Down-Regulation of the Insulin-like Growth Factor I Receptor by Antisense RNA Can Reverse the Transformed Phenotype of Human Cervical Cancer Cell Lines¹

In recent years, it has become clear that IGF-IR,³activated by its ligands, plays different roles in vitro or in vivo (1): (a) the IGF-IR is mitogenic, although it is required for optimal growth in conjunction with other growth factors, such as epidermal growth factor and platelet-derived growth factor (2, 3); (b) it is quasi-obligatory for the establishment and maintenance of the transformed phenotypes (4, 5, 6); and (c) it protects cells from apoptosis induced by a large variety of mechanisms(7, 8, 9, 10, 11). Interference with the expression and/or activation of the IGF-IR by antisense strategies (12, 13, 14, 15, 16, 17),dominant-negative mutants (18, 19, 20, 21), or triple-helix formation (22) reverse the transformed phenotype or inhibit tumorigenicity in many rodent and human cancer cells. Furthermore, antisense strategies (17), a myristylated COOH terminus of the IGF-IR (23), and dominant-negative mutants of the IGF-IR (20) cause massive apoptosis in vivo. Because substantial data support that normal cell growth is much less dependent on the IGF-IR (1),interference with the IGF-IR would be a good candidate for human gene therapy in many types of tumors.

R⁻ cells, originating from mouse embryo with a targeted disruption of IGF-I receptor gene (24, 25), are known to be refractory to transformation by several viral and cellular oncogenes. SV40 T antigen, even with activated ras(26, 27), bovine papillomavirus E5 protein(28), activated c-src (29), and other growth factor receptors that are overexpressed (30, 31) fail to transform R⁻ cells. However, it was reported that stable transfection of both E6 and E7 oncoproteins of HPV-16 could transform R⁻ cells in soft agar(32). By using R⁻ cells and their wild-type littermate W cells, it has become clear that the IGF-IR is a necessary requirement for the E7-induced transformation of these fibroblasts (32). In R⁻ cells,combination of both E6 and E7 transforms R⁻cells, indicating that E6 may be acting equally to the IGF-IR(32). In the presence of both E6 and E7 in cervical cancer cells, it is postulated that these two proteins might use an alternative pathway for transformation, and that the interference with the IGF-IR might result in failure to reverse the transformed phenotypes of cervical cancer cells.

In cervical cancer cells, it is well known that in >85% of the patients, the cells are positive for HPV (33, 34). In these cervical cancer cells with malignant-type HPVs, E6 protein may play an alternative role for IGF-IR and promote E7-induced transformation, even in the absence of IGF-IR (32). The present study is undertaken to elucidate whether the IGF-IR targeting can be used for reversal of the transformed phenotypes of cervical cancer cells in vitro by stable transfection of an antisense mRNA expression plasmid into HPV-positive and HPV-negative human cervical cancer-derived cell lines.

C33a, SiHa, and HeLa S3 are cell lines originally derived from human cervical carcinomas. All of these cell lines were obtained from the Department of Cell Biology, Institute of Cellular and Molecular Biology, Okayama University Medical School. All of the cell lines were passed in DMEM supplemented with 10% FBS. Status of HPV and major tumor suppressor genes p53 and pRB in these cell lines was characterized previously in detail. SiHa cells contain type 16 HPV, and HeLa S3 cells contain type 18 HPV. C33a cells do not contain any types of HPV (35, 36). SiHa and HeLa S3 cells carry wild-type p53 and pRB. Only C33a cells have p53 mutation at codon 273 and pRB in-frame deletion in exon 20 (37). R⁺ cells are R⁻ cells stably transfected with human IGF-IR and express about 9.0 × 10⁵ receptors/cell (27, 38).

The expression vectors HSP-IGF-IRS and HSP-IGF-IRAS (39)produce sense or antisense mRNA to the first 309 bp of cDNA fragment of IGF-IR including a 30-amino acid signal peptide sequence under the control of Drosophila HSP70 promoter (40). Transfections of plasmids were performed using Transfectam (Promega) under the strict selection of G418. To establish stable transformants with higher copy numbers, selection was performed at a permissive temperature of 34°C in 2.0, 1.5, and 1.0 mg/ml G418 for HeLa S3, SiHa, and C33a cells,respectively. Established clones were kept growing at 37°C until use.

Cells were plated at a concentration of 5 × 10⁴ per 35-mm plate in DMEM containing 10% FBS. After 24 h incubation at 37°C, cells were washed three times with Hank’s solution, and medium was replaced with SFM (DMEM supplemented with 0.1% BSA and 50 μg/ml transferrin), or SFM with 20 ng/ml IGF-I (Life Technologies, Inc.), or DMEM with 10% FBS. Cells were incubated for an additional 48 and 96 h, and cell numbers were counted in a hemocytometer. For antisense and sense clones, cells were passed several times at 39°C in advance, and then cells were incubated in each condition for 48 h at 39°C and counted. All of the points were the results of triplicate experiments.

Sense and antisense clones maintained at 37°C were shifted to either 34°C or 39°C. After several passages at each temperature, cells were washed three times with Hank’s solution and then incubated for an additional 72 h in SFM. Fresh SFM was renewed every 24 h. Subconfluent cells were lysed in a lysis buffer as described previously. Fifty μg of total cell lysates were separated and hybridized with anti-β-subunit of IGF-IR (C-20; Santa Cruz Biotechnology).

Sense or antisense clones were passed several times in media without G418 at 39°C prior to seeding in soft agar plates. HeLa S3-derived clones were seeded at 5 × 10³cells/35 mm plate in DMEM supplemented with 10% FBS and 0.2% agarose with 0.5% agarose underlay, and colonies >125 μm in diameter were counted after 2 weeks incubation at 39°C. SiHa- and C33a-derived clones were seeded at 3 × 10⁴cells/35-mm plate, and the same sized colonies were counted after 4–5 weeks incubation at 39°C.

Cells were washed three times with Hank’s solution and incubated in SFM for 24 h, trypsinized, and washed twice with PBS. Cells(5 × 10⁶) were resuspended in 100μl of sterile PBS and injected s.c. above the hind leg of male BALB/c nude mice (Charles River Breeders).

To determine expression levels of the IGF-IR in cervical cancer cell lines, whole-cell lysates were collected from each cell line incubated previously at 37°C. Fifty μg of lysates were separated on a 5–15%gradient acrylamide gel and stained with anti-β-subunit of IGF-IR(Fig. 1). Lanes 1 and 5 show negative and positive controls from lysates of R⁻ cells and R⁺ cells. The R⁺ lane was intentionally overexposed to show the low expression levels of cervical cancer cell lines. We used this Western blot for comparison rather than Scatchard plot analysis since Scatchard plots were often erratic because of the presence of abundant IGF binding proteins secreted from cancer cell lines. C33a shows only a slight signal of the IGF-IR at longer exposure (Fig. 1,B). IGF-IR numbers were roughly calculated from comparison by densitometry with several standard cell lysates, the receptor numbers of which were already determined(38) and summarized in Table 1.

Each cervical cancer cell line was incubated under various IGF-I concentrations in monolayer, and their IGF-I ligand dependency was evaluated (Fig. 2). Sole addition of IGF-I stimulated growth of all cell lines tested in the study, and it depended on IGF-IR expression levels, shown in Table 1. We repeated this experiment for the C33a cell line, which has the lowest IGF-IR number, and confirmed that IGF-I stimulation had a slight but definitely reproducible mitogenic effect.

Established stable clones with sense or antisense were shifted to 34°C or 39°C and passed several times. Because the transfected IGF-IR antisense is controlled under the thermosensitive HSP70 promoter(40), transcription of the antisense mRNA is on when clones are kept at 39°C, and it is off at 34°C. After serum deprivation for 72 h, lysates were collected from subconfluent cells, and the difference between the two temperatures was compared on Western blot. Fig. 3 shows the down-regulation of the IGF-IR expression in HeLa S3 antisense clones. There is a large variety of down-regulation levels among antisense clones. Because HeLa S3 cells express 7–44 times more IGF-IR than the other two cervical cancer cell lines, integration of multiple copy number seems necessary to get enough down-regulation for this cell line. In SiHa clones with antisense transfected, IGF-IR signals at 39°C could not be detected. C33a cells express only a little amount of receptor, and IGF-IR signal of either C33a antisense or C33a sense clones could not be clearly detected after 60 min exposure (data not shown).

Sense and antisense clones, as well as wild-type cells of the three cervical cancer cell lines, were examined for their mitogenicity in monolayer. Cells were incubated at 39°C without G418 and used for the study. Cell growth in SFM, SFM supplemented with 20 ng/ml IGF-I, and DMEM with 10% FBS was assessed as percentage of increase after 48 h of incubation (Fig. 4). Although there are some clonal variations, antisense clones of these three cell lines showed slightly inhibited monolayer growth than wild-type and sense transfected clones. Antisense clones showed ∼20%inhibition on the average but was not statistically significant.

Sense, antisense, and wild-type cells of these three cervical cancer cell lines were passed several times at 39°C, seeded into 0.2% soft agar plates, and incubated at 39°C. HeLa S3 cells were originally highly transforming, and 5 × 10³wild-type cells seeded in a 35-mm plate formed many colonies after 2 weeks of incubation. SiHa and C33a cells were much less transforming,so that cells at higher density (3 × 10⁴ per 35-mm plate) and longer incubation time(4 weeks for C33a and 5 weeks for SiHa) were necessary to obtain enough colonies, again confirming the much higher expression level of the IGF-IR in HeLa S3 cells. Colony number in soft agar is summarized in Table 2. In SiHa- and C33a-derived clones, the original IGF-IR expression levels of which were low, there is significant difference in colony formation in soft agar between antisense transfected clones and sense clones or wild-type cells. In HeLa S3 cells expressing much higher IGF-IR, colony numbers of antisense clones vary from 14 to 160. Suppression of transforming ability in soft agar and levels of the IGF-IR down-regulation estimated by Western blot are compared in Fig. 5. There is a good correlation between colony numbers in soft agar and IGF-IR expression levels, indicating that the integration of large copy numbers of antisense RNA-expressing plasmid is necessary to achieve enough down-regulation of the IGF-IR and to reverse transforming phenotype of these aggressive HeLa S3 cells.

The inhibitory effect of the IGF-IR antisense was then tested by injections of transfected clones into nude mice. Because C33a wild-type cells are not tumorigenic in nude mice, only HeLa S3- and SiHa-derived clones were tested. HeLa S3 wild-type and sense clones formed palpable tumors 2–3 days after injection, and balky tumors progressed rapidly. In contrast, representative antisense clones did not form any tumors until 1 week, and very small palpable tumors (<2 mm in diameter)remained unchanged for 4 weeks (Fig. 6). SiHa wild-type and sense clones formed palpable tumors after 1 week to 10 days and grew slowly; SiHa antisense clones did not form any tumors after 4 weeks.

Genomic integration of high-risk types of HPV has been strongly associated in cervical and other anogenital malignancies(33). It is well known that the oncogenic properties of malignant HPVs depend on the E6 and E7 proteins, which are capable of inactivating two major suppressor genes, p53 and pRB. E6 binds to p53 and promotes its ubiquitin-dependent proteolysis (41). E7 binds preferentially to hypophosphorylated pRB and inhibits binding of pRB to transcriptional factor E2F(42). Although E6 and/or E7 are known to transform NIH 3T3 cells and immortalize human primary cultured keratinocytes(43), E6 and E7 are not enough for carcinogenesis. Within the high-risk HPV-positive lesions, where both tumor suppressor proteins pRB and p53 are functionally inactivated by the expression of E6 and E7 proteins, only a small number of the lesions will progress to carcinomas. As another causative alteration, frequent loss of heterozygosity in chromosomal arms likely to harbor other tumor suppressor genes is reported from human cervical intraepithelial neoplasia patient specimens (44).

IGF-I and IGF-IR also play an important role in cell transformation induced by viral oncogenes (45). SV40 T antigen induces constitutive tyrosyl autophosphorylation of the IGF-IR(46). SV40 T antigen up-regulates IGF-I ligand expression and binds directly to IRS-1, the major downstream substrate of the IGF-IR (47). v-src also induces marked tyrosyl autophosphorylation of the IGF-IR (48). BPV E5 directly binds and activates platelet-derived growth factor β receptor(49) and might transactivate IGF-IR through the platelet-derived growth factor β receptor (31). The fact that R⁻ cells are refractory to transformation by these viral oncogenes is striking (26, 28). The only viral oncogene that is reported to transform R⁻cells by itself is the coexpression E6 and E7 of HPV-16 and v-src thus far (29, 32). Interestingly, SV40 T antigen, which does not transform R⁻ cells, is known to suppress both pRB and p53, similar to E6 and E7 of HPV. E6 binds to p53 and induces its degradation, whereas SV40 T antigen just binds stably to p53 and inactivates its function. The difference between HPV E6/E7 and SV40 T antigen in transforming ability in the absence of the IGF-IR still remains unknown. E6 was reported to have a protective effect from staurosporine-induced apoptosis, even in p53−/− cells, indicating the probability of a p53-independent pathway of E6 for cell survival signaling (32). In contrast, the IGF-IR is also reported to have strong protective effects against apoptosis induced by a large variety of causes (7, 8, 9, 10, 11). Steler et al.(32) hypothesized that the IGF-IR and E6 were functionally equivalent in transformation. It is likely that overexpressing E6 and E7 substitute the transforming ability of the IGF-IR, but a mechanism(s) of E6 and E7 for transforming R⁻cells involves not only the abrogation of two major suppressor gene functions but also substitution of some other properties of the IGF-IR. If E6 cooperates with E7 to send transforming and antiapoptotic signaling that is independent of the IGF-IR pathways, interference with the IGF-IR cannot be effective for transformed cells harboring malignant type HPVs, such as cervical cancer cells.

The present study shows that the down-regulation of the IGF-IR by antisense strategy is effective in the reversal of transformed phenotypes in vitro and in vivo for human cervical cancer cell lines in the presence (HPV-16 in SiHa and HPV-18 in HeLa S3) and absence (C33a) of HPV. Antisense oligonucleotide for HPV-16 E6 and E7 (50), antisense RNA against HPV-16 E6/E7(51), and single-chain antibody against HPV-16 E7(52) were tried previously in SiHa cells, and they all inhibited transformation. But all of them were tried in SiHa cells,which contain only one to two copies of integrated HPV-16 DNA and express relatively low amounts of IGF-IR (Table 1). In contrast, HeLa S3 cells contain ∼50 copies of integrated HPV-18 and express huge numbers of the IGF-IR (Table 1). HeLa S3 cells are highly transforming and make many more colonies in soft agar than SiHa cells. In SiHa and C33a cells, with lower IGF-IR expression and lower or no HPV integration, stable expression of the antisense mRNA against the IGF-IR resulted in marked suppression of colony formation in soft agar. In HeLa S3 clones with an antisense effect against the IGF-IR, some clones still formed colonies when the antisense effect was not overwhelming. But results from antisense clones with sufficient down-regulation of the IGF-IR were promising, and suppression of colony formation in soft agar was closely correlated with the down-regulation of the IGF-IR. Furthermore, it is clear that this biological effect on antisense clones is not attributable to the toxicity of overexpressing mRNAs,because no toxicity is observed in sense clones, and its action depends mainly on the condition of anchor independence. Although there is variation in the IGF-IR expression levels and integrated HPV copy numbers among human cervical cancer cells, our data from HeLa S3 clones clearly show that down-regulation of the IGF-IR, if sufficient, can reverse transformed phenotypes of cervical cancer cells, even with malignant-type HPVs.

Targeting the IGF-IR for cervical cancer cells seems ideal, now that it is confirmed that it can reverse transformed phenotypes, even in the presence of high copy number of malignant-type HPV, because:(a) it can be widely applied for all types of cervical cancer cells, regardless of specific HPV subtypes; (b) it can be also used in the absence of HPV; (c) normal cells are less dependent on the IGF-IR for growth; and (d)antimitogenic, antitransforming, and proapoptotic effects arising from interference with the IGF-IR pathways are usually more dramatically emphasized in vivo than in vitro.

We thank Dr. R. Baserga, Kimmel Cancer Center, for kindly providing HSP-IGF-IRS and AS (or HSP-EGF-ERAS) expression vectors,R⁻ and R⁺ cells.