Compensatory Stabilization of RII_β Protein, Cell Cycle Deregulation, and Growth Arrest in Colon and Prostate Carcinoma Cells by Antisense-directed Down-Regulation of Protein Kinase A RI_α Protein

Varying the ratio of two isoforms of cAMP² receptor protein, the PKAs type I (PKA-I) and type II (PKA-II), which are distinguished by different regulatory subunits, RI and RII, have been linked to cell growth and differentiation (1, 2). An enhanced expression of RI/PKA-I correlates with active cell growth and cell transformation, whereas a decrease in RI/PKA-I and an increase in RII/PKA-II are related to growth inhibition and differentiation-maturation (1, 2).

RI is the major R subunit of PKA detected in a variety of human cancer cell lines (2). The majority of primary human breast and colon carcinomas examined show an enhanced expression of RI and a higher ratio of PKA-I:PKA-II compared with normal counterparts (3, 4, 5, 6). Importantly, the relative overexpression of the RI_α subunit of PKA was associated with poor prognosis in patients with breast cancer (5). Quantitative PCR analysis of ovarian tumor tissue samples revealed that RI_α mRNA expression is elevated in serous tumors, compared with mucinous, endometroid, or clear cell tumors (7). The levels of RI_α mRNA were elevated in ovarian cancer cells, which are highly tumorigenic, whereas RI_α mRNA was expressed at lower levels in the nontumorigenic ovarian cancer cell lines and minimally expressed in the placental mRNA control (7).

We have demonstrated previously that a single-injection treatment with RI_α antisense results in the suppression of RI_α and PKA-I that precedes tumor growth inhibition (8). Growth inhibition persisted, even after RI_α suppression ceased, as long as PKA-I down-regulation was present. The PKA-I down-regulation was accompanied by a compensatory increase in the level of RII_β protein and PKA-II_β(RII_β-containing PKA-II) holoenzyme (8).

In this report, we demonstrate, using cultured cancer cells, that the loss of RI_α by antisense treatment results in biochemical compensation by RII_β and that this compensation is attributable to an increase in the half-life of RII_β protein without changes in the rate of RII_β protein synthesis or RII_β mRNA levels.

8-N₃-[³²P]cAMP(60 Ci/mmol; 1 Ci, 37 GBq) was obtained from ICN Pharmaceuticals. Pepstatin, antipain, chymostatin, leupeptin, aprotinin, and soybean trypsin inhibitor were from Sigma Chemical Co. Protein A-Sepharose CL-4B was purchased from Pharmacia-LKB. Polyclonal anti-RI_α, anti-RII_α,and anti-RII_β antibody was kindly provided by Professor Sang Dai Park (Seoul National University, Seoul, Korea).

The oligonucleotides used in the present study include the phosphorothioate oligodeoxynucleotide antisense targeted toward human RI_α (the NH₂-terminal codons 8–13, 18 mer; PS-DNA antisense RI_α; Ref. 8), and the second-generation PS-DNA-antisense RI_α of either the RNA-DNA hybrid (HYB0165) or the methylphosphonate (P-ME DNA; HYB0190). HYB0165,5′-[GCGU]GCCTCCTCAC[UGGC]-3′, is a PS-DNA-antisense RI_α containing segments of four 2′-O-methyl-oligoribonucleosides (RNA) at both the 5′ and 3′ends (bracketed). HYB0190, 5′-GCGTGCCTCCTCACTGGC-3′,contains six methylphosphonate oligodeoxynucleotides (underlined) at the center of the PS-DNA-antisense RI_α. The control oligonucleotides for HYB0165 are HYB0295,5′-[GCAU]GCTTCCACAC[AGGC]-3′,a 4-base mismatched (underlined) form of HYB0165, and HYB0674,5′[NNNN]NNNNNNN[NNNN]-3′, a random sequence oligonucleotide,with the bracketed segments representing 2′-O-methyl-RNA,and N = A, T, C, or G. All internucleotide linkages are phosphorothioate. As a control for HYB0190, we used HYB0191,5′-GCGCGCCTCCTCGCTGGC-3′, a two-base mismatched control DNA (two mismatches in bold). These oligonucleotides were kindly provided by Dr. Sudhir Agrawal (Hybridon, Inc.).

HCT-15 (MDR) human colon carcinoma cells were kindly provided by Dr. Dick Camelia, National Cancer Institute. Tumor cells (2 ×10⁶ cells) were inoculated s.c. into the left flank of athymic mice. The RI_α antisense PSODN (8) or MBOs of RNA-DNA hybrid HYB0165 and 4-base mismatched and random sequence control oligos were used. RI_α antisense or control oligos at daily dose(0.1 mg/0.1 ml saline/mouse, daily, five times/week), or saline (0.1 ml/mouse) was injected i.p. into mice when tumor size reached 30–50 mg, 7–10 days after cell inoculation. Tumor volumes were obtained from daily measurement of the longest and shortest diameters and calculation by the formula:

where

At indicated times, animals were sacrificed. Tumors were removed and immediately frozen in liquid N₂ and stored at−80°C until used.

Tumors were homogenized with a Teflon/glass homogenizer in ice-cold buffer 10 (Ref. 8; 20 mm Tris-HCl (pH 7.4),100 mm NaCl, 1% NP40, 0.5% sodium deoxycholate, 5 mm MgCl₂, 0.1 mmpepstatin, 0.1 mm antipain, 0.1 mm chymostatin,0.2 mm leupeptin, 0.4 μg/ml aprotinin, and 0.5 μg/ml soybean trypsin inhibitor filtered through a 0.45-μm-pore membrane)and centrifuged for 5 min in an Eppendorf microcentrifuge at 4°C. The supernatants were used as tumor extracts. The content of R subunits of PKA in tumors was determined by photoaffinity labeling with 8-N₃-[³²P]cAMP, followed by immunoprecipitation with anti-R antibodies as described previously (8).

LS-174T human colon carcinoma cells (a gift from John W. Greiner,National Cancer Institute, Bethesda, MD) were grown in Eagle’s MEM supplemented with 10% heat-inactivated fetal bovine serum, 0.2 mm Eagle’s MEM nonessential amino acids (Life Technologies, Inc., Gaithersburg, MD), 20 mm HEPES (pH 7.4), 2 mm glutamine (ICN Biomedicals), and antibiotic-antimycotic. LNCaP human prostate carcinoma cells (a gift from C. A. Stein, Columbia University, New York, NY) were grown in RPMI 1640 (Life Technologies, Inc.) supplemented with 1%heat-inactivated fetal bovine serum, 0.1 mm MEM nonessential amino acids (pH 7.4), and 1% antibiotic-antimycotic. Cells were grown in a humidified atmosphere of 95% air and 5%CO₂ at 37°C.

Fixed cells were treated with 1 mg/ml RNase (Sigma Chemical Co., St. Louis, MO) for 30 min and stained with 10 μg/ml propidium iodide(Sigma) for flow cytometry (FACScan, Becton Dickinson, San Jose, CA). The data were analyzed with the Modifit software (Becton Dickinson). Over 90% of confluent cells were accumulated at G₀-G₁ phase.

For cell growth experiments, 10⁵ cells were plated in 35-mm dishes (Corning Glass). At the indicated time points,the cells were trypsinized and counted using a ZI Coulter Counter(Coulter Electronics).

To increase the delivery of oligonucleotide into cells in culture,cationic lipid DOTAP (Boehringer Mannheim) was used in the oligonucleotide treatment. One day after seeding, the RI_α antisense and control oligonucleotides were added to the wells at indicated concentrations in the presence of DOTAP. After 1 day of incubation, the medium was removed, fresh medium was added in the absence of oligonucleotide and DOTAP, and cells were harvested at the indicated time points. Cells were also treated with DOTAP alone, and no cytotoxicity was observed under the experimental conditions used.

All procedures were performed at 0–4°C. After harvesting by scraping and centrifugation, cell pellets were washed once in NaCl/P_i (0.0017 mKH₂PO₄, 0.005 mNa₂HPO₄, and 0.15 m NaCl, pH 7.4). The final cell pellets were suspended in 500 μl of buffer 10 (8) supplemented with 0.5 mm phenylmethylsulfonyl fluoride and 1 mmbenzamidine, passed through a 20-gauge needle five times using a 1-ml syringe, allowed to stand at 4°C for 15 min, and then centrifuged for 5 min in an Eppendorf microcentrifuge at 4°C. The supernatants were used as lysates. Protein concentration (usually 1–5 mg/ml) was determined according to Lowry et al. (9) using BSA as a standard.

Cell extracts were prepared (see above), and total protein (40 μg)was run on 12% SDS-polyacrylamide gels and then transferred to nitrocellulose membranes. Blots were blocked overnight and probed with affinity purified polyclonal antibodies against RI_α, and RII_β. Blots were then washed and incubated with horseradish peroxidase-conjugated secondary antibodies and visualized using the Amersham enhanced chemiluminescence ECL system.

Total cellular RNA preparation, Northern blot analysis, and hybridization of RNA with ³²P-labeled human RI_α probes were described previously (10). DNA was labeled with[α-³²P]dCTP according to a standard protocol for nick translation reactions using an Amersham nick translation kit. The specific radioactivity of labeled DNA equaled 3.7 ×10⁶ cpm/μg DNA.

Total mRNA was incubated with a[³²P]CTP-labeled DNA probe for 16 h at 43°C. After hybridization, samples were digested with DNase (Sigma),precipitated in 10% trichloroacetic acid, and filtered onto Whatman CF/C filters. The amounts of DNase-resistant probe were determined by liquid scintillation counting.

Translation rate determination was performed according to Amieux et al. (11), with slight modifications. Both untreated cells and cells treated with RI_αantisense or control oligonucleotide (0.5 μm; 4 days) were washed twice in PBS [20 mmNaPO₄ (pH 7.0), 150 mmNaCl] and preincubated for 1 h in labeling medium(l-methionine +l-cysteine-free DMEM (Biofluids) with 10%heat-inactivated fetal bovine serum and antibiotic-antimycotic) and then incubated for 1 h at 37°C with 100μCi/ml[³⁵S]l-methionine+ l-cysteine protein labeling mix (Trans ³⁵S-Label; ICN). After labeling, cells were harvested by washing four times in cold PBS, followed by addition of lysis buffer (buffer 10; Ref. 8; see “Preparation of Cell Extracts”). Plates were scraped, and the cells were transferred to Eppendorf tubes and centrifuged. Supernatants were used as lysates. To determine ³⁵S-incorporation into total protein, 2 μl from each sample were spotted onto Whatman GF/C filters, and proteins were precipitated in 10% trichloroacetic acid,followed by three washes in 3% trichloroacetic acid/1% sodium PP_i. Filters were then dried and counted in liquid scintillation fluid. Samples containing equivalent total radioactivity were brought to a final volume of 100 μl in lysis buffer (buffer 10, see above) containing 100 mmNaCl and 40 μm cAMP and incubated for 2.5 h with affinity-purified polyclonal anti-RI_α or anti-RII_β antibodies, followed by 30 min with 3μl of a 10% suspension of protein A-Sepharose. Reactions were centrifuged to pellet the immunoprecipitates, which were stored at−70°C. Pellets were resuspended, solubilized, and centrifuged, and the supernatants were run on 10% SDS-PAGE. Gels were fixed for 30 min in 10% methanol and 5% acetic acid, followed by a 30-min incubation in Amplify. Gels were then dried and exposed to XAR Kodak film for 24 h.

After labeling the control and RI_α antisense treated cells for 1 h with 100 μCi/ml[³⁵S]l-methionine +l-cysteine protein-labeling mix, duplicate 10-cm plates were washed twice in DMEM containing 10% heat-inactivated fetal bovine serum and then incubated in the growth medium (see “Cell Culture”)containing 4 mm l-methionine +l-cysteine. At each time point, the cells were harvested and processed as described above. For the immunoprecipitation reactions, equivalent amounts of total protein were used rather than equivalent counts.

In the present study, we used a mixed-backbone antisense RI_α, i.e., the second-generation PS-DNA antisense RI_α of either an RNA-DNA hybrid(HYB0165) or a methylphosphonate DNA (P-ME DNA, HYB0190; see“Materials and Methods”). These oligonucleotides possess the favorable antisense properties of PS-oligo DNA (RNase H activation) and of RNA-DNA hybrid and P-ME DNA (increase in nuclease resistance and duplex stability). Such MBO antisense has been shown to have improved antisense activity over PS-oligonucleotides (12, 13). As a control, we used a 4-base mismatched oligonucleotide (HYB0295), a random sequence oligonucleotide (HYB0674), and a 2-base mismatched P-ME oligo (HYB0191; see “Materials and Methods”). To facilitate oligonucleotide delivery to the cell, cationic lipid DOTAP was used in the oligonucleotide treatment of cells (See “Materials and Methods”). Cells treated with DOTAP alone exhibited no cytotoxicity under all experimental conditions.

An enhanced expression of RI_α has been observed in MDR cancer cell lines from a variety of cell types compared with non-MDR parental cell lines (14). We examined the effects of RI_α antisense on MDR tumor growth. Studies of the pharmacokinetics of the PS-oligonucleotide indicate that a PS-oligo has a half-life of ∼40–50 h in plasma (15),which suggests that it may be given either daily or every other day. We therefore examined the effects of daily RI_αantisense treatment on in vivo growth of HCT-15 MDR human colon carcinoma.

RNA-DNA hybrid antisense (HYB0165), at a dose of 0.1 mg/mouse, i.p. daily, five times/week for 14 days, inhibited growth by 70%, but PS-oligo antisense at the same dose schedule promoted 50% growth inhibition (Fig. 1,A). Thus,the MBO antisense showed a greater potency of growth inhibition than the parental PS-oligo antisense. The four mismatched control oligos had no effect on tumor growth (Fig. 1 A).

The potency of antisense oligonucleotides in growth inhibition paralleled their efficacy in down-regulating RI_α protein levels in tumors. The RNA-DNA hybrid antisense, which inhibited growth severely, almost completely inhibited RI_α expression in tumors (Fig. 1,B, Lane 10). Down-regulation of RI_α was attributable to a sequence-specific effect of antisense, because the 4-base mismatched control oligo could not mimic the effect of antisense (Fig. 1,B, Lane 4). Importantly, concomitant with the suppression of RI_α expression was the appearance of RII_β protein, which is not expressed in untreated and mismatched oligo-treated tumors (Fig. 1 B, Lanes 9 and 12). Similar induction of RII_β has been observed previously in the single-injection RI_α antisense treatment of LS-174T colon carcinoma in nude mice (8).

We next examined whether the growth inhibition induced by RI_α antisense was attributable to changes in cell cycle phase distribution. Growth of LS-174T cells in culture was inhibited by 50% upon a single treatment (0.5 μm, day 5)with antisense RI_α of P-ME DNA or RNA-DNA hybrid (Fig. 2,A). Antisense inhibition of cell growth correlated with suppression of RI_α transcription (Fig. 3,B). Flow cytometric analysis demonstrated that within 18 h, antisense treatment provoked an increase in the population of cells in S phase of the cell cycle (Fig. 2,B). Four-base mismatched and random sequence control oligos, which exhibited no effect on cell growth and RI_α transcription (Figs. 2,A and 3),had no effect on cell cycle phase distribution (data not shown).

The growth of LS-174T colon carcinoma cells in culture was inhibited by RI_α antisense (HYB0165) in a sequence-specific and time-dependent manner. A single treatment (24 h) of the antisense(see “Materials and Methods”) at a concentration of 0.5μ m promoted 20 and 50% growth inhibition at days 3 and 5, respectively, as compared with the saline-treated control cells(Fig. 3,A). By comparison, the mismatched or random sequence oligonucleotide (HYB0295 or HYB0674) treatment (0.5μ m at day 5) had no effect on the cell growth(Fig. 3 A).

Northern blotting analysis showed that the RI_αantisense treatment resulted in a reduction of RI_α mRNA in a time-dependent manner (Fig. 3,B, Lanes 3, 6, and 9). By comparison, the mismatched control oligonucleotide had no effect on RI_α mRNA levels at any time point (Fig. 3 B, Lanes 2, 5, and 8).

Solution hybridization experiments confirmed the Northern blotting data. RI_α antisense treatment markedly reduced the level of RI_α mRNA, whereas the mismatched control oligonucleotide had no effect (Fig. 3,C). Importantly, RI_α antisense had no effect on mRNA levels of RII_α (data not shown) or RII_β (Fig. 3 C). Thus, growth inhibition induced by RIα antisense was correlated with sequence-specific inhibition of RI_α expression.

Our present and previous studies have shown that depletion of RIα by the RI_α antisense oligo up-regulates RII_β protein in LS-174T colon carcinoma cells(Fig. 4,B; Ref. 16), SKNSH neuroblastoma cells (16), MCF-7 breast carcinoma cells (17), and HL60 leukemia cells (18). Furthermore, RI_α antisense treatment of athymic mice bearing LS-174T colon carcinoma or HCT-15 MDR colon carcinoma results in inhibition of tumor growth, reduction of RI_α and PKA-I expression, the RI_α containing PKA holoenzyme and up-regulation of RII_β and PKA-II_β,the RII_β containing holoenzyme (Fig. 1; Ref. 8).

To address the mechanism of RIIβ compensation in RI_α antisense-treated cells, pulse-labeling experiments were performed in cultured LS-174T cells. No apparent difference was observed in the rate of synthesis of RII_β protein or RI_αprotein between untreated control cells, mismatched control oligonucleotide-treated cells, and antisense-treated cells after a 1-h pulse (Fig. 4,A). In contrast, Western blotting analysis showed that RII_β protein levels increased,whereas RI_α protein levels decreased in the antisense-treated cells, compared with untreated control cells (Fig. 4 B). The mismatched control oligonucleotide had no effect on RI_α or RII_β protein levels. These results indicate that the increased RII_β protein level must be attributable to stabilization of the protein.

Pulse-chase experiments were performed to determine the half-life of RII_β protein in untreated control, mismatched control oligonucleotide-treated, and antisense-treated LS-174T carcinoma cancer cells. The half-life of RII_βprotein in control and mismatched control oligonucleotide-treated cells was 2.0 h as measured by immunoprecipitation of ³⁵S-labeled RII_β protein from cell extracts after a chase with cold methionine. In contrast, the half-life of RII_β protein in antisense-treated cells was 11.0 h (Fig. 5,B). This represents a 5.5-fold increase in the half-life of the RII_βprotein upon treatment with RI_α antisense and is in good agreement with the 5-fold increase in RII_β protein observed in Western blotting analysis (Fig. 4,B). The half-life of RI_α was 17 h in control cells, and it decreased to 13 h in RI_α antisense-treated cells (Fig. 5,B). This explains the sharp decrease in the RI_α protein level found in vivowithin 24 h of antisense treatment (8). Importantly,the half-life of RII_α protein did not change,and the half-life of C_α decreased only slightly(18% decrease) in the antisense-treated cells (Fig. 5).

RII_β stabilization was also observed in LNCaP prostate cancer cells treated with RIα antisense. The half-life of RII_β protein in untreated control cells was 5.0 h, whereas the half-life of RII_βprotein in antisense-treated cells was 15.0 h, demonstrating a 3-fold increase in RII_β stability (Fig. 6). In contrast, the half-lives of RI_α, RII_α, and C_α did not change in the antisense-treated cells (Fig. 6). The antisense treatment resulted in sequence-specific reduction of RI_α expression and growth inhibition in LNCaP cells (data not shown). These results indicate that modulation of RII_β turnover rate may represent an important biological mechanism underlying the RI_α antisense-induced inhibition of colon and prostate cancer cell growth.

We have proposed previously that the RI_αregulatory subunit of PKA-I is an ontogenic growth-inducing protein and that its constitutive expression disrupts normal ontogenic processes,resulting in a pathogenic outgrowth such as cancer (2). The results presented here confirm this view and suggest that the RI_α antisense, which works through the Watson-Crick base pairing mechanism of action, can serve as a single gene-based therapeutic agent for cancer. The sequence-specific mechanism of action of RI_α antisense is strongly supported by the experimental data that MBO antisense, having an increased hybridizing capacity and nuclease resistance, increased the antisense effect of growth inhibition, whereas the mismatched MBO control oligos could not mimic the antisense effects.

In this study, minimization of the polyanionic nature of the PS-oligo and modifications of the immunostimulatory CpG motif were two important goals for us to demonstrate the sequence-specific antisense effects in the absence of the nonspecific toxicity and side effects of oligonucleotides. We have taken advantage of the reduced polyanionic characteristics of 2′-O-ME RNA and P-ME DNA and substituted a few deoxynucleosides with either 2′-O-ME RNA at both the 3′ and 5′ termini or P-ME DNA in the center of the PS-oligos,respectively. The overall results of these substitutions are increased affinity to target RNA, stability toward nucleases, and reduced polyanionic-related side effects. In addition, these oligonucleotides retain the capability to induce RNase H because of the presence of PS-oligo (12, 19).

The 18-mer RI_α antisense oligo (directed to codons 8–13 of human RI_α; Ref. 8)contains “GCGT,” a CpG immunostimulating motif (20) at the 5′ terminus. This “GCGT” was substituted with four 2′-O-ME RNAs in the 3′ and 5′ termini-modified RNA-DNA hybrid (HYB0165). This substitution enabled the RI_α antisense to be immunosuppressive rather than immunostimulatory. Substitution of the deoxynucleoside with 2′-O-ME RNA immediately next to the CpG motif at the 5′ end significantly suppressed immunostimulatory activity, whereas similar substitutions at the 3′ end had no suppressive effect (21). Thus, the immunostimulatory activity of RI_α antisense PS-oligo has been blocked in HYB0165, which contains a 2′-O-ME RNA modification at the 5′and 3′ ends.

The RI_α antisense-induced growth inhibition accompanied deregulation of cell cycle progression. In LS-174T human colon carcinoma cells (Fig. 2 B),RI_α antisense brought about an accumulation of cells in S phase of the cell cycle, indicating inappropriate entry of cells into S phase. This was shown previously for HL-60 human promyeloctic leukemia cells (22). In the antisense-treated HL-60 cells, the expression of cyclin E peaked throughout S and G₂-M phases of the cell cycle. It has been shown previously, using centrifugally elutriated HL-60 cells, that the amount of RI_α sharply increases in the fractions of cells enriched in the G₁-early S and late S-early G₂-M phases (22). Thus,RI_α/PKA-I may regulate the entry of cells from G₀-G₁ phase to the S phase and from S phase to G₂-M phase. Suppression of RI_α expression by the antisense oligo may therefore cause deregulation of the cell cycle at two critical points,one at the entry to S phase and the other at the entry to G₂-M phase, resulting in accumulation of cells in S phase and leading to apoptosis/differentiation. In fact, in HL-60 cells, RI_α antisense triggered granulocytic differentiation (22).

A unique feature of RI_α antisense was that its blockade action toward growth-stimulatory RI_αprotein expression led to a simultaneous increase in the competitive molecule RII_β, the growth-inhibitory protein of the cAMP signaling (2). This induction of RII_β caused either no change or a decrease in the expression of RII_α, an isoform of RII_β. Because RII_α is expressed constitutively in every type of cell and forms the more favored complex with the C subunit to form PKA holoenzyme as compared with other R subunits (23), it may serve as a reservoir to sequester the protein kinase in an inactive holoenzyme. Thus, the ratio of PKA-I:PKA-II would be regulated mainly by levels of RI_α and RII_β, because RII_α remains constant. Importantly, the RI_α antisense blockade of RI_α/PKA-I is unlikely to cause harmful side effects because it will simply reinforce the normal action of PKA-II by up-regulating RII_β.

In this report, we showed that the loss of the RI_α subunit of PKA in RI_α antisense-treated cancer cells in culture is compensated by the up-regulation of RII_βsubunit. This RII_β up-regulation resulted from an increase in the half-life of RII_β protein,with no change in the RII_β mRNA level or the rate of RII_β protein synthesis. These in vitro findings are consistent with observations made in tumors in vivo. We have shown in present and previous studies (Fig. 1; Ref. 8) that the growth inhibition of human colon carcinoma in nude mice induced by the RI_αantisense treatment is accompanied by the specific up-regulation of RII_β and PKA-II_β.

LS-174T colon carcinoma and LNCaP prostate carcinoma cells mainly express RI_α, RII_α, and C_α subunits of PKA and RII_β subunit at an undetectable level (10). The loss of RI_α induced by the antisense may result in increased concentrations of free “C”subunit. In the in vivo tumor experiments, we have shown that RII_β rapidly responds to this perturbation through association with “C” subunit to form a holoenzyme complex (8). RII_β in the holoenzyme complex (PKA-II_β) may be stabilized and therefore exhibit an increased half-life as shown in cultured cells(Figs. 5,6). This results in an increase in the level of RII_β protein in cancer cells that otherwise overexpress RI_α. Through this biochemical adaptation, the antisense-treated cancer cells change the ratio of PKA-I:PKA-II to a ratio similar to that of normal cells.

Experiments in cell culture have shown that C subunits preferentially assemble with RII subunits rather than RI subunits (10, 24, 25). Preferential binding of RII subunits to C subunits probably does not occur because of intrinsic differences between RI and RII in their affinity for C, because these affinities have been shown in vitro to be quite similar (26). Instead, it may be attributable to a lower K_a for cAMP-activation of PKA-I compared with PKA-II. Free RI subunits have been shown to have a higher affinity for cAMP than do RII subunits. The K_a values of cAMP binding to RI range from 0.1 to 1 nm (27, 28), whereas the K_a values for RII-cAMP binding range from 4 to 6 nm (29, 30). Given the enhanced sensitivity to activation of RI-containing holoenzyme, we predict that PKA-I, but not PKA-II, is the functional isozyme carrying out cAMP-mediated effects in cells under unstimulated physiological states.

Cancer cell lines and primary tumors have been shown to contain a higher level of RI/PKA-I than RII/PKA-II compared with normal cells (2, 5). HL-60 leukemia cells mainly express RI_α/PKA-I. Upon treatment with 8-Cl-cAMP, which selectively down-regulates PKA-I, RII_β and PKA-II_β are induced, and cells are arrested in growth (31). The PKA-I isozyme is present in transformed but not in untransformed NIH3T3 cells (32). Overexpression of RII_β in an expression vector in Ki-ras-transformed NIH3T3 cells results in up-regulation of PKA-II_β, elimination of PKA-I, and reversion of the phenotype to that of untransformed fibroblasts (33). During differentiation of Friend erythroleukemia cells, induction of RII_β coincides with the elimination of PKA-I and with increased PKA-II levels (34). These studies,together with our present study, suggest the cells’ capacity to maintain cAMP-mediated control of cellular functions, such as growth control. An important role for RII_β in this process is also suggested.

The stabilization of the RII_β protein via its prolonged half-life may also represent an important biological mechanism for cAMP-induced cell differentiation. This mechanism may maintain equivalent amounts of R and C subunits, thus protecting the cell from unregulated C subunit activity and rescuing the C subunit from rapid proteolysis. Importantly, this biochemical adaptation provides a very effective mechanism for regulating the ratio of PKA-I to PKA-II holoenzyme formed in a given tissue. For cancer cells, which deviate from the normal ratio of PKA-I to PKA-II and have increased type I PKA, this capacity of RII_β, which replaces RI_α in associating with the C subunit,may provide an important biological mechanism to normalize the PKA isozyme ratio and sustain inhibition of tumor growth. Importantly, that RI_α antisense triggers stabilization/up-regulation of a competitor molecule,RII_β, in cancer cells that otherwise overexpress RI_α represents a true antisense mechanism, i.e., sequence-specific inhibition of gene expression, leading to sustained inhibition of PKA-I expression and tumor growth.

We thank Sudhir Agrawal (Hybridon, Inc., Milford, MA) for providing the oligonucleotides.