Pyrrole-imidazole (Py–Im) polyamides are a class of programmable DNA minor groove binders capable of modulating the activity of DNA-binding proteins and affecting changes in gene expression. Estrogen receptor alpha (ERα) is a ligand-activated hormone receptor that binds as a homodimer to estrogen response elements (ERE) and is a driving oncogene in a majority of breast cancers. We tested a selection of structurally similar Py–Im polyamides with differing DNA sequence specificity for activity against 17β-estadiol (E2)–induced transcription and cytotoxicity in ERα positive, E2-stimulated T47DKBluc cells, which express luciferase under ERα control. The most active polyamide targeted the sequence 5′-WGGWCW-3′ (W = A or T), which is the canonical ERE half site. Whole transcriptome analysis using RNA-Seq revealed that treatment of E2-stimulated breast cancer cells with this polyamide reduced the effects of E2 on the majority of those most strongly affected by E2 but had much less effect on the majority of E2-induced transcripts. In vivo, this polyamide circulated at detectable levels following subcutaneous injection and reduced levels of ER-driven luciferase expression in xenografted tumors in mice after subcutaneous compound administration without significant host toxicity. Mol Cancer Ther; 12(5); 675–84. ©2013 AACR.
Estrogen receptor-alpha (ERα) is a member of the nuclear hormone receptor family of transcription factors and is active in a majority of breast adenocarcinomas (1, 2). Breast tumors that express ERα and are sensitive to circulating estrogens respond to therapeutics that modulate ERα activity (3). Such therapeutics include tamoxifen, a selective ER modulator that acts as a weak agonist/antagonist by binding to the ERα ligand-binding pocket and the aromatase inhibitors that inhibit synthesis of E2 (3). A different strategy for modulation of ERα activity is inhibition of the ERα–ERE interface by a DNA-binding molecule.
Pyrrole-imidazole (Py–Im) polyamides are a class of synthetic, minor groove-binding ligands inspired by the natural product distamycin A (4, 5). Py–Im polyamides are oligomers of aromatic amino acids linked in series to fold in an antiparallel fashion when bound in the minor groove of DNA (4, 5). Sequence specificity is programmed through side-by-side pairings of the Py and Im subunits that recognize differences in the shape and hydrogen bonding pattern presented by the edges of the Watson–Crick base pairs in the floor of the minor groove (6, 7). Binding specificity has been extensively characterized by DNAse I footprinting titrations and other methods. An Im:Py pair preferentially recognizes G:C; Py:Im prefers C:G, and Py:Py is degenerate for A:T and T:A (6, 7). Py–Im polyamide binding in the minor groove also induces allosteric changes to the major groove (8, 9) and binding affinity is sufficient to modulate the binding of DNA-binding proteins (8–11).
In cell culture, selected polyamides have been used to modulate expression of genes induced by testosterone (11), TNF-α (12), hypoxia (13, 14), and dexamethasone (10). The mechanisms by which polyamides affect gene expression changes in cell culture is still not well understood and may involve direct effects on multiple DNA-dependent processes including transcription factor occupancy, chromatin structure, RNA polymerases, and DNA replication (15). The pharmacokinetics and toxicity of a number of polyamides after intravenous and subcutaneous injection in mice and rats have been described (16–18). In mice, a selected polyamide was reported to induce changes in TGF-β expression in kidney glomeruli, and a fluorescent analog of this polyamide was observed in kidney glomeruli after tail vein injection in rats (19). Gene expression changes have also been observed in tumor xenografts in immune-compromised mice treated with a hairpin-polyamide (20).
In this study, our goal was to identify a Py–Im polyamide capable of affecting E2-stimulated gene expression in breast cancer cells and characterize its activity in cell culture and in tumor xenografts. To do this, we drew from an earlier study that reported a polyamide targeted to the estrogen response element (ERE) consensus half site 5′-WGGWCW-3′ inhibited ERα–binding to DNA in cell-free systems (21). The DNA-binding affinity and specificity of the ERE-targeted polyamide was characterized in this and other studies (21, 22). Since those publications, we have improved the nuclear uptake of polyamides via modification of the C-terminus (23). We have also shown that polyamides are bioavailable after intravenous and subcutaneous injection in mice (20, 24). We then decided to reexamine the activity of polyamides capable of disrupting ERE-driven gene expression for use in vivo. We have screened a focused library of polyamides for cytotoxicity and inhibition of luciferase activity using the breast cancer cell line T47D-KBluc that expresses luciferase under the control of 3 tandem, canonical EREs (25). The most active polyamide identified, which targets the consensus ERE, was further evaluated and showed a partial suppression of E2-stimulated gene expression in cell culture. This polyamide circulated in mouse serum after subcutaneous injection and showed activity against E2-induced luciferase expression in T47D-KBluc tumor xenografts in mice with minimal host toxicity. A fluorescent analog of this polyamide distributed widely in both tumor and mouse tissue after subcutaneous injection.
Materials and Methods
Polyamide synthesis and characterization
The polyamides 1 to 5 were synthesized following previously published solid phase synthesis protocols (26). Compound purities were confirmed by analytic high-performance liquid chromatography (HPLC) and matrix-assisted laser desorption/ionization–time-of-flight (MALDI-TOF) mass spectrometry. Melting temperature analysis was conducted on a Varian Cary 100 spectrophotometer with temperature control. Oligonucleotides (IDTDNA) were dissolved in 10 mmol/L sodium cacodylate, 10 mmol/L KCl, 10 mmol/L MgCl2, and 5 mmol/L CaCl2 at pH 7.0 at a concentration of 2 μmol/L. Polyamides were added to oligo solution to a final concentration of 4 μmol/L in 0.1% dimethyl sulfoxide (DMSO). Oligonucleotides were annealed from 25°C to 90°C and then back to 25°C at 5°C/min. Subsequently, the temperature was elevated at a rate 0.5°C/min between 25°C and 90°C. Melting temperatures are defined as a maximum of the first derivative of absorbance at 260 nm over the range of temperatures.
Cell culture and imaging
Cell lines used were purchased directly from American Type Culture Collection (ATCC) and used within 6 months. No subsequent authentications were done by the authors. All experiments were carried out with T47D-KBluc cells (ATCC), unless specifically mentioned otherwise. Cells were grown in RPMI-1640 held at 37°C in 5% CO2. Media was supplemented with 10% FBS and 1% penicillin/streptamycin. Before imaging, cells were plated in 35-mm optical dishes (MatTek) at 5 × 104 cells per dish in the presence of 10 nmol/L E2. Cells were dosed with polyamide for 24 hours. Cells were then washed twice with PBS and imaged on a confocal microscope (Exciter, Zeiss) using a ×63 oil immersion lens in a method previously described. Confocal imaging was conducted following our previously published protocols (27, 28).
Tissue processing for fluorescence imaging
The tissue sections for fluorescent imaging were obtained by fixing the tumors in 10% formaldehyde solution for 24 hours and subsequent cryoprotection in 15% sucrose (24 hours) and 30% sucrose (24 hours). The tumors were frozen in Tissue-Tek O.C.T. (Sakura Finetek) and 50 μm (for T47D-KBluc xenograft) or 10 μm (for other tissues) sections were obtained using a Leica CM 1800 cryotome. Imaging was conducted as described earlier.
Cell toxicity and luciferase assays
T47D-KBluc cells were plated at 3 × 103 cells per well in 96-well plates, incubated in standard growth media containing 10 nmol/L E2 for 48 hours, and then dosed with medium containing 10 nmol/L E2 and between 2 nmol/L and 50 μmol/L polyamides. The cells were then incubated for 96 hours and analyzed using either WST-1 assay (Roche) or luciferase assay system (Promega) according to the manufacturers' instructions.
Gene expression analysis by qRT-PCR
Cells were plated in 12 well-plates at 1.1 × 105 cells/well and incubated in the growth medium supplemented with 10 nmol/L E2 for 24 hours. Afterward, medium was exchanged with the growth medium supplemented with polyamides and 10 nmol/L E2. Quantitative real-time PCR (qRT-PCR) has been conducted according to previously established protocols (3–6). Confirmation of inhibition of TFF1 expression by polyamides 1 to 4 was carried out and qRT-PCR was conducted following the same timeline as cell toxicity and luciferase assays. Gene expression was normalized against GUSB as housekeeping gene. All primers yielded single amplicons as determined by both melting denaturation analysis and agarose gel electrophoresis. The following primer pairs were used. GUSB: forward 5′-CTC ATT TGG AAT TTT GCC GAT T-3′; reverse 5′-CCC AGT GAA GAT CCC CTT TTT A-3′. DOK7: forward 5′-GAC AAG TCG GAG CGT ATC AAG-3′; reverse 5′-ATG TCC TCT AGC GTC AGG CT-3′. WT1: forward 5′-CAC AGC ACA GGG TAC GAG AG-3′; reverse 5′-CAA GAG TCG GGG CTA CTC CA-3. TGFB2: forward 5′-CAG CAC ACT CGA TAT GGA CCA-3′; reverse 5′-CCT CGG GCT CAG GAT AGT CT-3′.
Chromatin immunoprecipitation experiments
T47D-KBluc cells were plated into 500-cm2 plates and grown in RPMI-1640 with 10% FBS until 75% confluence was reached. Plates were washed with RPMI-1640 with charcoal-treated 10% FBS and then the media replaced with RPMI-1640 with charcoal-treated 10% FBS with 2 μmol/L polyamide 1 and incubated for 48 hours. Plates were then treated with 10 nmol/L E2 or vehicle for 45 minutes. Cross-linked chromatin was obtained using the 2-step cross-linking methods previously described (29). Chromatin was isolated and sheared. Antibodies to ERα (AC-066-100; Diagenode) were used to immunoprecipitate ERα–bound DNA fragments. Cross-links were reversed and PCRs using primers targeted to the regions of interest were used to assess enrichment of bound fragments as compared with negative controls. TFF1 promoter: forward 5′-TCA GAT CCC TCA GCC AAG AT-3′; reverse 5′-TGG TCA AGC TAC ATG GAA GG-3′. Negative loci control: forward 5′-AAA GAC AAC AGT CCT GGA AAC A-3′; reverse 5′-AAA AAT TGC TCA TTG GAG ACC-3′.
Circulation and toxicity in vivo
All animal experiments were carried out according to approved Institutional Animal Care and Use Committee protocols at the California Institute of Technology (Pasadena, CA). Circulation studies were done as previously described (30). Briefly, 120 nmol of polyamide 1 was injected subcutaneously into the right flank of 4 female C57BL/6 mice in a total of 200 μL of a 20% DMSO/PBS vehicle. Blood was collected retroorbitally at serial time points. Serum was treated with methanol, analyzed via HPLC, and quantified against a standard curve of concentration versus peak area, all as previously described to determine approximate serum concentrations (24). For toxicity studies, 5 female C57BL/6 mice were injected with 20 nmol of polyamide 1 in a total of 200 μL of a 20% DMSO/PBS vehicle on days 1, 3, 5, 8, 10, 12, and then with 30 nmol on days 15, 17, 19, 22, 24, and 26 and were weighed before each treatment day. Mice were euthanized if weight loss was more than 15% of initial body weight, if dehydration was more than 10%, or moribund behavior was observed. None were observed in this experiment.
Engraftment of T47D-KBluc.
Experiments were carried out in appropriately shaved female NSG mice (JAX) between 8 and 12 weeks of age. Cells were injected into the left flank area of the animals as suspensions of 5.0 × 106 mL−1 in 50% RPMI-1640 growth medium and 50% Matrigel, 200 μL per injection. Mice also received a subcutaneous E2 pellet (0.72 mg, 60-day slow release; Innovative Research of America) implanted into the right flank on the day of engraftment.
Treatment and tumor monitoring.
Mice were treated with either 25 nmol of polyamide 1 or 50 nmol of polyamide 5. For the short-term and fluorescent imaging studies, they were treated for 8 days after engraftment, every second day for a total of 4 injections. For long-term treatment, injections started 16 days after engraftment and were continued twice a week for the following 4 weeks. Imaging was accomplished using the IVIS Imaging System (Caliper). The animals were anesthetized with 2% to 3% isoflurane and injected intraperitoneally with 150 μL of RediJect d-luciferin (Caliper) and subsequently transferred to the imaging chamber, whereas isoflurane levels were reduced to 1% to 2.5%. The floor of the imager was heated to +37°C to avoid animal hypothermia. Breathing frequency was monitored and not allowed to drop below 1 per second, adjusting the isoflurane levels accordingly at all times.
Endpoint criteria and euthanasia.
Animal endpoint criteria encompassed weight loss of more than 15%, restriction of motoric function by the engrafted tumor, dehydration of more than 10%, and moribund behavior. Where appropriate, the animals were euthanized by asphyxiation in a CO2 chamber.
Tumor tissue harvest.
Animals were resected and tumors excised using standard forceps, scissors and surgical blades. The tumors were weighed immediately afterward. For studies with fluorescein isothiocyanate (FITC)-conjugate 5, resected tumor tissue was homogenized via blunt force and then pushed through a microfilter to achieve single cell suspensions, which were plated on glass microscopy slides for 6 hours before imaging using a Zeiss Exciter fluorescence confocal microscope.
RNA-Seq sample preparation and data processing
Cells for gene expression analysis were plated in 10-cm diameter dishes at 1.1 × 106 cells per dish and incubated in the growth medium supplemented with 10 nmol/L E2 for 24 hours. Afterward, medium was exchanged with the growth medium supplemented with polyamides and 10 nmol/L E2 and incubated for 48 hours in 5% CO2 and 37°C. The RNA was then harvested using an RNEasy Kit (Qiagen). Subsequently, a Riboguard RNAse inhibitor was added and samples were treated with TurboDNA Free DNAse (Ambion), according to manufacturers' instructions. RNA-Seq libraries were prepared using standard Illumina reagents and protocols. Single read sequencing with the read length of 50 nucleotides were conducted on the Illumina HiSeq2000 sequencer, following manufacturers' instructions, producing 35 to 50 million reads per library. Sequencing data were mapped against the combined human (hg19) transcriptome, using the Bowtie program package 0.12.7 (31) and the refseq annotation. The open access processing package Cuffdiff was used to calculated differential gene expression. Inter-replicate statistical significance was calculated with the DEseq module (32).
Design of polyamides
We synthesized 4 8-ring hairpin Py–Im polyamides to screen for activity against E2-stimulated gene expression (Fig. 1 and Supplementary Fig. S1). Polyamide 1 targets 5′-WGGWCW-3′, which is the half site ERE consensus. Polyamide 2 was previously reported to inhibit a subset of dihydrotestosterone-induced gene expression in cultured prostate cancer cells (11). Polyamide 3 was recently characterized in cultured lung cancer cells and used to partially abrogate TNF-stimulated transcription (12). Polyamide 4 targets the sequence 5′-WGWCGW-3′. Polyamides 5 and 6 are FITC-conjugated analogs of polyamides 1 and 2, respectively, used to visualize cellular uptake and distribution in this study.
Evaluation of binding of polyamides to an ERE half site by DNA thermal denaturation assays
Polyamides 1 to 4 were incubated with duplex DNA 5′-CGATGGTCAAGC-3′, which contains an ERE half site consensus and melting temperatures measured (Fig. 2A). Duplex stabilization was greatest for polyamide 1, a polyamide that was predicted to bind this sequence based on established Py–Im polyamide pairing rules (6, 7). The other polyamides showed less stabilization of this duplex.
Luciferase activity and cytotoxicity in T47D-KBluc cells
The ERα–positive cell line T47D-KBluc expresses luciferase under the control of 3 tandem repeats of the sequence 5′-AGGTCACTTGACCT-3′ (25), which is the consensus sequence for the ERα–DNA homodimer (Fig. 2B). T47D-KBluc cells were grown in 10% FBS/RPMI-1640 media with 10 nmol/L E2 for 48 hours. Then, media was replenished with varying concentrations of polyamides 1 to 4 for 96 hours. An extended incubation time with E2 was used to approximate the in vivo condition of continued E2 circulation. Cell proliferation and viability was assayed using WST-1 (Roche) and luciferase output was measured (Fig. 2C). Both luciferase output and proliferation were affected most by treatment with polyamide 1 (IC50 0.47 μmol/L for viability, 0.14 μmol/L for luciferase suppression) and least by polyamide 3 (IC50 > 2.5 and 1.5 μmol/L, respectively). The representative data for luciferase and WST-1 assay are shown in Supplementary Fig. S2. We identified TFF1 as one of the most highly induced transcripts by E2 based on published reports (33). The effects of polyamides 1 to 4 on E2-stimulated TFF1 expression were measured to validate the luciferase screen. Polyamide 1 was again found most potent although polyamides 2 and 4 showed significant inhibition of TFF1 as well (Fig. 2D). Inhibition of TFF1 mRNA by polyamide 1 is dose responsive (Supplementary Fig. S3). In addition, polyamide 1 shows significantly less toxicity to LNCaP, U251, and A549 cell lines (Supplementary Fig. S4), which have low expression of ERα (34–37). Chromatin immunoprecipitation of ERα at the TFF1 promoter after E2 stimulation of cells pretreated with 1 showed reduced occupancy as compared with vehicle-treated cells (Supplementary Fig. S5).
Genome-wide polyamide effects on E2-induced gene expression
Effects of hairpin polyamide 1 at 0.3 and 1 μmol/L on the transcriptome of E2-induced cells were measured using RNA-Seq. Reads were mapped using Hg19 reference human genome and data were analyzed using the Bowtie and CuffDiff packages (38). Only the genes with fragments per kilobase of exon per million fragments mapped (FPKM) ≥ 20 and at least 2-fold change in gene expression upon treatment with either polyamide 1 or E2 were used in the analysis (Supplementary Table S1). Among those genes, at 1.0 μmol/L, polyamide 1 affected expression of 346 genes (0.7% of total) at least 2-fold as compared with E2-treated control. Of these genes, an equal number of genes were up- and downregulated (173 in each case). At the lower concentration of 0.3 μmol/L, expression of 127 genes (0.3% of total) was affected at least 2-fold, and a majority of these genes (77 vs. 50) were downregulated. At the same threshold, E2 upregulated 1,003 genes (2.0%; Fig. 3A) and downregulated 575 genes (1.2%; Fig. 3B). A fraction of expression changes induced by E2 were reversed by polyamide 1 (Supplementary Table S2), and this fraction was greater for E2-repressed genes. Among E2-upregulated genes, 43 (4.3%) were repressed by polyamide 1 at least 2-fold at 1.0 μmol/L. Among those 575 genes that were downregulated by E2, 95 (16.5%) were derepressed by 1 at 1.0 μmol/L at least 2-fold (Fig. 3A and B). Overall, of the 346 genes affected by polyamide 1 at 1.0 μmol/L, 138 (39.9%) represent genes whose up- or downregulation by E2 was abrogated by polyamide treatment. Genes whose expression was affected by polyamide 1 at a lower concentration (0.3 μmol/L) were largely a subset of the genes affected at 1.0 μmol/L, 103 of which (81.1%) were affected by polyamide 1 at both concentrations.
Further analysis was conducted using Euclidian distance clustering with complete linkage (Fig. 3C). Interestingly, while the majority of E2-affected genes are not affected by polyamide 1, out of the top 50 genes most strongly affected by E2, 28 (56%) are inhibited at least 2-fold and 38 of 50 (78%) genes are inhibited at least 1.5-fold by polyamide 1 (Fig. 3D).
Five transcripts were selected for verification by qRT-PCR and all 5 showed good reproducibility of the expression changes seen by RNA-Seq (Fig. 4). Four were upregulated by E2 (AREG, DOK7, TFF1, and WT1) and 1 downregulated by E2 (TGFB2).
Circulation and toxicity of polyamide 1 in mice
To assess serum concentrations of 1 after subcutaneous injection, 4 female C57BL/6 mice were injected subcutaneously into the left flank with 120 nmol of polyamide 1 in a 200 μL 20% DMSO/PBS vehicle. Serial serum samples were taken via retroorbital draw and processed by methods previously described (30). Polyamide 1 was detectable in serum for up to 24 hours after injections, and reached a maximum concentration of 3 μmol/L at 6 hours after injection (Supplementary Fig. S6A). Toxicity after repeated injections of 1 was assessed by daily weights and visual inspection of treated mice. Five female C57BL/6 mice were injected with 20 nmol of polyamide 1 subcutaneously to the left flank 3 times a week for 2 weeks without measurable weight loss. The regimen was then increased to 30 nmol for 2 weeks, again without measurable weight loss or changes in animal behavior (Supplementary Fig. S6B).
Effects on ERα–mediated transcription in T47D-KBluc tumor bearing mice after short-term treatment
To measure the efficacy of polyamide 1 in vivo against E2-induced transcription, T47D-KBluc cells were engrafted into female nonobese diabetic/severe combined immunodeficient (NOD/SCID)-gamma (NSG) immunocompromised mice supplemented with a slow-release subcutaneous E2 pellet in the right flank to facilitate E2-induced growth. After 1 week of growth, mice were imaged using the IVIS Imaging System (Caliper) and stratified into groups of 12 mice each for vehicle and polyamide treatment. Polyamide 1 (25 nmol) in 200 μL 20% DMSO/PBS was injected subcutaneously into the left shoulder every other day for a total of 4 injections. Vehicle-treated mice received 20% DMSO/PBS alone. After 3 injections, mice were reimaged. Luciferase output increased an average of 8-fold for the vehicle-treated mice and 3-fold for the mice treated with polyamide 1 (Fig. 5A). Mice were euthanized the day following the fourth injection for tumor resection. Tumors from vehicle-treated mice were 71 ± 12 mg and tumors from polyamide-treated mice 55 ± 11 mg (Fig. 5B), which does not explain the differences seen in luciferase expression. Representative images of mice treated with polyamide 1 or vehicle at day 6 are shown (Fig. 5C).
Effects on ERα–mediated transcription in T47D-KBluc tumor-bearing mice after long-term treatment
To investigate the effects of polyamide 1 in tumor-bearing mice after extended treatment, T47D-KBluc cells were again engrafted into female NSG mice supplemented with a subcutaneous E2 pellet in the right flank. Tumors were grown for 9 days before stratification of 5 mice each into polyamide 1 and vehicle treatment groups. Mice were treated with vehicle or 25 nmol of polyamide 1 in 20% DMSO/PBS, subcutaneously into the left shoulder twice a week for a course of 9 injections (25 days), beginning on day 16 after engraftment (Fig. 5D). Treated mice maintained their weights at more than 90% until the final days of treatment when their weights decreased to more than 85% before euthanasia. Luciferase was monitored weekly using the IVIS Imaging System. Luciferase output in the polyamide-treated mice was consistently lower than vehicle-treated mice (Fig. 6E). At the experimental endpoint, tumors from vehicle-treated mice were 165 ± 27 mg and tumors from polyamide-treated mice 128 ± 54 mg.
Tissue distribution of FITC-conjugated polyamide 5 in mice bearing T47D-KBluc xenografts
Py–Im polyamide 5 is a FITC-labeled conjugate of hairpin 1 that was synthesized to evaluate tissue and subcellular localization via fluorescence microscopy. T47D-KBluc cells cultured in vitro and then treated with 5 showed nuclear fluorescence similar to what has been reported in other cell lines (27, 28) treated with FITC-conjugated polyamides in cell culture (Supplementary Fig. S7). An NSG mouse engrafted with T47D-KBluc cells as described in the previous section was treated with polyamide 5 in a manner identical to that of polyamide 1, except at a dose of 50 nmol per injection. After 3 injections, the mouse was euthanized, the tumor resected, and internal organs dissected. Tissue was fixed, cryoprotected, sectioned, and imaged immediately. Fluorescence signal was evenly distributed throughout multiple sections of the tumor xenograft. A representative section is shown (Fig. 6A). High magnification reveals nuclear localization in tumor tissue (Fig. 6B). Sections of cardiac muscle show significant cytoplasmic fluorescence in a fibrous pattern (Fig. 6C). Sections of kidney and liver both show nuclear fluorescence localization, with minimal cytoplasmic fluorescence (Fig. 6D and E) Small bowel epithelia show diffuse cellular fluorescence (Fig. 6F).
To ensure that nuclear fluorescence in the xenografts was not an artifact of the fixation process, we extracted live cells from T47D-KBluc xenografted tumors from mice treated with polyamide 5 as earlier. In this experiment, cells were isolated via filtration and plated on microscope slides, and incubated for 6 hours before imaging. Cells derived from the tumor showed nuclear staining in a pattern similar to that seen in the fixed tumor sections as well as cultured cells treated with polyamide 5 in vitro (Supplementary Fig. S8).
In order for DNA-binding, Py–Im polyamides to be considered for therapeutic application, these molecules must possess favorable pharmacokinetic and pharmacodynamic properties and exert a desired effect in target tissues. In this study, ERα–induced transcription in xenografted breast cancer tumors was the target. A polyamide targeted to the ERE half site 5′-WGGWCW-3′ was identified from a focused screen for activity against ER-mediated transcription and cytotoxicity against ERα–positive breast cancer cells. This polyamide was further tested for its global effects on the transcriptome of E2-induced T47D-KBluc cells. Hairpin polyamide 1 showed limited toxicity and circulated at therapeutic levels in serum after subcutaneous injection. It also showed activity against ER-driven luciferase expression in xenografted tumors in immunocompromised mice. FITC-polyamide conjugate 5 shows widespread localization in body tissues including sections through the xenografted tumor, which reveal nuclear fluorescence.
Suppression of ER-induced gene expression
We screened for both suppression of E2-induced luciferase expression and for antiproliferation by WST-1 assay using T47D-KBluc cells. Polyamide 1 was the most active by both measures, whereas polyamide 3 was inactive by either measure. These molecules differ only at a single atom, which represents the difference between a Py and Im heterocycle. Although polyamides have been shown to have differing uptake properties depending on Py–Im content and sequence (27, 28), the differences in activity in this series is likely not explained by differing uptake efficiency as confocal microscopy of FITC-polyamide conjugates 5 and 6 are similar (Supplementary Fig. S7).
Global effects on the E2-stimulated transcriptome
E2 exerts its effects through direct DNA binding and less frequently extranuclear pathways that do not involve ERα–DNA binding (39). Our genome-wide transcriptome analysis (Fig. 4) revealed that a small fraction of gene expression changes induced by E2 treatment is suppressed by polyamide 1. From a total of 1,003 E2 upregulated genes, only 43 (0.43%) are repressed by polyamide 1 at 1.0 μmol/L, and from a total of 575 E2 downregulated genes, 95 (16.5%) were derepressed. However, among the 50 genes most strongly either induced or repressed by E2, a majority were significantly affected by polyamide 1. Among these top 50, 28 (56%) were downregulated by polyamide 1 (1.0 μmol/L) at least 2-fold. At a lower cutoff of 1.5-fold, 38 (76%) of E2-induced gene expression changes were abrogated by the action of polyamide 1 at 1.0 μmol/L. Many of these strongly E2-responsive genes play important roles in the development of tumors and are therapeutically relevant. Among them is Wilms tumor 1 (WT1), a gene originally identified as a tumor suppressor (40), however, more recently it has become apparent that can also act as an oncogene (41). WT1 expression is detectable in 90% of breast cancers (42) and high levels of WT1 expression are correlated with poor patient survival (43). TFF1 is a predictor for breast cancer patient survival (36). Transforming growth factor-β2 (TGF-β2) was observed among the most strongly E2-repressed genes and was also over 3-fold derepressed by polyamide 1 at 1.0 μmol/L. TGF-β2 is involved in cancer development that is also derepressed by traditional antiestrogens (44). We conclude that polyamide 1 acts in an antiestrogenic fashion among genes that are most potently affected by E2 but is less active for the majority of E2-responsive genes. If the mechanism by which polyamide 1 interferes with estrogen-driven gene expression is through direct interference with ER–DNA interfaces, we would not expect to affect ER-driven transcription at loci where ER signals through a tethering complex (45), such as with Ap1 and Sp1. Indirect interactions between ER and DNA through tethering with other proteins offer a partial explanation for the limited number of ER-driven transcripts affected by polyamide 1.
Most transcripts affected by polyamide 1 are not explained on the basis of E2 antagonism; 295 genes that are either up- or downregulated by polyamide 1, at least 2-fold that are not explained by effects on ER activity. Of these, 164 are upregulated and 131 downregulated by polyamide 1. To further characterize these effects, we used the DAVID functional annotation tool (46, 47). For the upregulated transcripts, enriched biologic processes include those involved with the regulation of apoptosis and cell-death, as well as responses to endogenous and hormone stimuli, whereas downregulated genes suggested that polyamide 1 is involved in regulation of GTPase-mediated signal transduction and protein transport and biosynthesis (Supplementary Table S3). The mechanisms of cell death may include the inhibition of transcription (10, 11, 13, 15), but other DNA-dependent processes may contribute and are an area of current investigation. Whether or not the effects of polyamide 1 on these biologic processes are specific to polyamide 1 or may represent a class-effect is unknown but is also under study.
Polyamide treatment suppresses E2-simulated luciferase expression in vivo
T47D-KBluc cells were chosen as the cell line for our study based on previous work using T47D cells as a model for ERα–positive breast cancer. Both vehicle- and polyamide-treated groups showed an increase in total luciferase expression from the baseline measurement immediately before treatment on day 1. However, on day 6, after 3 sequential injections, this increase was significantly blunted in the polyamide-treated group as compared with vehicle (from ∼8- to 3-fold), suggesting that 1 was able to reach sufficient concentrations in tumor tissue to affect luciferase expression. The approximately 2.5-fold difference in luciferase between polyamide- and vehicle-treated groups, if interpolated to the in vitro data, suggest an approximate concentration of 0.3 μmol/L within the xenograft tissue. Tumor masses did not differ significantly between polyamide- and vehicle-treated mice in this experiment. However, this 6-day experiment may be too brief to adequately assess effects on tumor growth.
Effects on tumor size
Although there are no published reports on the growth of T47D-KBluc xenografts in mice, data from parental T47D xenografts show a slow, linear growth pattern rather than exponential (48). To better assay for antitumor activity of polyamide 1, we conducted similar experiments over a longer period of time. T47D-KBluc xenografted tumors were grown for 2 weeks before initiating treatment, and treatment with polyamide 1 was conducted twice per week for a total of 4 weeks. We observed no significant change in tumor size at the experimental endpoint, although we found a sustained suppression of luciferase output in the polyamide-treated arm as compared with vehicle-treated, consistent with our initial observations. The IC50 for cytotoxicity of polyamide 1 in cell culture is 0.47 μmol/L, which we believe to be higher than the concentrations achieved within the tumor tissues in this study.
Tissue distribution of FITC-conjugate polyamide 5 in mice after repeated subcutaneous injections
Fixed, frozen sections through multiple internal organs harvested from T47D-KBluc engrafted mice treated with polyamide 5 reveal widespread organ distribution of fluorescent signal but with differing patterns of fluorescence between tissues, and with little obvious systemic toxicity. The tumor sections show nuclear fluorescence in a subcellular pattern that is similar to what is observed in cell culture (Fig. 6A). The liver and kidneys also show strong nuclear fluorescence (Fig. 6D and E), whereas sections through the intestinal epithelium and cardiac muscle show predominantly cytoplasmic and both cytoplasmic and nuclear fluorescence, respectively. A difference in the cellular uptake of polyamide–FITC conjugates between cell types has also been observed in vitro (27). Recent work has shown that polyamides can form aggregates in solution (49). Whether polyamide aggregation influences distribution in vivo is unknown. Tissue-specific targeting of small-molecule drugs is an area of current investigation that may become relevant for this class of molecules as additional in vivo experiments are planned.
Polyamide 1 delivered by subcutaneous injection in a simple DMSO/saline vehicle distributed widely in host and tumor tissue and showed adequate bioavailability to affect luciferase expression in xenografted tumor tissue, with acceptable toxicity. Future investigations will include optimization of polyamides for lower systemic toxicity without a compromise in efficacy.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: N.G. Nickols, J.O. Szablowski, J.A. Raskatov, P.B. Dervan
Development of methodology: N.G. Nickols, J.O. Szablowski, B.C. Li, P.B. Dervan
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): N.G. Nickols, J.O. Szablowski, A.E. Hargrove, B.C. Li
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): N.G. Nickols, J.O. Szablowski, A.E. Hargrove, J.A. Raskatov, P.B. Dervan
Writing, review, and/or revision of the manuscript: N.G. Nickols, J.O. Szablowski, A.E. Hargrove, B.C. Li, J.A. Raskatov, P.B. Dervan
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): N.G. Nickols, J.O. Szablowski, P.B. Dervan
Study supervision: P.B. Dervan
The authors thank Drs. Janet Baer, Dr. Karen L. Lencioni, and Gwen E. Williams for helpful discussions and technical assistance with animal experiments. Sequencing was conducted at the Millard and Muriel Jacobs Genetics and Genomics Laboratory at California Institute of Technology.
This work was supported in large part by the NIH Grant GM-51747. N.G. Nickols received support from the Jonsson Cancer Center Foundation at UCLA. J.O. Szablowski and B.C. Li were supported by NIH GM-51747. J.A. Raskatov received postdoctoral support from the Alexander von Humboldt foundation. A.E. Hargrove received postdoctoral support from the California Tobacco-Related Disease Research Program (19FT-0105) and the NIH (NRSA number 1F32CA156833).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.