Field of the Invention

The present invention relates generally to oligonucleotides which form triplexes with duplex DNA and which contain sites for the addition of DNA damaging agents. It further relates to methods of using these triplex forming oligonucleotides to specifically damage DNA. It more particularly relates to site selective DNA damaging using an m-gap triplex comprising two individual triple helix forming oligonucleotides connected by a polymeric linker. This polymeric linker is of sufficient length to allow the first and second TFO to bind at their respective sites.

Background of the Invention The studies of Dervan & Helene have shown that by targeting a polypurine tract within a naturally occurring DNA region, synthetic DNA oligonucleotides with the base composition (C _n , T _m ) form stable triple helices at acidic pH (1,2). Formation of those structures is based upon TAT and

C+GC triplets, where C+ refers to protonated C. Such structures bind so as to place the TFO parallel to the more purine-rich strand of the duplex target. The binding is stabilized by Hoogsteen binding within the major helix groove (3). Based upon work in several laboratories, it has also been shown that there exists a second triple helix motif, which is an antiparallel complex stabilized by GGC and TAT triplets (4,5). Binding occurs within the major groove and appears to be stabilized by reverse Hoogsteen H-bonding (6).

For both motifs, triple helix formation with standard base substituents has been shown to be maximally stable within homopurine-homopyridmidine duplex target sites. Recently, approaches have appeared to reduce that limitation. In particular, it has been shown by Johnston (7) and by Dervan (8,9) that triple helix formation can occur on a duplex target site in which a

homopurine region is fused to a homopyrimidine region. The Dervan group has addressed such binding site symmetry by introducing a 3'-3' phosphodiester "strand switch" linkage in the TFO at the site of the junction between the purine rich and the pyrimidine rich sequence (9). The Johnston and Dervan groups have accommodated the switch by employing both the parallel and antiparallel TFO motifs in the same TFO (7,8).

1. Moser H. and Dervan, P.B. (1987) Science 238: 645-650.

2. Le Doan T., Perroault L., Praseuth D., Habhoub N., Decout J.L., Thuong N. T., Lhomme J. and Helene C. (1987) Nucleic Acids Res.

15:7749-7760.

3. de los Santos C, Rosen M. and Patel D. (1989) Biochemistry 28:7282- 7289.

4. Durland, R.H., Kessler, D.J., Gunnell, S., Duvic, M.D., Pettitt, B.M. and Hogan, M.E. Binding of Triple Helix Forming Oligonucleotides to

Sites in Gene Promoters (1991) Biochemistry 30:9246-9255. 4. Beal, PA. and Dervan, P.B. A Second Structural Motif For Recognition of DNA by Triple Helix Formation (1991) Science 251: 1360-1363. 6. Radhakrishnan I, de los Santos C. and Patel D. (1991) J. Mol. Biol.

221:1413-1418.

7. Jayasena S.D. and Johnston B.J. (1992) PNAS 89:3526-3530.

8. Beal P.A. and Dervan P.N. (1992) JACS 114:4976-4982.

9. Home D.A. and Dervan P.B. (1990) JACS 112: 2435-2437. 10. Tedder D.G., Everett R.D., Wilcox K.W. Beard P. and Pizer L.I. (1989)

J. Virol. 63:2510-2520.

The present invention provides another novel method to accommodate duplex binding sites which fail to display simple homopurine/homopyrimidine asymmetry. The novel approach decomposes a duplex target site into a series of isolated homopurine or homopyrimidine sub-domains, linked by spacer

DNA elements with a random sequence.

Summary of the Invention

An object of the present invention is a triplex forming oligonucleotide for site selective DNA damaging.

An additional object of the present invention is a method for treating diseases which can be treated by damaging DNA.

A further object of the present invention is an "m-gap" TFO.

Thus in accomplishing the foregoing objects there is provided in accordance with one aspect of the present invention an m-gap triplex forming oligonucleotide comprising a first and second TFO, each capable of binding a duplex DNA to form a triplex, wherein said first TFO and said second TFO bind at sites within the duplex DNA which are contiguous and, a chemical linker connecting the first TFO to the second TFO, the linker of sufficient length to allow the first and second TFOs to bind the duplex DNA at the appropriate sites. As an alternative embodiment a plurality of contiguous TFOs can be linked.

In a further embodiment the m-gap TFO is used as a site selective DNA damaging agent. In this embodiment a DNA damaging agent is attached to the chemical linker and positioned upon the chemical linker for site selective DNA damage.

In specific embodiments of the present invention the chemical linker is selected from the group of consisting of a polyether, a peptide linker based upon the proline II helix, other peptides, branched polymers with amino or guanidium side chains. In an additional embodiment of the present invention the DNA damaging agent is selected from the group of psoralen, acridine derivatives, mustards, bromoacetates and others.

Another embodiment of the present invention is a method for site selection DNA damaging of duplex DNA comprising the step of delivering a pharmacological dose of an m-gap TFO to the duplex DNA to be damaged.

This can be used to treat a variety of diseases which can be treated by damaging DNA in animals and humans.

In a preferred embodiment of the present invention an m-gap TFO is used to treat HSV-1 infection. In this procedure a pharmacological dose of the m-gap TFO is delivered to an individual infected with HSV-1. The m-gap TFO specifically binds the HSV-1 genome to form a triplex. In the preferred embodiment the m-gap TFO binds the promoter region in the D glycoprotein gene.

Other and further objects, features and advantages will be apparent from the following description of the present preferred embodiments of the invention which are given for the purpose of disclosure when taken in conjunction with the accompanying drawings.

Brief Description of the Drawings

Figures 1A-C show schematically chemical linkers useful in the present invention. Figure 1A shows the L linker which is an 11 bond linker; Figure IB shows the S linker which is a 6 bond linker; and Figure 1C shows the K linker which is a 5 bond linker.

Figure 2 is a schematic representation of the Herpes Simplex Virus I gD promoter region.

Figures 3A and B are three dimensional representations showing the structure of a TFO of the present invention bound to duplex DNA. Figure 3A shows the TFO bound to the promoter region of the HSV-1 gD promoter targets when the target sites are an integer number of helix turns apart. Figure 3B represents TFO bound to duplex when the target sites are a non- integer distance apart. The left picture shows a molecular model with two isolated segments of TFO bound to the duplex target. The middle picture shows a molecular model with a 33 bond "L-linker" connecting the two TFO segments and in the process span, without distortion, a 1/2 helical turn spacer region of the duplex target. The right structure shows the TFO-linker hybrid in the major groove of the duplex target site. This is emphasized by removing the underlying duplex, so as to reveal the continuing helical structure which is assumed by the TFO-linker hybrid upon binding. Red: Oligonucleotide component of the TFO-hybrid. Yellow: Polymer component of the TFO-

hybrid. Blue: Purine-rich strand of the duplex binding site. Green: Pyrimidine-rich strand of the duplex binding site.

Figure 4 shows the duplex sequence at the HSV-1 gD promoter site of the human and the oligonucleotide with linkers which form triplexes. Figure 5 shows the HSV-1 gD promoter site in humans, as well as a variety of TFOs of the present invention.

Figure 6 shows the Hamster APRT site at Exon 2, as well as TFOs of the present invention which bind to the site.

Figure 7 shows the Exon 5 site of the Hamster APRT duplex, as well as TFOs which bind at this site.

Figure 8 shows the chemical structure of some reactive chemicals which can be used to damage the DNA. Figure 8A shows psoralen derivatives. Figure 8B shows azido rhodamine derivatives. Figure 8C shows Acridine 9 carboxylic acid derivatives. Figure 9 is a schematic of the core promoter region for the human progesterone receptor. This promoter possesses the CAAT box and the GC box elements. It shows the linking of three oligonucleotides to form a TFO of the present invention.

Figure 10 is an electrophoresis gel showing the bound species of the L series hybrid TFOs bound to the duplex DNA.

Figure 11 is a DNasel footprint of the 43bp fragment of the GT promoter showing TFO binding.

Figure 12 is a DNasel footprint of the 93bp fragment of the GT promoter showing TFO binding. Figure 13 shows the results of band shift analysis for the L series.

Figure 14 shows the primary structure of the triple helix in Figure 3B.

The drawings are not necessarily to scale. Certain features of the invention may be exaggerated in scale or shown in schematic form in the interest of clarity and conciseness.

Detailed Description of the Invention

It will be readily apparent to one skilled in the art that various substitutions and modifications may be made to the invention disclosed herein without departing from the scope and spirit of the invention. The term "oligonucleotide" as used herein is defined as a molecule comprising two or more deoxyribonucleotides or ribonucleotides, preferably more than 10. The exact size depends on many factors, including the specificity and binding affinity.

The term "TFO" or "triplex forming oligonucleotide" as used herein refers to synthetic oligonucleotides which are capable of forming a triple helix by binding the major groove of the duplex DNA structure. These TFOs can be referred to as the "Hogan" type and are prepared by the "Hogan" technique.

This technique generally comprises steps of identifying genomic DNA nucleotide target sequences of greater than about 20 nucleotides and having either about at least 65% purine bases or about at least 65% pyrimidine bases; and synthesizing the synthetic oligonucleotide complementary to the identified target sequence, the synthetic oligonucleotide having a G when the complementary location in DNA duplex has a GC base pair, having a T when the complementary location in the DNA duplex has an AT base pair. In specific embodiments the synthetic oligonucleotide can be selected from a group consisting of an oligonucleotide oriented 5' to 3' and binding parallel to the about at least 65% purine strand, or an oligonucleotide oriented 3' to 5' and binding antiparallel to the about 65% purine strand. As used herein, the term "major groove" refers to one of the grooves along the outer surface of the double stranded DNA helix which is formed because of the sugar phosphate backbone duplex DNA extends further from the axis than the bases. The major groove is important for binding regulatory molecules to DNA sequences. As used herein, the term "minor groove" refers to the other groove on the outer surface of the double stranded DNA helix.

In referring to "bases" herein, the term includes both the deoxyribonucleic acids and ribonucleic acids. The following abbreviations are

used. "A" refers to adenine as well as to its deoxyribose derivative, "T" refers to thymine, "U" refers to uridine, "G" refers to guanine as well as its deoxyribose derivative, "C" refers to cytosine as well as its deoxyribose derivative, "F" or "N" in sequences refers to other analogs, "X" refers to xanthene as well as its deoxyribose derivative and "I" refers to inosine.

The term "chemical linkers" as used herein refers to a group of chemical compounds which can be used to link two TFOs. Although a variety of chemical linkers are known in the art, they are usually selected from the group consisting of a peptide linker based on the proline II helix or a polyether or polymers with a branched side chain or linear polymers.

Examples of chemical linkers are shown in Figures LA, IB and lC.

The term "pharmacological dose" as used herein refers to the dose of an m-gap TFO which causes a pharmacological effect when given to an animal or human. The pharmacological dose introduced into the animal or human to be treated, will provide a sufficient quantity of m-gap TFO to the target site to bind to the target sequence of the duplex DNA and provide a specific effect either (1) preventing the target site from functioning or (2) damaging the duplex DNA at the specific site or (3) ablating the DNA at the site or (4) inhibiting the transcription/translation of the gene under the regulation of the site being bound. One skilled in the art will readily recognize that the dose will be dependent upon a variety of parameters, including the age, sex, height and weight of the human or animal to be treated, the organism or gene location which is to be attacked and the location of the target sequence within the organism. Given any set of parameters, one skilled in the art will be able to readily determine the appropriate dose.

One embodiment of the present invention is an "m-gap TFO." In one specific embodiment the m-gap TFO comprises a plurality of TFOs, each capable of binding a target site within a duplex DNA to form a triplex. The plurality of TFOs bind at target sites within the duplex DNA which are contiguous. Chemical linkers connect the plurality of TFOs. The linkers are of sufficient length to allow the plurality of TFOs to bind the duplex DNA at the target sites.

In an alternative embodiment the m-gap TFO comprises a first and second TFO each capable of binding a duplex DNA to form a triplex. The first TFO and the second TFO bind at sites within the duplex DNA which are contiguous. A chemical linker connects the first TFO to the second TFO. The linkers are of sufficient length to allow the first and second TFOs to bind to the duplex DNA at their appropriate sites.

As used herein, the term "contiguous" refers to sites which are near each other. It could include adjacent sites or sites separated by a distance. The distance can be of any size but is usually at least 1/5 of a helical turn and can be as large as seven helical turns. This is distinguishable from the prior art where the distance approaches zero.

Another embodiment of the present invention is an m-gap TFO used as a site selective DNA damaging agent. In this embodiment the m-gap TFO further comprises a DNA damaging agent attached to the chemical linker and positioned upon the chemical linker for site selective DNA damage. When the

TFOs bind at their appropriate sites the damaging agent is able to interact with the duplex DNA to damage it at a specific site.

In the present invention it is important to remember that the chemical linker must be of a sufficient length to allow for triplex formation by the TFOs. The length can be specifically selected such that the formed triplex will not distort the duplex DNA or the formed triplex will distort the duplex DNA. The length of the chemical linker can be increased or decreased and the m-gap TFO will still bind and form a triplex. The formation of the triplex results in some torsional stress or bending stress in the duplex DNA. Torsion or bending stress can, itself, help induce damage of the DNA. The linker provides a mechanism to link TFO binding sites to provide a co-operative binding unit. In the present invention it is found that binding takes place under different scenarios. If the distance spanned by a chemical linker between any two TFOs is an integral or nearly integral number of helical turns, the chemical linker will cross over the minor groove and connect the two TFOs through the minor groove. An example of a structure of such a complex is the HSV-1 gD promoter target, Figure 2. A three dimensional structure for this prototype is displayed in Figure 3A. On the other hand, if

the distance between the two TFOs is substantially less than one (for instance a half integral turn), the linker lays within the major groove and combines or connects the two TFOs as in Figure 3B. On the other hand, when the distance is a combination of these, that is, there is at least an integral number of helical turns in combination with the substantially less than integral number (for example, one and a half or two and a quarter helical turns), then both types of linkage of the TFOs occurs. Part of the chemical linker will cross over the minor groove and the other part will lay in the major groove connecting the two TFOs. When the m-gap linker crosses over the top of the minor helix groove, this is usually the path of closest approach between the bound TFO elements.

Further, when there is greater than two TFOs linked, any of the above combinations can be used between any two TFOs.

One skilled in the art will readily recognize that there are a variety of DNA damaging or ablation agents which can be used; for example, psoralen or acridine derivatives, bromoacetates, epoxides and mustards.

Specific m-gap TFOs which have been found to be useful are Nos. (I) to (XXXI) shown in Figures 2-7.

In another embodiment of the present invention, the m-gap TFOs are used in a method for site selective DNA damaging of a duplex DNA. This method comprises the step of delivering a pharmacological dose of an m-gap TFO to a duplex DNA to be damaged.

The m-gap TFO can also be used in a method for treating disease in humans and animals. This procedure is useful in any disease which can be treated by damaging native DNA in the animal or human or damaging foreign

DNA (viral, bacterial or other) which is present in the animal or human. The procedure comprises the step of delivering a pharmacological dose of an m- gap TFO to the human or animal to be treated wherein the m-gap TFO specifically binds to a duplex DNA and site specifically damages the duplex DNA.

In one specific embodiment of the present invention an m-gap TFO procedure is used for the treatment of HSV-1 infection. In one instance the m-gap specifically binds to the HSV-1 genome to form a triplex. A preferred

binding site is the promoter region of the D glycoprotein (gD) gene. A specific TFO useful in the treatment of HSV-1 infection is No. (II).

GENERAL PRINCIPLES OF M-GAP LINKER DESIGN Based upon data derived from the examples below, general principles can be derived to guide the design of the two limiting classes of m-gap linker.

MINOR GROOVE SPAN (Type 1) In this limiting motif, binding sites for TFO elements are separated by approximately integral multiples of one helix turn. Although the spacers are preferably 1 or 2 integral turns (10 or 20bp), one skilled in the art readily recognizes that it is possible to accommodate spacers as long as 5-7 turns. Upon triple helix formation of this type, the linker is positioned near the surface of the duplex and follows a path which is nearly parallel to the helix axis. The separation between sites of TFO connection is approximately 30-35 angstroms per turn of spacer duplex. Employing flexible linkers, this separation can be accommodated by a linker which provides approximately 25 rotatable bonds per turn of spacer duplex. Linkers shorter than this can still provide for triplex helix formation, but only upon the introduction of torsional or bending stress into the duplex. Linkers longer than this also provide for triple helix formation, but cooperativity between bound TFO elements is diminished.

MAJOR GROOVE OCCUPANCY (Type 2) In this limiting motif, binding sites for TFO elements are separated by significantly less than one helix turn. Upon triple helix formation, the linker is positioned in the major groove of the spacer duplex. In this instance, the separation between sites of TFO connection is approximately 3-4 angstroms per base pair equivalent of spacer duplex. Employing flexible linkers, this separation can be accommodated by a linker which provides approximately

5 rotatable bonds per base pair equivalent. Linkers shorter than this can still provide for triplex helix formation, but only upon the introduction of torsional or bending stress into the duplex. Linkers longer than this also provide for

triple helix formation, but cooperativity between bound TFO elements is diminished.

RIGID LINKERS The general structural design principles described above will also hold for rigid linkers, including ordered polypeptides and other molecules. However, in this instance, rotatable bonds are not the design principle of importance. Rather, the rigid linker must be capable of accommodating the length and shape of the minor groove path (Type 1) or the major groove path (Type 2).

MDCED OCCUPANCY

One skilled in the art can readily appreciate that some TFO elements will be separated by a distance which is greater than 1 helical turn but significantly less than an integral helical turn. (For example: 1-1/4, 2-1/2.)

In this situation, spacers incorporating both a Type 1 and Type 2 motif can be used.

Example 1 OLIGONUCLEOTIDE SYNTHESIS

Oligonucleotide and oligonucleotide-linker conjugates were prepared by the phosphoramidite method (Milligen Biosearch, Inc.) and purified by Sepaharose Q FPLC chromatography at pH 11. Subsequent to detritylation and exclusion chromatography on a G25 resin, the purity of the resulting product was confirmed by electrophoresis. Oligonucleotides were freeze-dried then redissolved in distilled water for binding analysis. ³² P 5'-end-labelled oligonucleotides were prepared by standard procedures, followed by G25 exclusion chromatography to remove unincorporated label.

Example 2

M-GAP TFO BINDING TO THE PROMOTER OF HERPES SIMPLEX VIRUS TYPE 1 (HSV-1) D GLYCOPROTEIN PROMOTER

It is possible to connect distant sites of TFO-DNA interaction by linking them with a chemical linker; this is referred to as an m-gap TFO binding motif. An example of this is the TFO binding to the promoter of the herpes simplex virus, Type 1 (HSV-1) D glycoprotein (gD) promoter. This example uses linked TFO half molecules which bind to a promoter site spanning -79 to -40 relative to the cap site of the HSV-1 glycoprotein D. The structure of this TFO series is displayed in Figure 2.

As seen in Figure 2, this gD binding domain comprises two 12bp homopurine runs, which serve as sites for local triplex formation. These two domains are coincident with sites Gl and G2 which are implicated in gD promoter function. Consequently, TFO binding upon this domain has functional significance. The two purine rich domains in this promoter are separated by a nine bp long, AT rich spacer, identified by underlining. As discussed below, if 12mer TFOs bind concurrently, the path of closest approach between these two 12bp segments is predicted to be a straight line, parallel to the helix axis and spanning the surface of the minor helix groove of the spacer duplex. Simple consideration of helix shape suggests that such a paired TFO structure would require 25-30 freely rotatable bonds in the linker, to span the 30-35A of intervening distance imposed by a single turn of B or A type helix. On the other hand, if the polymeric linker were to continue a helical path between the half molecules through the major groove, approximately 50-60 rotatable bonds would be required.

As seen in Figures 2 and 4, L=2 to 6 in this series form a 1-1 triple helix and simultaneously occupy both triple helix forming sites. All five TFO molecules display high binding affinity (K^ , but the TFO with the 22 rotatable bond linker (L=2) was the best for a single helix separation. Since at least 60 bond equivalents would be required to circumscribe a turn of a major groove (there are 66 rotatable bonds in 10 bases). The remarkable behavior of the L=2 complex and its analogues indicates that it connects half

molecules by spanning the minor helix groove, as shown in Figure 3. Similar results have been obtained with a linker (K series) comprising multiples of five rotatable bonds.

The molecular model which approximates the structure for the L=2 complex is displayed in Figure 3. In this molecular model, 25 bonds

(including the terminal bases) is sufficient to provide the required linker span.

The HSV-1 gD promotor complex is an example of a gene location susceptible to the m-gap TFO motif. By direct inspection of the molecular model, it is clear to one skilled in the art that ligands, such a psoralen, intercalators, etc., which can produce DNA damage by binding or attaching the DNA in through the minor groove can be attached to the linker of an m- gap TFO.

Example 3

VARIATION OF LINKER CHEMISTRY: THE HSV GENE The m-gap motif is enhanced to use it to port DNA damage into the minor helix groove by attachment of reactive chemicals to the m-gap linker. The HSV-1 gD promotor system of Example 1 was used. The HSV-1 gD region is used for chemical refinement and for chemical mapping analysis.

One skilled in the art recognizes that a variety of linkers are available. The choice of linker will depend on the triplex formed and the damaging agent used. One example of a linker is the flexible, inert polyether m-gap linker. Once the polyether linker was chosen, the next step was to introduce reactive groups into the polyether m-gap linker. Three approaches are taken:

(i) add an acridine intercalator (A) affixed to its own linker element; (ii) add a psoralen derivative (P) affixed to its own linker element; and (iii) add an amine linker element (NH ₂ ) as a simple alkylator attachment site.

These attachment chemistries are described in Example 4. The resulting m-gap TFO schemes are shown in Figure 5.

Based upon molecular modeling, it has been observed that the flexible polyether linkage can be improved by introducing additional conformational constraint. To achieve that end, a series of rigid m-gap linkers, based upon

the use of the polyproline II (pro II) helix as a rigid linear element are synthesized. The pro II helix is a linear, 3-fold helix with a 3.1 A rise per repeat. Hydrodynamic measurements show that it is among the most rigid linear polymers yet described (persistence length=150 amino acids). Based upon those physical properties, the following family of modified polypeptides are synthesized employing standard Fmoc solid phase chemistry.

A. SER-PRO-HB LINKER SERIES: INERT

OH-SER-(PRO) _n -HB-OH where n = 8,9,10

HB = 2-hydroxy butyrate

B. SER-PRO-HB LINKER SERIES: AMINE

SER - (PRO)„-K-(PRO) _n -HB-OH where n = 3,4

K = lysine

HB = 2-hydroxy butyrate

C. SER-PRO-HB LINKER SERIES: THIOL

OH-SER-(PRO) _n -C-(PRO) _n -HB-OH where n = 3,4

C = cysteine

HB = 2-hydroxy butyrate

Since the Pro II helix has a repeat near to 10A and is shaped like a corkscrew, the SER-PR0 ₉ -H m-gap linker possesses a shape which, although left handed, appears to be generally complementary to the surface features of the minor helix groove (which is approximately 10A wide). The model suggests that an alkylator affixed to the SER-PR0 ₉ -H linker near its center would be placed in close proximity to the duplex, sandwiched in a "hydrophobic pocket" between peptide and nucleic acid base planes in the

minor groove. For that reason, excellent proximity and dehydration effects in the minor groove at the m-gap linker are seen.

The SER-PRO _β -H m-gap linker and its derivatives possess a primary alcohol and a secondary alcohol at each end, as is the case for a nucleotide. Therefore, subsequent to synthesis and purification by reverse phase HPLC, the DMT phosphoramidate is synthesized, employing standard methods and then employed as a single synthon in solid phase oligonucleotide synthesis.

Once the correct span of the SER-PR0 ₉ -H m-gap linker is calculated for a given location in the genome, lysine (K) or cysteine (C) are introduced at the center as a site for postsynthetic attachment of reactive chemicals.

During oligonucleotide synthesis lysine and cysteine are protected with Fmoc and DMT, respectively, as is the case for simple, commercially available aliphatic linkers.

Example 4

M-GAP TFO DESIGN: APRT GENE: EXON 2

Based upon modeling, it appears that the Type 1, or minor groove spanning, m-gap motif is useful when the first and second TFO binding sites are separated by as little as 7bp (3/4 of a helix turn). Interestingly, in the m- gap binding motif a 7bp spacer region will require a longer m-gap linker than for a lObp spacing. By analogy with the HSV-1 prototype, it is estimated that 33 to 35 rotatable bonds (employing the polyether motif) are sufficient. Similarly, an increase to PRO ₁₀ is required for the rigid, polyproline m-gap linker series. Exon 2 of the Hamster APRT gene is an example of a site which possesses the characteristics suited for m-gap TFO binding. The structure of the site and the first family of m-gap TFOs to be synthesized are shown in Fig. 6.

The crucial features of the site are the following: 1. There are two purine-rich, GC-rich domains. The two TFO elements are good TFO binding sites. Modifying the Hogan type binding by placing a T in the TFO in opposition to the single CG inversion enhances triplex formation.

2. The TFO binding domains are displaced by approximately 3/4 of a helix turn. Although a 9bp spacer (lObp repeat) is probably ideal and can be accommodated by a 22-25 bond m-gap linker, modeling suggests that a 6bp spacer (7bp repeat) is accessible by increasing the size of the M-gap linker by about 50%. Therefore, three 11 bond "L" polyether linker elements (33 bonds) are introduced at this APRT site. This is in contrast to the HSV-1 site wherein two were sufficient. Also by analogy with the HSV-1 prototype, rigid SER-PRO _n -H linkers are introduced, but with an extended span (n = 10,11,12). One skilled in the art will readily recognize that the most favorable span chosen (ether/proline) will depend on the site of attachment of the two TFOs and the distance between the TFOs. Once the span is chosen the m-gap linker is used as the lattice for alkylator attachment.

3. The bound TFO is ideal for psoralen crosslinking and internal crosslinking. The sequence which flanks the left end of the binding site is ATAT. This comprises tandem consensus sites for psoralen chemistry.

Consequently, TFOs which display the highest affinity are coupled postsynthetically to psoralen, employing the polynucleotide kinase/gamma thio-ATP method, or equivalents.

Example 5

M-GAP TFO DESIGN: APRT GENE EXON 5:

This site is seen in Figure 7 and its crucial features of the site are the following:

1. The purine-rich GC-rich domains are expected to be very good TFO binding sites. Again a modification of the Hogan type TFO where T is placed in the TFO in opposition to the single CG inversion enhances binding.

2. TFO binding domains are displaced by approximately 1 helix turn and the second domain is a purine rich site. The lObp spacer in this site is probably ideal and can be accommodated by a 22-25 bond m-gap linker (see Example 1). However, the two TFO half sites are related by partial two fold symmetry requiring that the polarity of the two be reversed within the linker domain. This strand polarity switch is performed postsynthetically, by carbodiimide coupling of a terminal 5' amine linker to a 5' phosphate which

had been chemically synthesized into the "left" half of the TFO. This yields a stable phosphoramidate, which is referred to as -NP02-.

3. The bound TFO is ideal for internal crosslinking since the paired TFOs possess three CG inversions. Internal alkylation is obtained at these sites by postsynthetic modification.

Example 6 ADDITION OF REACTIVE CHEMICALS

1. Bromoacetate and iodoacetate addition to primary amine groups on the TFO. In this approach, the succimidate of bromoacetic acid is linked to a primary alkylamine moiety which has been synthesized into the TFO. Primary amines which will be used are: (i) primary amine side chains on the m-gap linker and (ii) amine groups placed at the 5' or 3' termini.

This post synthetic addition of bromoacetate to amines on an oligonucleotide is known in the art. An example is described in Meyer, R.B., et al. (1989) JACS 111:8517-8519. It is a convenient pathway to introduce a slowly reactive nucleophile onto an oligonucleotide. The rationale behind the method is that although a bromo or iodoacetyl group reacts with N7 of guanosine, it does so very slowly in dilute aqueous solution. Therefore, so long as a TFO remains unbound, in dilute solution, the TFO-alkylator conjugate does not react with itself.

One skilled in the art recognizes that in terms of nucleophilicity and the length of its linkage to the amine, the bromoacetate group works at this linker. However, as other chemical groups are implemented as oligo- conjugates they may provide better oligo-alkylator conjugates.

2. Psoralen derivatives. The primary photochemical approach of the present invention employs the iodohexane psoralen adduct. Its structure is displayed in Figure 8A.

This derivative is fixed to a 5' terminal phosphorothioate which was introduced chemically during synthesis, thereby generating the stable thioester. This reaction proceeds in high yield in H ₂ 0. The same approach can be used for large scale synthesis. For smaller scale studies, such as in

vitro mapping, the phosphorothioate is added enzymatically to the 5' TFO terminus, employing polynucleotide kinase and gamma-thio-ATP.

Psoralen is also linked to cysteine of the modified proline m-gap linker, a reaction which proceeds spontaneously in H ₂ 0 to yield the thioether. 3. Azido rhodamine 110 (RZ110). Rhodamine 123 and its derivative R110 bind to DNA. R123 is a vital stain and is used clinically as a photosensitization agent. By analogy to the facile synthesis of ethidium bromide and acridine diazides as photochemical crosslinkers, R110 can be efficiently converted into the bis-azido derivative RZ110 with quantitative yield. The corresponding RZ123 derivative is an excellent photochemical

DNA crosslinker. This chemistry is used for the purpose of initiating TFO- DNA crosslinks.

In the present invention RZ110 is coupled to primary amines in a TFO by means of carbodiimide mediated amine bond formation. Subsequent to purification, photochemistry is initiated with the 514 plus 485nm output of a 5W AR ⁺ ion laser. This crosslinker is used for attachment to the lysine moiety of the m-gap linker or to amines at the TFO terminus. An example is shown in Figure 8B.

4. Acridine 9 carboxylic acid (A9A). The carboxylic acid of acridine is a commercially available DNA intercalator and has been linked to form a synthon which can be introduced into the 3' or 5' terminus of a TFO or as part of an m-gap linker. It is used to increase the binding affinity of m-gap

TFOs. An example is shown on Figure 8C.

Example 7

PHYSICAL ANALYSIS OF TFO BINDING IN VITRO. 1. Structure and thermodynamic analysis on small duplex fragments. A variety of methods are widely used to analyze triple helix formation. These methods are used as needed to define the structure and stability of the TFOs described in this invention. In general, analysis at this level is performed on synthetic DNA binding sites. The methods are well known in the art. They have been described at length in Durland, et al.,

Biochemistry 30:9246-9255 (1991). In all instances in vitro binding studies are performed in 20mM Tris/HCl, lOmM Mg ⁺² , pH7.8, 37C:

Low resolution TFO structure methods include: (1) DNasel footprinting; (2) Copper phenanthroline footprinting; (3) OH mediated DNA cleavage by a DTPA/Fe chelate at the 3' TFO terminus; (4) Singlet oxygen mediated DNA cleavage by eosin at the 3' TFO terminus; (5) Electrophoretic band shift analysis; and (6) Restriction endonuclease protection.

Crosslinking analysis is measured with purified DNA. The formation of TFO-duplex DNA crosslinks is first assessed with purified DNA duplexes in vitro. The duplex targets include: (1) Figures 2, 4 and 5 the HSV-1 gD promoter: -40 to -79 relative to CAP; (2) Figure 6 the APRT Exon 2: 602 to 645; and (3) Figure 7 the APRT Exon 5: 2246 to 2300.

After crosslinking, Mg ⁺² is chelated from the binding buffer and analysis is performed by native 5% acrylamide electrophoresis in TBE. Under these conditions, only covalent TFO complexes will remain bound to the duplex target as a triple helix. Resolution of the covalent TFO-duplex product from free duplex is based upon the substantial retardation of duplex mobility associated with addition of a long oligonucleotide tail to a duplex. To assess selectivity, crosslinking is also performed using unrelated duplex sites in this series as controls.

One skilled in the art recognizes that this simple assay is used to determine the chemical modification and conditions of reaction which produce the highest crosslinking efficiency in vitro. The variables to be modified for the wet chemistry are time of reaction (1-5 hours) and temperature (20° C and 37° C). For the bromoacetate and iodoacetate chemistries, the variables to be manipulated are temperature and time of incubation. For the psoralen photochemistry, time and light intensity are varied in the near UV, employing the 330 to 360nm lines of the doubled argon laser (5-100 mW for 1-5 minutes). For RZllO photochemistry, the principle output of a 5W argon laser (1-5 min., .1 to 1 W) is employed.

One skilled in the art recognizes that once conditions of crosslinking have been optimized, the site of crosslinking is confirmed on the synthetic duplexes at high resolution. For bromo and iodoacetate crosslinks, this is

achieved by inducing beta elimination and strand cleavage at modified purine. Strand cleavage will be assessed on sequencing gels vs a "G ladder" generated by simple DMS chemistry).

To map the position of psoralen or RZllO crosslinks advantage is taken of the fact that such crosslinks serve as a chain terminator in a DNA polymerase reaction. DNA crosslinks are assessed at one base resolution employing a modification of the Sanger sequencing method.

Example 8 DELIVERY OF TFOs INTO CHO CELLS

Cell Uptake and Stability can be measured by a variety of methods.

One method to measure uptake rate and intracellular distribution analysis in

CHO cells employs the S-labelling method. TFOs and TFO-alkylator conjugates are ^S-labelled at the 5' terminus and uptake is assessed by scintillation counting of whole cell or nuclei over a 12 hours time range. For alkylator conjugates or conjugates with photo crosslinkers, the extent of covalent crosslinking to DNA is also assessed on DNA extracted from nuclei.

To evaluate TFO stability and the kinetics of TFO degradation in CHO cells the post-labeling technique to monitor TFO stability is used.

Example 9 ANALYSIS OF GENE SPECIFIC DNA CROSSLINKING IN VIVO: RECOMBINATION AND MUTAGENESIS TFO-alkylator conjugates which show site selective crosslinking activity are analyzed for their capacity to induce gene-selective mutation or recombination. In all instances, the TFOs are "capped" at the 3' terminus to confer good stability in cells and in growth medium. When oligonucleotide modification provides increased binding affinity, the cellular response to both the modified and unmodified TFO is measured to establish the relation between binding affinity and mutational or recombinational outcome.

In the preferred embodiment oligonucleotides are isolated as the sodium salt (alkylamines tend to be cytotoxic) and purified by exclusion

chromatography in PBS immediately prior to cell treatment. Samples of luM are sufficient for repetitive cellular studies.

It is well known in the art that crosslinks and other kinds of alkylator damage are both recombinagenic and mutagenic. Very sensitive assay systems to monitor homologous recombination and mutation within the hamster APRT gene are known.

The selectivity of any observed mutational or recombinational effect is of the greatest importance in these experiments. Consequently, the following treatment protocols are straightforward but provide rigorous cellular controls: (1) TFOs in the absence of a reactive conjugated chemical crosslinker;

(2) conjugates with "scrambled" TFO isomers; (3) TFOs and scrambled isomers added with unlinked reactive chemical; and (4) for photochemicals, cellular treatment in the dark.

In Example 3, Figure 6 and Example 4, Figure 7 TFOs are targeted against sites in Exons 2 and 5 of the APRT gene (the APRT2 and APRT5

TFO class). Alkylator induced DNA damage at these sites gives rise to the APRT-phenotype, which can be quantified for forward selection in medium containing 8-Azaadenine. In these experiments, the "wild type" Chinese hamster ovary cell line CHO-AT3-2 is employed. Cells are cultured under ordinary conditions in 10% fetal calf serum. At time zero, TFO conjugate is added in the growth medium over a concentration range from lO^-lO^M.

For psoralen or RZ100 conjugates, cultures are irradiated with either the visible (RZllO) or doubled output (psoralen) of the argon laser. Intensity and time of irradiation is adjusted to produce no greater than 25% cell death. Subsequent to irradiation, the cell is transferred to AA medium, to select for mutation to the AA ^r (APRT-) phenotype. Mutation rates are calculated as a function of dose. For bromoacetate or iodoacetate TFO conjugates, cells are treated with TFO for 24 hours, then transferred to AA selection medium to score for mutants.

Example 10

PHYSICAL ANALYSIS OF TFO BINDING:

IN CELLS: A PCR METHOD

Once it has been determined that TFO-alkylator conjugates have produced gene-selective and TFO selective mutation and/or recombination in

CHO cells, the site selectivity of TFO crosslinking is studied in the cell by physical methods.

A variety of methods can be used to amplify DNA sequences, including the ligation mediated PCR method of Mueller, et al., Science 246:810-813 (1989). DNA sequences which have been cleaved by treatment of living cells with the simple alkylator DMS are amplified. This method is used to map the sites at which TFO conjugates have delivered alkylation by bromoacetate or iodoacetate or photocrosslinkingby psoralen or RZllO. The advantage of this method is that only about 10 ⁵ cells are needed for sample analysis as compared with 10 ⁸ for DNase-hypersensitivity analyses, and the fact that this technique provides binding analysis at one nucleotide resolution.

Briefly, this procedure involves treating CHO-ATS-49tg cells bearing a single copy of APRT with TFO-alkylator conjugate, as described for recombination assays. Any of the target sites are amenable to recombination and/or mutagenesis. Controls for this analysis are: (1) cells treated with

TFO alone; (2) cells treated with alkylator alone; and (3) cells treated with scrambled oligonucleotide conjugates.

After the cells are treated with an alkylator, the DNA is purified and heated to produce strand breaks at the site of alkylation. The DNA is then annealed to an APRT-specific primer which is extended by action of

Sequenase. The primer extends out to the site of alkylator damage or crosslinking. Because of the crosslinking or damage the polymerization extension ends. A linker is then added by ligation to the terminus of the newly synthesized fragment. Following denaturation, the fragments, which now possess unique primers at either end, are amplified by PCR.

Subsequent to these analytical reactions, crosslinking or alkylator damage at a single site gives rise to a discrete amplified fragment. One skilled in the art recognizes that this PCR method is used routinely to

generate "in vivo footprints" of protein binding sites by a subtractive method where the observable arises from local reduction of DMS cleavage rate. This is a reactive method for mapping a positive alkylator interaction.

Example 11

M-GAP TFO BINDING TO SITE 1 WITH THE HUMAN PROGESTERONE RECEPTOR

This second class of m-gap TFO (Figure 9) employs both limiting classes of linker technology. At spacer region A (Figure 9), TFO sites (in bold) are separated by roughly 1/2 of a helical turn. Therefore, a 25 rotatable bond polymer has been employed to link the leftmost pair of TFOs, with the limiting bound structure in which the linker occupies the major groove, much like the TFOs to either side. At spacer region B (Figure 9), TFO sites (in bold) are separated by roughly 9/10 of a helical turn. Therefore, a 30 rotatable bond polymer has been employed to link the rightmost pair of TFO's, with the alternative bound structure in which the linker polymer traverses the minor helix groove at the spacer region, much as described in Figure 3.

Example 12 BINDING STOICHIOMETRY In order to estimate the stoichiometry of hybrid TFO binding, increasing concentration of TFO was added to a synthetic 43bp duplex fragment, spanning promoter sites -38 to -82 relative to the CAP site. Duplex concentration has been held constant at 0.5uM. This concentration appears to be at least lOOx the apparent dissociation constant for the triplex under study.

Within the "L series" of hybrid TFOs, it is observed that linkers in the range of 22 to 66 added rotatable bonds (2AP-6AP) give rise to a unique bound species, as detected by electrophoresis (Figure 10). As expected from the model in Figure 2, this bound species appears to saturate at 0.5uM or one added hybrid TFO equivalent per duplex.

The ligand with only 11 rotatable bonds (1AP), appears to bind by a different mechanism, leading to multiple, slowly migrating electrophoretic species. Preliminary data suggests that this complicated equilibrium includes concentration dependent duplex-duplex interactions, mediated by binding of TFO half molecules.

Although they possess a zwiterionic linker, the binding of hydrid TFOs of the "K series" (multiples of 5 bonds) and "S series" (6 bond elements) appears to be very similar to that of the anionic "L series." Binding of K series TFOs with linkers greater than 25 bonds (1065, 1066) or the "S series" TFO with 24 bonds (9AP) appears to be characterized by a single bound species which saturates at one added TFO equivalent per duplex. As the linker becomes shorter than 25 bonds, additional slow mobility species appear, at the expense of the species which results from simple 1:1 stoichiometry (see, for example, binding of 1061-1064 and 8AP). This behavior of these three sets of TFO hybrids is consistent. Thus, linkers which are 25 bonds or longer give rise to a simple 1:1 bound TFO- DNA complex on the gD promoter target.

Example 13 Footprinting

DNasel footprinting upon a 43 bp fragment of the gD promoter, with its purine rich strand radiolabelled at the 3' terminus was used to delineate the structure of the complex formed between these hybrid TFOs and the duplex target. Cleavage was assessed at lxlO^M of added TFO, which is sufficient to drive the binding interaction to completion for all ligands. In

Figure 11, attention has been focused upon those members of the "L series" which display simple 1:1 binding stoichiometry at greater than 22 rotatable bonds.

The span of the 1:1 bound complex from Figure 2 has been superimposed to the right of DNasel cleavage profiles. DMS cleavage of this fragment (last lane) has been included to provide a series of size standards

As seen, a control oligonucleotide (EG36AP) had no effect on DNAsel cleavage, under conditions where all "L series" TFOs produce a pair of well

defined DNasel footprints at or near the expected site of triple helix formation. As predicted from the simple model in Figure 2, DNasel cleavage is detected within the spacer region between TFO binding sites, indicating that the spacer is not engaged in local triple helix formation. Footprinting analysis was repeated on a cloned 93bp fragment of the gD promoter to investigate the specificity of hybrid TFO binding. This fragment was labelled on the 3' end of its pyrimidine rich strand by the action of Klenow polymerase at a convenient Hinfl terminus. As seen in Figure 12 (right panel) the "L series TFOs display a well defined footprint at lO^M, over the domain predicted from Figure 2, with no evidence for binding elsewhere in the 93 bp region. Data for the correspondence "K series" complex are displayed in the left panel. Again, a well defined footprint is detected, with no evidence for non-specific binding.

Example 14

Binding Constants

Although the DNasel footprinting data suggested tight, relatively specific binding for these hybrid TFO molecules, their affinity and site selectively were monitored more qualitatively by means of band shift data which are displayed in Figure 13 for the "L series." These were monitored at

10 ^'7 M in duplex equivalents. As seen, under conditions where binding is complete by 10 ^'7 M of added hybrid TFO (top), binding of a control oligonucleotide cannot be detected (middle), nor can binding of the hybrid TFO be detected on an unrelated duplex target (bottom). In order to explore the binding affinity of these hybrid TFOs in more detail, band shift analysis was performed at a duplex concentration of 10 ^"10 M in the standard binding buffer. Band shift data of this kind were catalogued in Table 1, along with binding affinity for unlinked 12mer TFO half molecules, for the binding of the EG36 control TFO and also for hybrid TFO binding to an unrelated duplex.

TABLE I

HYBRID TFO APPARANT Kdiss APPARENT Kdiss gD TARGET CONTROL TARGET

(M/L) (M/L)

1AP UNDEFINED 2AP <10 ¹⁰ >10 ^'6 4AP < 10 ¹⁰ 5AP <10 ¹⁰ 6AP <10 ¹⁰ 7AP <10 ¹⁰

1065 UNDEFINED 1066 UNDEFINED 1067 UNDEFINED 1068 <10 ¹⁰ >10 ^"β M 1069 <10 ¹⁰

HSVA 10 ⁹ HSVB 10 ⁸

EG36 >10 ⁶

In such a titration experiment, titration midpoints approximate a dissociation constant in the limit that Kdiss is greater than the duplex concentration. As seen in Table 1, titration midpoints for hybrid TFOs with linkers greater than 22-25 bonds are very near to the concentration of duplex target. It is believed that these values provide an artificially high estimate of the true dissociation constant (true values are lower than 10 ^'10 M). Thus, the binding data shows the upper limit.

Further, it can be seen in Table 1 that, in the absence of a linker, the individual 12mer TFO half molecules display apparent dissociation constants in the 10 ^'9 M to 10 ^"8 M range. Based upon that value, an appropriately chosen linker at the gD binding site can affect a TFO binding constant enhancement of at least 10 to 100 in this assay system.

All patents and publications mentioned in this specification are indicative of the levels of those skilled in the art to which the invention pertains. All patents and publications are herein incorporated by reference to the same extent as if each individual publication is specifically and individually indicated to be incorporated by reference.

One skilled in the art will readily appreciate that the present invention is well adapted to carry out the objects and obtain the ends and advantages mentioned, as well as those inherent therein. The triplex forming oligonucleotides, ligands along with the methods, procedures, treatments, molecules, and specific compounds described herein are presently representative of preferred embodiments, are exemplary and are not intended as limitations on the scope of the invention. Changes therein and other uses will occur to those skilled in the art which are encompassed within the spirit of the invention and are defined by the scope within the claims.

What we claim is: