itself is the function. Recently, we have developed a novel anti-gene molecule "Zorro LNA", which seems to strand-invade into super-coiled DNA, simultaneously hybridizes to both strands and potently inhibits RNA polymerase II and III derived transcription. The original Zorro LNA construct consists of a 14-mer LNA oligonucleotide that binds to the coding strand, while a connected 16-mer LNA binds to the opposite template strand. Both LNA oligonucleotides are synthesized as continuous oligonucleotides with a mixture of LNA and DNA bases. They hybridize to one another and to the dsDNA by Watson-Crick base pairing (Figure 21). In this report, we have been able to further optimize the transcriptional inhibitor. We found that lengthening of the DNA binding regions enhanced the silencing, whereas extension of the linker sequence only diminished the effect.

Several technical issues were identified. Aggregates were observed in the wells in the gel retardation experiments probably caused by concatamerization of the dsDNA oligonucleotides. The tendency to Figure 21: Zorro LNA, hybridized to dsDNA aggregate increased with the length of the by strand-invasion linker. The further the dsDNA strands were

44 forced apart by the Zorro, the more prone they became to hybridize to other complexes instead reannealing and thereby closing the loop over the Zorro construct. This was verified by AFM and is a concern for the fluorescence resonance energy transfer (FRET) measurements. We believe that the length of the linker decreases when constricted by the open loop formed by the DNA strands, in comparison to free ends. However, the concatamerization probably underestimates the condensing effect.

Expression studies using a single binding site in the plasmid resembled the effect in the oligonucleotide experiment using targets without mismatches. In both assays, the 7 bp linker was found to be of optimal length. The constructs were designed to have linkers, differing in size of 2 to 3 bp and owing to the optimal effect of 7 bp linkers 6 and 8 bp linkers also have to be tested. It could turn out that the 7 bp linker has the perfect “curvature” once assembled into a dsDNA:Zorro complex. The hypothesis is supported by the fact that 7 bp and 12 bp linkers, which were designed to have a half and a full helical turn, respectively, were found to be the two extremes in the oligonucleotide strand-invasion assay. The curvature is dependent on linker length, sequence and LNA content. The 3’ nucleotides in the 7 bp linker were LNAs, which resulted in strong hybridization and, additionally, to a more open helical structure. If LNAs were to be removed from the ends of the current 7 bp linker, the ends would likely change their conformation.

Another interesting finding was that by extending the DNA binding regions of the Zorro construct, the potency could be further increased. The rationale might be that the melting temperature and the binding affinity to the DNA increases with lengthened sequence or it might be the relative direction and positioning of the linker regions. All Zorro2:0:X constructs have 1 bp overlap between the DNA binding region of LNA oligonucleotides whereas Zorro2:2:7 has 3 bp overlap. This fact results in an additional separation of the last base of the DNA biding region of the LNAs by approximately 7 Å and 70° in rotation along the DNA-axis. Probably, all parameters are of importance and, in addition, the sequence and the nature Figure 22: In vivo effect of Zorro LNA. of the bases composing the linker are not Plasmid coding for the luciferase reporter irrelevant as assumed here. gene was hydrodynamically injected alone (left mouse) or pre-hybridized with Zorro LNA (right mouse). The expression of We have found that Zorro LNA potently luciferase was measured after 3 days. The silences the expression of specific genes expression of the pre-hybridized plasmid transcribed by RNA polymerase II and was silenced by 90%. III(98,99). The Zorro LNAs inhibit transcription of the gene downstream of the binding site, without influencing the expression of the genes driven by uncompromised promoters on the same plasmid. The Zorro construct can be co-injected with the plasmid, thus it is potentially able to strand-

45 invade in vivo, but higher molar excess is needed. Co-injections have been tested in mice, both by intravenous (Figure 22) and intramuscular administration with 90% inhibition of transcription elongation (98,99).

In all assays tested previously, two consecutive binding sites were needed for efficient silencing and this might be the case for pronounced suppression even if the Zorro construct is further optimized. However, for the first time we could detect silencing, albeit only at the 20% level, with a single site when the DNA binding regions of the Zorro were extended and with a 3 bp overlap between the DNA binding regions. It is not known whether structural limitations of the open loop within the dsDNA, caused by the strand-invasion, might be thermodynamically unfavorable for the single Zorro construct when being attacked by the polymerase. The Zorros are experiencing cooperative hybridization in the two binding site plasmid, which facilitates strand- invasion and the stability of the complex. Furthermore, sequences flanking the binding sites in the one and two binding site plasmids are not identical. The sum of all these factors are probably the reason for that saturated hybridization was obtained with one binding site but only moderate gene silencing by the optimal Zorro2:2:7. We do not believe that simply extending the DNA binding sequences of the Zorro construct is the solution but rather trying other novel designs as Ω-shaped constructs.

Before drawing any conclusions, new binding site sequences have to be evaluated. The Zorro LNA concept will mainly be of general interest if it can be applied to any binding site sequence, targeting chromosomal DNA, regardless the number of consecutive constructs needed for potency.

4 POPULÄRVETENSKAPLIG SAMMANFATTNING

Genen är en enhet för information om ärftliga egenskaper, som ärvs från föräldrarna till barnet. En gen består av en specifik DNA sekvens som översätts till RNA i cellkärnan och förs ut i cellkroppen där RNA slutligen används som ett recept för att bygga proteiner. Proteiner utför nästan alla aktiviter som sker i levande varelser. De består av byggstenar som kallas aminosyror och det är aminosyrornas ordning och samman- sättning som genens DNA kod beskriver.

Många genetiska sjukdomar orsakas av att DNA koden i en gen förändras, muteras. Det "sjuka" anlaget beror då på att genen muterats på ett sådant sätt att den inte kodar för ett funktionellt protein. I vissa fall kodar den muterade genen för ett förändrat protein som är skadligt för kroppen. En sådan gen måste således stängas av.

Genterapi har blivit ett samlingsnamn för medicinska verktyg som påverkar arvsmassan och översättningen från DNA till RNA till protein. Istället för att medicinera kroppen med konventionella läkemedel så kan man tillföra kroppen en fungerande kopia av den gen vars avsaknad är orsak till det grundläggande problemet. Från början spordes genterapi vara framtidens bot för genetiska sjukdomar men det visade sig dock vara mer komplicerat än väntat.

Problemet med genterapi är att föra in genen i cellen. Gener kan föras in med hjälp av biologiskt inaktiva virus som modifieras till att istället föra in en terapeutisk gen. Virusets enda naturliga funktion är just att ta sig in i celler och sedermera få cellerna att producera virusets egna proteiner. De är väldigt effektiva men dessvärre finns alltid säkerhetsaspekter med virus och helst vill man använda andra alternativ.

Kroppen försöker skydda sig mot virus och andra mikroorganismer. Alla medicinska försök att föra in DNA kommer också att motarbetas av kroppen. Det finns många låsta dörrar på vägen vilket gör genöverföring till en komplex process. I vårt laboratorium har en syntetisk genöverföringsmetod utvecklats, Bioplex, där nyckel-molekyler sätts fast direkt på den terapeutiska genen. Dessa molekyler ska ledsaga genen genom kroppens låsta dörrar. Först från blodet till rätt vävnad och rätt celltyp. Genen måste tas upp av cellen och undgå nedbrytning och slutligen ta sig in i kärnan där den terapeutiska genen översätts till det saknade proteinet. Nyckel-molekylerna fästs med hjälp av en modifierad form av DNA som fungerar som ett ”genetiskt klister”.

Den första artikeln beskriver hur det genetiska klistret kan användas för att bygga upp större förgrenade komplex, allt för att kunna fästa så många nyckel-molekyler som möjligt. Ju fler kopior av varje nyckel-molekyl, desto mer troligt är det att nyckeln träffar låset.

Det finns tillfällen då nyckel-molekylerna bör klyvas av från den terapeutiska genen för att aktiveras eller efter det att deras funktion är utagerad. Den andra artikeln visar hur man kan använda kroppens enzymer för att klyva av nyckel-molekylerna specifikt och vid korrekt tidpunkt.

Nyckel-molekylernas uppgift är att föra in den terapeutiska genen effektivt och specifikt. Vissa nyckel-molekyler fungerar på många celltyper emedan andra bara passar en speciell cell. I många fall krävs korrekta kombinationer av molekyler för att genöverföringen ska vara både effektiv och specifik. Dessvärre finns det väldigt många olika nyckel-molekyler och att experimentellt pröva sig fram är väldigt mödosamt. Den tredje artikeln beskriver hur automatiska robotsystem kan används för att testa mängder av kombinationer på ett snabbt och kostnadseffektivt sätt på olika typer av odlade mänskliga celler.

Som tidigare nämnt finns det många sjukdomar där man behöver stänga av en gen, då ett felaktigt arvsanlag orsakar sjukdom. Utöver genetiska sjukdomar gäller det även för vissa virusinfektioner och cancertyper. I den sista artikeln beskrivs en ny form av genetiskt klister som binder tillräckligt hårt till DNA för att ”stänga av” den sjuka genen. Konstruktet heter Zorro LNA på grund av att det har formen av Z när det är bundet i DNA spiralen.

Schematisk skiss av Zorro LNA (i rött och lila) som sitter fast i DNA spiralen (i grönt). (Reproduktion med tillstånd från tidskriften Medicinsk Vetenskap).

5 ACKNOWLEDGEMENTS

A number of people have contributed to this work and I would like to express my sincere gratitude and appreciation to all of you.

My supervisor, Lars J. Brandén, who convinced me to begin my PhD studies at Karolinska Insitutet and for that I am immensely happy - today. An energetic know-it- all guy and everyone’s hi-guy. The most enthusiastic person I ever met, he finds a positive angle to the most hopeless result.

To my co-supervisor and professor, C.I. Edvard Smith who was not necessarily aware of the extent of the tutoring required when accepting me. He has the profound ability to squeeze out all ideas and aspects of a problem, which you may or may not be aware of having. A guardian of spoken and written Swedish and English.

To professor James Rothman and the people at Columbia Genome Center, Columbia University, for the hospitality and all the help. Especially Geoffery Barger for repeatedly changing the scripts for the robots without as much as a sigh.

Kersten Rabe and prof. Christof Niemeyer, University of Dortmund. Vineet Pande and prof. Lennart Nilsson, Karolinska Institutet. Juan-José Valle-Delgado and Mark Rutland, Royal Institute of Technology. Alex Peralvarez-Marin and prof. Astrid Gräslund, Stockholm University. Prof. Jesper Wengel, University of Southern Denmark. Björn Tidbeck and co-workers at Isosep. Prof. Roger Strömblad, Karolinska Institutet. Valeria Sigot and prof. Thomas Jovin, Max Plank Institute in Göttingen.

All the present MCG colleagues, particularly Karin Lundin for introducing me to the field during my degree project and Abdi, Beston, Emelie, Lotta, Leonardo, Rongbin, Liang, Pedro, Iulian, Maroof, Joana, José and former people, Magnus and Jessica. Oscar, the teenager who became middle-aged overnight, roomie and longtime partner in crime. Anna for always keeping the discussion strictly professional. Juhana, for his amusing illusion of that testosterone and pink match. Co-workers at CRC and Novum. In particular Christian, Evren, Tolga and Alexandra. The administration for being understanding and Micke for miscellaneous. Ulf for having a fantastic day and all the Finns for fruitful lunch discussions and a quick introduction to the language of love.

Samir EL-Andaloussi for consultation and soothing inspiration in the preparation of the thesis. Henrik, Maarja and Johanna for help, peptides and pleasure.

To my non-indoctrinated friends. Fredrik, Joakim, Mickael and Ulf for keeping me broke but happy. Värnamo Daniel for showing me Stockholm

To my mother and farther for all their support and understanding from the very beginning. To my sisters, Lotta & Lisa, and their young families. To my relatives by marriage, the interesting Rasch family. Mormor och morfar som lite oväntat är de enda i släkten som har full koll på vad jag ägnar mig åt.

To the love of my life, my wife Emma. To Hilda and her unborn sibling for keeping me from overdoing my dissertation. You are my everything!

6 REFERENCES

1. Anderson, W.F. (1984) Prospects 8. Rogers, S. and Pfuderer, P. for human gene therapy. Science, (1968) Use of viruses as carriers 226, 401-409. of added genetic information. 2. Anderson, W.F. (1992) Human Nature, 219, 749-751. gene therapy. Science, 256, 808- 9. Blaese, R.M., Culver, K.W., 813. Miller, A.D., Carter, C.S., 3. Friedmann, T. and Roblin, R. Fleisher, T., Clerici, M., Shearer, (1972) Gene therapy for human G., Chang, L., Chiang, Y., genetic disease? Science, 175, Tolstoshev, P. et al. (1995) T 949-955. lymphocyte-directed gene 4. Miller, A.D. (1992) Human gene therapy for ADA- SCID: initial therapy comes of age. Nature, trial results after 4 years. Science, 357, 455-460. 270, 475-480. 5. Reed, J.C. (2002) Apoptosis- 10. Anderson, W.F. (2000) Gene based therapies. Nat Rev Drug therapy. The best of times, the Discov, 1, 111-121. worst of times. Science, 288, 627- 6. Somia, N. and Verma, I.M. 629. (2000) Gene therapy: trials and 11. Cavazzana-Calvo, M., Hacein- tribulations. Nat Rev Genet, 1, Bey, S., de Saint Basile, G., 91-99. Gross, F., Yvon, E., Nusbaum, P., 7. Aghi, M., Hochberg, F. and Selz, F., Hue, C., Certain, S., Breakefield, X.O. (2000) Prodrug Casanova, J.L. et al. (2000) Gene activation enzymes in cancer therapy of human severe gene therapy. J Gene Med, 2, combined immunodeficiency 148-164.

(SCID)-X1 disease. Science, 288, 19. Wells, D.J. (2004) Gene therapy 669-672.12. Hacein-Bey- progress and prospects: Abina, S., Fischer, A. and electroporation and other physical Cavazzana-Calvo, M. (2002) methods. Gene Ther, 11, 1363- Gene therapy of X-linked severe 1369. combined immunodeficiency. Int 20. Liu, F. and Huang, L. (2002) J Hematol, 76, 295-298. Development of non-viral vectors 13. Hacein-Bey-Abina, S., von Kalle, for systemic gene delivery. J C., Schmidt, M., Le Deist, F., Control Release, 78, 259-266. Wulffraat, N., McIntyre, E., 21. Felgner, P.L. and Ringold, G.M. Radford, I., Villeval, J.L., Fraser, (1989) Cationic liposome- C.C., Cavazzana-Calvo, M. et al. mediated transfection. Nature, (2003) A serious adverse event 337, 387-388. after successful gene therapy for 22. Niidome, T. and Huang, L. X-linked severe combined (2002) Gene therapy progress and immunodeficiency. N Engl J prospects: nonviral vectors. Gene Med, 348, 255-256. Ther, 9, 1647-1652. 14. Hacein-Bey-Abina, S., Von 23. Felgner, J.H., Kumar, R., Sridhar, Kalle, C., Schmidt, M., C.N., Wheeler, C.J., Tsai, Y.J., McCormack, M.P., Wulffraat, N., Border, R., Ramsey, P., Martin, Leboulch, P., Lim, A., Osborne, M. and Felgner, P.L. (1994) C.S., Pawliuk, R., Morillon, E. et Enhanced gene delivery and al. (2003) LMO2-associated mechanism studies with a novel clonal T cell proliferation in two series of cationic lipid patients after gene therapy for formulations. J Biol Chem, 269, SCID-X1. Science, 302, 415-419. 2550-2561. 15. Graham, F.L. and van der Eb, 24. Gershon, H., Ghirlando, R., A.J. (1973) A new technique for Guttman, S.B. and Minsky, A. the assay of infectivity of human (1993) Mode of formation and adenovirus 5 DNA. Virology, 52, structural features of DNA- 456-467. cationic liposome complexes 16. Capecchi, M.R. (1980) High used for transfection. efficiency transformation by Biochemistry, 32, 7143-7151. direct microinjection of DNA 25. Lukacs, G.L., Haggie, P., Seksek, into cultured mammalian cells. O., Lechardeur, D., Freedman, N. Cell, 22, 479-488. and Verkman, A.S. (2000) Size- 17. Wolff, J.A., Malone, R.W., dependent DNA mobility in Williams, P., Chong, W., Acsadi, cytoplasm and nucleus. J Biol G., Jani, A. and Felgner, P.L. Chem, 275, 1625-1629. (1990) Direct gene transfer into 26. Laemmli, U.K. (1975) mouse muscle in vivo. Science, Characterization of DNA 247, 1465-1468. condensates induced by 18. Greco, O., Scott, S.D., Marples, poly(ethylene oxide) and B. and Dachs, G.U. (2002) polylysine. Proc Natl Acad Sci U Cancer gene therapy: 'delivery, S A, 72, 4288-4292. delivery, delivery'. Front Biosci, 27. Felgner, P.L., Barenholz, Y., 7, d1516-1524. Behr, J.P., Cheng, S.H., Cullis, P., Huang, L., Jessee, J.A.,

Seymour, L., Szoka, F., Thierry, 35. Yoneda, Y., Semba, T., Kaneda, A.R. et al. (1997) Nomenclature Y., Noble, R.L., Matsuoka, Y., for synthetic gene delivery Kurihara, T., Okada, Y. and systems. Hum Gene Ther, 8, 511- Imamoto, N. (1992) A long 512. synthetic peptide containing a 28. Branden, L.J. and Smith, C.I.E. nuclear localization signal and its (2002) Bioplex technology: novel flanking sequences of SV40 T- synthetic gene delivery system antigen directs the transport of based on peptides anchored to IgM into the nucleus efficiently. nucleic acids. Methods Enzymol, Exp Cell Res, 201, 313-320. 346, 106-124. 36. Collas, P., Husebye, H. and 29. Pierschbacher, M.D. and Alestrom, P. (1996) The nuclear Ruoslahti, E. (1984) Cell localization sequence of the attachment activity of fibronectin SV40 T antigen promotes can be duplicated by small transgene uptake and expression synthetic fragments of the in zebrafish embryo nuclei. molecule. Nature, 309, 30-33. Transgenic Res, 5, 451-458. 30. Erbacher, P., Remy, J.S. and 37. Sebestyen, M.G., Ludtke, J.J., Behr, J.P. (1999) Gene transfer Bassik, M.C., Zhang, G., Budker, with synthetic virus-like particles V., Lukhtanov, E.A., Hagstrom, via the integrin-mediated J.E. and Wolff, J.A. (1998) DNA endocytosis pathway. Gene Ther, vector chemistry: the covalent 6, 138-145. attachment of signal peptides to 31. Chen, J., Skehel, J.J. and Wiley, plasmid DNA. Nat Biotechnol, D.C. (1998) A polar octapeptide 16, 80-85. fused to the N-terminal fusion 38. Zanta, M.A., Belguise-Valladier, peptide solubilizes the influenza P. and Behr, J.P. (1999) Gene virus HA2 subunit ectodomain. delivery: a single nuclear Biochemistry, 37, 13643-13649. localization signal peptide is 32. Kalderon, D., Roberts, B.L., sufficient to carry DNA to the Richardson, W.D. and Smith, cell nucleus. Proc Natl Acad Sci A.E. (1984) A short amino acid U S A, 96, 91-96. sequence able to specify nuclear 39. Branden, L.J., Mohamed, A.J. location. Cell, 39, 499-509. and Smith, C.I.E. (1999) A 33. Michaud, N. and Goldfarb, D.S. peptide nucleic acid-nuclear (1993) Most nuclear proteins are localization signal fusion that imported by a single pathway. mediates nuclear transport of Exp Cell Res, 208, 128-136. DNA. Nat Biotechnol, 17, 784- 34. Miyamoto, Y., Imamoto, N., 787. Sekimoto, T., Tachibana, T., 40. Branden, L.J., Christensson, B. Seki, T., Tada, S., Enomoto, T. and Smith, C.I.E. (2001) In vivo and Yoneda, Y. (1997) nuclear delivery of Differential modes of nuclear oligonucleotides via hybridizing localization signal (NLS) bifunctional peptides. Gene Ther, recognition by three distinct 8, 84-87. classes of NLS receptors. J Biol 41. Lundin, K.E., Ge, R., Svahn, Chem, 272, 26375-26381. M.G., Tornquist, E., Leijon, M., Branden, L.J. and Smith, C.I.E.

(2004) Cooperative strand 48. Kuhn, H., Demidov, V.V., Frank- invasion of supercoiled plasmid Kamenetskii, M.D. and Nielsen, DNA by mixed linear PNA and P.E. (1998) Kinetic sequence PNA-peptide chimeras. Biomol discrimination of cationic bis- Eng, 21, 51-59. PNAs upon targeting of double- 42. Bukanov, N.O., Demidov, V.V., stranded DNA. Nucleic Acids Nielsen, P.E. and Frank- Res, 26, 582-587. Kamenetskii, M.D. (1998) PD- 49. Lundin, K.E., Hasan, M., loop: a complex of duplex DNA Moreno, P.M., Tornquist, E., with an oligonucleotide. Proc Oprea, I., Svahn, M.G., Natl Acad Sci U S A, 95, 5516- Simonson, E.O. and Smith, C.I.E. 5520. (2005) Increased stability and 43. Demidov, V.V., Cherny, D.I., specificity through combined Kurakin, A.V., Yavnilovich, hybridization of peptide nucleic M.V., Malkov, V.A., Frank- acid (PNA) and locked nucleic Kamenetskii, M.D., Sonnichsen, acid (LNA) to supercoiled S.H. and Nielsen, P.E. (1994) plasmids for PNA-anchored Electron microscopy mapping of "Bioplex" formation. Biomol oligopurine tracts in duplex DNA Eng, 22, 185-192. by peptide nucleic acid targeting. 50. McNamara, J.O., 2nd, Nucleic Acids Res, 22, 5218- Andrechek, E.R., Wang, Y., 5222. Viles, K.D., Rempel, R.E., 44. Demidov, V.V. and Frank- Gilboa, E., Sullenger, B.A. and Kamenetskii, M.D. (2004) Two Giangrande, P.H. (2006) Cell sides of the coin: affinity and type-specific delivery of siRNAs specificity of nucleic acid with aptamer-siRNA chimeras. interactions. Trends Biochem Sci, Nat Biotechnol, 24, 1005-1015. 29, 62-71. 51. Miescher, F. (1871) Ueber die 45. Smulevitch, S.V., Simmons, chemische Zusammensetzung der C.G., Norton, J.C., Wise, T.W. Eiterzellen. Med.-Chem. Unters., and Corey, D.R. (1996) 4, 441-460. Enhancement of strand invasion 52. Avery, O.T., MacLeod, C.M. and by oligonucleotides through McCarty, M. (1944) Studies of manipulation of backbone the chemical nature of the charge. Nat Biotechnol, 14, 1700- substance inducing 1704. transformation of pneumococcal 46. Bentin, T. and Nielsen, P.E. types. In duction of (2002) PNA-mediated transformation by a immobilization of supercoiled deoxyribonucleic acid fraction DNA. Methods Mol Biol, 208, isolated from pneumococcus type 111-118. III. J. Exp. Med., 79, 137-158. 47. Kaihatsu, K., Braasch, D.A., 53. Watson, J.D. and Crick, F.H. Cansizoglu, A. and Corey, D.R. (1953) Genetical implications of (2002) Enhanced strand invasion the structure of deoxyribonucleic by peptide nucleic acid-peptide acid. Nature, 171, 964-967. conjugates. Biochemistry, 41, 54. Matthaei, J.H. and Nirenberg, 11118-11125. M.W. (1961) Characteristics and stabilization of DNAase-sensitive

protein synthesis in E. coli functional assay development. extracts. Proc Natl Acad Sci U S Biochemistry, 37, 6235-6239. A, 47, 1580-1588. 63. EL Andaloussi, S., Guterstam, P. 55. Paterson, B.M., Roberts, B.E. and and Langel, Ü. (2007) Assessing Kuff, E.L. (1977) Structural gene the delivery efficacy and identification and mapping by internalization route of cell- DNA-mRNA hybrid-arrested penetrating peptides. Nat Protoc, cell-free translation. Proc Natl in press. Acad Sci U S A, 74, 4370-4374. 64. Fire, A., Xu, S., Montgomery, 56. Stephenson, M.L. and Zamecnik, M.K., Kostas, S.A., Driver, S.E. P.C. (1978) Inhibition of Rous and Mello, C.C. (1998) Potent sarcoma viral RNA translation by and specific genetic interference a specific by double-stranded RNA in oligodeoxyribonucleotide. Proc Caenorhabditis elegans. Nature, Natl Acad Sci U S A, 75, 285- 391, 806-811. 288. 65. Manche, L., Green, S.R., 57. Zamecnik, P.C. and Stephenson, Schmedt, C. and Mathews, M.B. M.L. (1978) Inhibition of Rous (1992) Interactions between sarcoma virus replication and cell double-stranded RNA regulators transformation by a specific and the protein kinase DAI. Mol oligodeoxynucleotide. Proc Natl Cell Biol, 12, 5238-5248. Acad Sci U S A, 75, 280-284. 66. Minks, M.A., West, D.K., 58. Tatum, E.L. (1966) Molecular Benvin, S. and Baglioni, C. biology, nucleic acids, and the (1979) Structural requirements of future of medicine. Perspect Biol double-stranded RNA for the Med, 10, 19-32. activation of 2',5'-oligo(A) 59. Crooke, S.T. (2004) Progress in polymerase and protein kinase of antisense technology. Annu Rev interferon-treated HeLa cells. J Med, 55, 61-95. Biol Chem, 254, 10180-10183. 60. Dominski, Z. and Kole, R. (1993) 67. Samuel, C.E. (2001) Antiviral Restoration of correct splicing in actions of interferons. Clin thalassemic pre-mRNA by Microbiol Rev, 14, 778-809, table antisense oligonucleotides. Proc of contents. Natl Acad Sci U S A, 90, 8673- 68. Elbashir, S.M., Harborth, J., 8677. Lendeckel, W., Yalcin, A., 61. Sierakowska, H., Sambade, M.J., Weber, K. and Tuschl, T. (2001) Agrawal, S. and Kole, R. (1996) Duplexes of 21-nucleotide RNAs Repair of thalassemic human mediate RNA interference in beta-globin mRNA in cultured mammalian cells. mammalian cells by antisense Nature, 411, 494-498. oligonucleotides. Proc Natl Acad 69. Elbashir, S.M., Lendeckel, W. Sci U S A, 93, 12840-12844. and Tuschl, T. (2001) RNA 62. Kang, S.H., Cho, M.J. and Kole, interference is mediated by 21- R. (1998) Up-regulation of and 22-nucleotide RNAs. Genes luciferase gene expression with Dev, 15, 188-200. antisense oligonucleotides: 70. Ketting, R.F., Haverkamp, T.H., implications and applications in van Luenen, H.G. and Plasterk, R.H. (1999) Mut-7 of C. elegans,

required for transposon silencing 78. Morris, K.V., Chan, S.W., and RNA interference, is a Jacobsen, S.E. and Looney, D.J. homolog of Werner syndrome (2004) Small interfering RNA- helicase and RNaseD. Cell, 99, induced transcriptional gene 133-141. silencing in human cells. Science, 71. Tabara, H., Sarkissian, M., Kelly, 305, 1289-1292. W.G., Fleenor, J., Grishok, A., 79. Chow, J.C., Yen, Z., Ziesche, Timmons, L., Fire, A. and Mello, S.M. and Brown, C.J. (2005) C.C. (1999) The rde-1 gene, Silencing of the mammalian X RNA interference, and chromosome. Annu Rev transposon silencing in C. Genomics Hum Genet, 6, 69-92. elegans. Cell, 99, 123-132. 80. Kim, D.H., Villeneuve, L.M., 72. Mattick, J.S. (2001) Non-coding Morris, K.V. and Rossi, J.J. RNAs: the architects of (2006) Argonaute-1 directs eukaryotic complexity. EMBO siRNA-mediated transcriptional Rep, 2, 986-991. gene silencing in human cells. 73. Mattick, J.S. (2003) Challenging Nat Struct Mol Biol, 13, 793-797. the dogma: the hidden layer of 81. Meister, G., Landthaler, M., non-protein-coding RNAs in Patkaniowska, A., Dorsett, Y., complex organisms. Bioessays, Teng, G. and Tuschl, T. (2004) 25, 930-939. Human Argonaute2 mediates 74. Peters, L. and Meister, G. (2007) RNA cleavage targeted by Argonaute proteins: mediators of miRNAs and siRNAs. Mol Cell, RNA silencing. Mol Cell, 26, 15, 185-197. 611-623. 82. Besch, R., Giovannangeli, C. and 75. Elbashir, S.M., Martinez, J., Degitz, K. (2004) Triplex- Patkaniowska, A., Lendeckel, W. forming oligonucleotides - and Tuschl, T. (2001) Functional sequence-specific DNA ligands anatomy of siRNAs for as tools for gene inhibition and mediating efficient RNAi in for modulation of DNA- Drosophila melanogaster embryo associated functions. Curr Drug lysate. Embo J, 20, 6877-6888. Targets, 5, 691-703. 76. Zilberman, D., Cao, X. and 83. Faria, M., Wood, C.D., Jacobsen, S.E. (2003) Perrouault, L., Nelson, J.S., ARGONAUTE4 control of Winter, A., White, M.R., Helene, locus-specific siRNA C. and Giovannangeli, C. (2000) accumulation and DNA and Targeted inhibition of histone methylation. Science, transcription elongation in cells 299, 716-719. mediated by triplex-forming 77. Castanotto, D., Tommasi, S., Li, oligonucleotides. Proc Natl Acad M., Li, H., Yanow, S., Pfeifer, Sci U S A, 97, 3862-3867. G.P. and Rossi, J.J. (2005) Short 84. Faria, M., Wood, C.D., White, hairpin RNA-directed cytosine M.R., Helene, C. and (CpG) methylation of the Giovannangeli, C. (2001) RASSF1A gene promoter in Transcription inhibition induced HeLa cells. Mol Ther, 12, 179- by modified triple helix-forming 183. oligonucleotides: a quantitative

assay for evaluation in cells. J 90. Janowski, B.A., Kaihatsu, K., Mol Biol, 306, 15-24. Huffman, K.E., Schwartz, J.C., 85. Brunet, E., Alberti, P., Ram, R., Hardy, D., Mendelson, Perrouault, L., Babu, R., Wengel, C.R. and Corey, D.R. (2005) J. and Giovannangeli, C. (2005) Inhibiting transcription of Exploring cellular activity of chromosomal DNA with antigene locked nucleic acid-modified peptide nucleic acids. Nat Chem triplex-forming oligonucleotides Biol, 1, 210-215. and defining its molecular basis. 91. Beane, R.L., Ram, R., Gabillet, J Biol Chem, 280, 20076-20085. S., Arar, K., Monia, B.P. and 86. Vasquez, K.M., Christensen, J., Corey, D.R. (2007) Inhibiting Li, L., Finch, R.A. and Glazer, Gene Expression with Locked P.M. (2002) Human XPA and Nucleic Acids (LNAs) That RPA DNA repair proteins Target Chromosomal DNA. participate in specific recognition Biochemistry, 46, 7572-7580. of triplex-induced helical 92. Hu, J. and Corey, D.R. (2007) distortions. Proc Natl Acad Sci U Inhibiting Gene Expression with S A, 99, 5848-5853. Peptide Nucleic Acid (PNA)- 87. Boffa, L.C., Morris, P.L., Peptide Conjugates That Target Carpaneto, E.M., Louissaint, M. Chromosomal DNA. and Allfrey, V.G. (1996) Biochemistry, 46, 7581-7589. Invasion of the CAG triplet 93. Demidov, V.V., Yavnilovich, repeats by a complementary M.V., Belotserkovskii, B.P., peptide nucleic acid inhibits Frank-Kamenetskii, M.D. and transcription of the androgen Nielsen, P.E. (1995) Kinetics and receptor and TATA-binding mechanism of polyamide protein genes and correlates with ("peptide") nucleic acid binding refolding of an active to duplex DNA. Proc Natl Acad nucleosome containing a unique Sci U S A, 92, 2637-2641. AR gene sequence. J Biol Chem, 94. Hanvey, J.C., Peffer, N.J., Bisi, 271, 13228-13233. J.E., Thomson, S.A., Cadilla, R., 88. Boffa, L.C., Scarfi, S., Mariani, Josey, J.A., Ricca, D.J., Hassman, M.R., Damonte, G., Allfrey, C.F., Bonham, M.A., Au, K.G. et V.G., Benatti, U. and Morris, al. (1992) Antisense and antigene P.L. (2000) Dihydrotestosterone properties of peptide nucleic as a selective cellular/nuclear acids. Science, 258, 1481-1485. localization vector for anti-gene 95. Peffer, N.J., Hanvey, J.C., Bisi, peptide nucleic acid in prostatic J.E., Thomson, S.A., Hassman, carcinoma cells. Cancer Res, 60, C.F., Noble, S.A. and Babiss, 2258-2262. L.E. (1993) Strand-invasion of 89. Cutrona, G., Carpaneto, E.M., duplex DNA by peptide nucleic Ulivi, M., Roncella, S., Landt, O., acid oligomers. Proc Natl Acad Ferrarini, M. and Boffa, L.C. Sci U S A, 90, 10648-10652. (2000) Effects in live cells of a c- 96. Bentin, T., Larsen, H.J. and myc anti-gene PNA linked to a Nielsen, P.E. (2003) Combined nuclear localization signal. Nat triplex/duplex invasion of double- Biotechnol, 18, 300-303. stranded DNA by "tail-clamp"

peptide nucleic acid. mice. Biopharm Drug Dispos, 13, Biochemistry, 42, 13987-13995. 221-227. 97. Kaihatsu, K., Shah, R.H., Zhao, 104. Hoke, G.D., Draper, K., Freier, X. and Corey, D.R. (2003) S.M., Gonzalez, C., Driver, V.B., Extending recognition by peptide Zounes, M.C. and Ecker, D.J. nucleic acids (PNAs): binding to (1991) Effects of duplex DNA and inhibition of phosphorothioate capping on transcription by tail-clamp PNA- antisense oligonucleotide peptide conjugates. Biochemistry, stability, hybridization and 42, 13996-14003. antiviral efficacy versus herpes 98. Ge, R., Svahn, M.G., Simonson, simplex virus infection. Nucleic O.E., Mohamed, A.J., Lundin, Acids Res, 19, 5743-5748. K.E. and Smith, C.I.E. (2007) 105. Brown, D.A., Kang, S.H., Sequence-specific inhibition of Gryaznov, S.M., DeDionisio, L., RNA polymerase III-dependent Heidenreich, O., Sullivan, S., Xu, transcription using Zorro Locked X. and Nerenberg, M.I. (1994) Nucleic Acid (LNA). submitted. Effect of phosphorothioate 99. Ge, R., Heinonen, J.E., Svahn, modification of M.G., Mohamed, A.J., Lundin, oligodeoxynucleotides on K.E. and Smith, C.I.E. (2007) specific protein binding. J Biol Zorro locked nucleic acid induces Chem, 269, 26801-26805. sequence-specific gene silencing. 106. Myers, K.J. and Dean, N.M. Faseb J, 21, 1902-1914. (2000) Sensible use of antisense: 100. McCaffrey, A.P., Meuse, L., how to use oligonucleotides as Karimi, M., Contag, C.H. and research tools. Trends Pharmacol Kay, M.A. (2003) A potent and Sci, 21, 19-23. specific morpholino antisense 107. Pilowsky, P.M., Suzuki, S. and inhibitor of hepatitis C translation Minson, J.B. (1994) in mice. Hepatology, 38, 503- ANTISENSE oligonucleotides: a 508. new tool in neuroscience. Clin 101. Herdewijn, P. (2000) Exp Pharmacol Physiol, 21, 935- Heterocyclic modifications of 944. oligonucleotides and antisense 108. Levin, A.A. (1999) A review of technology. Antisense Nucleic the issues in the Acid Drug Dev, 10, 297-310. pharmacokinetics and toxicology 102. Campbell, J.M., Bacon, T.A. and of phosphorothioate antisense Wickstrom, E. (1990) oligonucleotides. Biochim Oligodeoxynucleoside Biophys Acta, 1489, 69-84. phosphorothioate stability in 109. Cossum, P.A., Sasmor, H., subcellular extracts, culture Dellinger, D., Truong, L., media, sera and cerebrospinal Cummins, L., Owens, S.R., fluid. J Biochem Biophys Markham, P.M., Shea, J.P. and Methods, 20, 259-267. Crooke, S. (1993) Disposition of 103. Goodarzi, G., Watabe, M. and the 14C-labeled phosphorothioate Watabe, K. (1992) Organ oligonucleotide ISIS 2105 after distribution and stability of intravenous administration to phosphorothioated rats. J Pharmacol Exp Ther, 267, oligodeoxyribonucleotides in 1181-1190.

110. Phillips, J.A., Craig, S.J., Bayley, Sequence-selective recognition of D., Christian, R.A., Geary, R. and DNA by strand displacement Nicklin, P.L. (1997) with a thymine-substituted Pharmacokinetics, metabolism, polyamide. Science, 254, 1497- and elimination of a 20-mer 1500. phosphorothioate 117. Nielsen, P.E. (2002) PNA oligodeoxynucleotide (CGP technology. Methods Enzymol, 69846A) after intravenous and 208, 3-26. subcutaneous administration. 118. Lundin, K.E., Good, L., Biochem Pharmacol, 54, 657- Stromberg, R., Graslund, A. and 668. Smith, C.I.E. (2006) Biological 111. Henry, S.P., Zuckerman, J.E., activity and biotechnological Rojko, J., Hall, W.C., Harman, aspects of peptide nucleic acid. R.J., Kitchen, D. and Crooke, Adv Genet, 56, 1-51. S.T. (1997) Toxicological 119. Hyrup, B. and Nielsen, P.E. properties of several novel (1996) Peptide nucleic acids oligonucleotide analogs in mice. (PNA): synthesis, properties and Anticancer Drug Des, 12, 1-14. potential applications. Bioorg 112. Crooke, S.T. (2000) Progress in Med Chem, 4, 5-23. antisense technology: the end of 120. Egholm, M., Buchardt, O., the beginning. Methods Enzymol, Christensen, L., Behrens, C., 313, 46-56. Freier, S.M., Driver, D.A., Berg, 113. Monia, B.P., Lesnik, E.A., R.H., Kim, S.K., Norden, B. and Gonzalez, C., Lima, W.F., Nielsen, P.E. (1993) PNA McGee, D., Guinosso, C.J., hybridizes to complementary Kawasaki, A.M., Cook, P.D. and oligonucleotides obeying the Freier, S.M. (1993) Evaluation of Watson-Crick hydrogen-bonding 2'-modified oligonucleotides rules. Nature, 365, 566-568. containing 2'-deoxy gaps as 121. Jensen, K.K., Orum, H., Nielsen, antisense inhibitors of gene P.E. and Norden, B. (1997) expression. J Biol Chem, 268, Kinetics for hybridization of 14514-14522. peptide nucleic acids (PNA) with 114. Summerton, J. (1999) DNA and RNA studied with the Morpholino antisense oligomers: BIAcore technique. Biochemistry, the case for an RNase H- 36, 5072-5077. independent structural type. 122. Tomac, S., Sarkar, M., Biochim Biophys Acta, 1489, Ratilainen, T., Wittung, P., 141-158. Nielsen, P.E., Nordén, B. and 115. Zelphati, O., Liang, X., Nguyen, Gräslund, A. (1996) Ionic Effects C., Barlow, S., Sheng, S., Shao, on the Stability and Z. and Felgner, P.L. (2000) PNA- Conformation of Peptide Nucleic dependent gene chemistry: stable Acid Complexes. J Am Chem coupling of peptides and Soc, 118, 5544-5552. oligonucleotides to plasmid 123. Peyret, N., Seneviratne, P.A., DNA. Biotechniques, 28, 304- Allawi, H.T. and SantaLucia, J., 310, 312-304, 316. Jr. (1999) Nearest-neighbor 116. Nielsen, P.E., Egholm, M., Berg, thermodynamics and NMR of R.H. and Buchardt, O. (1991) DNA sequences with internal

A.A, C.C, G.G, and T.T 131. Betts, L., Josey, J.A., Veal, J.M. mismatches. Biochemistry, 38, and Jordan, S.R. (1995) A nucleic 3468-3477. acid triple helix formed by a 124. Zhu, J. and Wartell, R.M. (1997) peptide nucleic acid-DNA The relative stabilities of base complex. Science, 270, 1838- pair stacking interactions and 1841. single mismatches in long RNA 132. Lohse, J., Dahl, O. and Nielsen, measured by temperature P.E. (1999) Double duplex gradient gel electrophoresis. invasion by peptide nucleic acid: Biochemistry, 36, 15326-15335. a general principle for sequence- 125. Eriksson, M. and Nielsen, P.E. specific targeting of double- (1996) PNA-nucleic acid stranded DNA. Proc Natl Acad complexes. Structure, stability Sci U S A, 96, 11804-11808. and dynamics. Q Rev Biophys, 133. Svinarchuk, F., Debin, A., 29, 369-394. Bertrand, J.R. and Malvy, C. 126. Eriksson, M. and Nielsen, P.E. (1996) Investigation of the (1996) Solution structure of a intracellular stability and peptide nucleic acid-DNA formation of a triple helix formed duplex. Nat Struct Biol, 3, 410- with a short purine 413. oligonucleotide targeted to the 127. Leijon, M., Graslund, A., murine c-pim-1 proto-oncogene Nielsen, P.E., Buchardt, O., promotor. Nucleic Acids Res, 24, Norden, B., Kristensen, S.M. and 295-302. Eriksson, M. (1994) Structural 134. Bowater, R.P., Aboul-ela, F. and characterization of PNA-DNA Lilley, D.M. (1994) Large-scale duplexes by NMR. Evidence for opening of A + T rich regions DNA in a B-like conformation. within supercoiled DNA Biochemistry, 33, 9820-9825. molecules is suppressed by salt. 128. Cherny, D.Y., Belotserkovskii, Nucleic Acids Res, 22, 2042- B.P., Frank-Kamenetskii, M.D., 2050. Egholm, M., Buchardt, O., Berg, 135. Gueron, M., Kochoyan, M. and R.H. and Nielsen, P.E. (1993) Leroy, J.L. (1987) A single mode DNA unwinding upon strand- of DNA base-pair opening drives displacement binding of a imino proton exchange. Nature, thymine-substituted polyamide to 328, 89-92. double-stranded DNA. Proc Natl 136. Kochoyan, M., Leroy, J.L. and Acad Sci U S A, 90, 1667-1670. Gueron, M. (1987) Proton 129. Arnott, S., Bond, P.J., Selsing, E. exchange and base-pair lifetimes and Smith, P.J. (1976) Models of in a deoxy-duplex containing a triple-stranded polynucleotides purine-pyrimidine step and in the with optimised stereochemistry. duplex of inverse sequence. J Nucleic Acids Res, 3, 2459-2470. Mol Biol, 196, 599-609. 130. Moser, H.E. and Dervan, P.B. 137. Krueger, A., Protozanova, E. and (1987) Sequence-specific Frank-Kamenetskii, M.D. (2006) cleavage of double helical DNA Sequence-dependent base pair by triple helix formation. Science, opening in DNA double helix. 238, 645-650. Biophys J, 90, 3091-3099.

138. Yakovchuk, P., Protozanova, E. Ribonucleosides. J Org Chem, and Frank-Kamenetskii, M.D. 63, 2778-2781. (2006) Base-stacking and base- 146. Petersen, M., Nielsen, C.B., pairing contributions into thermal Nielsen, K.E., Jensen, G.A., stability of the DNA double Bondensgaard, K., Singh, S.K., helix. Nucleic Acids Res, 34, 564- Rajwanshi, V.K., Koshkin, A.A., 574. Dahl, B.M., Wengel, J. et al. 139. Orum, H., Nielsen, P.E., Egholm, (2000) The conformations of M., Berg, R.H., Buchardt, O. and locked nucleic acids (LNA). J Stanley, C. (1993) Single base Mol Recognit, 13, 44-53. pair mutation analysis by PNA 147. McTigue, P.M., Peterson, R.J. directed PCR clamping. Nucleic and Kahn, J.D. (2004) Sequence- Acids Res, 21, 5332-5336. dependent thermodynamic 140. Good, L., Awasthi, S.K., parameters for locked nucleic Dryselius, R., Larsson, O. and acid (LNA)-DNA duplex Nielsen, P.E. (2001) Bactericidal formation. Biochemistry, 43, antisense effects of peptide-PNA 5388-5405. conjugates. Nat Biotechnol, 19, 148. SantaLucia, J. and Hicks, D. 360-364. (2004) THE 141. Nielsen, P.E. (1999) Peptide THERMODYNAMICS OF DNA nucleic acids as therapeutic STRUCTURAL MOTIFS. agents. Curr Opin Struct Biol, 9, Annual Review of Biophysics and 353-357. Biomolecular Structure, 33, 415- 142. Kumar, R., Singh, S.K., Koshkin, 440. A.A., Rajwanshi, V.K., 149. Saenger, W. (1984). Springer, Meldgaard, M. and Wengel, J. New York. (1998) The first analogues of 150. Hunter, C.A. (1993) Sequence- LNA (locked nucleic acids): dependent DNA structure. The phosphorothioate-LNA and 2'- role of base stacking interactions. thio-LNA. Bioorg Med Chem J Mol Biol, 230, 1025-1054. Lett., 8, 2219-2222. 151. Luo, R., Gilson, H.S., Potter, M.J. 143. Obika, S., Nanbu, D., Hari, Y., and Gilson, M.K. (2001) The Andoh, J.-I., Morio, K.-I., Doi, T. physical basis of nucleic acid and Imanishi, T. (1998) Stability base stacking in water. Biophys J, and structural features of the 80, 140-148. duplexes containing nucleoside 152. Kurreck, J., Wyszko, E., Gillen, analogues with a fixed N-type C. and Erdmann, V.A. (2002) conformation, 2′-O,4′-C- Design of antisense methyleneribonucleosides. oligonucleotides stabilized by Tetrahed Lett, 39, 5401-5404. locked nucleic acids. Nucleic 144. Petersen, M. and Wengel, J. Acids Res, 30, 1911-1918. (2003) LNA: a versatile tool for 153. Wahlestedt, C., Salmi, P., Good, therapeutics and genomics. L., Kela, J., Johnsson, T., Trends Biotechnol., 21, 74-81. Hokfelt, T., Broberger, C., 145. Koshkin, A.A. and Wengel, J. Porreca, F., Lai, J., Ren, K. et al. (1998) Synthesis of Novel 2',3'- (2000) Potent and nontoxic Linked Bicyclic Thymine antisense oligonucleotides containing locked nucleic acids.

Proc Natl Acad Sci U S A, 97, locked nucleic acids to improve 5633-5638. aptamer in vivo stability and 154. Elmen, J., Thonberg, H., targeting function. Nucleic Acids Ljungberg, K., Frieden, M., Res, 32, 5757-5765. Westergaard, M., Xu, Y., 160. Swayze, E.E., Siwkowski, A.M., Wahren, B., Liang, Z., Orum, H., Wancewicz, E.V., Migawa, M.T., Koch, T. et al. (2005) Locked Wyrzykiewicz, T.K., Hung, G., nucleic acid (LNA) mediated Monia, B.P. and Bennett, C.F. improvements in siRNA stability (2007) Antisense and functionality. Nucleic Acids oligonucleotides containing Res, 33, 439-447. locked nucleic acid improve 155. Crinelli, R., Bianchi, M., potency but cause significant Gentilini, L. and Magnani, M. hepatotoxicity in animals. (2002) Design and Nucleic Acids Res, 35, 687-700. characterization of decoy 161. Hertoghs, K.M., Ellis, J.H. and oligonucleotides containing Catchpole, I.R. (2003) Use of locked nucleic acids. Nucleic locked nucleic acid Acids Res, 30, 2435-2443. oligonucleotides to add 156. Crinelli, R., Bianchi, M., functionality to plasmid DNA. Gentilini, L., Palma, L., Nucleic Acids Res, 31, 5817- Sorensen, M.D., Bryld, T., Babu, 5830. R.B., Arar, K., Wengel, J. and 162. Svahn, M.G., Hasan, M., Sigot, Magnani, M. (2004) V., Valle-Delgado, J.J., Rutland, Transcription factor decoy M.W., Lundin, K.E. and Smith, oligonucleotides modified with C.I.E. (2007) Self-assembling locked nucleic acids: an in vitro supramolecular complexes by study to reconcile biostability single-stranded extension from with binding affinity. Nucleic plasmid DNA. Oligonucleotides, Acids Res, 32, 1874-1885. 17, 80-94. 157. Vester, B., Lundberg, L.B., 163. Ng, P.S. and Bergstrom, D.E. Sorensen, M.D., Babu, B.R., (2005) Alternative nucleic acid Douthwaite, S. and Wengel, J. analogues for programmable (2002) LNAzymes: incorporation assembly: hybridization of LNA of LNA-type monomers into to PNA. Nano Lett, 5, 107-111. DNAzymes markedly increases 164. Egholm, M., Christensen, L., RNA cleavage. J Am Chem Soc, Dueholm, K.L., Buchardt, O., 124, 13682-13683. Coull, J. and Nielsen, P.E. (1995) 158. Roberts, J., Palma, E., Sazani, P., Efficient pH-independent Orum, H., Cho, M. and Kole, R. sequence-specific DNA binding (2006) Efficient and persistent by pseudoisocytosine-containing splice switching by systemically bis-PNA. Nucleic Acids Res, 23, delivered LNA oligonucleotides 217-222. in mice. Mol Ther, 14, 471-475. 165. Shiraishi, T. and Nielsen, P.E. 159. Schmidt, K.S., Borkowski, S., (2004) Down-regulation of Kurreck, J., Stephens, A.W., MDM2 and activation of p53 in Bald, R., Hecht, M., Friebe, M., human cancer cells by antisense Dinkelborg, L. and Erdmann, 9-aminoacridine-PNA (peptide V.A. (2004) Application of

nucleic acid) conjugates. Nucleic Antennapedia homeodomain Acids Res, 32, 4893-4902. translocates through biological 166. Lee, Y.C. (1992) Biochemistry of membranes. J Biol Chem, 269, carbohydrate-protein interaction. 10444-10450. Faseb J, 6, 3193-3200. 174. Pooga, M., Hällbrink, M., Zorko, 167. Wu, J., Nantz, M.H. and Zern, M. and Langel, Ü. (1998) Cell M.A. (2002) Targeting penetration by transportan. Faseb hepatocytes for drug and gene J, 12, 67-77. delivery: emerging novel 175. Soomets, U., Lindgren, M., approaches and applications. Gallet, X., Hällbrink, M., Front Biosci, 7, d717-725. Elmquist, A., Balaspiri, L., 168. Hautanen, A., Gailit, J., Mann, Zorko, M., Pooga, M., Brasseur, D.M. and Ruoslahti, E. (1989) R. and Langel, Ü. (2000) Effects of modifications of the Deletion analogues of RGD sequence and its context on transportan. Biochim Biophys recognition by the fibronectin Acta, 1467, 165-176. receptor. J Biol Chem, 264, 1437- 176. Mitchell, D.J., Kim, D.T., 1442. Steinman, L., Fathman, C.G. and 169. Frankel, A.D. and Pabo, C.O. Rothbard, J.B. (2000) (1988) Cellular uptake of the tat Polyarginine enters cells more protein from human efficiently than other polycationic immunodeficiency virus. Cell, homopolymers. J Pept Res, 56, 55, 1189-1193. 318-325. 170. Green, M. and Loewenstein, 177. Richard, J.P., Melikov, K., Vives, P.M. (1988) Autonomous E., Ramos, C., Verbeure, B., functional domains of chemically Gait, M.J., Chernomordik, L.V. synthesized human and Lebleu, B. (2003) Cell- immunodeficiency virus tat trans- penetrating peptides. A activator protein. Cell, 55, 1179- reevaluation of the mechanism of 1188. cellular uptake. J Biol Chem, 278, 171. Vives, E., Brodin, P. and Lebleu, 585-590. B. (1997) A truncated HIV-1 Tat 178. Morris, M.C., Chaloin, L., Mery, protein basic domain rapidly J., Heitz, F. and Divita, G. (1999) translocates through the plasma A novel potent strategy for gene membrane and accumulates in delivery using a single peptide the cell nucleus. J Biol Chem, vector as a carrier. Nucleic Acids 272, 16010-16017. Res, 27, 3510-3517. 172. Derossi, D., Calvet, S., 179. Morris, M.C., Gros, E., Aldrian- Trembleau, A., Brunissen, A., Herrada, G., Choob, M., Chassaing, G. and Prochiantz, A. Archdeacon, J., Heitz, F. and (1996) Cell internalization of the Divita, G. (2007) A non-covalent third helix of the Antennapedia peptide-based carrier for in vivo homeodomain is receptor- delivery of DNA mimics. Nucleic independent. J Biol Chem, 271, Acids Res, 35, e49. 18188-18193. 180. Morris, M.C., Vidal, P., Chaloin, 173. Derossi, D., Joliot, A.H., L., Heitz, F. and Divita, G. Chassaing, G. and Prochiantz, A. (1997) A new peptide vector for (1994) The third helix of the efficient delivery of

oligonucleotides into mammalian conserved (K/R)XTQT motif. J cells. Nucleic Acids Res, 25, Biol Chem, 276, 14059-14066. 2730-2736. 189. Stewart, M. (2007) Molecular 181. Langel, Ü. (2006) Handbook of mechanism of the nuclear protein Cell-Penetrating Peptides. CRC import cycle. Nat Rev Mol Cell PRESS. Biol, 8, 195-208. 182. Abes, S., Moulton, H.M., Clair, 190. Fontes, M.R., Teh, T., Toth, G., P., Prevot, P., Youngblood, D.S., John, A., Pavo, I., Jans, D.A. and Wu, R.P., Iversen, P.L. and Kobe, B. (2003) Role of flanking Lebleu, B. (2006) Vectorization sequences and phosphorylation in of morpholino oligomers by the the recognition of the simian- (R-Ahx-R)4 peptide allows virus-40 large T-antigen nuclear efficient splicing correction in the localization sequences by absence of endosomolytic agents. importin-alpha. Biochem J, 375, J Control Release, 116, 304-313. 339-349. 183. Moulton, H.M. and Moulton, J.D. 191. Lyonnais, S., Hounsou, C., (2004) Arginine-rich cell- Teulade-Fichou, M.P., Jeusset, J., penetrating peptides with Le Cam, E. and Mirambeau, G. uncharged antisense oligomers. (2002) G-quartets assembly Drug Discov Today, 9, 870. within a G-rich DNA flap. A 184. Lundberg, P., El-Andaloussi, S., possible event at the center of the Sutlu, T., Johansson, H. and HIV-1 genome. Nucleic Acids Langel, Ü. (2007) Delivery of Res, 30, 5276-5283. short interfering RNA using 192. Arhel, N.J., Souquere-Besse, S., endosomolytic cell-penetrating Munier, S., Souque, P., peptides. Faseb J. Guadagnini, S., Rutherford, S., 185. Lemkine, G.F. and Demeneix, Prevost, M.C., Allen, T.D. and B.A. (2001) Polyethylenimines Charneau, P. (2007) HIV-1 DNA for in vivo gene delivery. Curr Flap formation promotes Opin Mol Ther, 3, 178-182. uncoating of the pre-integration 186. Venugopalan, P., Jain, S., Sankar, complex at the nuclear pore. S., Singh, P., Rawat, A. and Embo J, 26, 3025-3037. Vyas, S.P. (2002) pH-sensitive 193. Chen, X., Wu, J.M., Hornischer, liposomes: mechanism of K., Kel, A. and Wingender, E. triggered release to drug and gene (2006) TiProD: the Tissue- delivery prospects. Pharmazie, specific Promoter Database. 57, 659-671. Nucleic Acids Res, 34, D104-107. 187. Dowty, M.E., Williams, P., 194. Bai, M., Campisi, L. and Zhang, G., Hagstrom, J.E. and Freimuth, P. (1994) Vitronectin Wolff, J.A. (1995) Plasmid DNA receptor antibodies inhibit entry into postmitotic nuclei of infection of HeLa and A549 cells primary rat myotubes. Proc Natl by adenovirus type 12 but not by Acad Sci U S A, 92, 4572-4576. adenovirus type 2. J Virol, 68, 188. Lo, K.W., Naisbitt, S., Fan, J.S., 5925-5932. Sheng, M. and Zhang, M. (2001) 195. Wickham, T.J., Mathias, P., The 8-kDa dynein light chain Cheresh, D.A. and Nemerow, binds to its targets via a G.R. (1993) Integrins alpha v beta 3 and alpha v beta 5 promote

adenovirus internalization but not functional peptides from virus attachment. Cell, 73, 309- bioplexes. J Control Release, 98, 319. 169-177. 196. Deng, H., Liu, R., Ellmeier, W., 204. Nir, H., Eichen, Y. and Schuster, Choe, S., Unutmaz, D., Burkhart, G. (2005) The construction of M., Di Marzio, P., Marmon, S., DNA molecules of figure-eight Sutton, R.E., Hill, C.M. et al. structure. Anal Biochem, 344, 86- (1996) Identification of a major 91. co-receptor for primary isolates 205. Frank-Kamenetskii, M. and of HIV-1. Nature, 381, 661-666. Mirkin, S. (1995) Triplex DNA 197. Arteaga, H.J., Hinkula, J., van structures. Annu Rev Biochem., Dijk-Hard, I., Dilber, M.S., 64, 65-95. Wahren, B., Christensson, B., 206. Lee, J.S., Latimer, L.J., Haug, Mohamed, A.J. and Smith, C.I.E. B.L., Pulleyblank, D.E., Skinner, (2003) Choosing CCR5 or Rev D.M. and Burkholder, G.D. siRNA in HIV-1. Nat Biotechnol, (1989) Triplex DNA in plasmids 21, 230-231. and chromosomes. Gene, 82, 198. Qin, X.F., An, D.S., Chen, I.S. 191-199. and Baltimore, D. (2003) 207. Rothemund. (2006) Folding Inhibiting HIV-1 infection in DNA to create nonscale shapes human T cells by lentiviral- and patterns. Nature, 440, 297- mediated delivery of small 302. interfering RNA against CCR5. 208. Brucale, M., Zuccheri, G. and Proc Natl Acad Sci U S A, 100, Samori, B. (2006) Matering the 183-188. complexity of DNA 199. Pooga, M., Hällbrink, M., Zorko, nanostructures. TRENDS in M. and Langel, Ü. (1998) Cell Biotechnology, 24, 235-243. penetration by transportan. Faseb 209. Seeman, N.C. and Lukeman, P.S. J, 12, 67-77. (2005) Nucleic acid 200. Niemeyer, C.M. (2007) nanostructures: bottom-up control Functional devices from DNA of geometry on the nanoscale and proteins. NanoToday, 2, 42- Reports on Progress in Physics, 52. 237-270. 201. Seeman, N.C. (2003) DNA in a 210. Mao, C., Sun, W., Shen, Z. and material world. Nature, 421, 427- Seeman, N.C. (1999) A 431. nanomechanical device based on 202. Simonson, O.E., Svahn, M.G., the B-Z transition of DNA. Tornquist, E., Lundin, K.E. and Nature, 397, 144-146. Smith, C.I.E. (2005) Bioplex 211. Seeman, N.C. (2006) DNA technology: novel synthetic gene enables nanoscale control of the delivery pharmaceutical based on structure of matter. Q Rev peptides anchored to nucleic Biophys, 1-9. acids. Curr Pharm Des, 11, 212. Alivisatos, A.P., Johnsson, K.P., 3671-3680. Peng, X., Wilson, T.E., Loweth, 203. Svahn, M.G., Salih, G., C.J., Bruchez, M.P., Jr. and Simonson, E.O., Smith, C.I.E. Schultz, P.G. (1996) and Branden, L.J. (2004) Organization of 'nanocrystal Protease-induced release of

molecules' using DNA. Nature, activity. J Biol Chem, 277, 382, 609-611. 22725-22733. 213. Braun, E., Eichen, Y., Sivan, U. 221. Arteaga, H.J., Hinkula, J., van and Ben-Yoseph, G. (1998) Dijk-Hard, I., Dilber, M.S., DNA-templated assembly and Wahren, B., Christensson, B., electrode attachment of a Mohamed, A.J. and Smith, conducting silver wire. Nature, C.I.E.E. (2003) Choosing CCR5 391, 775-778. or Rev siRNA in HIV-1. Nat 214. Braun, E. and Keren, K. (2004) Biotechnol, 21, 230-231. From DNA to transistor. 222. Zhang, J.H., Chung, T.D. and Advances in Physics, 53, 441- Oldenburg, K.R. (1999) A 196. Simple Statistical Parameter for 215. Winfree, E., Liu, F., Wenzler, Use in Evaluation and Validation L.A. and Seeman, N.C. (1998) of High Throughput Screening Design and self-assembly of two- Assays. J Biomol Screen, 4, 67- dimensional DNA crystals. 73. Nature, 394, 539-544. 223. de Diesbach, P., Berens, C., 216. Paukstelis, P.J., Nowakowski, J., N'Kuli, F., Monsigny, M., Birktoft, J.J. and Seeman, N.C. Sonveaux, E., Wattiez, R. and (2004) Crystal structure of a Courtoy, P.J. (2000) continuous three-dimensional Identification, purification and DNA lattice. Chem Biol, 11, partial characterisation of an 1119-1126. oligonucleotide receptor in 217. Lund, K., Liu, Y., Lindsay, S. membranes of HepG2 cells. and Yan, H. (2005) Self- Nucleic Acids Res, 28, 868-874. assembling a molecular pegboard. J Am Chem Soc, 127, 17606-17607. 218. Wetzer, B., Byk, G., Frederic, M., Airiau, M., Blanche, F., Pitard, B. and Scherman, D. (2001) Reducible cationic lipids for gene transfer. Biochem J, 356, 747-756. 219. Tang, F. and Hughes, J.A. (1998) Introduction of a disulfide bond into a cationic lipid enhances transgene expression of plasmid DNA. Biochem Biophys Res Commun, 242, 141-145. 220. Hsu, C.H., Wu, S.H., Chang, D.K. and Chen, C. (2002) Structural characterizations of fusion peptide analogs of influenza virus hemagglutinin. Implication of the necessity of a helix-hinge-helix motif in fusion