Proteins that associate with DNA vary in structure, sequence and structural binding motifs, degree of interaction, regulation, and role in regulating nuclear metabolism. Protein targets for drug design that will be discussed in this review include topoisomerase proteins, DNA methyltransferases, HMGs, TBP and histone deacetylases. Other transcription factors, repair proteins and other important nuclear proteins involved in replication and recombination are also important candidates to consider, but will not be discussed here.
Topoisomerase enzymes are involved in resolving topological problems in DNA, such as superhelical tension, that arise during most nuclear activities involving DNA. Topoisomerase I (TOPO I) acts by introducing one break in one strand of the DNA, whereas Topoisomerase II (TOPO II) acts by making a double strand break. Aside from several intercalating agents such as doxorubicin, TOPO I and II are the targets of numerous drugs, some of which function as poisons and others as competitive inhibitors.
The major class of TOPO I poisons is exemplified by camptothecin (CPT) [32,154], a cytotoxic alkaloid containing an essential six membered lactone ring. This is a natural product, originally isolated from Camptotheca acuminata. It binds to TOPO I in association with DNA and stabilizes the single-stranded break, preventing strand passing through the break and subsequent resealing. Structure-activity studies have lead to the development of a number of closely related compounds that are also used clinically. TOPO II poisons can be classified into two major classes. The DNA intercalators that bind primarily to DNA are exemplified by the anthracycline doxorubicin , while those that bind primarily to the protein can be exemplified by the epipodophyllotoxin etoposide . Both classes of compounds work by formation of a ternary drug/DNA/enzyme complex that inhibits the DNA resealing activity of the enzyme, stabilizing the cleavable complex and resulting in DNA double-strand breaks. Other inhibitors, such as fostriecin  and derivatives, inhibit the catalytic activity of topoisomerase without stabilizing the cleavable complex.
Changes in the nuclear environment that result from topoisomerase inhibitors vary. TOPO II is cell-cycle dependent and expression is higher in actively proliferating cells leading to their greater susceptibility[158,159]. TOPO II inhibitors generally disrupt the cell-cycle during S phase because the increased concentration of DNA double-strand breaks interferes with DNA replication and triggers apoptosis[153,160]. In the case of CPT, the generation of DNA breaks in addition to the collision of the stabilized cleavable complex with the DNA replication machinery is necessary for cell killing [160-162]. CPT also blocks the elongation step of transcription.
DNA methylases and DNA methylation
Nuclear DNA methylation patterns are often altered in human malignancies. In some cases, the genome-wide degree of methylation is reduced and in certain circumstances may be responsible for abnormal gene activation. In other cases, promoters of certain genes are hypermethylated and this leads to gene silencing . There are at least three enzymes responsible for DNA methylation of CpG dinucleotides (the transfer of methyl groups from S-adenosyl-L-methionine to cytosine to form 5-methylcytosine), DNA methyltransferase (DNMT) 1, DNMT3A, and DNMT 3B. DNMT1 is predominantly responsible for maintaining methylation patterns during replication, whereas DNMT3A and 3B are required for de novo methylation.
CpG islands are concentrated regions of CpG dinucleotides generally found in promoter regions[166,167]. Methylation of these regions in normal cells generally occurs on genes located on the inactivated X chromosome and on selectively silenced alleles[168,169]. In neoplastic tissue however, methylation can lead to silencing of key tumour suppressor genes. Tumour suppressor and growth inhibitory genes, such as p15INK4b, p16INK4a, and p21WAF1/CIP1, are often silenced in cancer cells by DNA hypermethylation of their promoter regions. Reversing the degree of DNA methylation using azanucleoside DNA methyltransferase inhibitors such as 5-azacytidine  and 5-aza-2'-deoxycytidine  (decitabine) reactivates transcription at previously silenced tumour suppressor promoters. The mechanism by which these azanucleotides act is via incorporation into the cellular DNA and inhibition of DNMT1 via covalent interaction (thioether bond) at C-6 which forces replication to proceed without DNA methylation. Thus far these compounds have shown some success in clinical trials [174-177].
Two alternate strategies that have been employed to inhibit DNMT1 are the use of DNMT1 anti-sense oligonucleotides and the use of hairpin-structured oligonucleotide substrate mimics. Fournel et al. ) demonstrated that use of antisense oligonucleotides led to loss of DNMT1 protein, decrease in methylation at the p16INK4a promoter, and expression of p16INK4a RNA. Although the hairpin mimics effectively inhibit purified DNMT1 in vitro, they have not been able to induce methylation changes in cells.
Histone acetylation/deacetylation (HDACs)
The regulation of chromatin structure and DNA sequence accessibility is the subject of extensive research because they play important roles in governing numerous nuclear activities including transcription, recombination, and replication. The basic nuclear DNA unit is the nucleosome; a 146 bp stretch of nuclear DNA wrapped around an octamer of histone (H) proteins (two each of H2A, H2B, H3 and H4) [180-182]. Arrays of nucleosomes are then folded into higher-order structures such as chromatin fibres. The three known mechanisms influencing chromatin structure are ATP-dependent chromatin remodelling, histone replacement, and covalent modification of either DNA or associated histones. In the last case, modifications include cytosine methylation and histone acetylation, phosphorylation, methylation, ubiquitination, and ADP-ribosylation. The regulation of these mechanisms and their cross-talk in the nucleus affects whether DNA sequences will be accessible for nuclear processes.
Active transcription is associated with lysine acetylation of histones and methylation at Lys-9 of H3; whereas, transcriptional repression is associated with deacetylated histones, methylation of Lys-4 of H3 and methylated DNA[180,181,183,184]. The main mechanism by which HDACs repress transcription is likely to be the result of functional linkage between HDACs and DNA methylation via recruitment of HDAC to DNA by methyl binding proteins (for example MeCP2 and MBPs1–4). This in turn inhibits the recruitment transcriptional activators such as SWI/SNF, SAGA, and TBP proteins to the promoter regions preventing gene expression. In addition, the deacetylated histone tails may actively recruit transcriptional silencers. For example in yeast, Sir3 preferentially binds to non-acetylated DNA and associates with Sir2/Sir4 to form a "silencing" complex that spreads heterochromatin formation [180,183,186]. Although Sir3 and Sir4 human homologues have not been identified, the potential for preferential binding to non-acetylated DNA is still a feasible mechanism to consider. In addition, it is important to consider that HDACs may act on other molecules, such as specific transcription factors and alter their activation states.
There are three major classes of HDACs, although known inhibitors only work on two of the three classes. The HDAC I class is composed of proteins that are homologous to the yeast RPD3 gene; HDAC1,2,3,8, and 11. HDAC class II is subdivided into two; IIA proteins include HDAC4,5,7, and 9, IIB proteins are HDAC 6 and 10 and are characterized by sequence homology to yeast HDAI. Class III proteins are similar to the yeast repressor protein, Sir2, although they have not been extensively characterized in mammals. The HDAC inhibitors of both HDAC I and II classes bind and inhibit HDAC activity, induce acetylation of histones in cells, inhibit tumour cell proliferation in vitro, and several compounds have shown promising results in human xenograft experiments [187-190].
The HDACs are attractive targets for anti-cancer drug design because they are involved in gene silencing similar to the DNA methylase targets mentioned above, and a number of structurally diverse classes of inhibitors are in development. However, only the first two types of HDACs are sensitive to the inhibitors developed thus far. These fall into several structural classes; the short chain fatty acids, the hydroxamic acids, the cyclic tetrapeptides, the benzamides, and the epoxides.
A number of these HDAC inhibitors are at various stages of clinical study. Phase I studies with suberoylanilide hydroxamic acid (SAHA)  concluded that it was well tolerated, inhibited the HDAC activity in vivo (in peripheral blood mononuclear cells and in post-therapy tumour biopsy), and has anti-tumour activity in solid (bladder) and haematological tumours. SAHA is in phase II development for treatment of cutaneous T cell lymphoma, peripheral T cell lymphoma, and recurrent or metastatic squamous cell carcinoma of the head and neck. The depsipeptide FR901228 (FK228)  is in trials for a variety of leukemias and lymphomas, and refractory or progressive small cell and non-small cell lung cancer.
The cellular response to HDAC inhibitors is becoming better characterized. For example, SAHA binds directly to the HDAC catalytic site, inhibits enzyme activity, inhibits cellular proliferation and promotes an apoptotic cellular response [193-195]. Oligonucleotide microarray analysis of multiple myeloma cells treated with SAHA demonstrated that commitment to apoptosis is associated with suppression of genes involved in promoting cell growth and survival, drug resistance, cell cycle control, DNA replication and repair, and proteosome function.
Combination therapy using HDAC inhibitors and other agents such as DNA methylase inhibitors and heat shock protein 90 (HSP90) antagonists have demonstrated a synergistic effect on cellular responses such as activation of gene expression and induction of apoptosis [197-202]. For example, using depsipeptide and 5-aza-2'-deoxycytidine treatment of breast cancer cell lines, Primeau et al. demonstrated synergistic anti-neoplastic activity and the activation of mapsin and gelsolin gene expression, two metastatic tumour suppressor genes that are silenced by epigenetic mechanisms in breast cancer lines. In the study by Rahmini et al, co-administration of the HSP90 antagonist 17-allylamino-17-demethoxy-geldanamycin (17-AAG) and SAHA induced synergistic induction of mitochondrial damage, caspase activation (-3 and -8), and apoptosis in several different human leukaemia cell lines. These investigators found that co-administration of 17-AAG with SAHA blocked the SAHA mediated induction of the cyclin-dependent kinase inhibitor, p21WAF1/CIP1, and that this novel finding might explain why the combination treatment was more effective. When cells were forced to express p21WAF1/CIP1, the combination treatment was far less effective.
Proteins that recognize distorted DNA structures: Tata Box Binding Protein and High Mobility Group proteins
In many cases the interaction of small molecules with DNA results in a distortion in the normal DNA structure. This distortion may result in the inhibition or induction of protein binding at or in the vicinity of the perturbation. The outcomes of the alteration vary and may be more or less dependent on the primary nucleotide sequence. In this subsection, TBP and HMG proteins are discussed.
Cis-diamminedichloroplatinum  (cisplatin) is a DNA alkylator that forms predominantly intrastrand crosslinks by binding to the N7 nitrogen of adjacent guanidine residues or guanidine-adenine residues. Cisplatin binding causes a structural distortion in the DNA (bend of 40–60°, and twist of 25–32°) as revealed by crystallography and NMR spectroscopy. The cisplatin-distorted DNA activates binding of a number of nuclear proteins such as TBP, p53, HMG1, and H1[23,24,204-208]. In the case of TBP, binding is somewhat sequence dependent, although the preferred sequence is not the TBP consensus sequence. It has been suggested that the affinity of TBP for cisplatin-modified DNA is because the DNA distortion caused by cisplatin crosslinking is similar to TATA box DNA structure when bound to TBP.
The effects of cisplatin treatment include inhibition of DNA synthesis and RNA transcription, effects on the cell cycle, and activation of apoptosis. The binding of HMG1 to the cisplatin adduct has been shown to block translesion DNA synthesis in vitro . Studies by He et al. suggest that the interaction of HMG1 and the DNA adduct increases the cytotoxic potential of cisplatin. In these experiments, breast cancer cells were induced to over express HMG1 by pre-treatment with estrogen and were shown to be more sensitive to cisplatin than uninduced cells. This sensitization could be because DNA replication is halted either directly or because DNA adducts are no longer effectively recognized by cellular nucleotide excision repair machinery[212,213].
While cisplatin has been effective in treating a number of different cancer types including testicular cancer, ovarian cancer, head and neck cancer, the onset of resistance has limited its use[207,214]. In general, the degree of cytotoxicity of cisplatin is correlated with the formation and duration of DNA adducts, therefore processes that interfere with these will contribute to the development of resistance. The pathways that have been implicated in the development of resistance to cisplatin include pharmacologic-based mechanisms (reduced drug uptake and increased drug inactivation) as well as mechanisms that inhibit apoptotic signalling. Some of these anti-apoptotic mechanisms include increased repair of the cisplatin-DNA adducts and/or masking of the DNA damage by creating unnatural protein binding sites and/or inducing binding of proteins such as HMG1, and H1 [1,14,203,215]. Additionally, many studies point to the DNA adduct masking (repair shielding) as a process strongly involved in potentiating cytotoxicity.