Uses for DNA structure-specific recognition protein

This is a divisional of application Ser. No. 08/328,809 filed on Oct. 25, 1994, now U.S. Pat. No. 5,705,334, which is a continuation-in-part of U.S. Ser. No. 08/258,442, filed Jun. 9, 1994, now U.S. Pat. No. 5,670,621, which is a divisional of U.S. Ser. No. 07/814,964 filed Dec. 26, 1991 now U.S. Pat. No. 5,359,047. U.S. Ser. No. 07/814,964 is a continuation-in-part of U.S. Ser. No. 07/539,906 filed Jun. 18, 1990, now abandoned, which was a continuation-in-part of U.S. Ser. No. 07/410,981 filed Sep. 22, 1989, now abandoned which was a continuation-in-part of U.S. Ser. No. 07/247,774 filed Sep. 22, 1988, now abandoned. The teachings of each of the foregoing documents are incorporated herein by reference.

Work described herein was supported by grants from the National Cancer Institute, the National Science Foundation, the National Institutes of Health, the Human Frontier Science Program Organization and a National Research Service Award. The United States Government has certain rights in the invention.

The present invention relates generally to risk-assessment of suspected genotoxins, evaluation of novel chemotherapeutic agents, and novel chemotherapeutic methods for cancer management.

Cancer arises when a normal cell undergoes neoplastic transformation and becomes a malignant cell. Transformed (malignant) cells escape normal physiologic controls specifying cell phenotype and restraining cell proliferation. Transformed cells in an individual's body thus proliferate, forming a tumor (also referred to as a neoplasm). When a neoplasm is found, the clinical objective is to destroy malignant cells selectively while mitigating any harm caused to normal cells in the individual undergoing treatment. Currently, three major approaches are followed for the clinical management of cancer in humans and other animals. Surgical resection of solid tumors, malignant nodules and or entire organs may be appropriate for certain types of neoplasia. For other types, e.g., those manifested as soluble (ascites) tumors, hematopoeitic malignancies such as leukemia, or where metastasis of a primary tumor to another site in the body is suspected, radiation or chemotherapy may be appropriate. Either of these techniques also is commonly used as an adjunct to surgery. Harrison's Principles of Internal Medicine, Part 11 Hematology and Oncology, Ch. 296, 297 and 300-308 (12th ed. 1991).

Chemotherapy is based on the use of drugs that are selectively toxic (cytotoxic) to cancer cells. Id. at Ch. 301. Several general classes of chemotherapeutic drugs have been developed, including drugs that interfere with nucleic acid synthesis, protein synthesis, and other vital metabolic processes. These generally are referred to as antimetabolite drugs. Other classes of chemotherapeutic drugs inflict damage on cellular DNA. Drugs of these classes generally are referred to as genotoxic. Two widely used genotoxic anticancer drugs that have been shown to damage cellular DNA by producing crosslinks therein are cisplatin [cis-diamminedichloroplatinum(II)] and carboplatin [diammine(1,1-cyclobutanedicarboxylato)platinum(II)]. Bruhn et al. (1990), 38 Prog. Inorg. Chem. 477, Burnouf et al. (1987), 84 Proc. Natl. Acad. Sci. USA 3758, Sorenson and Eastman (1987), 48 Cancer Res. 4484 and 6703, Pinto and Lippard (1985), 82 Proc. Natl. Acad. Sci., USA 4616, Lim and Martini (1984), 38 J. Inorg. Nucl. Chem. 119, Lee and Martin (1976), 17 Inorg. Chim. Acta 105, Harder and Rosenberg (1970), 6 Int. J. Cancer 207, Howle and Gale (1970), 19 Biochem. Pharmacol 2757. Cisplatin and/or carboplatin currently are used in the treatment of selected, diverse neoplasms of epithelial and mesenchymal origin, including carcinomas and sarcomas of the respiratory, gastrointestinal and reproductive tracts, of the central nervous system, and of squamous origin in the head and neck. Harrison's Principles of Internal Medicine ( 12th ed. 1991) at Ch. 301. Cisplatin currently is preferred for the management of testicular carcinoma, and in many instances produces a lasting remission. Loehrer and Einhorn (1984), 100 Ann. Int. Med. 704. Susceptibility of an individual neoplasm to a desired chemotherapeutic drug or combination thereof often, however, can be accurately assessed only after a trial period of treatment. The time invested in an unsuccessful trial period poses a significant risk in the clinical management of aggressive malignancies.

The repair of damage to cellular DNA is an important biological process carried out by a cell's enzymatic DNA repair machinery. Unrepaired lesions in a cell's genome can impede DNA replication, impair the replication fidelity of newly synthesized DNA or hinder the expression of genes needed for cell survival. Thus, genotoxic drugs generally are considered more toxic to actively dividing cells that engage in DNA synthesis than to quiescent, nondividing cells. Indeed, cells carrying a genetic defect in one or more elements of the enzymatic DNA repair machinery are extremely sensitive to cisplatin. Fraval et al. (1978), 51 Mutat. Res. 121, Beck and Brubaker (1973), 116 J. Bacteriol 1247. Normal cells of many body tissues, however, are quiescent and commit infrequently to re-enter the cell cycle and divide. Greater time between rounds of cell division generally is afforded for the repair of DNA damage in normal cells inflected by chemotherapeutic genotoxins. As a result, some selectivity is achieved for the killing of cancer cells. Many treatment regimes reflect attempts to improve selectivity for cancer cells by coadministering chemotherapeutic drugs belonging to two or more of these general classes.

In some tissues, however, normal cells divide continuously. Thus, skin, hair follicles, buccal mucosa and other tissues of the gut lining, sperm and blood-forming tissues of the bone marrow remain vulnerable to the action of genotoxic drugs, including cisplatin. These and other classes of chemotherapeutic drugs can also cause severe adverse side effects in drug-sensitive organs, such as the liver and kidneys. These and other adverse side effects seriously constrain the dosage levels and lengths of treatment regimens that can be prescribed for individuals in need of cancer chemotherapy. Harrison's Principles of Internal Medicine (12th ed. 1991) at Ch. 301. See also Jones et al. (1985), 52 Lab. Invest. 363-374 and Loehrer and Einhorn (1984), 100 Ann. Int. Med. 704-714. Such constraints can prejudice the effectiveness of clinical treatment. For example, the drug or drug combination administered must contact and affect cancer cells at times appropriate to impair cell survival. Genotoxic drugs are most effective for killing cancer cells that are actively dividing when chemotherapeutic treatment is applied. Conversely, such drugs are relatively ineffective for the treatment of slow growing neoplasms. Carcinoma cells of the breast, lung and colorectal tissues, for example, typically double as slowly as once every 100 days. Harrison's Principles of Internal Medicine (12th ed. 1991) at Table 301-1. Such slowly growing neoplasms present difficult chemotherapeutic targets.

Moreover, cancer cells can acquire resistance to genotoxic drugs through diminished uptake or other changes in drug metabolism, such as those that occur upon drug-induced gene amplification or expression of a cellular gene for multiple drug resistance (MDR). Harrison's Principles of Internal Medicine (12th ed. 1991) at Ch. 301. Resistance to genotoxic drugs also can be acquired by activation or enhanced expression of enzymes in the cancer cell's enzymatic DNA repair machinery. Therapies that employ combinations of drugs, or drugs and radiation, attempt to overcome these limitations. The pharmacokinetic profile of each chemotherapeutic drug in such a combinatorial regime, however, will differ. In particular, permeability of neoplastic tissue for each drug will be different. Thus, it can be difficult to achieve genotoxically effective concentrations of multiple chemotherapeutic drugs in target tissues.

Needs remain for additional chemotherapeutic drugs with improved selectivity for destroying transformed cells in situ, without significantly impairing viability of untransformed cells. Needs remain also for enhancing effectiveness of chemotherapeutic drugs, such that satisfactory cell killing can be achieved with lower doses thereof than are currently needed. Thus, needs remain for improved, more accurate methods of testing whether a given chemotherapeutic drug will be effective for killing a particular colony of transformed cells in situ. Poignant needs remain for chemotherapeutic drugs with improved selectivity for destroying transformed cells. Particularly poignant needs remain for ways to render transformed cells selectively more vulnerable to killing through chemotherapy.

It is an object of this invention to provide a method for assessing whether a suspected genotoxic agent forms lesions in DNA that are bound (recognized) by a DNA structure specific recognition protein (SSRP). Thus, it is an object of this invention to provide an in vitro assay for predicting whether a suspected genotoxic agent forms persistent genomic lesions in eukaryotic cellular DNA.

Another object of this invention is to provide a method for assessing whether a eukaryotic cell contains a DNA structure specific recognition protein that binds to DNA lesions formed by a genotoxic agent. Thus, it is an object of this invention to provide a method for predicting susceptibility of a eukaryotic cell to killing by a genotoxic agent.

Yet another object of this invention is to provide a method of screening new genotoxic drug candidates for the ability to form DNA lesions that are bound by a DNA structure specific recognition protein. Thus, it is an object of this invention to provide a screening method for the rational design of new genotoxic drugs that form persistent genomic lesions in eukaryotic cells. Accordingly, it is an object of this invention to provide new genotoxic drugs identified from the screening method described herein.

Still another object of this invention to provide a method of causing a eukaryotic cell to express a DNA structure specific recognition protein encoded by a heterologous nucleic acid. Thus, it is an object of this invention to provide a method for enhancing persistence of DNA lesions in the genome of eukaryotic cells. The objects of this invention accordingly include providing a method for sensitizing eukaryotic cells to killing by a genotoxic agent. A further object of this invention therefore includes providing an improved method for killing eukaryotic cells, based on rendering the cells sensitive to a genotoxic agent by causing said cells to express a DNA structure specific recognition protein, and then exposing the cells to the genotoxic agent.

These and other objects, along with advantages and features of the invention disclosed herein, will be apparent from the description, drawings and claims that follow.

The invention described herein rests on the discovery that eukaryotic cells contain one or more intracellular structure specific recognition proteins (SSRPs) that bind to sequence-independent structural motifs in cellular DNA produced by the binding thereto of genotoxic agents. Genotoxic agents or genotoxins bind to or otherwise physically or chemically interact with cellular DNA, causing injury thereto. A site of injury (a lesion) in cellular DNA is referred to herein as a genomic lesion. DNA lesions can include disruptions of the nucleotide sequence, nucleotide basepairing, or distortions of the structure of the DNA double helix. Structural distortion lesions produce three-dimensional DNA structural motifs (e.g., bends, kinks, unwinding, overwinding, non-B helical forms such as A- or Z-DNA, junctions between different helical forms, stem-loop structures, cruciforms, local melting, crossover junctions and the like). Genomic lesions in cellular DNA that are not repaired before the cell commits itself to the cycle of cell division contribute to cell death. Thus, one determinant of a genotoxic agent's cytotoxicity (propensity for contributing to cell death) is the resistance of genomic lesions formed therefrom to cellular repair. Genotoxic agents that form persistent genomic lesions, e.g., lesions that remain in the genome at least until the cell commits to the cell cycle, generally are more effective cytotoxins than agents that form transient, easily repaired genomic lesions. Hence, genotoxic agents that form persistent genomic lesions are preferred for use as chemotherapeutic agents in the clinical management of cancer.

The invention rests more precisely on the discovery, recounted in U.S. Pat. No. 5,359,047 (incorporated herein by reference), that eukaryotic cells contain one or more SSRPs that bind to 1,2-dinucleotide intrastrand adducts of genotoxic metal coordination compounds currently used as chemotherapeutic agents in the clinical management of cancer. Such genotoxic metal coordination compounds include noble metal compounds, such as platinum(II) and platinum(IV) compounds. Typically, the compounds comprise a platinum atom linked to a pair of cis-configured substitutionally labile moieties and a pair of cis-configured electron donor moieties. Binding of the noble metal coordination compounds to nucleic acids occurs upon substitution of the cis-configured labile moieties with atoms of the nucleotide bases, usually adenosine (A) or guanine (G) residues. This produces a crosslink, bridged by the noble metal atom (e.g., platinum) between two vicinal, adjacent or paired nucleotide bases. Platinum-bridged crosslinks between adjacent adenosine and/or guanine residues within a single nucleotide strand (1,2-intrastrand dinucleotide adducts or lesions) of double stranded DNA are abbreviated herein as 1,2-d(A{circumflex over ( )}G) and 1,2-d(G{circumflex over ( )}G) lesions. The class of genotoxic noble metal coordination compounds that form SSRP-recognized genomic lesions includes cisplatin (cis-diamminedichloroplatinum(II) or cis-DDP), carboplatin (diammine(1,1-cyclobutane-dicarboxylato)platinum(II), cis-diamminetetrachloroplatinum(IV), iproplatin (CHIP), DACCP, malonatoplatin, cis-dichloro(ethylenediamine)platinum(II), cis-dichloro(1,2-diaminocyclohexyl)platinum(II), and the like. For convenience, SSRP recognized 1,2-intrastrand dinucleotide adducts formed by any member of this class are referred to herein as cisplatin-type lesions (or adducts).

SSRPs have been shown to bind to the 1,2-d(A{circumflex over ( )}G) or 1,2-d(G{circumflex over ( )}G) intrastrand DNA adducts of cisplatin irrespective of the 5′ or 3′ orientation of the lesion site and irrespective of the nucleotide sequence adjacent to or comprising the lesion site. Hence, SSRP binding is understood to be sequence-independent, in contrast to the binding properties of other, known nucleic acid binding proteins. SSRP binding to the 1,2-intrastrand dinucleotide adduct (lesion) of a cisplatin-type genotoxic agent results in the formation of a lesioned DNA/SSRP complex. This complex can be detected visually using techniques described in U.S. Pat. No. 5,359,047, including modified Western (Southwestern) blotting and electrophoretic mobility shift analysis (EMSA, also known as bandshift analysis).

SSRPs thus far reported to bind to 1,2-intrastrand cisplatin-type lesions in DNA comprise at least one structural domain generally referred to as an HMG domain. Exemplary, preferred SSRP HMG domains include the HMG domains of human and Drosophila SSRP1, having the sequences set forth, respectively, in amino acid residues 539-614 of Seq. ID No. 2 and residues 547-620 of Seq. ID No. 6. Other useful SSRP HMG domains are encoded by nucleic acids that hybridize specifically, at least under low stringency hybridization conditions such as described in U.S. Pat. No. 5,359,047, to nucleic acid encoding the HMG domain of human or Drosophila SSRP1. SSRPs comprising such HMG domains and occurring in non-human or non-Drosophila eukaryotes are considered homologs of human or Drosophila SSRP1. SSRP-encoding homologous nucleic acids have been detected in diverse eukaryotes, including arthropods (represented by the fruitfly Drosophila melanogaster ) and vertebrates including mammals (e.g., human, chimpanzee, monkey, elephant, pig, dog, rabbit, mouse and opossum), aves (e.g., chicken) and fish. It is deduced that homologs of the human and/or Drosophila SSRP occur in numerous eukaryotes, including at least arthropods and vertebrates. A mouse protein comprising an SSRP HMG domain and considered to be a homolog of human SSRP1 has been referred in the literature as T160. SSRP variants occurring within a given eukaryotic species (e.g., humans) that are encoded by nucleic acids comprising sequences similar but not identical to, e.g., residues 539-614 of Seq ID No. 2 (human SSRP1), are understood to be polymorphic or allelic SSRP1 variants. Homologous and polymorphic SSRP1 variants also are useful in the invention described herein.

Proteins comprising still other useful SSRP HMG domains can be identified empirically, based upon their ability to form detectable cisplatin-lesioned DNA/protein complexes. Such other useful SSRP HMG domains need not be encoded by nucleic acid that hybridizes specifically to nucleic acid encoding the HMG domain of human or Drosophila SSRP1. At least one such empirically identified, useful SSRP is fractional yeast SSRP (fySSRP), Seq. ID No. 8. This SSRP has been referred to in publications as IXR-1 (intrastrand crosslink recognition protein 1). Additional useful SSRP HMG domains occur in such known HMG proteins as HMG-1, HMG-2, UBF, LEF-1, SRY, mtTFA, ABF2 and the like. These and other known HMG domain SSRPs have been isolated, variously, from diverse eukaryotes, including human, rodent, Xenopus, Drosophila and yeast.

The consequence of SSRP binding to a genomic lesion is that the sterically large SSRP (or a fragment thereof comprising an HMG domain) becomes localized in the immediate vicinity of the genomic lesion. The SSRP is large enough to sterically obscure (cover) a region of cellular DNA extending from the lesion site in either the 5′ and 3′ direction for at least about five base pairs, preferably at least about eight base pairs, more preferably at least about twelve base pairs. As a result, lesion-bound SSRP shields the genomic lesion from repair by the cell's enzymatic DNA repair machinery. SSRP-shielded lesions persist in the genome longer than unshielded lesions. SSRP-shielded lesions accordingly are more effective for prejudicing the fidelity of DNA replication, hindering the expression of genes relevant to cell survival, and otherwise contributing to disarray of the cell's nuclear architecture. One or more of the foregoing can contribute to cell death, e.g., by triggering apoptosis.

Certain HMG domain proteins useful herein as SSRPs have been characterized in the literature as transcription factors that control or modulate the expression of one or more cellular genes, including genes that are relevant to cell metabolism or cell secretory function. One such transcription factor is upstream binding factor (UBF), which controls the expression of ribosomal RNA genes and thus is pivotal to the function of the cell's protein synthesis machinery. It is thought that cisplatin-type lesions to which such transcription factors bind as SSRPs mimic or resemble the factor's natural genomic binding site. Binding of such transcription factors to cisplatin-type genomic lesions in effect sequesters the transcription factors at sites other than the natural genomic binding site. Titration of the transcription factors away from their natural genomic binding sites contributes to dysregulation of the controlled genes and therefore contributes to disarray of cellular processes and functions directed by the products (generally proteins, e.g., enzymes) of the controlled genes. For example, sequestration or “hijacking” of the HMG domain transcription factor UBF by cisplatin-type lesions contributes to disarray of cellular protein synthesis, a process needed for cell survival.

The invention described herein accordingly features, in one aspect, a method for predicting cytotoxicity of an agent that binds to DNA (a genotoxic agent or genotoxin). In this method, a sample of double-stranded DNA bearing a lesion formed by the genotoxic agent is contacted with a DNA structure-specific recognition protein, such that a lesioned DNA/SSRP complex forms. This complex is detected or visualized, and optionally quantitated e.g., relative to a standard genotoxic agent known to form a DNA lesion bound by the SSRP. Capacity of the genotoxic agent to form SSRP-shielded DNA lesions in vitro is considered reasonably predictive of competence of the agent to form persistent genomic lesions in cellular DNA, rather than transient, easily repaired lesions.

In another aspect, the invention features a method for assessing cytotoxicity of an agent that inflicts genomic lesions on cellular DNA. That is, the invention features a method for predicting susceptibility of eukaryotic cells to the cytotoxic effects of a genotoxin. In this method, a sample comprising eukaryotic cells is treated so as to release intracellular proteins. The released intracellular proteins are assessed for the presence of one or more DNA structure-specific recognition proteins that bind to DNA lesioned by the genotoxin. Thus, released intracellular proteins are contacted with probe DNA comprising at least one lesion formed by the genotoxin, such that a lesioned probe DNA/cellular SSRP complex forms. This complex is detected or visualized, and optionally quantitated e.g., relative to a standard SSRP known to bind DNA lesions formed by the genotoxic agent. Presence within the eukaryotic cells of one or more SSRPs that bind to the lesioned probe DNA is considered reasonably predictive of formation of persistent genomic lesions in cellular DNA. Accordingly, the presence and amount of SSRPs within the eukaryotic cells can be used to confirm whether a desired genotoxic agent will be cytotoxic to the cells, as well as to assist in the calculation of the dose of genotoxic agent needed to produce the desired degree or rapidity of cell killing.

In yet another aspect, the invention features a method for identifying novel cytotoxic agents that bind to DNA to form genomic lesions. That is, the invention features a screening method for assessing new, genotoxic drug candidates for the ability to form SSRP-recognizable and thus persistent genomic lesions. This method involves contacting a sample of DNA, optionally comprising a detectable moiety, with one or more candidate cytotoxic agents, then incubating the DNA with the candidate under conditions sufficient for DNA binding of genotoxic agents. The DNA bearing a genomic lesion formed by a candidate genotoxin is separated from the incubation mixture comprising unlesioned DNA and unbound candidate. Successfully lesioned DNA is contacted an SSRP under conditions sufficient for the formation of a lesioned DNA/SSRP complex, which is thereupon detected. Optionally, SSRP can be used as an affinity separation agent to isolate successfully lesioned DNA from the incubation mixture. This rational drug screening method can be automated for high-throughput screening of numerous candidate compounds. It is suitable for screening random libraries of compounds, e.g., libraries produced by random or directed combinatorial synthesis of inorganic, organic or biological compounds. The invention accordingly encompasses new cytotoxic agents identified according to the present screening method.

Suitable methods for detecting lesioned DNA/SSRP complexes formed in the above aspects of the present invention include EHSA and Southwestern blotting, both generally according to U.S. Pat. No. 5,359,047. In these and other methods described herein, detection can optionally be facilitated through the use of lesioned probe DNA. Probe DNA is a fragment (e.g., a restriction fragment) of naturally occurring or recombinant DNA, or is a synthetically constructed DNA, of a size suitable for use in standard analytical procedures. For example, the probe DNA is at least about 60 basepairs (bp), preferably at least about 80 bp, more preferably at least about 100 bp in length. Lesioned probe DNA contains at least one structural motif (lesion) produced by the binding thereto of a genotoxic agent. Optionally, the probe DNA also comprises a detectable moiety, such as a radioisotope, chromophore, fluorophore, hapten or other high affinity ligand (e.g., biotin). Other methods for detecting lesioned DNA/SSRP complexes, optionally involving the use of a suitable probe DNA, include nitrocellulose filter retention assay and excinuclease protection assay, both described herein. The nitrocellulose filter retention assay is based upon the selective retention or filter-binding of proteins such as SSRPs. Lesioned probe DNA binds to the SSRP and thus is retained by the filter, whereas unlesioned probe DNA (or probe DNA bearing an unrecognized lesion) flows through or is not retained by the filter. If desired, the filter can be blocked or treated to reduce nonspecific retention. Nitrocellulose filter retention assays can be carried out, e.g., using a standard dot blotting apparatus. The selective retention principle of the nitrocellulose filter retention assay can be enlarged to other affinity based separation or analytical systems, including affinity chromatography systems and the like, through no more than routine experimentation. The excinuclease protection assay is based directly on the steric hindrance, by bound SSRP, of DNA lesion repair by a eukaryotic DNA repair enzyme. In this assay, the lesioned DNA/SSRP complex is contacted with excinuclease and incubated therewith under conditions sufficient for the excinuclease-catalyzed removal of lesions from DNA. If a DNA lesion is accessible to the excinuclease, a single-stranded nucleic acid fragment comprising the lesion is removed from the double-stranded DNA. Typically, the fragment is less than 30 bp long. The resulting gap is filled with a patch of newly synthesized DNA complementary to the sequence of the unlesioned strand. Using appropriate nucleic acid labeling techniques, described herein, one or more of the nucleic acid products of successful excinuclease repair can be detected. Failure to excise a lesion from DNA, or the degree (e.g., percent) of inhibition thereof indicates SSRP shielding and thus is reasonably correlated with persistence of lesions in the genome.

To facilitate detection of lesioned DNA/SSRP lesions according to the foregoing methods, the invention also provides kits comprising, as applicable, one or more SSRPs, optionally formulated as a cocktail, probe DNA bearing a defined cisplatin-type lesion or in which such a lesion can be produced, a DNA labeling reagent, and optionally a detection or separation reagent selected from an excinuclease preparation and a nitrocellulose filter. Kit components are conveniently packaged for either manual or automated practice of the foregoing methods.

In still another aspect, the invention features a method of sensitizing eukaryotic cells to a genotoxic agent, including a method of rendering eukaryotic cells naturally resistant to cell killing by the genotoxic agent vulnerable thereto. Thus, this aspect of the invention features a method of enhancing cytotoxic effectiveness of a genotoxic agent that normally inflicts only transient lesions on cellular DNA. In this method, eukaryotic cells are contacted with nucleic acid encoding an SSRP that binds to genomic lesions produced by the genotoxic agent, under conditions sufficient for the nucleic acid to be internalized and expressed within said cells. The SSRP-encoding nucleic acid is a foreign (heterologous) nucleic acid, optionally a plasmid, cosmid, expression vector, or virus, e.g., a retrovirus. Intracellular expression of the encoded SSRP enhances persistence of genomic lesions, as the expressed SSRP shields lesions produced by the genotoxic agent from repair by cellular excinuclease. Nucleic acid encoding the SSRP can be caused to internalize within the cells by electroporation or microinjection. Alternatively, where the nucleic acid is present in an expression vector, it can be caused to internalize by transfection according to standard techniques or routine modifications thereof. Optionally, the internalized nucleic acid becomes integrated into the cellular genome. Preferably, the encoded SSRP is overexpressed within the cell, such that an excess of SSRP accumulates, thermodynamically favoring the formation of lesioned DNA/SSRP complexes at the sites of genomic lesions.

Accordingly, yet a further aspect of the invention features an improved method for killing eukaryotic cells. This improved method involves contacting the cells to be killed with nucleic acid encoding an SSRP that binds to lesions in DNA produced by a selected genotoxic agent, under conditions sufficient for the internalization and expression (preferably, overexpression) of the SSRP-encoding nucleic acid within the cells. The method further involves contacting the cells expressing the encoded SSRP with the selected genotoxic agent, under conditions sufficient for the formation of persistent and therefore cytotoxic lesions in the cell genome. Advantageously, then, the invention may allow the use of low doses of the genotoxic agent, formerly considered poorly effective or ineffective for cell killing. The invention also may enhance the effectiveness of additional genotoxins, including genotoxins formerly considered poorly effective or ineffective as cytotoxins. Further, the invention may reconstitute the cytotoxic susceptibility of cells that are refractory to killing by genotoxins, including cells that express a gene for multiple drug resistance.

Eukaryotic cells with which the foregoing methods can be practiced can be cells of a unicellular or multicellular organism. The cells can be maintained in or adapted to culture ex vivo, or can be cells withdrawn from a multicellular organism (e.g., a body fluid sample or tissue biopsy). Alternatively, the cells can be present in vivo in tissue or organs of a multicellular eukaryotic organism. The term, multicellular eukaryotic organism, embraces at least arthropods and vertebrates, including fish, amphibians, birds and mammals, particularly humans. The eukaryotic cells can exhibit either normal or transformed phenotypes. Thus, the eukaryotic cells can be transformed (neoplastic or malignant) cells, including carcinoma cells and sarcoma cells. Transformed mammalian cells with which the present invention can be practiced include transformed cells arising within any body tissue or body compartment, including transformed cells of central or peripheral nervous system, mammary, lymphoid, myeloid, cutaneous, respiratory tract, gastrointestinal tract, and urogenital tract origin. To assess susceptibility of transformed cells to killing by a desired chemotherapeutic genotoxin, a sample comprising the transformed cells can be withdrawn from an individual to be treated with the chemotherapeutic agent by standard biopsy techniques and processed for the release of intracellular proteins comprising endogenous SSRPs as described above. If desired, transformed cells can be sensitized to cell killing in situ by the genotoxic agent by causing them to internalize foreign nucleic acid encoding SSRP. Nucleic acid encoding SSRP can be administered to the individual using standard techniques or modifications thereof, appropriate to deliver the nucleic acid to the body compartment, organ or tissue harboring transformed cells. Preferably, the SSRP encoding nucleic acid is internalized by dividing cells, including transformed cells that have escaped normal physiologic and molecular restraints on cell proliferation and cell differentiation. Subsequent exposure of the SSRP-expressing transformed cells to a genotoxic agent according to accepted chemotherapeutic protocols or routine modifications thereof results in preferential killing in situ of the transformed cells.

The foregoing and other objects, features and advantages of the present invention, as well as the invention itself, will be more fully understood from the following description of preferred embodiments, when read together with the accompanying drawings, in which:

FIG. 1A is a schematic illustration of the steric shielding, by SSRP of a cisplatin-type genomic lesion from repair by the cellular enzymatic DNA repair machinery.

FIG. 1B is a schematic illustration of the titration of SSRP away from its endogenous genomic binding site by cisplatin-type genomic lesion.

FIG. 2 is a schematic illustration, prepared from the predicted amino acid sequence of the structure specific recognition protein (Seq. ID No. 2) encoded by human SSRP1 gene, showing various domains thereof.

FIG. 3 is a schematic illustration, prepared from the predicted amino acid sequences of the structure specific recognition protein products of the homologous human (Seq. ID No. 2) and D. melanogaster (Dmel, Seq. ID No. 6) SSRP1 genes, showing various domains thereof.

FIG. 4 is a kinetic plot showing the differential rates of excision repair of the 1,3-d(GTG) and 1,2-d(G{circumflex over ( )}G) dinucleotide intrastrand adducts of cisplatin by human cell extracts with excinuclease activity.

FIG. 5 is a quantitative plot showing that an HMG domain protein (HMG1) differentially shields the cisplatin 1,2-d(G{circumflex over ( )}G) and 1,3-d(GTG) dinucleotide intrastrand adducts from repair by human cell extracts with excinuclease activity.

Broadly, the invention capitalizes on the principle that DNA structure specific recognition proteins (SSRPS) contribute to the cytotoxic efficacy of chemotherapeutic genotoxins by binding to toxin-associated genomic lesions and sterically shielding the lesions from repair. That is, lesion-bound SSRP hinders access to the lesion site by elements of the cell's enzymatic DNA repair machinery, including the multisubunit enzyme, excinuclease. This principle is illustrated schematically in FIG. 1 A. SSRP-shielded lesions persist in the genome and are more likely than unshielded lesions to contribute to the disarray of cellular metabolism and thus cell death. It is thought that SSRP recognized genomic lesions, although produced by the binding of genotoxic agents to cellular DNA, resemble naturally occurring structural motifs in the genome. Such naturally occurring motifs may be associated with the packaging of cellular DNA in chromatin, or the participation of chromatin in higher ordered aspects of nuclear architecture. Alternatively, such naturally occurring motifs may be associated with DNA replication, gene transcription, transcriptional repression, and like processes involving gene expression.

It has been observed that cisplatin and several of the clinically effective platinum coordination drugs developed subsequent to the discovery of cisplatin comprise a pair of cis-configured, substitutionally labile chloride moieties. Hence, cisplatin-like drugs most likely form DNA adducts that are similar to the well-characterized adducts of cisplatin itself. Lippard et al. (1983), 22 Biochemistry 5165. Other platinum coordination compounds, including the trans stereoisomer of cisplatin, lack this pair of cis-configured labile moieties and are largely clinically ineffective. The adduct or lesion formed most frequently by the binding of cisplatin to cellular DNA is the 1,2-intrastrand dinucleotide adduct, in which adjacent nucleotide bases become crosslinked directly through a platinum bridge. 1,2-d(A{circumflex over ( )}G) and 1,2-d(G{circumflex over ( )}G) adducts account together for approximately 90% of the DNA lesions produced in vivo by cisplatin and cisplatin-type drugs. The 1,2-intrastrand cisplatin-type adduct structurally comprises an unwinding element of about 13° at the site of a fairly inflexible bend in the double helix of 32-34° toward the major groove. Bellon and Lippard (1990), 35 Biophys. Chem. 179, Rice et al. (1988), 85 Proc. Natl. Acad. Sci. U.S.A. 4158. The platinum bridge itself, together with substituents of the platinum atom located trans to the substitutionally labile moieties, projects into the major groove. Efforts have been made to characterize the local unwinding element of cisplatin-type lesions using antinucleoside antibodies. Rice et al. (1988), 85 Proc. Natl. Acad. Sci. USA 4158, Sherman and Lippard (1987), 87 Chem. Rev. 1153, Sundquist et al. (1986), 25 Biochemistry 1520.

Methods described herein arose from the appreciation that eukaryotic proteins comprising one or more HMG domains (Grosschedl et al. (1994), 10 Trends Genet. 94, Jantzen et al. (1990), 344 Nature 830) bind specifically to 1,2-intrastrand d(G{circumflex over ( )}G) and d(A{circumflex over ( )}G) cisplatin-type DNA adducts, but not to other types of lesions in DNA, even when produced by cisplatin. Bruhn et al. (1992), 89 Proc. Natl. Acad. Sci USA 2307; Pil and Lippard (1992), 256 Science 234. Copending U.S. Ser. No. 08/258,442 (now U.S. Pat. No. 5,670,621) and Ser. No. 07/814,964 (now accorded U.S. Pat. No. 5,359,047), the teachings of each of which have been incorporated by reference herein, describe the use of probe DNA bearing cisplatin-type lesions to identify structure specific recognition proteins in eukaryotic cells. A cellular SSRP present in mammalian (human (HeLa) and hamster (V79)) cell extracts bound to double stranded probe DNA bearing lesions produced by cisplatin, cis-dichloro(ethylenediamine)platinum(II) and cis-dichloro(1,2-diaminocyclohexane)platinum(II). The cellular SSRP did not bind unlesioned double stranded or single stranded DNA having the same sequence as the lesioned probe, nor to probe DNA bearing lesions produced by transplatin (the trans isomer of cisplatin) or chloro(diethylenetriamine)platinum(II), neither of which is capable of producing 1,2-intrastrand dinucleotide adducts.

Salient features of the eukaryotic SSRP present in human and hamster cell extracts were characterized by EMSA and Southwestern blotting, both of which allow visualization of lesioned DNA/SSRP complexes in vitro. As described in U.S. Pat. No. 5,359,047, the relative electrophoretic mobility of cellular SSRP, when resolved under standard conditions, was consistent with a molecular mass (Mr) of about 100,000 daltons (d). Further physicochemical characterization confirmed that the cellular SSRP has a sedimentation coefficient, in a standard sucrose gradient, of 5.6S, corresponding to a molecular mass of 91,000 d for an ideal globular protein. Agreement of these two determinations is within acceptable technical limits. The binding constant for formation of the cellular SSRP/lesioned DNA complex was estimated, using competitive EMSA studies, to lie within the range of (1-20)×10 −10 M, indicating a physiologically relevant degree of specificity for cisplatin-type lesions. Thus, functional SSRP occurs and can be detected in cell extracts of eukaryotic cells, including human cells. Standard techniques, or routine modifications thereof, can be applied for releasing intracellular proteins, including functional SSRP, from eukaryotic cells for diagnostic and other purposes within the ambit of this invention.

U.S. Pat. No. 5,359,047 describes the identification, using a Southwestern blot screening technique, of nucleic acid fragments encoding functional regions of the human SSRP. Isolation of two cloned nucleic acid fragments, designated λPt1 and λPt2 (Seq. ID Nos. 3 and 4, respectively), also is reported in Toney et al. (1989), 86 Proc. Natl. Acad. Sci. USA 8328. The fragment encoding an SSRP sequence in λPt1 (Seq ID No. 3) is 1.88 kilobases (kb) in length; that in λPt2 (Seq. ID No. 4) is 1.44 kb long. E. coli lysogens (Y1089) comprising the longer λPt1 insert (Seq. ID No. 3) were deposited on Sep. 22, 1988 at the American Type Tissue Culture Collection, 12301 Parklawn Drive, Rockville, Md. 20852, USA, under the terms of the Budapest Treaty and assigned accession number 40498. Upon grant of U.S. Pat. No. 5,359,047, all restrictions upon access to this deposit will be removed. Southern blotting and sequencing studies have confirmed that the λPt1 and λPt2 inserts (Seq. ID Nos. 3 and 4, respectively) are aligned at their 5′ ends; thus, the entire insert of λPt2 lies within the λPt1 sequence. Either SSRP-encoding nucleic acid fragment can be used as a hybridization probe to detect human or other, homologous SSRPs or variants thereof. The shorter clone λPt2 (Seq. ID No. 4), however, is preferred as it more narrowly encompasses the SSRP domain that participates in structure specific recognition of cisplatin-type lesions in DNA. Clone λPt2 (Seq. ID No. 4) was used as a probe in standard Northern blot studies of human and rodent cells to confirm that the eukaryotic cellular SSRP is encoded by homologous 2.8 kb messenger RNAs in at least these representative organisms. From the length of these mRNA transcripts, the intact encoded cellular SSRP should have a molecular mass of 100,000 d. This predicted size correlates well with estimates derived from Southwestern blotting. Northern blotting studies of SSRP expression in various rodent body tissues revealed a pattern consistent with that of a critical gene involved in basic eukaryotic cellular metabolism or survival. SSRP expression did not fluctuate between tissues resistant or susceptible to cisplatin therapy, and did not correlate with the appearance of a cisplatin resistant phenotype in cultured cells.

Clone λPt2 (Seq. ID No. 4) also has been used to identify additional SSRP sequences in several human cDNA libraries. As described in U.S. Pat. No. 5,359,047, these overlapping sequences have been aligned as a composite sequence, reconstructing the complete coding sequence for human SSRP1 (Seq. ID No. 1), reported also in Bruhn et al. (1992), 89 Proc. Natl. Acad. Sci. USA 2307, the teachings of which are incorporated by reference herein. The composite nucleic acid sequence, spanning 2839 bp of DNA, comprises a continuous open reading frame of 2310 bp, extending from nucleotide position 275. This open reading frame encodes a protein, human SSRP1, predicted to have the amino acid sequence set forth in Seq. ID No. 2. The λPt1 insert corresponds to nucleotides 725 to 2612 of Seq. ID No. 1, whereas the λPt2 insert corresponds to nucleotides 725 to 2162. The polypeptide expression product of the λPt2 insert corresponds to amino acid residues 149 to 627 of the encoded human SSRP1. The full-length human SSRP1 polypeptide is predicted to be a 710 amino acid protein of molecular weight 81,068 d. Human SSRP1 is predicted to include several highly charged domains, shown schematically in FIG. 2 . The acidic domain spanning amino acid residues 440-496, contains 26 negatively charged and 4 positively charged amino acids. This domain has similarity to nucleolin, a factor involved in transcriptional control of rRNA genes. Srivastava et al. (1989), 250 FEBS Lett. 99. Two basic domains (Basic I and Basic II) occupy residues 512-534 and 623-640, respectively. Another highly charged series of amino acid residues lies at the carboxyl terminus, spanning residues 661-709. This domain contains 14 negative and 9 positively charged residues. The hydropathy profile of the encoded protein indicates that the entire region from amino acid residue 400 to the carboxyl terminus is highly hydrophilic.

The predicted sequence of human SSRP1 also comprises a domain spanning amino acid residues 539 to 614, referred to herein as an HMG domain, that has been found to share significant levels of sequence similarity with high mobility group (HMG) 1 and 2 proteins from several eukaryotic species, and with upstream binding factor (UBF), a eukaryotic transcription factor known to comprise an HMG domain and to activate transcription of ribosomal RNA genes. Jantzen et al. (1990), 344 Nature 830, Bustin et al. (1990), 1049 Biochim. Biophys. Acta 231, van Holde (1988) Chromatin (Springer-Verlag, NY), Eink and Bustin (1985), 156 Exp. Cell Res. 295. Optimal alignment between the HMG domain of hSSRP1 and human HMG1, ignoring any sequence discontinuities, revealed a 47% amino acid identity in the regions compared. Comparable levels of sequence similarity also exist between the hSSRP1 HMG domain and the corresponding regions of other HMG domain proteins, including sex-determining region Y (SRY), mitochondrial transcription factor II (mtTFII), lymphoid enhancer binding factor I (Lef-1), the T-cell specific transcription factor TCF-1α, the yeast autonomously replicating sequence factor ABF2, and a mouse protein, T160, said to bind to V(D)J recombination signal sequence (RSS) probes. Sinclair et al. (1990), 346 Nature 240, Gubbay et al. (1990), 346 Nature 245, Parisi and Clayton (1991), 250 Science 965, Travis et al. (1991), 5 Genes & Dev. 880, Waterman et al. (1991), 5 Genes & Dev. 656, Diffley and Stillman (1991), 88 Proc. Natl. Acad. Sci. USA 7864, Shirakata et al. (1991), 11 Mol. Cell. Biol. 4528. Of these, the T160 protein, which shares 95.5% similarity with hSSRP1, is considered to be the murine homolog of human SSRP1. The expression product of clone λPt2 (Seq. ID No. 4), which binds effectively to cisplatin-type lesions in DNA, includes the acidic domain, Basic I, and the HMG domain of hSSRP1. Of these, the HMG domain is considered to be the functional domain of hSSRP1 that specifically binds to cisplatin-type genomic lesions.

This view is supported by reports that human HMG-1 binds strongly and specifically to cisplatin-modified oligonucleotides. Pil and Lippard (1992), 256 Science 234, Hughes et al. (1992), 267 J. Biol. Chem. 13520. HMG-1 and -2 are strongly evolutionarily conserved, with homologs identified in diverse eukaryotic genomes, including the human, bovine, porcine, rodent, fish, yeast, maize and protozoan genomes. Wen et al. (1989), 17 Nucl. Acids Res. 1197, Pentecost and Dixon (1984), 4 Biosci. Rep. 49, Kaplan and Duncan (1988), 16 Nuc. Acids Res. 10375, Tsuda et al. (1988), 27 Biochemistry 6159, Paonessa et al. (1987), 15 Nucl. Acids Res. 9077,; Lee et al. (1987), 15 Nucl. Acids Res. 5051, Pentecost et al. (1985), 13 Nucl. Acids Res. 4871, Kolodrubetz and Burgum (1990), 265 J. Biol. Chem. 3234, Grasser and Feix (1991), 19 Nucl. Acids Res. 2573, Roth et al. (1987), 15 Nucl. Acids Res. 8112, Hayashi et al. (1989), 105 J. Biochem. 577. HMG-1 and -2 have been implicated in DNA processing, particularly in transcriptional regulation, e.g., as reported in Watt and Molloy (1988), 16, Nucl. Acids Res. 1471 and Tremethick and Molloy (1986), 261 J. Bio. Chem. 6986. Other properties of HMG-1 are consistent with a role in DNA packaging in chromatin. For example, HMG-1 suppresses nucleosome core particle formation, and selectively unwinds negatively supercoiled DNA. Waga et al. (1989), 1007 Biochim. Biophys. Acta 209, Sheflin and Spaulding (1989), 28 Biochemistry 5658. HMG-1 and -2 also have been shown to bind specifically to structural distortions to DNA such as B-Z junctions and cruciforms. Bianchi et al. (1989), 243 Science 1056, Hamada and Bustin (1985), 24 Biochemistry 1428. Indeed, HMG-1 has been shown to modify these DNA structural motifs such that transcription in vitro proceeds past these otherwise blocking structures. Waga et al. (1990), 265 J. Biol. Chem. 19424, Waga et al. (1988), 153 Biochem. Biophys. Res. Comm. 334. Recent studies have established that the HMG-1 protein comprises two domains, each of which is capable independently of binding to four-way junction DNA. Bianchi et al. (1992), 11 EMBO J. 1055. This confirms earlier reports that HMG-domain fragments of UBF, Lef-1 and TCF-1α retain the specific DNA binding properties of the corresponding intact transcription factors. Jantzen et al. (1990), 344 Nature 830, Giese et al. (1991), 5 Genes & Devel. 2567, Waterman et al. (1991), 5 Genes & Dev. 656.

Diffley and Stillman (1991), 88 Proc. Nat. Acad. Sci. USA 7864, upon considering the known DNA binding properties of HMG-1 and related proteins, suggested that HMG-domain proteins recognize DNA structural motifs rather than sequences. Indeed, HMG-1 and isolated HMG domains thereof now have been shown actively to bend linear duplex DNA, facilitating the in vitro cyclization of substrate DNA by phage T4 DNA ligase. Pil et al. (1993), 90 Proc. Natl. Acad. Sci. USA 9465. Similarly, the HMG domain proteins SRY and Lef-1 both have been shown to induce sharp bends, e.g., of approximately 130°, toward the major groove of DNA substrates comprising their respective cognate binding sites. Ferrari et al. (1992), 11 EMBO J. 4497, Giese et al. (1992), 69 Cell 185. SRY binding to four-way junction DNA is viewed as predominantly structure-specific, rather than sequence dependent. Ferrari et al. (1992), 11 EMBO J. 4497. The HMG domains of Lef-1 and SRY contact the DNA double helix predominantly on the minor groove side and actively bend the helix toward the major groove. Giese et al. (1992), 69 Cell 185. King and Weiss (1993), 90 Proc. Natl. Acad. Sci. USA 11990, have established that the HMG domain of SRY partially intercalates into the widened minor groove at the apex of the recognized or induced bend in substrate duplex DNA. Bending of DNA by the HMG domain spatially approximates linearly distant regions of the double helix. HMG-1, UBF, SRY, Lef-1 and related HMG domain proteins accordingly now are viewed as participating in higher ordered aspects of chromatin structure and nuclear architecture. Wolffe (1994), 264 Science 1100, King and Weiss (1993), 90 Proc. Natl. Acad. Sci. USA 11990, Ferrari et al. (1992), 11 EHBO J. 4497 and Giese et al. (1992), 69 Cell 185, the teachings of each of which are incorporated herein by reference. These studies confirm the view consistently expressed in U.S. Ser. Nos. 08/258,442 and 07/814,964 and prior related applications that the 1,2-d(A{circumflex over ( )}G) and 1,2-d(G{circumflex over ( )}G) intrastrand lesions of cisplatin resemble DNA structures that arise naturally within the eukaryotic genome.

As for HMG-1 and -2, homologs of human SSRP1 occur throughout the eukaryotic phyla. Standard Southern blotting techniques involving detectably labeled λPt2 insert DNA (Seq. ID No. 4) as a probe established that gene sequences encoding homologous SSRPs exist at least in chimpanzee, monkey, elephant, pig, dog, rabbit, mouse, opossum, chicken, fish, and the fruitfly, Drosophila melanogaster. T he isolation and cloning of the Drosophila SSRP1 homolog are reported in U.S. Pat. No. 5,359,047 and in Bruhn et al. (1993), 21 Nucl. Acids Res. 1643, the teachings of which are incorporated by reference herein. The full length composite nucleic acid sequence encoding Drosophila SSRP1, produced from the alignment of two overlapping cDNA clones, is set forth in Seq. ID No. 5. The Drosophila sequence corresponds to 2384 bp of DNA, and contains large open reading frames in both directions, spanning nucleotides 123-2291 and 2300-600. The larger of the two open reading frames predicts a 723 amino acid protein having a molecular mass of 81,524 d (Seq. ID No. 6). This protein shares extensive sequence similarity with hSSRP1 at both the DNA and protein levels. Sequence similarity also was observed between Drosophila SSRP1 and the above-mentioned members of the eukaryotic HMG domain protein family. Alignment of the human and Drosophila SSRP1 homologs in particular revealed that the presence and order of charged domains therein have been conserved across the 600 million years of evolutionary distance that separate humans from fruitflies. In particular, both phylogenetic counterpart proteins include HMG domains at corresponding locations, shown in FIG. 3 . Thus, homologs or phylogenetic counterparts of the human SSRP1 can be isolated as taught in U.S. Pat. No. 5,359,047 and are suitable for use in the present invention.

Homologous SSRPs were not, however, observed upon Southern blot analysis of DNA from the nematode Caenorhabditis elegans, yeast, the parasite Giardia (which retains both prokaryotic and eukaryotic characteristics), or the prokaryotic organisms Pseudomonas and Streptomyces. Bruhn et al. (1993), 21 Nucl. Acids Res. 1643. Nonetheless, as taught in U.S. Pat. No. 5,359,047, additional useful structure specific recognition proteins can be isolated empirically, based upon their binding to cisplatin-lesioned probe DNA. The yeast structure specific recognition protein, initially referred to as ySSRP (in U.S. Pat. No. 5,359,047) and later as Ixr-1 (intrastrand crosslink recognition protein 1, Brown et al. (1993), 261 Science 603), was isolated in this manner. Seq. ID No. 7 sets forth the nucleic acid sequence of the cloned fragment, λyPt, encoding a functional fragment of the Ixr-1 protein. The encoded amino acid sequence is shown in Seq. ID No. 8. Standard Northern blot analysis of yeast messenger RNAs, using detectably labeled λyPt as a probe, established that Ixr-1 is encoded by a 2.1 kb mRNA, consistent with a translated polypeptide of molecular mass 78,000 d. The central, non-glutamine rich portion of the functional Ixr-1 fragment encoded by λyPt (amino acid residues 282-510) shares sequence similarity with other HMG domain family members, particularly the yeast protein ABF2. ABF2 contains two HMG-boxes and is highly related (37% identical, 65% similar) to ySSRP over 151 of its 183 amino acids. Diffley (1991), 88 Proc. Nat. Acad. Sci. USA 7864, has suggested that ABF2 binds to DNA structures, rather than to specific sequences. The isolation of Ixr-1 illustrates that the invention described herein is not limited to the use of hSSRP1 and homologs thereof, but can be practiced with any DNA structure specific recognition protein that specifically binds to cisplatin-type lesions in DNA.

Initially, it was thought that the binding of HMG-domain proteins might aid in damage recognition and thus promote repair of lesions in the genome of eukaryotic cells. Donahue et al. (1990), 29 Biochemistry 5872, and subsequent related publications acknowledged that SSRP binding instead might impede the repair of genomic lesions by sterically obscuring the lesion site, hindering access thereto by elements of the cell's enzymatic DNA repair machinery, as depicted in FIG. 1 A. From the high estimated binding affinity of eukaryotic cellular SSRP for cisplatin-type lesions, it can be predicted that SSRP-shielded lesions will persist in the genome for longer periods than unshielded genomic lesions. Lesions that are unrepaired for significant periods, e.g., at least until the cell commits to enter or reenter the cell division cycle, contribute to cell death by processes such as apoptosis.

A third possible model (shown in FIG. 1B) for the role of SSRPs in cisplatin mediated cell killing, not inconsistent with the repair shielding concept, also was proposed in Donahue et al. (1990), 29 Biochemistry 5872. According to this third model, cisplatin-binding SSRPs naturally regulate the function of genes implicated in the emergence of malignancies, or conversely in the maintenance of normal eukaryotic phenotypes. Cisplatin lesions thus provide DNA structural motifs that mimic the natural regulatory sequences of such genes, thereby titrating SSRP away from its natural binding sites in the genome. See also Scovell (1989), A26 J. Macromol. Sci. Chem. 455. Effective sequestration of SSRP can be achieved where the protein's binding affinity for cisplatin lesions is within about {fraction (1/100)} of its binding affinity for the natural genomic binding site. Preferably, the cisplatin lesion affinity is within about {fraction (1/10)} of that for the natural site; more preferably SSRP binds at least as tightly to cisplatin lesions as to its natural site in the genome. Sequestration by cisplatin lesions therefore “hijacks” SSRP from its natural site in the genome, inducing disarray in cellular processes normally controlled by SSRP. Sequence-independent hijacking of the HMG domain transcription factor hUBF by cisplatin lesions is demonstrated in copending U.S. Ser. No. 08/239,428 (filed May 4, 1994) and reported in Trieber et al. (1994), 91 Proc. Natl. Acad. Sci. USA 5672, the teachings of both of which are incorporated herein by reference. The sequence-independent binding affinity of hUBF for cisplatin lesions in DNA was shown to exceed that of HMG-1, and, more pointedly, to far exceed that of the XPAC protein, which recognizes damaged DNA and is essential for human nucleotide excision repair. Jones and Wood (1993), 32 Biochemistry 12096. XPAC, therefore, should not displace hUBF from cisplatin lesions. The numbers of intracellular hUBF molecules, and of cisplatin genomic lesions formed per cell during a typical round of chemotherapy, have been calculated. Both are in the range of about 5×10 4 /cell. Bell et al. (1988), 241 Science 1192, Reed et al. (1993), 53 Cancer Res. 3694. Biologically significant and synergistic assaults on the survival of eukaryotic cells should therefore follow from the cisplatin-hUBF interactions predicted by both the hijacking and shielding models for cisplatin genotoxicity.

The repair recruitment and shielding models originally proposed in Donahue et al. (1990), 29 Biochemistry 5872, are based on opposing views of the natural function of SSRP. Gene disruption studies with the yeast SSRP, Ixr-1, provided data consistent with the shielding concept. Thus, Brown et al. (1993), 261 Science 603, reported that a mutant strain of yeast carrying homozygous ixr-1 null alleles became two-fold less sensitive to cisplatin killing than the parent (wildtype Ixr-1) strain. Szymkowski et al. (1992), 89 Proc. Natl. Acad. Sci. USA 10772, incorporated herein by reference, also supports the view that a cellular SSRP shields 1,2-intrastrand cisplatin-type DNA lesions from repair in eukaryotic cells. HeLa cell extracts, demonstrably competent to repair genomic lesions induced by 2-acetyl-aminofluorine, were unable to repair a cisplatin 1,2-d(G{circumflex over ( )}G) intrastrand DNA lesion. Prior nicking of substrate DNA comprising the 1,2-d(G{circumflex over ( )}G) lesion, however, rendered this cisplatin lesion susceptible to repair by elements of the HeLa enzymatic DNA repair machinery present in the cell extracts. Id.

Studies disclosed herein, in contrast to earlier circumstantial reports, establish directly that DNA structure specific recognition proteins comprising at least one HMG domain can sterically shield cisplatin-type 1,2-intrastrand DNA adducts from repair by elements of the eukaryotic enzymatic DNA repair machinery. The present studies assess the effects of HMG1 and the human mitochondrial transcription factor h-mtTFA on the excision repair of two defined cisplatin-DNA adducts, the 1,2-intrastrand d(G{circumflex over ( )}G) lesion and the 1,3-intrastrand d(GTG) lesion. Bustin et al. (1990), 1049 Biochim. Biophys. Acta 231, Kao et al. (1993), 90 Proc. Natl. Acad. Sci. USA 5598, Megraw and Chae (1993), 268 J. Biol. Chem. 12758, and Parisi et al. (1993), 13 Molec. Cell. Biol. 1951.

Genomic lesions formed by the covalent interaction of DNA with genotoxic drugs, such as cisplatin, are removed from DNA by excinuclease, an ATP-dependent multisubunit enzyme system. Sancar and Tang (1993), 57 Photochem. Photobiol. 905, Tanaka and Wood (1994), 19 Trends Biochem. Sci. 83. In humans, excinuclease removes covalent lesions in cellular DNA by hydrolyzing the 5th phosphodiester bond 3′, and the 22nd-24th phosphodiester bonds 5′, to the lesion site, such that 27-29 nucleotide-long oligomers comprising the covalent adduct are excised from the genome. Huang et al. (1992), 89 Proc. Natl. Acad. Sci. USA 3664, Svoboda et al. (1993), 268 J. Biol. Chem. 1931.

The study described more fully below in EXAMPLE 1 established that, in contrast to the findings of Szymkowski et al. (1992), 89 Proc. Natl. Acad. Sci. USA 10772, crude human (HeLa) cell extracts comprising excinuclease can excise cisplatin lesions from DNA. Excision activity was not detected, however, in cell-free extracts from cell lines of two different complementation groups of xeroderma pigmentosum (XP), a human disease characterized by loss of nucleotide excision repair capability. Cells from XP patients are sensitive to DNA lesions including thymine dimers and cisplatin adducts. Cleaver and Kraemer (1989), in Metabolic Basis of Inherited Disease, 2849 (Scriver et al., Eds.). Mixing of the cell-free extracts from the two different complementation groups restored excision capability (Reardon et al. (1993), 58 Cold Spring Harbor Symp. Quant. Biol. 605), a characteristic of the specific action of the multisubunit human excinuclease on damaged DNA. These results indicate that cisplatin lesion repair proceeds by the same enzymatic mechanism as the removal of pyrimidine dimers and psoralen monoadducts from human cellular DNA.

Next, timecourse studies were conducted to establish the kinetics of 1,2-d(G{circumflex over ( )}G) and 1,3-d(GTG) cisplatin lesion removal by human exinuclease. EXAMPLE 2, below, discloses that both lesions were repaired, but with different efficiencies. The 1,3-intrastrand crosslink consistently was excised from substrate DNA 30-50% faster than excision of the 1,2-intrastrand crosslink. A kinetic plot of these excision results is shown in FIG. 4 . An excinuclease reaction time (30 minutes) within the linear portion of the kinetic plot was selected for further studies of the impact of HMG domain proteins on repair kinetics. It should be noted that the kinetic data shown in FIG. 4 contradict the earlier report of Szymkowski et al. (1992), 89 Proc. Natl. Acad. Sci. USA 10772, human (HeLa) cell extracts were unable to repair 1,2-d(G{circumflex over ( )}G) cisplatin lesions in DNA. The excision assay described herein is, however, significantly more sensitive than the repair synthesis assay relied upon in Szymkowski et al. (1992).

Pil and Lippard (1992), 256 Science 234, reported that HMG1 binds to the 1,2-d(G{circumflex over ( )}G) cisplatin adduct with an affinity significantly in excess of its affinity for the 1,3-d(GTG) cisplatin adduct. Further, the affinity of HMG1 for the latter crosslink was essentially the same as that for unmodified DNA. EXAMPLE 3, below, established that HMG1 not only binds specifically to 1,2-intrastrand adducts: it specifically alters the rate of removal of these DNA lesions by exinuclease. Excision of the cisplatin 1,2-intrastrand crosslink by the exinuclease in HeLa cell-free extract was monotonically inhibited by escalating concentrations of purified HMG1. In contrast, excision of the 1,3-crosslink was modestly stimulated in the presence of HMG1 concentrations up to about 4 μM and then inhibited at higher concentrations. A concentration course plot of these results is shown in FIG. 5 . At 4 μM HMG1, both the 1,2- and the 1,3-intrastrand crosslinked substrates were saturated with HMG1, yet only excision of the 1,2-d(G{circumflex over ( )}G) crosslink was inhibited. Therefore, only the specific mode of HMG binding interferes with the excision repair of cisplatin lesions. The stimulatory effect observed for excision of the 1,3-intrastrand crosslink might be attributable to HMG1-mediated bending of the DNA (Bianchi et al. (1989), 243 Science 1056) in a manner favorable for the excinuclease. Only at very high concentrations of HMG1 (in excess of 4 μM) did non-specific binding interfere with repair.

Another HMG domain protein, human mitochondrial transcription factor A (h-mtTFA) also has been shown to bind, in a sequence independent manner, to cisplatin-type 1,2-intrastrand dinucleotide lesions in DNA, as assessed by EMSA and Southwestern blotting. This member of the SSRP family also is shown herein to selectively shield the 1,2-d(G{circumflex over ( )}G) intrastrand adduct from repair by human excinuclease. Indeed, h-mtTFA bound to the 1,2-d(G{circumflex over ( )}G) lesion with greater affinity than HMG1. TABLE 1, below, discloses that 0.5 μM concentrations of h-mtTFA inhibited excision of the 1,2-intrastrand cisplatin adduct by 95% while blocking removal of the 1,3-d(GTG) crosslink by only 40%. The same molar concentration of HMG1 achieved only about 20% repair shielding.

EXAMPLES 3 and 4 directly demonstrate that two distinct HMG domain proteins within the SSRP family bind with specificity to the major DNA adduct of cisplatin produced in vivo (the 1,2-intrastrand d(G{circumflex over ( )}G) adduct; see Fichtinger-Schepman et al. (1985), 24 Biochemistry 707; Eastman (1987), 34 Pharmacol. Ther. 155. Specific binding produces a DNA-protein complex at the lesion site that effectively shields the lesion from excinuclease activity. These results confirm early indications in Brown et al. (1993), 261 Science 603 and Trieber et al. (1994), 91 Proc. Natl. Acad. Sci. USA 5672, that any eukaryotic protein that comprises at least one HMG domain that binds with high, specific affinity to cisplatin-type 1,2-intrastrand lesions in DNA can shield these genomic lesions from repair, thereby enhancing persistence of cisplatin-type lesions in the genome in vivo. Shielded lesions persist in the genome, disrupting the transcription of cellular genes and impeding DNA synthesis associated with cell replication. Shielding by the full range of HMG domain proteins present intracellularly, rather than by hSSRP1 or a homolog thereof alone, therefore contributes to the sensitivity of tumors to cisplatin chemotherapy. Accordingly, the entire spectrum of intracellular cisplatin lesion binding SSRPs can be assessed to gauge the cytotoxic efficacy of cisplatin and related chemotherapeutic drugs.

Results presented herein further explain the longstanding conundrum presented by observations that hSSRP1 expression patterns do not correlate with the sensitivities of various eukaryotic tissues and cell lines to cisplatin. Bruhn et al. (1992), 89 Proc. Natl. Acad. Sci USA 2307. The concentrations of HMG domain SSRPs used in the EXAMPLE 3 and 4 studies were significantly higher than the endogenous cellular concentration of each respective protein. However, there are many known HMG domain SSRPs, and possibly many yet to be discovered. Many of these endogenous HMG domain proteins may act in concert to shield cisplatin lesions from repair in vivo. Thus, measurement of the total concentration of intracellular HMG domain proteins should provide a reasonably accurate predictor of the sensitivity of particular cell and tissue types to cisplatin. Preferably, this value can be weighted for the relative affinities of major endogenous SSRPs for the 1,2-intrastrand crosslink, or for intracellular compartmentalization affects on the accessibility of cisplatin lesions to particular endogenous SSRPs. The total endogenous concentration of HMG domain proteins that bind cisplatin-type genomic lesions may match or even exceed the concentrations of purified individual HMG domain proteins used in EXAMPLES 3 and 4.

Further, the in vitro assay system described in EXAMPLES 3 and 4 can be used as the basis for preliminary testing, in cell free systems, of novel cisplatin analogs or other chemotherapeutic drug candidates. It has been established that cisplatin and related drugs produce lesions in DNA through covalent bonds formed upon hydrolysis of the bonds linking the chlorine and platinum atoms, and subsequent establishment of bonds linking the platinum atom to two purine bases. Following DNA adduct formation, the ammine groups bonded to the platinum atom in cisplatin protrude into the major groove. HMG domain SSRPs, however, bind to recognized DNA lesions predominantly on the minor groove side, opposite to the platinum bridge. Neither the platinum atom itself nor electron donor substituents thereof participate directly in SSRP/DNA lesion binding. Therefore, significant structural variation can be tolerated in the drug moieties located trans to the substitutionally labile chlorine atoms.

The invention thus provides a convenient in vitro screening procedure for assessing whether novel cisplatin analogs, or other chemotherapeutic candidates, form DNA lesions that are recognized and shielded from repair by SSRP family members. Novel genotoxic drug candidates can be synthesized individually or selected from a library of random products of combinatorial, e.g., synthesis as disclosed in Needels et al. (1993), 90 Proc. Natl. Acad. Sci. USA 10700 or Ohlmeyer et al. (1993), 90 Proc. Natl. Acad. Sci. USA 10922, the teachings of each of which are incorporated herein by reference. For example, a random combinatorial library of cisplatin-type drug candidates can be produced by allowing a cis-blocked platinum(II) preparation to react randomly with an amino acid preparation, such as a cocktail comprising some or all of the naturally occurring α-amino acids. New cisplatin-type drugs accordingly can be selected and further refined for their ability to form lesions that are bound with high affinity by a preferred SSRP, or that are bound by a panel of HMG domain SSRPs, tested singly or as a cocktail. Accordingly, the invention described herein encompasses new cisplatin-type chemotherapeutic drug candidates identified using the present screening assay. Further, the invention encompasses methods of making such new cisplatin-type chemotherapeutic drugs, comprising the steps of synthesizing a candidate cisplatin-type chemotherapeutic candidate and assessing the candidate for competence to produce SSRP-recognized genomic lesions. SSRP shielding of nucleic acid lesions can be assessed by any of the techniques disclosed herein. Thus, shielding can be assessed in vitro using a excinuclease assay generally according to Examples 3 and 4. Drug candidates which produce a high exinuclease shielding index in the present cell-free screening assay should be selected for further evaluation of antitumor efficacy, e.g., using transformed cell lines, primary cells in culture or animal models. Thus, the present invention offers the capacity to make more effective use of traditional preclinical screening procedures.

The invention further provides new methods for sensitizing eukaryotic cells, e.g., transformed cells, to the cytotoxic effects of cisplatin-type genotoxins. These methods can be used to enhance the effectiveness of cisplatin-type chemotherapy by achieving a greater degree of cell killing than would be observed with current protocols at similar drug dosage levels. Further, these methods can be used to sensitize transformed cells that are refractory to killing by cisplatin-type genotoxins. In the present methods, nucleic acid encoding an SSRP or fragment thereof (e.g., an HMG domain) that binds to cisplatin-type DNA lesions, is contacted with the eukaryotic cells under conditions sufficient for uptake of foreign nucleic acids. SSRP encoding nucleic acids can be inserted into a viral vector, such as a retroviral vector, or packaged into liposomes, injected directly or transfected into eukaryotic cells using known techniques such as those reviewed in Mulligan (1993), 260 Science 926, the teachings of which are incorporated by reference herein. As mentioned previously, the eukaryotic cells can comprise cells of a unicellular or multicellular organism, and can comprise cells maintained in culture, cells withdrawn from a multicellular organism, or cells present in the tissues or organs of a multicellular organism. That is, the method can be practiced in vitro, ex vivo (using a sample, such as a biopsy, withdrawn from a multicellular organism such as a mammal, e.g., a human), or in vivo, by local or systemic administration to a multicellular organism.

For the treatment of malignancy in situ, retroviral vectors are preferred, as they will be selectively internalized by actively dividing cells, such as transformed cells. If desired, cells can be withdrawn from the individual, caused to internalize SSRP-encoding retroviral nucleic acid and reintroduced to the individual, preferably by local injection or infusion in the vicinity of transformed cells. The cells caused to internalize SSRP-encoding nucleic acid ex vivo accordingly provide a localized sustained release of SSRP-encoding nucleic acid that can be internalized and expressed by transformed cells in situ. See, e.g., Culver et al. (1992), 256 Science 1550, the teachings of which are incorporated by reference herein. Such methods can be used to achieve selective killing of neoplastic (transformed) cells throughout the body, e.g., cells of reproductive tract, cutaneous, mammary, or neurologic origin. Cells that are of reproductive tract origin can be more specifically, of ovarian, uterine, endometrial, cervical, vaginal, prostate, or testicular origin. Cells that are of mammary origin can be more specifically, of breast origin. Cells that are of cutaneous origin comprise basal cells, melanocytes, dendritic cells and cutaneous T lymphocytes. Cells that are of neurologic origin can be of central or peripheral nervous system origin, and can be neurons, glia, Schwann cells and the like. Transformed cells of each of the foregoing tissue-specific origins can be caused to internalize SSRP encoding nucleic acids and express the encoded heterologous SSRP. Use of suitable expression control sequences will optionally cause overexpression of the heterologous SSRP, which is expected to significantly enhance susceptibility of the eukaryotic cells to cisplatin-type drug cytotoxicity, by ensuring an intracellular excess of SSRPs to shield cisplatin-type lesions from repair.

For in situ use to destroy transformed cells in the tissues of a multicellular organism (e.g., a mammal) the protocols for administering nucleic acid encoding SSRP will vary depending on the location of cells to be destroyed, replicative rate of the cells, level of repair proficiency of the cells, dose of genotoxic drug to be administered, route of delivery thereof, and pharmacokinetic profiles of clearance and tissue uptake of both the SSRP encoding nucleic acid and the genotoxic drug. Variables affecting the amounts needed thus include, but are not limited to, the nature (e.g., species or tissue type), quantity and accessibility (i.e., body compartment location) of eukaryotic cells to be destroyed, and the nature, genotoxicity, and affinity of the cisplatin-type genotoxin. Thus, as appropriate, SSRP encoding nucleic acid can be combined with a pharmaceutically acceptable carrier or excipient for formulation as a liquid, suspension, solid, salve, ointment or the like, suitable for oral, nasal, intravenous, intracerebral, intraspinal, intraperitoneal, topical, subdermal, intramuscular, or other routes of administration. SSRP encoding nucleic acid can be administered in a single dose (e.g., a bolus injection), a series of doses of equivalent, escalating, decreasing or intermittently varied quantity, or infused over a period of time (e.g., by intravenous drip or infusion), or by release from a slow-release delivery vehicle. The appropriate dose will of course be dictated by the precise circumstances under which the invention is practiced, but will generally be in the range of 0.01 ng to 10 g per kg body weight, preferably in the range of 1 ng to 0.1 g per kg, and more preferably in the range of 100 ng to 10 mg per kg.

Following uptake and expression intracellularly of the encoded SSRP, both of which can be monitored if preferred using standard techniques or routine modifications thereof, the cisplatin-type genotoxic agent is administered. Thereafter, the degree of eukaryotic cell killing achieved can be ascertained through standard, widely available techniques, such as visual or microscopic inspection, biochemical, chromogenic or immunologic methods for detecting products of eukaryotic cell lysis, and the like. Such techniques can be used to establish both the dose and time period effective to accomplish objectives of the present invention under particular circumstances. Once effective doses and time periods are established, it may be no longer necessary to monitor the progress of cell killing.

Practice of the invention will be still more fully understood from the following examples, which are presented herein for illustration only and should not be construed as limiting the invention in any way.