Small molecules targeting histone H4 as potential therapeutics for chronic myelogenous leukemia

We recently identified a polyamide-chlorambucil conjugate, 1R-Chl, which alkylates and down-regulates transcription of the human histone H4c gene and inhibits the growth of several cancer cell lines in vitro and in a murine SW620 xenograft model, without apparent animal toxicity. In this study, we analyzed the effects of 1R-Chl in the chronic myelogenous leukemia cell line K562 and identified another polyamide conjugate, 6R-Chl, which targets H4 genes and elicits a similar cellular response. Other polyamide conjugates that do not target the H4 gene do not elicit this response. In a murine model, both 1R-Chl and 6R-Chl were found to be highly effective in blocking K562 xenograft growth with high-dose tolerance. Unlike conventional and distamycin-based alkylators, little or no cytotoxicities and animal toxicities were observed in mg/kg dosage ranges. These results suggest that these polyamide alkylators may be a viable treatment alternative for chronic myelogenous leukemia. [Mol Cancer Ther 2008;7(4):769–78]

Therapies targeting specific genes or gene products are a major aim of modern cancer biology and intervention (1–4). Notable recent successes in the field include both monoclonal antibodies (5, 6) and small molecules that target enzymes and receptors that are overexpressed or mutated in various cancers (3, 7, 8). Notwithstanding these important developments, DNA-alkylating agents remain the most common drugs for treatment of several solid and hematologic malignancies (9–11). Myelotoxicity and concomitant cytotoxicity due to limited DNA sequence selectivity is often the dose-limiting factor for use of these compounds in humans. Alkylating agents based on the minor groove binder distamycin A, including tallimustine and brostallicin, have shown improvements over nonconjugated DNA-alkylating agents in terms of affinity and specificity at the nucleotide level (12–15). However, these compounds possess only limited specificity (16), and myelotoxicity is still the dose-limiting factor in establishing an effective chemotherapy (12, 13).

Pyrrole-imidazole polyamides are a class of sequence-specific DNA-binding small molecules that have been shown to have high binding specificity and affinity, with some molecules even exceeding the binding affinities of transcription factors (17). Many polyamides are cell permeable and readily localize to the nuclei of cultured cells (18–21). Polyamides are effective inhibitors of RNA transcription when they disrupt essential protein-DNA interactions at promoter and enhancer elements (17, 22–25), and they may also modulate gene expression by modifying chromatin structure (26–28). Linked with DNA-alkylating agents, such as chlorambucil (29) or CC-1065/CBI derivatives (30, 31), polyamide conjugates react covalently with DNA at specific sites and inhibit transcription by stalling RNA polymerase during elongation.

Recently, α-diaminobutyric acid–linked hairpin polyamide-chlorambucil conjugates have been shown to bind and alkylate DNA sequences both in vitro and in cell culture models, with good sequence specificity (32–34). Initial screens of various hairpin polyamide-chlorambucil conjugates have shown the polyamide 1R-Chl (Fig. 1A, top) to be an inhibitor of cell proliferation in various cancer cell lines with no apparent cytotoxicity and little or no murine animal toxicity (32, 33, 35). This molecule binds within the coding region of the histone H4c gene both in vitro and in SW620 human colon carcinoma cells and down-regulates H4c transcription. Polyamides with similar pyrrole and imidazole compositions targeted to different DNA sequences failed to alkylate the coding region of the histone H4c gene and were found to be inactive both in cell culture and in a SW620 xenograft cancer model (32).

Studies suggest a two-hit mechanism for the observed cellular effects of 1R-Chl: down-regulation of histone gene transcription causes nucleosome depletion, followed by widespread alkylation of open chromatin, which elicits cell cycle arrest through the DNA repair pathway (35). Although our initial results point to the histone H4c gene as the major target of 1R-Chl, microarray studies in the SW620 cancer cell line indicate that the mRNA levels of several other genes are also affected (32). Thus, down-regulation of other genes may be involved in the cellular response to 1R-Chl.

In the present study, we extend our analysis to the well-established chronic myelogenous leukemia (CML) cell line K562. If histone H4 genes are the primary targets of 1R-Chl that lead to a block in cancer cell proliferation (35), other polyamide-chlorambucil conjugates targeting H4 genes would be predicted to elicit the same cellular response. We describe the synthesis and characterization of a small library of constitutional isomers of 1R-Chl. These molecules have the same chemical composition as 1R-Chl but would be expected to bind different DNA sequences. One conjugate, 6R-Chl (Fig. 1A, bottom), which targets sites adjacent to and overlapping the binding site for 1R-Chl (Fig. 1B) in the H4c gene, was found to have biological properties similar to 1R-Chl both in K562 cell culture and in a mouse xenograft model established with K562 cells. Other polyamide-chlorambucil alkylators that did not bind within the H4c gene or down-regulate histone H4 expression had no effect on cell proliferation. Microarray analysis in K562 cells reveals that the histone H4 genes H4c and H4j/H4k are down-regulated by 1R-Chl treatment. Transcripts for the H4k and H4j genes cannot be distinguished due to similarity in sequence. In addition, we explored the pharmacokinetic properties of 1R-Chl, the results of which point to this class of molecules as potential human cancer therapeutics.

Pyrrole-imidazole polyamides were synthesized by standard solid-phase methods (36) using α-(R)-2,4-diaminobutyric acid or α-(S)-2,4-diaminobutyric acid as the hairpin turn unit (33). Polyamide-chlorambucil conjugates were generated as previously described (29), whereby the carboxylic acid of chlorambucil (Sigma-Aldrich) is activated and coupled to the free amine of the hairpin turn. The identity and purity of the compounds were established by mass spectrometry (MS) analysis (matrix-assisted laser desorption/ionization time-of-flight mass spectroscopy and electrospray mass spectroscopy) and analytic high-performance liquid chromatography, respectively.

Binding affinities of the parent polyamides (which lack chlorambucil) for target match and mismatch sites were determined by quantitative DNase I footprinting (37) using a radiolabeled DNA fragment of the plasmid pMFST6 (Supplementary Fig. S1A).³

The plasmid was constructed by annealing the oligonucleotide pair 5′-AGCTGTAGTATCTATAGTGCTTACTATCTATAGTCCTTACTATCTATAGCTCTTACTATCTATAGCTGTTACTATCTATACTATCTAC-3′ and 5′-CATCATAGATATCACGAATGATAGATATCAGGAATGATAGATATCGAGAATATAGATATCGACAATGATAGATAGATAGATGCTAG-3′. Annealed oligonucleotides were ligated into the BamHI/HindIII restriction fragment of pUC19 using T4 DNA ligase, and the plasmid was transformed into Escherichia coli JM109 competent cells. Ampicillin-resistant white colonies were selected from 25 mL Luria-Bertani agar plates containing 50 mg/mL ampicillin treated with 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside and isopropyl-l-thio-B-d-galactopyranoside solutions and grown overnight at 37°C. Cells were harvested the following day, and purification of the plasmid was done with a Wizard Plus Midiprep DNA Purification kit (Promega). Alkylation experiments done on the pMFST6 insert were done as previously described (29). Alkylation experiments were also done on a 240-bp region of the H4c mRNA coding sequence, which was amplified from genomic DNA with the following PCR primers: 5′-GTGCTAAGCGCCATCGTAAG-3′ and 5′-CCCTGACGTTTTAGGGCATA-3′. These experiments were conducted using 10 ng of the PCR product incubated for 16 h at 37°C in 100 μL of 20 mmol/L NaCl and 10 mmol/L Tris-Cl (pH 7.4), with each of the polyamide-chlorambucil conjugates at 10, 100, and 1,000 nmol/L concentration followed by thermal cleavage and primer extension labeling, as described (32).

The human CML lymphoblast cell line K562 (purchased from the American Type Culture Collection), which contains the b3a2 Bcr-Abl translocation, was used in this study. Cells were grown and maintained in RPMI 1640 containing 10% fetal bovine serum under standard mammalian cell culture conditions as recommended by the American Type Culture Collection. Direct phase-contrast microscopic visualization was used to monitor the effects of polyamide-chlorambucil conjugates on cell growth rates and cell morphology. Promega CellTiter 96 Aqueous One Solution Cell Proliferation assay [using 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H tetrazolium (MTS) conversion to formazan to examine mitochondrial activity], trypan blue exclusion (Vi-Cell XR viability analyzer, Beckman Coulter), and Annexin V-FITC/propidium iodide apoptosis staining (BD PharMingen) were used to determine cell proliferation (EC₅₀), viability, and initiation of apoptosis, respectively.

The effects of polyamide-chlorambucil conjugates on cell cycle progression were monitored by flow cytometry analysis in the Scripps fluorescence-activated cell sorting core facility. Polyamide-treated cells (250 nmol/L of polyamide in culture medium for 24 h) were collected by centrifugation (200 × g for 5 min). Cell pellets were resuspended in 500 μL PBS and fixed with addition of 4.5 mL of prechilled 70% ethanol, stained with propidium iodide (50 μg/mL), and analyzed for DNA content, reflecting the fraction of cells at each point in the cell cycle (G₀-G₁, S, and G₂-M). Cells with less than a 2C DNA content are indicative of DNA fragmentation and apoptosis.

K562 CML cells were incubated in culture medium for 24 h with each of the polyamide-chlorambucil conjugates at a 250 nmol/L concentration followed by digestion of purified DNA with DraI, thermal cleavage, and ligation-mediated PCR (LM-PCR) using the primers for the H4c gene listed above, as previously described (32).

RNA from polyamide-treated cells was extracted using the Absolutely RNA Miniprep kit (Stratagene). Reverse transcription-PCR (RT-PCR) was done using iScript One-Step RT-PCR kit with SYBR Green (Bio-Rad Laboratories) in accordance with the manufacturer's instructions. Levels of H4c, H4k/H4j, SMG1, RPL13, HNRPD, RAB35, and H2AFY transcripts were quantified by amplifying a segment of their respective mRNAs with appropriate primer sets (Supplementary Table S1). The reverse transcription reaction was carried out at 50°C for 10 min followed by iTaq hot-start DNA polymerase activation by heating at 95°C for 15 min. Three-step cycling was done: denaturation of 15 s at 95°C, annealing of 30 s at 55°C, and extension of 30 s at 72°C for 45 cycles. All gene expression levels were normalized by parallel amplification and quantification of mRNA levels for the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase mRNA as an endogenous reference with the following primers: 5′-GAGTCAACGGATTTGGTCGT-3′ and 5′-GAGGTCAATGAGGGGTCAT-3′.

Equal numbers of cells were lysed with radioimmunoprecipitation assay buffer [50 mmol/L Tris-HCl (pH 7.4), 1% NP40, 0.25% deoxycholate, 1 mmol/L EDTA] with 1× complete mini protease inhibitor cocktail (Roche Diagnostics) for 30 min at 4°C followed by 15-s sonication pulses (Branson Sonifier 150 at 3 W). Cell lysates were then centrifuged at 14,000 rpm (12,000 × g) for 15 min and the supernatants were removed and combined with lithium dodecyl sulfate sample loading buffer (Invitrogen). SDS-PAGE was done with the Invitrogen NuPAGE system using 4% to 12% Bis-Tris gels and MES running buffer (Invitrogen). Electrophoresis was carried out at 200 V for 45 min, and gel contents were transferred at 30 V for 1 h to a 0.2 μm nitrocellulose membrane. Membranes were then blocked with 5% bovine serum albumin for 1 h at 4°C and probed with histone H4 or glyceraldehyde-3-phosphate dehydrogenase (Abcam) primary antibodies. Protein-antibody complexes were then visualized by enhanced chemiluminescence using Amersham ECL system (GE Healthcare) with either anti-rabbit or anti-goat horseradish peroxidase–conjugated secondary antibodies (Santa Cruz Biotechnology).

Female athymic nude mice were purchased from The Scripps Research Institute Division of Animal Resources. Experimental protocols were approved by the Scripps Institutional Animal Welfare Committee. K562 cells were suspended to 50 million cells/mL in Matrigel (BD Biosciences), 0.2 mL of which was s.c. injected into the rear left flank of each mouse (6–8 wk of age). Mice were monitored and tumor sizes were measured daily; tumor volumes were calculated as 0.5 × length × (width)². Tumors were staged for 7 to 14 d to enter growth phase. At this point, compounds were given at a dose of 7.5 mg/kg per injection via the tail vein thrice over a 5-d period.

Pharmacokinetic studies were done in normal BALB/c mice to determine plasma levels of the compounds 1R-Chl and 1S-Chl over time and to calculate their constant of elimination (k_el), half-life (t_1/2), and volume of distribution (V_d). Forty-eight female BALB/c mice were divided into three groups of 16 mice each. Group 1 received 100 μL (100 nmol) of 1 mmol/L 1R-Chl, group 2 received the same amount of 1S-Chl, and group 3 received an equal volume of PBS (control) via i.v. bolus injection. Blood (0.5 mL) was collected from each animal at specific time points (0, 5, 15, and 30 min and 1, 4, 8, and 24 h) in heparinized tubes, then allowed to coagulate at ambient temperature for 24 h, and centrifuged at 3,100 rpm for 10 min to separate the serum from the clot.

Concentrations of 1R-Chl and 1S-Chl in the serum were determined by MS (see below). Each time point had two animals per group. Data collected were plotted as concentration (Y axis) versus time (X axis) and subjected to first-order kinetic analysis to calculate the pharmacokinetic variables. From this graph, t_1/2 was calculated, and from the natural logarithm of the concentration versus time, k_el (slope) and V_d (dose/anti-logarithm Y-intercept) were determined (Y-intercept represents the theoretical concentration at t = 0). V_d for humans was predicted by multiplying the V_d by 70 kg (average human weight).

1R-Chl distribution in the mouse body was analyzed as follows: mice were injected with 500 nmol of 1R-Chl or PBS and euthanized after either 2 or 24 h; organs were fixed in formalin. A small piece from each major organ was weighed, homogenized in PBS buffer, and sonicated by two pulses of 15 s each. 1R-Chl concentration was determined by MS (see below).

A control polyamide-chlorambucil conjugate, with the sequence ImImPyIm-(R)^Chlγ-PyPyPyPy-β-Dp, was used as an internal standard for concentration determination by MS. The molecular mass of this molecule differs from 1R-Chl and 1S-Chl by 79.1 a.m.u. From each sample, a 200 μL serum aliquot was taken to which 2 μL of 100 pmol/μL control polyamide solution were added along with 800 μL of chilled 100% methanol. Samples were vigorously vortexed for 1 min and then incubated for 15 min at 4°C. They were subsequently centrifuged at 12,000 rpm for 10 min, and the supernatant was harvested and concentrated to 50 μL by methanol extraction. Standards with concentrations from 100 to 12,500 fmol/μL of 1R-Chl or 1S-Chl were prepared in serum and extracted by the above procedure. Concentrations were determined by electrospray MS with an Agilent 1100 single quadrupole instrument coupled to an Agilent 1100 liquid chromatography system and a 50 mm × 2.0 mm C18 column for injection and separation.

To investigate the effects of pyrrole-imidazole polyamide-chlorambucil conjugates on K562 cells, we synthesized a small library of constitutional isomers of 1R-Chl, along with its stereoisomer, 1S-Chl (Fig. 2A). All molecules targeted the general DNA sequence 5′-WGNNNW-3′ (where W = A or T; N = A, C, G, or T); the binding site for 1R-Chl is 5′-WGGWGW-3′. Binding affinities for nonalkylating parent molecules were determined via quantitative DNase I footprinting on plasmid inserts containing match sites (Fig. 2A). It has been previously shown that attachment of the chlorambucil moiety does not seem to affect DNA-binding properties (29). Affinities for 1R, 3R, 5R, and 1S have been previously reported (33–35); polyamides 4R and 6R were analyzed on plasmid insert pMFST6 (Supplementary Fig. S1B). All polyamides were found to bind their expected match sites.

The ability of each polyamide-chlorambucil conjugate to alkylate nucleotides adjacent to its match binding sites on synthetic DNA constructs was also evaluated. Alkylation profiles of polyamides 1R-Chl, 3R-Chl, 5R-Chl, and 1S-Chl have been previously reported (33–35); polyamides 4R-Chl and 6R-Chl were analyzed on plasmid insert pMFST6 (Supplementary Fig. S1C). All polyamides alkylated DNA in the nanomolar concentration range. H4c gene-specific DNA alkylation activities were monitored with the polyamide-chlorambucil conjugates on a PCR product derived from the H4c gene. Significantly, only 1R-Chl and 6R-Chl have binding sites within the H4c gene and only these compounds effectively alkylate the H4c PCR product in vitro (Supplementary Fig. S2). 6R-Chl targets the DNA sequence 5′-WGWGCW-3′; the sequence 5′-TGTGCT-3′ is found both adjacent to and overlapping the binding site for 1R-Chl on the top strand of the H4c and H4k/H4j genes (5′-AGGTGT-3′; Fig. 1B).

Polyamide-chlorambucil DNA alkylators were then tested for effects on growth and morphology of K562 cells. Only cells treated with either 1R-Chl or 6R-Chl exhibited growth inhibition at 250 nmol/L; all others were inactive against K562 proliferation at the same concentration. The control polyamide 5R-Chl has EC₅₀ of 4.2 μmol/L, and 3R-Chl, 4R-Chl, and 1S-Chl have EC₅₀s of >10 μmol/L (Fig. 2A and B; Supplementary Fig. S3). This reflects a minimum 16-fold difference in EC₅₀ values for the active polyamide conjugates relative to the controls. 1R-Chl and 6R-Chl also caused an enlargement of cell volume at 250 nmol/L and induced G₂-M cell cycle arrest (Fig. 2C and D). Following treatment with 1R-Chl or 6R-Chl for 3 to 6 days, an increase in the number of apoptotic cells was observed via Annexin V-FITC/propidium iodide fluorescence-activated cell sorting analysis (Fig. 2D). Other polyamide-chlorambucil conjugates had no effect on apoptosis (Supplementary Fig. S4).

LM-PCR (38) was used to determine sites of alkylation by 1R-Chl and 6R-Chl on the H4c gene in K562 cells. Cells were incubated with each of the hairpin polyamide-chlorambucil conjugates for 24 h at a concentration shown to cause growth arrest with 1R-Chl and 6R-Chl (250 nmol/L). After purification of genomic DNA and digestion with the restriction enzyme DraI, the DNA was heated to induce strand breakage at sites of alkylation (29). LM-PCR with H4c gene-specific primers showed that both 1R-Chl and 6R-Chl alkylate the H4c gene in cultured K562 cells. Consistent with in vitro alkylation results, other molecules in the library failed to alkylate the H4c gene in cell culture (Fig. 3).

Quantitative real-time PCR was used to verify the results of expression arrays for several significant genes from K562 microarray studies (Gene Expression Omnibus accession number GSE8832). The expression of histone H4 mRNAs H4c and H4k/H4j were decreased >60% compared with untreated control cells (Fig. 4A). Identical DNA-binding and alkylation sites for 1R-Chl are present in the H4c, H4k, and H4j genes. Western blot analysis for total histone H4 protein in whole-cell lysate shows significant down-regulation of total H4 protein levels after 24 h of incubation with 1R-Chl (Fig. 4B). Surprisingly, several of the most down-regulated and highly significant genes, based on P values and extent of down-regulation in microarray analysis (i.e., SMG1, RPL13, HNRPD, RAB35, and H2AFY), seemed not to be significantly affected by 1R-Chl when analyzed by real-time quantitative RT-PCR (qRT-PCR; Supplementary Fig. S5).

K562 cells were injected into athymic nude mice, and tumors were staged to >100 mm³ (approximately 7–14 days after injection of cells). In a first experiment, mice were i.v. injected with three doses of 1R-Chl, 1S-Chl, or PBS vehicle control (Fig. 5A, left); for a second experiment, mice were injected with 1R-Chl, 6R-Chl, or PBS vehicle (Fig. 5A, right). The molecule 1S-Chl was used as a control polyamide; although it targets the same sequence as 1R-Chl, the opposite stereochemistry of its turn unit results in a greatly reduced ability to bind and alkylate DNA (32, 33). The dosage routine was 7.5 mg/kg per injection and the polyamides were given on treatment days 0, 2, and 5. The treatment dose of 1R-Chl was determined based on the LD₂₀ of tallimustine and brostallicin, both of which have shown severe dose-limiting toxicities in the murine model and during phase II clinical trials (9, 39).

Mice treated with 1R-Chl and 6R-Chl showed immediate growth regression of the tumor xenografts, whereas 1S-Chl and PBS vehicle had no effect on tumor growth (Fig. 5A and B). Importantly, all mice treated with either 1R-Chl or 6R-Chl appeared healthy with minimal weight loss, whereas all mice treated with 1S-Chl and PBS vehicle showed signs of wasting with significant weight loss due to exponential tumor growth. No obvious toxicity was associated with polyamide treatment. Based on the T/C% (tumor volume analysis) and log cell kill (LCK) values (tumor growth delay), 1R-Chl and 6R-Chl treatments are highly effective: according to the National Cancer Institute standards, T/C < 42% and LCK > 0.7 are minimal levels for activity (39). For the two experiments conducted, the T/C% for 1R-Chl–treated mice versus untreated control were 24.4% and 12.52%, and LCK was 1.41 and 1.57, respectively. For 6R-Chl, the T/C% was 11.30% and LCK was 1.63. In contrast, treatment with 1S-Chl gives results outside of the therapeutic range (T/C% = 78.13 and LCK = 0.20).

Twenty-four hours following the final 1R-Chl treatment, K562 xenograft tumors were dissected and their mRNA was isolated for qRT-PCR analysis. Both H4c and H4k/H4j transcripts were observed to be down-regulated in tumors from 1R-Chl–treated mice. Analysis of tumor samples from 1S-Chl–treated and PBS-treated mice revealed no effect on histone H4c/H4k/H4j mRNA levels (Fig. 5C), consistent with in vitro results.

A Kaplan-Meier survival plot for mice treated with 1R-Chl or 6R-Chl versus PBS shows a significant extension in life span even after only one dosing regimen (Fig. 5D). We expect that prolonged administration of these polyamide-chlorambucil conjugates would extend life span even further and will be determined in future studies.

To predict the behavior and likely dosage regimen of the polyamides in humans, pharmacokinetic variables for 1R-Chl and 1S-Chl were determined in BALB/c mice. MS was used to determine polyamide concentration in serum following injection of 1R-Chl or 1S-Chl. Both compounds seemed to exhibit first-order decay kinetics after a brief lag time (Fig. 6A), indicating that the rate of elimination from serum is proportional to the concentration of the compound in serum at a particular time point. From these results, we determined that both 1R-Chl and 1S-Chl have t_1/2 values of ∼2 h. k_el was determined by plotting the natural logarithm of 1R-Chl serum concentration versus time (during the clearance phase) and calculating the slope of the best-fit straight line; k_el was found to be ∼0.45 h⁻¹. The concentration of the compound achieved in blood at t = 0 was determined by extrapolating the line to the Y-intercept. V_d was calculated by dividing the dose injected by this theoretical initial concentration (t₀). 1R-Chl was predicted to have a V_d of ∼0.48 L/kg, corresponding to ∼33.8 L for a 70 kg human (Fig. 6B). Pharmacokinetic variables for 1S-Chl were also determined and are similar to those of 1R-Chl: k_el of ∼0.55 h⁻¹; V_d of ∼0.68 L/kg; V_d of ∼47.62 for a 70 kg human.

The distribution of 1R-Chl in various mouse organs was determined by MS (Fig. 6C). Following a 500 nmol dose, equal amounts of 1R-Chl were detected in the brain, muscle, stomach, liver, heart, and kidney; it is likely that 1R-Chl crosses the blood-brain barrier. Extensive amounts of 1R-Chl were found in the lung, spleen, small intestine, and pancreas after 2 h after injection and at t_1/2, as well as 24 h after injection, at which point 1R-Chl is undetectable in serum (Fig. 6A). These results indicate that 1R-Chl can be readily eliminated from serum and persist in peripheral organs after tissue entry.

In the library of polyamide-chlorambucil conjugates described herein, 1R-Chl and 6R-Chl are the only molecules expected to bind and alkylate within histone H4c/H4k/H4j genes. Sites of alkylation within the H4c gene are on the template strand, at the 5′ ends of the 1R-Chl and 6R-Chl targeted sequences (Figs. 1 and 3; Supplementary Fig. S2). Alkylation of this strand would be predicted to block transcription elongation by RNA polymerase II. Reporter gene assays have shown that alkylation of the template strand produces a more pronounced effect on transcription inhibition than on the noncoding strand (40). Real-time qRT-PCR and Western blot analyses confirm that K562 cells treated with 1R-Chl or 6R-Chl (250 nmol/L for 24 h) exhibit down-regulation of H4c and H4k/H4j transcripts and total histone H4 protein levels. Other polyamide-chlorambucil conjugates do not alkylate H4 genes in vitro or in living cells and neither affect H4c and H4k/H4j mRNA and histone H4 levels nor inhibit K562 cell growth.

Previous studies involving 1R-Chl have shown that down-regulation of the H4c gene in SW620 cells is required for growth inhibition, but changes in H4k/H4j transcription were not observed (32). Further inspection of Affymetrix microarray data for SW620 cells revealed that H4k/H4j genes are not highly expressed in SW620 cells relative to H4c. However, K562 cells have similar H4c and H4k/H4j expression levels as confirmed by real-time qRT-PCR. The primary targets of down-regulation by 1R-Chl in K562 cells were the histone H4 genes H4c and H4j/H4k. The only gene down-regulated by 1R-Chl in both K562 and SW620 cells is H4c, indicating it to be the key target that elicits growth effects. Furthermore, only cultured cell lines that express the histone H4c gene at high levels are affected by 1R-Chl (32, 35).

Polyamides 1R-Chl and 6R-Chl are the first sequence-specific DNA alkylators reported to show little animal toxicity at mid to high mg/kg range with three consecutive injections every other day over 5 days. This dosing regimen is highly effective in blocking the growth of established tumors (Fig. 5). No behavior or significant weight changes were observed during experiments with normal BALB/c or xenograft mice. In comparison, tallimustine and brostallicin, both of which are micromolar DNA alkylators targeting AT-rich sequences, have shown significant animal toxicity at 4 and 0.8 mg/kg, respectively (9, 39).

We have shown here that 1R-Chl possesses desirable pharmacokinetic variables: it has a t_1/2 of ∼2 h, enough time to allow the drug to effectively reach its target while minimizing prolonged presence in the bloodstream, which can result in toxic effects. k_el of 1R-Chl is relatively low (∼0.45 h⁻¹) compared with those of most anticancer drugs, indicating that, should a steady concentration of the drug be necessary to achieve efficacy, a regimen of constant infusion may be considered. The k_el of 1R-Chl is similar to that of capecitabine (0.5 h⁻¹), a nonalkylating anticancer drug used to treat breast cancer. In addition, V_d > 30 for 1R-Chl, indicating that it is very well distributed in the body, an observation that is consistent with the presence of the molecule in the brain, muscle, stomach, liver, heart, and kidney in our biodistribution analysis (Fig. 6C). The pharmacokinetic variables for 1R-Chl are close to those of some commonly used chemotherapeutic drugs, including both alkylating agents and other types of compounds.

The development of the selective p210 Bcr-Abl tyrosine kinase inhibitor Gleevec (STI 571, imatinib mesylate; Novartis) as a directed therapeutic for CML has been a major advance in cancer therapy (41, 42). However, many Gleevec patients acquire resistance mutations to the drug or acquisition of Bcr-Abl–independent genetic abnormalities during the course of treatment (43–46). Thus, development of additional therapeutic approaches for CML and other malignancies is worthwhile. Here, we describe two sequence-specific DNA alkylators, 1R-Chl and 6R-Chl, which are capable of inhibiting K562 CML cell growth both in vitro and in vivo. 1R-Chl and 6R-Chl treatments are also highly effective in blocking K562 xenograft growth, with high-dose tolerance in the murine model. Based on these observations and the finding that 1R-Chl blocks SW620 xenograft growth in nude mice (32), and the growth of two other xenograft models (Calu-1 lung cancer cells and 22Rv1 prostate cancer cells; data not shown), polyamide-chlorambucil conjugates seem to be promising cancer therapeutics. The inhibitory mechanism of 1R-Chl and 6R-Chl in K562 cells is likely to be independent of Bcr-Abl because no significant down-regulation of Bcr-Abl transcripts was observed in either Affymetrix GeneChip analysis or real-time qRT-PCR (data not shown). These results suggest that 1R-Chl maybe an effective treatment alternative for CML or may be useful in combination with Gleevec.

We thank the staff of The Scripps Research Institute Department of Animal Resources, and The Scripps Mass Spectrometry and DNA Microarray facilities for technical assistance.