Specific Inhibition of ICAM-1 Expression Mediated by Gene Targeting with Triplex-forming Oligonucleotides*

Selected sequences in the DNA double helix can be specifically recognized by oligonucleotides via hydrogen bonding interactions. The resulting triple helix can modulate DNA metabolism and especially interfere with transcription in a gene-specific manner. To explore the potential of triplex-forming oligonucleotides (TFOs) as gene repressors, a TFO was designed to target a 16-bp sequence within the third intron of the human intercellular-adhesion molecule-1 (ICAM-1) gene, which plays a key role in initiating inflammation. TFO binding to its ICAM-1 target sequence was characterized in vitro and also demonstrated in cell nuclei with the set-up of a novel magnetic capture assay, which represents a general experimental approach to the detection of specific TFO binding and to the determination of the accessibility of a given genomic DNA locus. In a human keratinocyte cell line (A431), we observed that: (i) the ICAM-1 target sequence in the chromatin context within the nuclei is still available for triplex formation and (ii) TFO inhibits sequence and gene-specific interferon-γ-induced ICAM-1 surface expression. Collectively, the data demonstrate effective and specific inhibition of ICAM-1 expression by TFO treatment and support the view that triplex-mediated gene targeting might be a valuable technique for anti-inflammatory or anticancer strategies.

Intercellular-adhesion molecule 1 (ICAM-1, 1 CD54) is a cell surface molecule that mediates adhesion processes involving leukocytes and other cell types (1). It is a glycosylated transmembrane protein of the immunoglobulin superfamily and possesses five extracellular immunoglobulin-like domains (2,3). Although under physiologic conditions ICAM-1 is expressed on only a few cell types including vascular endothelial cells, it is induced on many other cells by cytokine-like tumor necrosis factor-␣, IL-1, or IFN-␥ (4, 5) but also by exogenous inflammatory stimuli (4,6,7). Binding of up-regulated ICAM-1 to its ligands on leukocytes, the integrins LFA-1 (CD11a/CD18) or Mac-1 (CD11b/CD18), is involved in several immunologic events that are pivotal in the establishment of inflammatory responses, especially in leukocyte extravasation, target cell lysis by cytotoxic T cells, or antigen presentation to T cells (1,8). ICAM-1 contributes to the adhesion between leukocytes and endothelial cells and thus, plays a role in ischemia/reperfusion injury (9). In addition, it may also facilitate the establishment or progression of certain infections, such as rhinovirus infections (1). Thus, the inhibition of ICAM-1 has become an attractive pharmacologic concept. It is being evaluated, using monoclonal anti-ICAM-1 antibodies or ICAM-1 antisense oligonucleotides, for the treatment of inflammatory diseases (e.g. Crohn's disease or psoriasis vulgaris) and in transplantation research for the prolongation of graft survival and reduction of ischemia/reperfusion injury (10 -12).
The specific interference with gene expression has wideranging applications in experimental biology and is an attractive approach to the development of therapeutics. Several genespecific inhibitory strategies have been established. Among the most advanced technologies are antisense oligonucleotides, which operate at a post-transcriptional level via binding to specific sequences in RNAs (13). As an alternative to this approach, gene expression may be modulated at an earlier stage, at the level of transcription, by molecules that specifically interact with the DNA double helix (14). Among them, triplex-forming oligonucleotides (TFOs) bind in the major groove of the double helix via Hoogsteen hydrogen bonds between oligonucleotide bases and purine bases in the DNA target that are already engaged in Watson-Crick hydrogen bonds. Such binding can occur at sequences containing a stretch of pyrimidines on one DNA strand and complementary purines on the other DNA strand (oligopyrimidineoligopurine sequences). Depending on their base composition, TFOs bind in a parallel or antiparallel orientation to the purine-containing strand (15).
A number of studies have reported TFO-associated activities in cell culture experiments including the modulation of gene expression and the induction of genomic modifications at selected sites (16). Triplex-mediated inhibition of transcription may be achieved by interfering with transcription initiation (competition of TFOs with transcription factors at the promoter) or by interfering with transcription elongation (RNA synthesis arrest at triplex structures). The arrest of RNA synthesis in cells has been only recently described, and a limited number of studies are presently available for plasmid-harbored genes (17,18) or integrated foreign sequences (19). In addition to the inhibition of transcription, the possibility of forming specific and stable complexes at selected genomic sites might induce drastic cellular responses such as modifications of cell cycle checkpoints, especially in malignant cells (20).
The demonstration of triplex formation at a targeted chromosomal locus is an important step to confirm the oligonucleo-tide mode of action. Accessibility of chromosomal target sequences for triplex-forming oligonucleotides within the cellular context has been demonstrated in different experimental systems: (i) by direct detection of specific binding of the triplexforming oligonucleotide using either competitive PCR or restriction enzyme protection assays (21) or primer extension assays (22,23) and (ii) by indirect data with the detection of site-specific genomic modifications caused by triplex-forming oligonucleotides in cells (24,25) but also in mice (26). There is, however, only one study demonstrating both chromosomal binding of TFO and an associated biologic activity in cells (19), an observation that was made with a chromosomally integrated foreign gene.
In this study, we present the characterization of a TFO that inhibits ICAM-1 expression in a sequence-and gene-specific manner. In addition, we demonstrate the TFO binding to its target sequence within its nuclear chromosomal organization using a newly established method. This is, to our knowledge, the first demonstration of in situ binding of an inhibiting TFO that targets an endogenous gene. These results raise the possibility that TFOs may be used in cells as high affinity specific DNA ligands, especially for gene-directed arrest of transcription elongation.

EXPERIMENTAL PROCEDURES
Oligodeoxynucleotides-The triplex-forming oligonucleotide, TFOgt, and control oligonucleotides, COgt1 and COgt2, were obtained from Qbiogene (Heidelberg, Germany); sequences are displayed in Fig. 1. Either 3Ј ends (TFOgt and COgt1) or 5Ј ends (COgt2) were modified with 4,5Ј-8-trimethylpsoralen via a six-carbon linker. Ends that were not psoralen-modified were conjugated to biotin (to be used in magnetic capture assays) or to triethyleneglycol (to prevent nuclease-mediated degradation). A 5Ј-fluoresceinated TFOgt was used for microscopy. Oligonucleotides for PCR were obtained from Metabion (Munich, Germany).
UVA Treatment-UVA was applied with a PUVA 200 light arch (Waldmann, Villingen-Schwenningen, Germany) with F8T5 PUVA bulbs, whose emission spectrum is mainly between 315 and 365 nm. Irradiation was done at room temperature through window glass to eliminate traces of UVB. The irradiation dose delivered to the samples was determined by a UV meter (Waldmann), and the fluence was 2 milliwatts/cm 2 at a 19 cm source-to-target distance.
Restriction Enzyme Protection Assay-Specific triplex formation was evaluated in vitro in a restriction enzyme protection assay. The 16-bp ICAM-1 target sequence was included in a 28-bp sequence to create an EcoNI recognition site that overlaps with the 16-bp oligopyrimidineoligopurine ICAM-1 target sequence (see Fig. 2A). This sequence was inserted into a 4.5-kb plasmid that had been constructed from the vectors pCAT-Basic TM and pSV-␤ Gal TM (Promega, Mannheim, Germany). 2 g (0.66 pmol) of plasmid DNA were incubated with increasing amounts of oligonucleotides for 90 min at 37°C in 10 l of a triplexforming buffer (10 mM Tris-HCl, pH 7.5; 10 mM MgCl 2 ; 1 mM spermidine). Reactions were UVA-irradiated (5 J/cm 2 ) to allow photoadduct formation between the psoralen moiety of the oligonucleotide and the ICAM-1 DNA sequence. Plasmid DNA was purified prior to restriction enzyme analysis with EcoNI and EcoRI (New England Biolabs, Beverly, MA) at 37°C overnight. DNA purification (and thus psoralen/UVAmediated cross-linking) was necessary prior to restriction enzyme digest because the triplex-forming buffer interferes with EcoNI enzyme cleavage. The extent of inhibition of EcoNI cleavage by triplex-mediated photoadducts was assessed by gel electrophoresis and quantitated densitometrically (GelDoc system, Bio-Rad).
Mobility Shift Assays-Triplex formation was additionally analyzed in mobility shift assays. A 23-bp double-stranded DNA fragment representing the ICAM-1 target locus containing the 16-bp oligopyrimidine-oligopurine sequence (see Fig. 1) was produced by annealing equimolar amounts of complementary single-stranded oligodeoxynucleotides. Both 3Ј ends were labeled with digoxigenin (DIG Gel Shift kit, Roche Molecular Biochemicals). 30 fmol of labeled target DNA were incubated with 1 M TFOgt or COgt1 in 10 l of triplex-forming buffer. After incubation for 90 min at 37°C, samples were either not irradiated or irradiated with various UVA doses. TFOs bound to the target duplex during irradiation were covalently linked to the duplex via psoralen/ UVA photoreactions, and covalent complexes were assessed as mobility shifts by denaturing gel electrophoresis (16% polyacrylamide gel containing 7 M urea; 10% glycerine in 44.5 mM Tris; 44.5 mM boric acid; 1 mM EDTA, pH 8.3).
Preparation of Naked Genomic DNA for Capture-Genomic DNA isolated from A431 cells was digested with PstI (Roche Molecular Biochemicals) and purified. PstI digestion results in an 825-bp ICAM-1 fragment containing the triplex target sequence. 2 g of genomic DNA were incubated with 10 M of various oligonucleotides (biotinylated or non-biotinylated TFOgt, biotinylated COgt1) in 10-l triplex-forming buffer for 90 min at 37°C. Biotinylated COgt1 and non-biotinylated TFOgt oligonucleotides were used as controls. UVA irradiation (5 J/cm 2 ) was performed to allow photoadduct formation. Oligonucleotide bound to the ICAM-1 target sequence was separated from unbound oligonucleotide by agarose gel electrophoresis because unbound oligonucleotide would compete with target-bound oligonucleotide for streptavidin on magnetic beads. After gel purification, aliquots of each sample were: (i) saved for control of successful gel extraction of the 825-bp ICAM-1 target fragment (named ''before capture'' portion) and (ii) subjected to magnetic separation (named ''after capture'' portion).
Preparation of Isolated Nuclei for Capture-Nuclei were prepared by incubation of 3 ϫ 10 6 A431 cells in a buffer containing 0.32 M sucrose; 10 mM Tris-HCl, pH 7.5; 5 mM MgCl 2 ; and 0.5% Triton X-100 for 10 min on ice. Nuclei were then washed in the same buffer, transferred to the triplex-forming buffer, incubated with 10 M of various oligonucleotides for 90 min at 37°C in a 100-l reaction volume with gentle resuspension to avoid sedimentation of nuclei, and UVA-irradiated (5 J/cm 2 ). The integrity of nuclei was checked microscopically. DNA was then prepared, digested with PstI, gel-isolated, and subjected to further analysis.
Magnetic Capture and PCRs-To reduce unspecific DNA binding to magnetic beads (Dynabeads, Dynal, Hamburg, Germany), beads were preincubated with salmon testes DNA (5 g/l, Sigma) in 2 M LiCl for 1 h at 37°C. 10 g of magnetic beads were then incubated with 2 g of DNA in a 20-l reaction volume containing 2 M LiCl and 2 g/l salmon testes DNA for 30 min at 37°C. Separation was carried out with a magnetic separator (MPC-E, Dynal) for 2 min. To remove any nonbiotinylated DNA fragments from the beads, three washes with 300 l of redistilled water were performed for 5 min at 60°C. Beads were subsequently resuspended in 50 l of redistilled water, and an aliquot was subjected to PCR. Amplification was carried out using primers P1 and P2 (250 nM; P1, 5Ј-GAACTGGCACCCCTCCCCTCTT-3Ј and P2, 5Ј-CCGGGGCCACACCCATCTCAAA-3Ј) specific for a 388-bp portion of the 825-bp PstI ICAM-1 fragment. Reaction conditions were: 1.0 mM MgCl 2 , 1.5 units of Taq polymerase (Peqlab, Erlangen, Germany), 32 cycles with 30 s at 94°C, 30 s at 63°C, 30 s at 72°C. For a more accurate assessment of triplex formation, captured DNA was subjected to semiquantitative real-time PCR based on fluorescence detection of amplified double-stranded DNA in the LightCycler TM system (Roche Molecular Biochemicals). An aliquot of resuspended beads harboring captured DNA was subjected to a 10-cycle conventional PCR with primers P3 and P4 (250 nM; P3, 5Ј-CCAACCTCACCGTGGTGCTGCT-3Ј and P4, 5Ј-CCCACCTTCTCCCTGCTGGCTT-3Ј), which amplify a 220-bp fragment within the 825-bp ICAM-1 fragment. PCR conditions were 1 mM MgCl 2 , 60 s at 94°C, 60 s at 63°C, 90 s at 72°C. PCR amplificates were purified (PCR purification kit, Qiagen, Hilden, Germany) to remove beads because they interfere with fluorescence detection of PCR products. Aliquots were then analyzed by real-time PCR in a LightCycler TM apparatus using the FastStart DNA Master SYBR Green I kit (Roche Molecular Biochemicals) for fluorescence detection of PCR products. PCR conditions were: 500 nM of primers P3 and P4, 2 mM MgCl 2 , 35 cycles with 15 s at 95°C, 10 s at 63°C, 15 s at 72°C, 15 s at 84°C. The last temperature step was implemented to reduce fluorescence produced by primer dimers. Specific amplification of the 220-bp fragment was controlled by melting curve analysis (27) and also by gel electrophoresis.
Cell Culture-The human squameous cell carcinoma-derived cell line A431 was from American Type Culture Collection, Manassas, VA. Cells were cultured in Dulbecco's modified Eagle's medium supplemented with 2 mM L-glutamine, 100 units/ml penicillin, 100 g/ml streptomycin, 1 g/ml amphotericin B (all from Invitrogen), and 10% heat-inactivated fetal calf serum (ccpro, Neustadt, Germany) at 37°C and 5% CO 2 . Cells were used at subconfluency and kept in serum-free medium during experiments. UVA irradiation of cells (2 J/cm 2 ) was done in phosphate-buffered saline at room temperature. After irradiation, cells were provided with fresh serum-free medium and further incubated at 37°C and 5% CO 2 . Non-irradiated cells were removed from the incubator during irradiation procedures but kept from UVA exposure. To increase ICAM-1 transcription, A431 cells were treated by IFN-␥ (500 units/ml; Amersham Biosciences) as indicated.
Transfection of Cells with Oligonucleotides-A431 cells were transfected with oligonucleotides using Superfect TM (cationic-activated dendrimer, Qiagen). Various amounts of TFO were mixed with a fixed amount of Superfect TM (10 l) in a total volume of 400 l of serum-free medium without antibiotics/antimycotics. After a 10-min incubation at room temperature, the transfection mixture was diluted with serumfree medium to 1 ml, added to cells at 70% confluency (ϳ300,000 cells/3-cm well), and cells were incubated for 4 h.
In experiments in which cellular oligonucleotide uptake was investigated, cells were transfected with fluorescein-conjugated TFOgt, and transfection efficiency was assessed (i) by flow cytometry analysis to evaluate the proportion of fluorescent cells and (ii) by fluorescence microscopy (Axioskop, Zeiss, Jena, Germany) to determine the cellular localization of the fluorescence. In experiments aimed at determining the effect of the triplex-forming oligonucleotide on expression of cell surface molecules, cells were transfected with oligonucleotides, washed with phosphate-buffered saline, and irradiated as indicated. After irradiation, phosphate-buffered saline was replaced with serum-free medium, and IFN-␥ was added or not (500 units/ml). Expression of cell surface molecules was analyzed by flow cytometry 16 h after transfection.
Flow Cytometry-ICAM-1 and HLA-DR cell surface expressions were assessed by a one-step staining procedure and subsequent flow cytometry (FACS analysis). Cells were incubated with an fluorescein isothiocyanate-coupled murine anti-human ICAM-1 monoclonal antibody (MedSystems, Vienna, Austria) and/or with a R-phycoerythrin-coupled murine anti-human HLA-DR monoclonal antibody (BD PharMingen). Control stainings for ICAM-1 were performed with an fluorescein isothiocyanate-coupled isotype-matched control monoclonal antibody (mouse IgG1, Beckmann Coulter), and control stainings for HLA-DR were performed with a R-phycoerythrin-coupled isotype-matched control monoclonal antibody (mouse IgG1, BD PharMingen). Stained cells were analyzed in a FACScan TM II flow cytometer (BD PharMingen).
As a parameter for viability, cell membrane permeability was assessed by dye exclusion using 7-amino-actinomycin (BD PharMingen). 150 ng of 7-amino-actinomycin were added to cells along with specific antibodies, and three-color FACS analysis was performed. Typically, less than 10% of dead cells were detected in transfection experiments.
RNA Extraction and Northern Blot Analysis-Total cellular RNA was isolated from A431 cells using the RNA-Clean kit (Hybaid-AGS, Heidelberg, Germany). RNA (10 g of each sample) was size-fractioned and blotted to positively charged nylon membranes using standard procedures. A 1.8-kb EcoNI/EcoRI fragment of the human ICAM-1 cDNA (2) was radiolabeled with [␣-32 P]dATP (Amersham Biosciences) via random hexamer primer extension and used as a hybridization probe. A 0.21-kb PCR product amplified from a human glycerolaldehyde-3-phosphate dehydrogenase (GAPDH) cDNA was labeled and used as a control probe as described previously (6). Hybridization, stringent washes, and exposure of films to hybridized blots was performed as described previously (28). Specific signals were densitometrically quantitated using the GelDoc system and software (Bio-Rad). Signals generated by the ICAM-1 hybridization probe were normalized to the respective GAPDH mRNA signals. Different exposure times were used to ensure that the signal strength of the analyzed bands on the autoradiographs was in the linear range.

RESULTS
Design of Triplex-forming Oligonucleotides Targeted to the ICAM-1 Gene-The ICAM-1 gene sequence was screened for suitable targets for triplex formation. A 16-bp oligopyrimidineoligopurine sequence in the third intron of the ICAM-1 gene was selected as a target (Fig. 1) with a neighboring 5Ј-TpA motif suitable for psoralen/UVA-induced photoadduct generation. An oligonucleotide was designed to form a triple helix at this site: TFOgt contains guanosine and thymidine nucleosides, allowing formation of T⅐AxT and C⅐GxG base triplets (with (⅐) standing for Watson-Crick interactions and (x) standing for reverse Hoogsteen interactions) after binding in an antiparallel orientation with respect to the purine strand of the double helical target sequence. We chose to use psoralen-conjugated oligonucleotides because: (i) they can be photo-induced to become irreversibly bound to their target DNA sequence and (ii) in the absence of photoactivation of the psoralen moiety, such conjugates have also been described to have increased antigene activity in cells as compared with the corresponding unconjugated oligonucleotides (17). For the demonstration of sequencespecific binding activity of the TFOgt oligonucleotide, two control oligodeoxynucleotides were designed, COgt1 and COgt2 (Fig. 1), both composed of (G,T) nucleotides with short G-runs (G 3 or G 2 ) similar to TFOgt.
In Vitro Characterization of Triplex Formation-Specificity and efficacy of triplex formation were examined with a restriction enzyme protection assay that is based on the ability of triplex structures to interfere with restriction enzyme cleavage (29,30). We used a plasmid containing the triplex target sequence overlapping with an EcoNI cleavage site ( Fig. 2A). The plasmid was incubated with oligonucleotides, UVA-irradiated, and digested with EcoNI and EcoRI. Three fragments of 3804-, 453-, and 270-bp length were generated in the absence of oligonucleotides or in the presence of control oligonucleotides ( Fig. 2A). In contrast, a 723-bp fragment composed of the two smaller fragments was obtained with TFOgt, suggesting that the EcoNI site was protected by the TFOgt photoadduct. A 50% inhibition was observed with a 3-10-fold molar excess of TFOgt, and almost complete inhibition was observed at a 30fold molar excess ( Fig. 2A). Inhibition was also observed when a buffer containing an additional 140 mM potassium mimicking more physiologic ionic conditions was used. However, a 6-fold higher amount of TFOgt was required for comparable inhibition (data not shown), suggesting that triplex formation is less efficient in the presence of potassium, as described (31).
Triplex-induced photoproducts were further characterized in mobility shift assays (Fig. 2B). A 23-bp ICAM-1 DNA fragment containing the target sequence was incubated with TFOgt or COgt1. Samples were UVA-irradiated or not and subjected to denaturing gel electrophoresis to characterize triplex-mediated photoadducts (Fig. 2B). Adducts were detected in irradiated samples containing TFOgt. Two shifted species were observed, and the amount of one of them increased with increasing UVA irradiation doses. These results are consistent with the formation of: (i) monoadducts (TFOgt covalently linked to one strand of the double helix; Fig. 2B, species b) and (ii) cross-links (TFOgt linked to both strands of the double helix; Fig. 2B, species a), as described previously (29). Typically, a UVA dose of 5 J/cm 2 produced ϳ60% cross-links and 40% monoadducts. No photoadducts were detected if TFOgt was replaced with the control oligonucleotide COgt1, confirming triplex binding prior to photoadduct formation. Collectively, these data demonstrate the ability of the oligonucleotide TFOgt to interact and photoreact with the 16-bp ICAM-1 target sequence in a sequencespecific manner.  (38). A 23-bp gene segment is depicted that contains a 16-bp oligopyrimidineoligopurine sequence (underlined) and a 5Ј-TpA site suitable for psoralen/UVA-cross-links (boxed). Sequences of the triplex-forming oligonucleotide TFOgt and the control oligonucleotides COgt1 (scrambled) and COgt2 are shown below the ICAM-1 gene sequence. Oligonucleotides are conjugated to psoralen as indicated (Pso).
Demonstration of TFO Binding to the ICAM-1 Gene-As a next step, we tested the ability of TFOs to selectively bind to the ICAM-1 target sequence within the context of genomic DNA. Triplex formation has been detected at different genomic sequences by the use of several methods including restriction enzyme protection assay, quantitative competitive PCR (21), or primer extension (22,23). In this study, restriction enzyme protection assay was not applicable due to the lack of a suitable flanking restriction site at the ICAM-1 genomic target sequence. We have, therefore, established a new assay for the detection of specific TFO binding to genomic DNA. If a triplex is formed by a psoralen-and biotin-conjugated oligonucleotide at its genomic target sequence, the oligonucleotide can be psoralen/UVA-cross-linked with the double helix. After appropriate digestion, a triplex-containing restriction fragment can be captured using a biotin/streptavidin-based mechanism and identified by PCR-amplification of a portion of the captured fragment (see Fig. 3).
This experimental approach was used to evaluate TFOgt binding to the ICAM-1 gene, initially in naked genomic DNA. A PCR product as evidence for triplex formation was found when biotinylated TFOgt was reacted with naked genomic DNA (Fig. 4A). No PCR fragment was detected (i) when UVA irradiation was omitted (data not shown); (ii) when the biotinylated control oligonu- cleotide COgt1 was used, consistent with sequence specificity of triplex formation; or (iii) when non-biotinylated TFOgt was used. The latter finding allows us to exclude nonspecific, biotinindependent binding of the targeted genomic DNA fragment to the magnetic beads during the separation procedure, which would produce false-positive PCR results. When samples that had not been subjected to magnetic capture (before the capture portion) were used for PCR, a signal was observed with all three oligonucleotides (biotinylated TFOgt, biotinylated COgt1, or nonbiotinylated TFOgt), indicating that the 825-bp ICAM-1 fragment was gel-extracted and intact before magnetic separation.
In a further step, the ability of TFOgt to bind to the genomic target in its intact supranucleosomal structure was tested (Fig.  4B). When biotinylated TFOgt was reacted with isolated nuclei of A431 cells, a PCR product was detected. As observed for naked genomic DNA, no PCR product was detected (i) in the absence of UVA irradiation (data not shown) and (ii) in the presence of non-biotinylated oligonucleotide. Surprisingly, PCR signals were detected with biotinylated control oligonucleotides, COgt1 (Fig. 4B) and COgt2 (data not shown). However, these signals were weak when compared with the PCR signal obtained in samples with TFOgt. This observation was made with conventional PCR (Fig. 4B) but was also confirmed by semiquantitative PCR (Fig. 4C). In addition to the 5Ј-TpA motif neighboring the target triplex sequence, the ICAM-1 PstI fragment harbors 11 more 5Ј-TpA motifs. In their vicinity, we did not identify evident sites for triplex formation. At those motifs, cross-links might, thus, be generated by the photoaddition of the psoralen groups independently of triplex formation. These data suggest that TFOgt formed a substantial amount of triple helix dependent cross-links in cell nuclei, indicating the accessibility of the selected ICAM-1 target sequence in the intact supranucleosomal structure.

Triplex-forming Oligonucleotides Can Be Delivered into A431
Cells and Interfere with ICAM-1 Expression-Conditions for an effective delivery of TFOs into A431 cells were established using fluoresceinated TFOgt. When uptake was microscopically monitored 3 h after transfection of TFOgt with a cationic transfection reagent, bright nuclear fluorescence was detected in a large proportion of cells along with a weaker cytoplasmatic fluorescence (data not shown). In addition, FACS analysis revealed cellular fluorescence in 90% of cells exposed to 1 M fluoresceinated TFOgt (data not shown). No cellular uptake could be detected without transfection reagent even with high oligonucleotide concentrations (data not shown).
Subsequent experiments were aimed at studying the effects of triplex oligonucleotides on ICAM-1 expression by flow cytometry analysis. Untreated cells displayed a low ICAM-1 surface expression, which was markedly increased after a 16-h exposure to 500 units/ml IFN-␥ (Fig. 5A), confirming previous observations (32). In the presence of TFOgt, the IFN-␥-mediated ICAM-1 induction was inhibited in a dose-dependent manner (Fig. 5A). The control oligonucleotide COgt1 failed to affect IFN-␥-induced ICAM-1 expression, and so did the second control oligonucleotide, COgt2 (data not shown). The inhibitory effect of TFOgt was also observed at later time points (after up to 27 h of exposure to IFN-␥, data not shown) with the same level of inhibition. In the presented experiments, psoralenconjugated oligonucleotides were used, and UVA irradiation was performed, but interestingly, the same inhibitory activity was observed if UVA was omitted (data not shown).
At the mRNA level (Fig. 6), ICAM-1 mRNA was strongly induced by a 3 h exposure to IFN-␥. This induction was markedly reduced (ϳ60%) by the presence of TFOgt, but not COgt, confirming the concept of a sequence-specific inhibition of ICAM-1 expression. Collectively, these data demonstrate that the triplex-forming oligonucleotide TFOgt efficiently inhibits ICAM-1 expression in a sequence-specific manner.
Triplex-forming Oligonucleotides Do Not Affect the Expression of Another IFN-␥-induced Gene, HLA-DR-As a control for gene-specific inhibition, we assessed the activity of the triplexforming oligonucleotide TFOgt on the expression of a different IFN-␥-inducible gene, HLA-DR. Untreated A431 cells showed a basal HLA-DR surface expression, which was increased after a 16-h exposure to 500 units/ml IFN-␥ (Fig. 5B). Although exposure of cells to TFOgt markedly inhibited ICAM-1 expression, TFOgt did not significantly affect HLA-DR expression, nor did the control oligonucleotide COgt1. These data suggest a genespecific inhibitory effect of TFOgt on ICAM-1 expression and also support the view that the inhibiting effect was not due to any interference with IFN-␥ signaling pathways (33). DISCUSSION This study describes effective inhibition of ICAM-1 cell surface expression induced by an oligonucleotide (TFOgt) designed to form a triplex on a 16-bp oligopyrimidine-oligopurine sequence present in the third intron of the ICAM-1 gene. Importantly, binding of the TFOgt oligonucleotide to the ICAM-1 target region was demonstrated within the chromosomal context in cell nuclei.
We developed a novel experimental approach of general applicability based on magnetic capture of triplex-associated genomic DNA and subsequent PCR detection. This capture assay detects those genomic triple helix structures in which the oligonucleotide is permanently cross-linked to the target sequence via psoralen/UVA. Triplex-forming oligonucleotides that are not cross-linked are removed from the target site during preparatory steps of the assay. Thus, although the capture assay positively identifies cross-linked specific chromosomal triplex structures (and distinguishes them from nonspecific cross-links), it does not detect non-covalent chromosomal triplex structures and thus, it does not allow estimation of the extent of triplex-bound chromosomal target sequences.
The inhibition of ICAM-1 protein expression by the triplexforming oligonucleotide TFOgt was sequence-specific; two control oligonucleotides did not exhibit any activity. Such specific decrease can be caused by: (i) TFOgt binding to the 16-bp target DNA sequence by triplex formation and subsequent inhibition of transcription elongation or (ii) TFOgt binding to the ICAM-1 RNA by duplex formation and subsequent RNase H cleavage of ICAM-1 RNA. Binding to ICAM-1 RNA appears unlikely because we did not identify a matching sequence in the ICAM-1 RNA for the formation of a stable duplex with the TFOgt oligonucleotide or even a portion of it. Additional support for triplex formation comes from the observation that TFOgt was demonstrated to bind the targeted ICAM-1 DNA locus in cell nuclei. This set of data strongly suggests that TFOgt inhibits ICAM-1 expression by a triplex-mediated blockage of transcription elongation.
In our experiments, the low basal ICAM-1 expression on A431 cells was increased using the potent transcriptional inducer IFN-␥ to more sensitively detect ICAM-1 inhibition by TFOgt. We were, however, concerned by the fact that an oligonucleotide purely composed of (G,T) nucleotides was reported to inhibit IFN-␥-induced ICAM-1 expression (33), probably because this oligonucleotide prevents IFN-␥ from binding to its cell surface receptor. To exclude the possibility that TFOgt, exclusively composed of (G,T), inhibited ICAM-1 expression by interfering with IFN-␥ signaling, we evaluated whether TFOgt interfered with the expression of HLA-DR, which is also induced by IFN-␥. However, HLA-DR expression was neither affected by TFOgt nor by COgt1 and COgt2, all composed of (G,T) nucleotides.
We utilized psoralen-TFO conjugates, which can be crosslinked to the target DNA after photoactivation. Interestingly, the TFOgt oligonucleotide treatment induced the same inhibitory activity in the presence or in the absence of photoactivation. These data suggest two possibilities: (i) triplex-directed photoproducts were frequently formed, but the level of inhibition associated with non-covalent sequence-specific binding of TFOgt oligonucleotide was the same as with triplex-induced cross-links or (ii) the yield of targeted photoadducts was low, possibly because the TFOgt transfection and the timing and/or conditions of the UVA-irradiation were not optimal. Additional data (not shown) support the second possibility: if the UVA dosage used to irradiate oligonucleotide-exposed cell nuclei was reduced from 5 J/cm 2 to 2 J/cm 2 , we detected a strongly reduced amount of photoadduct-associated ICAM-1 DNA by our capture assay, consistent with the view that a 5 J/cm 2 dose was necessary to achieve sufficient TFOgt photoaddition (see also Fig. 2B for demonstration of UVA dose dependence of cross-link formation in experiments with naked DNA). Since cell cultures were irradiated with a low UVA dose (2 J/cm 2 ) to limit cell mortality, such conditions may have led to the formation of only few intracellular photoproducts. Similar results (lack of effect of photoactivation on cellular antigene activity) have been reported previously (24,34), but in these studies, direct measurements of third strand-directed photoadducts at the target locus were not provided, precluding the distinction between the two possibilities proposed above. Nevertheless, the TFOgt-mediated inhibition of ICAM-1 expression demonstrates that triplex-forming oligonucleotides capable of high affinity binding can modulate gene expression even without the generation of covalent modifications.
We report that ICAM-1 knock-down is likely by the physical arrest of RNA synthesis caused by the formation of a stable triplex at the DNA target locus. The processing of such stable complexes remains to be fully determined, as well as the cellular responses associated with transcription arrest, for example, the possibility to act as a poison for transcription-coupled nucleotide excision repair (20).
Besides monoclonal antibodies and antisense oligonucleotides, triplex-forming oligonucleotides (also called antigene oligonucleotides) represent an alternative approach to the selective inhibition of a specific gene by targeting the gene itself with the potential to generate longer lasting pharmacologic effects. We anticipate that a substantial increase in the intracellular efficacy of triplex-induced effects will be achieved (i) by chemical modifications of the oligonucleotide that could improve triplex stability (for example, phosphoramidate oligonucleotides (19,25) or (ii) by delivery methods enhancing cellular uptake after systemic or topical applications of the oligonucleotide, still a major issue in oligonucleotide-based developments (1,13,35,36). Of note, the topical delivery to the skin of oligonucleotides using suitable vehicles, e.g. liposomal formulations, has become feasible (36) and may constitute an attractive anti-ICAM strategy in inflammatory dermatoses in which ICAM-1 is considered pathophysiologically relevant (37), e.g. psoriasis. The present findings support the possibility that triplex-forming oligonucleotides together with other high affinity specific DNA-binding molecules, such as peptide nucleic acids and polyamides (16), may prove useful for the development of therapeutics and as tools for further understanding of DNA-associated functions in the cellular context.