Changes in Gene Expression Induced by Phosphorothioate Oligodeoxynucleotides (Including G3139) in PC3 Prostate Carcinoma Cells Are Recapitulated at Least in Part by Treatment with Interferon-β and -γ

G3139 is an antisense 18mer phosphorothioate oligodeoxyribonucleotide that is targeted to codons 1–6 of the Bcl-2 mRNA (1). Phosphorothioate oligonucleotides have found extensive use in antisense experiments (2, 3), and, more recently, as therapeutic candidates (4, 5), having progressed as far in the clinical setting as registration Phase III trials in advanced melanoma, myeloma, and chronic lymphocytic leukemia.

Nevertheless, despite the relatively favorable properties of phosphorothioate oligonucleotides, the interpretation of experimental data obtained from the use of these molecules is beset with a number of problems. Many, but probably not all of these problems, stem from their high affinity interactions with heparin binding proteins (6, 7, 8). These interactions strongly suggest that for virtually any cellular phenotype observed after G3139 treatment, the contribution of nonsequence-specific effects must be considered. For example, in T24 bladder and PC3 prostate carcinoma cells, we demonstrated recently (9) that G3139, in addition to down-regulating Bcl-2 protein and mRNA expression, also down-regulated the expression of protein kinase C (PKC)-α protein and mRNA expression but not that of other PKC isoforms.

Subsequently (10), it was demonstrated in PC3 cells that the ability of G3139 to down-regulate bcl-2 expression, to inhibit cell growth in tissue culture, and to cause persistence of reactive oxygen species production in treated cells, could also be recapitulated by sequence variants of G3139, provided that the “bis CpG motif” was preserved. Taken together, it is clear that in addition to the Watson-Crick-based, RNase H-dependent, antisense mechanism of down-regulation of Bcl-2 expression, additional mechanisms exist that contributes greatly to the observed phenotype.

In this work, we demonstrate, by microarray technology experiments, that G3139 (and in fact all of the phosphorothioate oligonucleotides we evaluated, but to a variably lesser extent) are capable of inducing the up-regulation of the expression of mRNA species usually associated with the treatment of cells by IFNs. Furthermore, we also demonstrate that inhibition of cell proliferation, and at least some of the changes in gene expression observed after G3139 treatment of prostate carcinoma cells, are recapitulated by treatment of the cells with IFN-β and -γ. Thus, under the conditions of our experiments, it may not be necessary to evoke an antisense mechanism to entirely account for the observed phenotype in PC3 cells after G3139 treatment.

Mycoplasma-free PC3 cells were obtained from the American Type Culture Collection (Rockville, MD) and were grown in RPMI (Invitrogen, Grand Island, NY), containing 10% (v/v) heat-inactivated (56°C) fetal bovine serum supplemented as described previously (10). Stock cultures were maintained at 37°C in a humidified 5% CO₂ incubator. PC3 cells were not used after the passages 10–12.

mRNA was isolated by standard procedures after treatment of PC3 prostate cancer cells with the appropriate oligonucleotides or IFNs (3 days of continuous treatment; IFN-β: 210 pm; IFN-γ; 700 pm). The corresponding cDNA was synthesized (Superscript Choice Double-Stranded cDNA Synthesis kit; Life Technologies, Inc.). Biotin-labeled cRNA was synthesized from the cDNA by using T7 RNA polymerase and the BioArray High Yield RNA Transcript Labeling kit (Enzo). This cRNA was fragmented at 94°C (pH 8.1), in the presence of Mg²⁺. Hybridization to an Affymetrix U95A Gene Chip was performed for 16 h at 45°C. The hybridized biotinylated-cRNA on the chip was stained with 10 μg/ml streptavidin-phycoerythrin, restained by a biotinylated antistreptavidin monoclonal antibody (mAb; 3 μg/ml), and then again with streptavidin-phycoerythrin. After completion of staining, the chip was scanned by the laser scanning Hewlett Packard G2500A Gene Array Scanner (excitation wavelength = 488 nm).

Algorithms embedded in the Gene Data Expressionist (Basel, CH) software that take all of the signals from the perfect match probe sets and calculate the expression values were used for condensing data. Global normalization was applied (scaling all values of mRNA expression) so that the log mean of the average signal was arbitrarily set to 10³. Each 3-day experiment was performed in triplicate; each 1-day experiment was performed once. A Student’s t test was performed comparing each triplicate set to every other triplicate set. A threshold of significant differential expression (P < 10⁻⁴) was applied. In an additional comparison, a Student’s t test performed comparing each triplicate to the sum of all of the others. In this test a significance threshold P < 10⁻⁵ was applied. In addition, a Kruskal-Wallis test (detects genes differentially expressed in one group compared with one or more of the other groups) was applied for all of the triplicate groups (significance threshold P < 0.01 was applied). We identified 590 probe sets by these methods.

For the 1-day analysis, control (untreated) cells were compared with treatment of cells with G3139 (10 μm) without Lipofectin. Their expression profiles were very similar and were used to generate a t test for comparison to the rest of the groups.

These two were also compared against G4126, G3139, 2009, G4232, and C20, all transfected with Lipofectin. In total, 41 genes were obtained (P < 10⁻⁴). These genes are the ones that are affected in common either by transfection or by the oligonucleotides.

For the 1-day versus 3-day analysis, a Student’s t test was performed (P < 10⁻⁵), and 1303 probe sets were obtained, of which 39 had a >2-fold (arbitrarily chosen) change between the group means. (A principal component analysis selected an additional 68 genes with a strong association that did not fulfill one of the two criteria).

To cluster all of the probe sets, 132 were subtracted from the original 590 as being specific for C20 at 3 days (either up or down-regulated). A total of 458 probe sets remained. We added 41 probe sets from the day 1 analysis and 107 from the day 1 versus day 3 analysis to the total. The total number of probe sets, therefore, after subtraction of all of the overlap, was 558.

A self-organizing map clustering was then performed. We could assign 520 probe sets to one of the following four clusters: (a) expression higher after 1 day than 3 days: 51 probe sets; (b) expression higher after 3 days than 1 day: 58 probe sets; (c) expression higher in oligonucleotide-transfected cells at 3 days versus controls at 3 days (C20, which is included in the analysis, does not cause higher expression and is similar to control): 356 probe sets; (d) same as 3, but higher expression at 1 day versus 3 days: 49 probe sets.

Because the genes obtained from clusters 1 and 2 were altered to approximately the same extent for all of the oligonucleotides, they were felt to be related either to the transfection process, the oligonucleotide backbone, or both, and will not be discussed further.

The anti-bcl-2 mAb was from Dako (Carpinteria, CA). The anti-PKC mAb was from Transduction Laboratories (Lexington, KY) and the anti-PKC-β1, ε, and ζ polyclonal antibodies were from Invitrogen (Carlsbad, CA). The anti-α-tubulin mAb was from Sigma-Aldrich (St. Louis, MO). The anti-p21 mAb was from Santa Cruz Biotechnology (Santa Cruz, CA). FITC-conjugated mouse anti-human mAbs MHC class I (HLA-ABC) and class II (HLA-DR) were from BD PharMingen (San Diego, CA). Lipofectin was from Invitrogen. Human IFN-γ (specific activity = 4.24 × 10⁶ units/mg) was from PBL Biomedical Laboratories (Piscataway, NJ). Human IFN-β-1a (200 million units/mg) was from Biogen, Inc. (Cambridge, MA) and was generously donated by Paul Fisher. Phosphorothioate oligonucleotides were synthesized and purified via standard procedures and kindly supplied by Genta (Berkeley Heights, NJ). A list of the oligomers used in these experiments is presented in Table 1.

Cells were seeded the day before the experiment in six-well plates at a density of 25 × 10⁴ cells per well, to be 60–70% confluent on the day of the experiment. All of the transfections were performed in Opti-MEM (Invitrogen). The incubation time for oligonucleotide/Lipofectin complexes was 5 h. The total incubation time before cell lysis and protein isolation was 72 h at 37°C.

Cells treated with oligonucleotide-lipid complexes were extracted in lysis buffer (10) at 4°C for 1 h. Aliquots of cell extracts, containing 25–50 μg of protein, were resolved by SDS-PAGE, and then transferred to Hybond ECL filter paper (Amersham, Arlington Heights, IL). After treatment with appropriate primary and secondary antibody, enhanced chemiluminescence was performed.

Briefly, 15–20 × 10⁴ cells were seeded in six-well plates and allowed to attach overnight. Cells were then treated with the appropriate concentrations of oligonucleotide complexed to Lipofectin (15 μg/ml) for 5 h 37°C. After 3–7 days of incubation, cells were trypsinized and harvested, and their viability was determined by trypan blue exclusion. Experiments were performed in triplicate, and data are presented as the average ± SD. In no experiment did >10–15% of cells stain with trypan blue.

Cells treated for 72 h with oligonucleotides were harvested after trypsinization, and apoptotic cells were assayed by flow cytometry using the Annexin V-FITC apoptosis detection kit (Becton Dickinson, San Jose, CA). Early apoptotic cells bound to Annexin V-FITC but excluded propidium iodide. Cells in late apoptotic stages were labeled with both Annexin V-FITC and propidium iodide.

Total RNA was isolated from PC3 cells using the RNeasy kit (Qiagen, Valencia, CA). RNA was reverse-transcribed with Platinum Taq (Invitrogen). All of the RT-PCR reactions (50 μl final volume) were performed as follows: 50°C for 35 min, 94°C for 2 min, 37 cycles consisting of 60°C for 30 s, 72°C for 1 min, 94°C for 30 s, and finally 72°C for 10 min. The forward and reverse primers for amplification of bcl-2 were 5′-GGTGCCACCTGTGGTCCACCTG-3′ and 5′-CTTCACTTGTGGCCCAGATAGG-3′, respectively (11). The primers (5′-GAGCTGCGTGTGGCTCCCGAGG-3′, forward; 5′-CGCAGGATGGCATGGGGGAGGGCATACCCC-3′, reverse; Ref. 11) designed to amplify a 246-bp fragment of β-actin were used to normalize for RNA concentration.

For TaqMan PCR analysis, total RNA was reverse transcribed for 1 h at 37°C using random hexanucleotide primers and RAV-2 reverse transcriptase (Amersham Biosciences, Chalfont St. Giles, United Kingdom). Amplification was performed on an ABI Prism 7900HT sequence detection system (Applied Biosystems, Weiterstadt, Germany) using AmpliTaq Gold DNA polymerase. Primers were obtained from Applied Biosystems (Assays-on-Demand program). For normalization of RNA concentrations, TaqMan PCR analysis of the 18S rRNA was used.

Total cellular RNA was isolated using TRIzol Reagent (Life Technologies, Inc.). Twenty-five to 30 μg was resolved on 1.2% agarose gel containing 1.1% formaldehyde and transferred to Hybond-N nylon membranes (Amersham). The membranes were probed for bcl-2, washed, and exposed to X-ray film as described previously (10).

PC3 cells were transfected with either G3139 or G4126 for 5 h in Opti-MEM as described above, or treated continuously with IFN-γ (700 pm) for 2, 3, or 4 days. Then, the cells were incubated with FITC-conjugated anti-MHC class I or class II antibody at 4°C for 30 min and analyzed by flow cytometry.

As has been demonstrated previously (10), G3139 and related oligonucleotides down-regulate the expression of bcl-2 protein in PC3 prostate cancer cells. As demonstrated by Northern and RT-PCR analysis (Fig. 1), the expression of bcl-2 mRNA is also down-regulated. However, bcl-2 protein is predominately membrane-bound. Thus, whereas bcl-2 mRNA is maximally down-regulated 24 h after a 5-h transfection with G3139 (but far less so with control, 2-base mismatch G4126), it took 72 h for bcl-2 protein to achieve maximal down-regulation (10). Furthermore, recovery of mRNA expression to control levels occurred by 3 days.

The expression levels of mRNAs defined by Cluster 3 is shown in Fig. 2. Twenty-four h after treatment, the scaled expression values of these genes in control, untreated cells, and in cells exposed to 10 μm G3139 in the absence of the Lipofection transfection reagent, are approximately identical.

Some of the genes that are found in Cluster 3 are presented in Table 2, where the scores represent fold up-regulation of mRNA expression in oligonucleotide/Lipofectin-treated PC3 cells after 3 days. As can be seen in Fig. 2, the expression of virtually all of these mRNAs was also up-regulated on day 1 by all of the oligomers tested. By day 3, however, up-regulation of mRNA expression in this cluster persisted most strongly for G3139 (Fig. 2), but less so in the order 2009 > 4126 ≫ C20. (It should be noted that 2009 and C20 are both two bases longer than G3139).

These data demonstrate that effects observed at 3 days with G3139 and the other oligonucleotides are not purely the result of transfection, of Lipofectin, or of the oligonucleotide backbone, but rather, alterations in the intensity of gene expression induced by these molecules. Strikingly, as demonstrated in Table 2 (arbitrarily truncated at a score of a 2.3-fold increase in mRNA expression), up-regulation of expression of many of the same genes can also be observed when cells are treated with IFNs, or are exposed to lipopolysaccharide or double-stranded (ds) RNA (12, 13). Note also that what distinguishes G3139-treated cells from those treated with other oligonucleotides is not the identity of the up-regulated genes but rather the persistence of up-regulation of gene expression in cluster 3.

IFN-related mRNAs found to be up-regulated in Cluster 3 of microarray assay include: MxA (human Mx1), as well as MxB, IFI27, IFI35, 9–27 (IFITM-1), 6–16 (IFI16), 2′-5′ oligoadenylate synthetase (OAS) 1, OAS 2, IFI41, the transcription factor ISGF3 (also known as IRF-9) and its subunit STAT1 (which is required by type I IFNs; Ref. 14), and the cyclin-dependent kinase inhibitor p21/Waf1/Cip1. However, we found no evidence either at 1 or 3 days for the up-regulation of either IFN-β or -γ mRNA expression.

Confirmation of the oligonucleotide microarray data were provided by TaqMan real-time RT-PCR. In G3139-treated cells, MxA, ISG15, OAS1, and OAS2 mRNAs are all up-regulated approximately 4–6-fold. As expected, the expression of these mRNAs was also highly up-regulated (30–60-fold) by treatment of cells with either IFN-β or -γ (data not shown).⁶ The relationship between the expression of genes changed by phosphorothioate oligonucleotides (400 nm; 5-h treatment; RNA isolated 67 h later) as compared with that changed by IFN-β and -γ (72 h of continuous treatment before RNA isolation) is shown in the Venn diagram in Fig. 3. There is a significant overlap (338 genes) in the gene expression changes induced by IFN-β and -γ. A change in expression of 138 of these 338 genes is also induced by phosphorothioate oligonucleotides (virtually all are up-regulated). Some of these genes are listed individually in Table 2A. For virtually all of them, it should be noted that induction by IFN-β tends to peak after 1 day, whereas induction by IFN-γ is maximal only after 3 days of continuous treatment. Furthermore, whereas the induction of gene expression can be qualitatively similar for phosphorothioates and IFNs (although it should also be noted that 199 genes induced by IFNs-β and -γ are not induced by phosphorothioates), treatment of cells with IFNs (Table 2A) produces far more dramatic effects on induction of mRNA expression. A similar situation exists with a set of 14 genes of which the expression is up-regulated by IFN-β and by phosphorothioates, but not by IFN-γ. Most of these genes are only minimally induced (4.4-fold maximum induction) by phosphorothiotes (Table 2B), but some, such as IFIT1, vipirin, and OASL are induced, respectively, 24.8, 48.4, and 20.4-fold by IFN-β.

The expression of an additional 14 genes is changed (all are up-regulated) by phosphorothioate oligonucleotides but not by IFNs (Table 2C); these changes for the most part appear relatively minimal. For 2 genes, adrenomedullin and IL24, no increase in protein expression could be observed by Western blotting (data not shown).

Finally, the expression of 11 genes minimally induced by phosphorothioates is also induced by IFN-γ but not by IFN-β (Table 2D).

As demonstrated in Fig. 4, A and B, respectively, IFN-β and -γ treatment of PC3 cells, similar to G3139 treatment, also produced a dramatic down-regulation (maximal 95% for IFN-β at 250 pm) of bcl-2 protein expression (after 3 days). This was also observed in cells treated with IFN-γ, with inhibition (∼80%) found at 700 pm. Bcl-2 mRNA expression, as assessed by RT-PCR, was down-regulated 24 h after treatment, by both IFN-β (data not shown) and -γ (Fig. 5). However, by 3 days after IFN treatment, again as determined by RT-PCR, bcl-2 mRNA levels recover to approximately those found in untreated cells. This is unlike what was observed in G3139-treated cells, in which bcl-2 mRNA levels remain suppressed for as long as 5–6 days.

Treatment of PC3 cells with both IFN-β and -γ also produced significant down-regulation of PKC-α protein expression (∼75%; Fig. 6, B and C, respectively) at the same concentrations that produced maximal bcl-2 protein down-regulation. Again, as with G3139, which also causes down-regulation of PKC-α protein expression (Fig. 6,A), there was no down-regulation of expression of other PKC isoforms, including PKC-β1, -ε, and -ζ (Fig. 5 C)

By 1 day after oligonucleotide transfection, levels of p21/Waf1/Cip1 protein expression appear to be approximately equal to those observed in the control, untreated cells. However, 2 days later, levels of p21 protein expression in untreated cells declined to almost undetectable levels. In contrast, in G3139-treated cells (Fig. 7), the level of p21 protein expression is some 28-fold higher versus untreated cells and 14-fold higher than in G4126 treated cells. A similar pattern of expression of p21 is also seen 3 days after treatment of PC3 cells with IFN-β (250 pm) and IFN-γ (700 pm).

IFN-γ also dramatically up-regulated the cell-surface expression of MHC-I but not MHC-II (data not shown) on PC3 cells, as detected flow cytometrically after fluorescent mAb staining (Fig. 8). Similarly, G3139, but not G4126, up-regulated MHC-I (∼60% increase in mean channel number) but not MHC-II expression versus control, untreated cells, albeit to a much smaller extent than did IFN-γ. Nevertheless, this increase was highly statistically significant, as determined by a Student’s two-sided t test (Fig. 8). The expression of MHC-I mRNA was not up-regulated in the oligonucleotide microarray assay.

Similar to what we observed after G3139 treatment of PC3 cells, continuous treatment of PC3 cells with IFN-β and -γ caused almost complete cessation of cell growth. However, this was associated with minimal loss in cellular viability as judged by trypan blue exclusion for at least 5 days, again similar to what was observed with G3139 treatment (Fig. 9). However (10), cessation of cellular proliferation was not associated with a significant increase in apoptosis, as measured by Annexin V binding. In control, untreated cells, 4.9% of the population was in early apoptosis, with 4.2% in late apoptosis/early necrosis. This increased to 5.9% and 5.8%, respectively, after G4126 treatment and 10.4% and 10.6% with G3139 treatment. After 3 days of continuous treatment with IFN-β (1250 pm) or -γ (700 pm), no more than 0.3% of cells were in early apoptosis, and no more than 8.9% could be found in late apoptosis/early necrosis.

After treatment of PC3 cells with G3139, relatively high levels of reactive oxygen species are produced relative to the low rate of cellular proliferation (10). Nevertheless, unlike after treatment with G3139, IFN-β or -γ- treatment of PC3 cells did not promote the production of high levels of reactive oxygen species relative to the cellular proliferation rate (data not shown).

We then determined, via ELISA assay, whether G3139 was inducing PC3 cells to secrete either type I or II IFNs. The lower limit of detection of the ELISA assays is 12.5 pm for IFN-β and 0.6 pm for IFN-γ. However, we could not detect any IFN in the culture media at concentrations above these lower limits (data not shown). On the other hand, both IFN-β and -γ could easily be detected by ELISA in both treated and untreated PC3 cell lysates (∼2.0 ± 0.3 × 10⁻² and 6 ± 2 × 10⁻⁴ pmol/mg protein for IFN-β and -γ, respectively). However, no statistically significant induction of endogenous IFN protein production was observed by treatment of PC3 cells for 3 days with either G3139 or G4126. In fact, we observed a small decrease (to 1.5 ± 0.1 × 10⁻² pmol/mg protein) in the concentration of IFN-β, but not -γ. IFN-α could be detected in the lysates of untreated cells (∼1.0 ± 0.2 × 10⁻³ pmol/mg protein), but its concentration decreased to ∼4.0 ± 0.2 × 10⁻⁴ pmol/mg protein after treatment with either G4126 or G3139.

We have demonstrated, by oligonucleotide microarray and TaqMan real-time RT-PCR, that the expression of certain cellular genes usually associated with exposure of immunocompetent cells to dsRNA, IFN, and viruses, are also up-regulated by transfection of phosphorothioate oligonucleotides in PC3 prostate carcinoma cells. Furthermore, this up-regulation strongly persists for as long as 3 days after a five-hour treatment with G3139. The up-regulation of the same genes (cluster 3) can also be observed, but at lower amplitude, with all of the other oligonucleotides used in this study. However, it is critical to note that the data presented herein should not be extrapolated to other cell systems, nor should it be uncritically assumed that these mechanisms as described are necessarily clinically relevant.

The molecular basis for the relatively long-term up-regulation of the IFN cascade induced by G3139 is unclear. However, it is possible that up-regulation of IFN-related growth suppressive genes leads to the long-term suppression of cell growth in complete media and loss of ability to clone in soft agar (10). This strongly growth suppressed phenotype might, in turn, be associated with the presence of the “bis-CpG” motif found in G3139 (10), which is not present in any of the other oligomers evaluated in that study.

In murine bone marrow-derived macrophages, the TLR3 receptor regulates the response to naked dsRNA (12, 15), which ultimately leads to activation of nuclear factor κB. However, increased nuclear localization of nuclear factor κB was not observed in electrophoretic gel mobility shift assays 3 days after treatment of PC3 cell with G3139 (data not shown). The TLR4 receptor, on the other hand, mediates the response to lipopolysaccharide (12, 16) and like copolymer of polyinosinic and polycytidylic acids, also leads to nuclear factor κB activation, making it unlikely that either of these two receptors is involved in the changes in gene expression caused by oligonucleotide treatment. In addition, the lack of any effect of naked G3139, even at a 10 μm concentration, indicates that cellular uptake of the oligonucleotide is required. Furthermore, the effects described in this work are independent of TLR9, the CpG receptor (17, 18), as the expression of its mRNA was not detected in PC3 cells in either an oligonucleotide microarray assay or by TaqMan RT-PCR. Rather, it seems likely that significant extraendosomal oligonucleotide internalization, and possibly nuclear localization, must occur.

Human prostate cancer cells are highly sensitive to the antiproliferative effects of IFNs (19). Some of the IFN-related mRNA species up-regulated by phosphorothioate oligodeoxynucleotides in general, but especially by G3139, appear to be highly growth suppressive. For example, IFIT1 mRNA, the transcription product of the 561 gene that is responsible for the translation of p56, was identified by the microarray. Its expression may be strongly induced by dsRNA (as short as 85 bp; Ref. 20) or IFN-β (13). P56 appears to be growth suppressive (14), acting by inhibiting protein synthesis (14). This pathway of protein synthesis inhibition does not seem to require dsRNA, unlike the protein kinase R and 2–5 (A) synthetase pathways, both of which are also activated by IFN.

Although the expression of mRNA for both OAS1 and OAS2 was found to be increased in the microarray assay, the protein product of translation of these mRNAs may be inactive, because it requires dsRNA to polymerize ATP to produce RNase l-activating 2′-5′ oligoadenylates (21). However, no clear source of dsRNA in these experiments exists, and, thus, it cannot be the OAS that is inducing growth suppression.

We have also demonstrated that at least some of the changes in gene expression induced by G3139, including up-regulation of p21/Waf1/Cip1 protein expression and down-regulation of PKC-α and bcl-2 protein expression itself, were also observed after treatment with IFN-β and -γ. Interestingly, however, although the expression of bcl-2 mRNA was down-regulated by IFN-γ at 24 h after treatment, it recovered by 72 h, unlike bcl-2 protein, of which the expression remained low. Thus, it appears likely that the decrease in bcl-2 protein expression in IFN-treated cells is a result of block of translation of the mRNA, an increased rate of degradation, or both. This hypothesis is supported by the fact that G4126-treated cells demonstrate the identical pattern of changes in expression of bcl-2 mRNA with time, yet bcl-2 protein expression is not diminished by 3 days after treatment. However, this is unlike what was observed after G3139 treatment, where at 72 h both bcl-2 protein and mRNA expression remain low. Indeed, we believe that G3139 treatment of PC3 cells is so highly effective at the down-regulation of bcl-2 protein expression precisely because it does so by two distinct mechanisms: an antisense RNase H-dependent mechanism, which is active at the level of the bcl-2 mRNA, and an “IFN-mimetic” mechanism, which appears to be active post-translationally.

We also observed up-regulation of the expression of p21/Waf1/Cip1 mRNA expression in the oligonucleotide microarray analysis after treatment of PC3 cells with G3139. This correlates with the very large increase in p21 protein expression also observed in these cells. It also appears that type I IFNs, similar to G3139, can block the progression of many cell types from the G₁ into the S phase of the cell cycle (19). In DU145 cells, this has been associated previously with an IFN-induced doubling in levels of p21/Waf1/Cip1 protein expression, and presumably at least in part results from the ability of this protein to bind cyclin E-cyclin-dependent kinase 2 and block its activity.

Nevertheless, the mechanism by which IFN-response genes are activated is somewhat unclear. We had originally hypothesized that G3139 was inducing the up-regulation of production of IFN-β or -γ protein, which could then activate IFN receptors in an autocrine/paracrine fashion. This possibility is ruled out by the ELISA data that demonstrate a 50% decrease in intracellular IFN concentration after oligonucleotide treatment. Added to this, there is insufficient IFN-γ in the culture medium (<10 pm) to bind to and activate the cell surface IFN-γ receptor (K_d = 100 pm). Moreover, the oligonucleotides cannot be directly binding to and activating IFN receptors, because treatment of cells with naked oligonucleotides does not recapitulate the observed phenotype.

Teleologically, the origin of the hypothesized “dsRNA-like” or “IFN-mimetic” mechanism observed in this work is also unclear. The only single-stranded DNA genome is found in parvoviridae, which appear to be very inefficient at inducing type I IFNs in mammalian cells (22). Likewise, neither single stranded oligodeoxyribonucleotides nor their phosphorothioate analogs have been known previously to be inducers of IFN or of IFN-related proteins in carcinoma cells (23), as opposed to their induction in immunocompetent cells.

Finally, it should be noted that although the experiments presented above are highly representative, not all of the genes described here were induced under all of the transfection conditions. The transfection process is complex, and differences in particle size, rate of adsorption to the cell surface, rate of cellular endocytosis, and rate of endosomal egress may all contribute to differences in the intranucleotide concentration of oligonucleotide, and hence to the observed phenotype.