The EMBO Journal vol.10 no.5 pp.1 111 -1118, 1991

Short modified antisense oligonucleotides directed against Ha-ras point mutation induce selective cleavage of the mRNA and inhibit T24 cells proliferation T.Saison-Behmoaras, B.Tocque&, I.Rey1, M.Chassignol2, N.T.Thuong2 and C.Helene Laboratoire de Biophysique, INSERM U.201 - CNRS UA.481, Museum National d'Histoire Naturelle, 43 rue Cuvier, 75005 Paris, 'Rh6ne-Poulenc Sante, Centre de Recherche de Vitry, 13 Quai Jules Guesde, 94403 Vitry sur Seine and 2Centre de Biophysique Moleculaire, CNRS - 45071, Orleans Cedex 02, France Communicated by C.Helene

We have used derivatized antisense oligodeoxynucleotides both in vitro and in vivo specifically to inhibit translation of the activated human oncogene Ha-ras. The oligonucleotides (5'-CCACACCGA-3') were targeted to a region of Ha-ras mRNA including the point mutation G - T at the 12th codon which leads to a Gly -Val substitution in the ras p21 protein. They were linked to an intercalating agent and/or to a hydrophobic tail, both to increase their affinity for their mRNA target and to enhance their uptake by tumor cells. A cell-free translation system was used to demonstrate an RNase Hdependent specific inhibition of activated ras protein synthesis. 50% inhibition was observed at a concentration of 0.5 yM of the most efficient oligonucleotide (5'-substitution with an acridine derivative and 3'-substitution by a dodecanol chain). This inhibitory effect stems from a point mutation-sensitive cleavage of the mRNA and it mirrors the growth inhibition obtained with T24 bladder carcinoma cells, which carry activated Ha-ras. The proliferation of HBL100 cells (non tumorigenic human mammary cell line) which carry two copies of normal Ha-ras was unaffected. This study shows that it is possible to design antisense agents that will inactivate the mutated oncogene but not the protooncogene which is generally essential to cell survival. Key words: antisense agent/Ha-ras/oligonucleotide/oncogene/ translation inhibition

Introduction It has been found that 10-20% of human tumors have a mutation in one of the three ras genes (Ha-ras, Ki-ras, N-ras) leading to the production of p21 ras oncoproteins, which are thought to play an important role in the transformed phenotype (Bos, 1989). The ras proteins bind GTP and GDP, and have intrinsic GTPase activity (McCormick, 1989). They may control cell proliferation by regulating a signal transduction pathway as do the regulatory G proteins. However the biochemical mode of action and the biological target molecules of the ras proteins are unknown. In order to study the biological effects of ras expression in the context of the molecular biology of ras-dependent pathway and to provide a rational basis for the development of antitumor drugs we are investigating the use of 'antisense' oligonucleotides and (C Oxford University

Press

their modified analogues, which upon hybridization to complementary mRNA sequences, interfere with translation and thus can be employed for sequence-specific control of gene expression. In an attempt to inhibit the expression of an oncogene, application of antisense oligonucleotides has proved to be a powerful tool. Antisense oligodeoxynucleotides targeted to c-myc mRNA, when added to cultures of normal T lymphocytes inhibit proliferation in response to mitogen (Heikkila et al., 1987; Harel-Bellan et al., 1988) and when added to HL-60 cells, inhibit proliferation and induce differentiation (Holt et al., 1988). Oligomers complementary to c-myb encoded mRNA were shown to inhibit human hematopoiesis in vitro (Jaskulski et al., 1988) and proliferation of human myeloid leukemia cell lines (Anfossi et al., 1989). In this report, we provide data showing that it is possible to inhibit selectively the expression of activated Ha-ras with a modified 9-mer whereas normal Ha-ras expression is not markedly affected. Such antisense oligomers could be very useful in examining the role of mutated ras in the initiation and maintenance of tumor development.

Results Design of modified antisense oligonucleotides We have previously shown that rather short antisense oligonucleotides could maintain a high selectivity of binding to complementary sequences on mRNAs provided their binding affinity was increased by attaching an intercalating agent to one or both ends of the oligonucleotide (Asseline et al., 1984; Helene and Toulme, 1989). The intercalating agent provides an addditional binding energy by inserting its aromatic ring between the last two base pairs of the oligonucleotide - RNA hybrid. If the target sequence of the oligonucleotide is mutated in such a way that the last base pair of the oligonucleotide -RNA hybrid cannot form, then intercalation should be lost and the hybrid strongly destabilized. In the ras gene family, acivation of the protooncogene to form the oncogene is due to point mutations, most often in the 12th and 61st codons. We chose to investigate the Ha-ras gene of a bladder carcinoma (T24/EJ) which is activated by a G - T transversion in the 12th codon (Reddy et al., 1982). The antisense oligonucleotide sequence was chosen to be complementary to the mRNA sequence downstream of the mutation with the last base pair formed with U on the mutated mRNA so that an A -G mismatch should be present when the oligonucleotide binds to the protooncogene. An acridine derivative (2-methoxy-6-chloro-9-aminoacridine) was chosen as an intercalating agent and attached to either the 3'- or the 5'-end of the oligonucleotide. In order to improve cell uptake, a hydrophobic substituent was also attached to the 3'-end of some of the antisense oligonucleotides (see Scheme 1 and Materials and methods). As will be described below 1111

T.Saison-Behmoaras et al.

HO-(CH2-)120H -Dmt-O-(CH2-)120H 1 2 Dmt-O-(CH2-)120-C-(CH2-)2 COH

1

available rabbit reticulocyte lysate has very low levels of RNase H activity whereas wheat germ extracts have high activity (Walder and Walder, 1988). Therefore we performed translation assays with oligonucleotides in rabbit reticulocyte lysate similar to those described above with the wheat germ extract. Figure 2B shows that the 9-mer linked via its 3'-end to the acridine had no effect on p21 synthesis at 20 yM (Figure 2B, lane b). However addition of RNase H from E. coli to the translation medium restored the inhibition observed with the same oligonucleotide in the wheat germ extract (Figure 2B, lane c). The presence of a high RNase H activity in wheat germ extracts undoubtedly contributed to the inhibition of p21 synthesis observed in this assay. Table I shows the concentrations of oligomers required to inhibit 50% of p21 synthesis in rabbit reticulocyte lysate complemented with RNase H from E.coli. Similar results were obtained in wheat germ extract. An unmodified oligonucleotide complementary to 9 nucleotides spanning the 12th codon of Ha-ras inhibited by 50% the translation of ras mRNA at 22 ,uM. When an intercalator such as an acridine derivative was tethered to the oligonucleotide, hybridization with the target was previously found to be more stable as a result of the additional binding energy brought about by intercalation of the dye within the duplex (Asseline et al., 1984). Linkage of acridine to the 3'- or 5'-end of the 9-mer gave an advantage to the modified anti-messenger since 13 and 16 1tM, respectively, were required to obtain 50% inhibition. The most striking result was obtained with the 9-mer linked to a dodecanol at its 3'-end and to acridine at the 5'-end. Even though the hydrophobic tail was attached to the oligonucleotide in order to increase oligomer uptake by cells, this oligonucleotide was 30 times more efficient in vitro than the 9-mer carrying only the acridine at the 5'-end. It is noteworthy that the 9-mer linked only to the dodecanol tail at the 3'-end was also more efficient than the unmodified or acridine-substituted oligomers. Letsinger et al. (1986) and Jager et al. (1988) have shown that lipophylic substituents may stabilize DNA helical structures. In addition more efficient induction of RNase H cleavage by this oligonucleotide may also result from facilitation of RNase H binding to the hybrid duplex. The presence of an acridine derivative linked to the 10-mer did not give much advantage to the modified oligonucleotide suggesting that in the case of the 10-mer hybrid stability was enough to induce RNase H-mediated translation inhibition. The 9-mer linked to acridine and targeted to the second codon of Ha-ras (oligomer (7) in Figure 1) was not a very efficient

1 2

11

o

o

1 3,4 Dmt-O-(CH2-)12 O- C -(CH2-)2 C NH- )

11

3

11

o

0

Scheme 1 - Dmt: dimethoxytrityl; 1: dimethoxytrityl chloride, pyridine; 2: succinic anhydride, 4-dimethylaminopyridine, pyridine; 3: p-nitrophenol, pyridine, dicyclohexylcarbodiimide, dioxane; 4: aminopropyl-Fractosil 500 (H2N-(4, triethylamine. See Materials and methods for synthesis of oligonucleotides.

this hydrophobic 'tail' not only improved cell uptake but also enhanced the inhibitory effect of the oligonucleotide on mRNA translation even in in vitro translation assays. Inhibition of ras-p21 cell free synthesis by modified oligonucleotides in wheat germ extract and rabbit reticulocyte lysate Figure 1 shows the sequences of the oligomers and their complementary binding sites on the messenger RNA of human activated Ha-ras. In order to evaluate the efficiency of these oligonucleotides to selectively inhibit translation of Ha-ras mRNA we prepared total cytosolic mRNA from cells overexpressing activated Ha-ras and translated them in a wheat germ extract in the presence of Ha-ras targeted or unrelated oligomers. Figure 2A shows such an example of inhibition of activated p21 translation in wheat germ extract with 20 .tM of a 9-mer substituted by an acridine derivative at its 3'-end (lane c). There was no effect of the antisense oligonucleotides on general protein synthesis as determined when the samples were run onto a gel prior to immunoprecipitation (data not shown). There was also no difference in the levels of proteins which co-immunoprecipitated with ras p21 (Figure 2A,B). Only ras p21 was significantly affected, when anti-ras modified 9-mer was added. Addition at the same concentration of an unspecific 9-mer bearing the same modifications as the specific one had no effect on p21 synthesis (Figure 2A, lane b). It was imporant to determine whether an endogenous enzymatic activity such as ribonuclease H (RNase H) was mediating hybrid arrest of translation by oligodeoxyribonucleotides in this system. RNase-H cleaves a RNA-DNA hybrid and its activity has been previously implicated in hybrid arrest experiments (Minschull and Hunt, 1986; Cazenave et al., 1987). It is known that commercially

G +1

.10

+20

+30

.40

+50

5,'...AGGAGCGAUGACGGAAUAUAAGCUGGUGGUGGUGGGCGCCGUCGGUGUGGGCAAGAGUGCGCUGAC.. (7) 3.CCTTATATT- ms. Acr (0) 3.AGCCACACC5.

(1) (2) 3,

(8)

HO

5.

ml 2' CCCAACGCA - m. Acr

Control

Acr-ms-AGCCACACC5. 3.AGCCACACC- m5. Acr

(3)

HO

m1 2'AGCCACACC5.

(4)

HO

m12'-AGCCACACC - ms. 3CAGCCACACC5.

(5) (6)

Acr.

ms- CAGCCACACC5,

9 mer

Acr

I

10 mer

Fig. 1. Sequences and terminal modifications introduced into oligonucleotide targeted to activated Ha-ras mRNA (see Methods and materials description of the chemical modifications introduced in the 9-mer). The numbers corresponding to each oligomer are indicated on the left of sequences. Oligomer (8) used as a control had the same base composition as oligonucleotides 0-4 but the sequence has been scrambled. 1112

for

Oncogene inhibition by antisense oligonucleotides

inhibitor of translation (25 ltM were required for 50% inhibition of translation). It is quite possible that the complementary sequence of this oligomer was not accessible for hydrogen bonding due to short or long distance interactions within the mRNA. Indeed messenger RNAs, like other RNAs, are folded back on themselves to form secondary structures (such as hairpins) and tertiary structures arising from interactions between elements of the secondary structures. We did not pre-anneal oligomers to mRNA by heat treatment in our in vitro experiments. This treatment was previously shown to improve the accessibility of different target sequences (Yu et al., 1989) but does not represent the situation prevailing inside cells. In the case of the 9-mer directed to the second codon it is noteworthy that this oligonucleotide can form only two GC base pairs as

A (WG, kDa 30-

20

-

14.4_ a

b

c

B (RRL) kDa 43 _ 30

-

compared to six GC base pairs with the 9-mer directed to the 12th codon. Therefore stability is expected to be weaker. Oligonucleotides discriminate between mutated and wild-type ras mRNAs Capped mRNA of activated (Val) and wild type (Gly) Ha-ras obtained by in vitro transcription (see Methods and materials) were incubated in a rabbit reticulocyte lysate at 37°C with E.coli RNase H and the 9-mer linked to the acridine and dodecanol substituents. After 1 h of incubation the RNA was analysed on a 6% polyacrylamide sequencing gel. 70% of the 'Val' mRNA was cleaved into two fragments of 100 and 700 nucleotides (Figure 3, lane 3). The larger cleavage product represents the 3' part of the RNA transcript while the smaller fragment represents a capped 5 '-fragment of the mRNA located upstream from the binding site of the oligonucleotide. Only 7% of the 'Gly' mRNA was cleaved by the modified antisense 9-mer under the same conditions (Figure 3, lane 2). Using chain termination of nascent RNA transcripts by incorporation of 3'-0-methyl guanosine we determined the exact site (Figure 3, lane 5) of oligonucleotide directed RNase H cleavage of Ha-ras messenger RNA. The cleavage reaction required both the specific oligonucleotide and the enzyme, since the RNA remained intact in control samples missing either one of these components (Figure 3, lanes 1 and 4). We conclude from these experiments that (i) the oligomer Acr-m5-5'-CCACACCGA-3'-m12OH formed a sequence-specific duplex with the 'Val' mRNA. RNase H cleaved the 'Val' mRNA engaged in the duplex at two sites situated on the 3' side of the target sequence. At 0.2 ItM oligonucleotide after 1 h at 37°C 70% of the 'Val' mRNA was specifically degraded; (ii) the oligomer Acr-m5-5'-CCACACCGA-3'-m12OH did not induce strong RNase H cleavage on the 'Gly' mRNA (only 7% under our experimental conditions). Since we know that mRNAoligonucleotide duplex is a substrate for RNase H the explanation for the low cleavage efficiency is that the 9-mer does not hybridize strongly enough to 'Gly' mRNA. Although these results show that a short modified oligonucleotide induces RNase H cleavage of a perfect RNAoligonucleotide duplex and not of a duplex with one mismatch they have been obtained with RNase H from E. coli. In the previous section we have shown that no -

Table I.

Oligonucleotide modification

Concentration for 50% inhibition (uM)

-

Acr-m5-5'.3'-m12OH

22 13 16 1.5 0.5

7

Acr-m5-5'

25

5 6

-

3'-m5-Acr

10 8

No.

9 mer

a

b

c

Fig. 2. A. Effect of oligonucleotides on translation of activated Ha-ras mRNA in a wheat germ extract (WG). 20 Ag of total cytosolic RNA from Chinese lung fibroblast transformed with T24 Ha-ras was translated for 30 min at 25°C (a), with 20 zM of unspecific oligomer 5'-TAACCCTGA-3'm5Acr (b) or with 20 jtM of anti-ras 9-mer 5'-CCACACCGA-3'm5Acr (oligonucleotide 1) (c). [35S]methionine labelled ras p21 was immunoprecipitated with a ras-specific monoclonal antibody and analysed by SDS-PAGE. B. Effect of modified 9-mer on translation of activated Ha-ras mRNA in a rabbit reticulocyte lysate (RRL). The same RNA as in Figure 2A was translated for 45 min at 30°C in the absence of oligomer (a), in the presence of 20 ytM oligonucleotide 1 (b) or 20 yM of oligonucleotide 1 and 5 units of RNase H from E.coli (c). Arrows in Figure A and B indicate the position of Ha-ras p21 protein.

10 mer

0 1 2 3 4

3'-m5-Acr Acr-m5-5'

3'-mI2OH

Concentration of oligonucleotides required for 50% inhibition of translation of Ha-ras mRNA in rabbit reticulocyte lysate supplemented with RNase H. Similar results were obtained with wheat germ extract. p21 protein synthesis was quantified from densitometer tracing of autoradiographs. The numbers in the first column refer to the oligonucleotides described in Figure 1.

1113

T.Saison-Behmoaras et al.

A i.

a

-. _

bI

4k

4ftw!.,

...

4MM

100%

2112

1 1() I

OW

"fM

1R ;.

V G V- G V G V G V G p'

A

| |I!

.. .........

Fig. 4. A. Short (5'-terminal) fragments generated upon cleavage of duplexes formed by anti-ras 9-mers (10 uM) with normal Ha-ras mRNA (G) or mutated Ha-ras mRNA (V) in rabbit reticulocyte lysate. Samples were incubated for 30 min at 37°C. (-) means unmodified 9 mers, other modifications are indicated at the top of the figure. The Ha-ras mRNA was uniformly labelled during in vitro transcription (see Materials and methods). The sequence of the two mRNAs (V and G) in the cleavage region are shown on the right part (x indicates the position of the G - U mutation). B. Same conditions as in A but E.coli RNase H (1.6 units) was added and doubly modified oligonucleotide concentration was decreased to 1 AM.

...

. . . . .................

B -

I, _____

i; 1-i,~-~ I .- .~, ,. , ;1 A f' , I'

of substituted 9-mers to 10 ,tM we obtained cleavage of activated ras mRNA without addition of E.coli RNase H (Figure 4A). As we can see in Figure 4B, 1 itM of doubly substituted 9-mer induces highly specific and efficient RNA cleavage in the presence of E.coli RNase H. A detailed ;, !comparison of the cleavage sites induced by the eucaryotic bacterial RNase H will be described in another -and publication. -

Fig. 3. A. Cleavage by Ecoli RNase H of normal Ha-ras (Gly) and mutated Ha-ras (Val) in duplexes with the 9-mer substituted by acridine (5') and dodecanol (3') (oligonucleotide 4, lanes 2, 3 and 4). Oligonucleotide 8 was used as a control (lane 1) (see Figure 1 for oligonucleotide sequences). In vitro transcribed uniformly labelled RNAs were incubated at 37°C in rabbit reticulocyte lysate with 0.2 itM of oligomers in the presence (+) or absence (-) of 0.8 units of E.coli RNase H. The resulting RNA products were separated on a 6% polyacrylamide sequencing gel. Arrow (a) indicates the full length mRNA, (b) and (c) the position of the two RNA fragments obtained after RNase H cleavage (also see Figure 4). M denotes DNA size markers. Chain termination of nascent RNA transcripts at G was obtained by incorporation of 3'-O-methyl guanosine (see Methods and materials) (lane 5). The RNA sequence illustrates the target sequence of the doubly substituted 9-mer oligonucleotide (lane 5). B. The overall organization of the Ha-ras transcript. Vertical arrows mark the sites of RNase H cutting in Ha-ras 'Val' mRNA hybridized to doubly substituted 9-mer.

inhibition of p21 synthesis occurred in rabbit reticulocyte lysate in the presence of the acridine linked 9-mer at low (,uM) concentrations. When we increased the concentration 1114

B

Oligodeoxynucleotide uptake and stability Exogenously introduced oligodeoxynucleotides would not be very effective agents for hybrid arrest of translation if they were rapidly hydrolyzed. The potential therapeutic value of antisense oligomers will depend on their resistance to degradation and their efficiency to permeate the plasma membrane of living cells. We have tested the stability of unmodified and modified oligonucleotides in the cell culture medium with 7% fetal calf serum, heat-inactivated at 65 °C to destroy nuclease activity. Anti-ras oligomers (an unmodified 10-mer and a 10-mer linked to acridine at the 3' end) were 5'end-labelled and added to the T24 cell culture medium. At times indicated in Figure 5A oligomers were extracted from the culture medium and applied to denaturing gels. The unmodified 10-mer was fully degraded after 23 h (lane f). Acridine substitution at the 3' end protected the oligonucleotide from exonucleases (lane g -1). The small degradation observed with the 3' acridine-linked oligonucleotide arises from contamination with unmodified 10-mer (lane g).

Oncogene inhibition by antisense oligonucleotides

F

.saw

.

v 0

A ..........,..-

,a

.

. ___~4

.4 f..

a b c de f g h

i

j k I

Fig. 5. A. Stability of unmodified and acridine-linked 10-mer in a T24 cell culture medium (see Methods and materials). 5' 32P-labelled 10-mer (a-f) and 10-mer linked to acridine at the 3' end (g to 1) were incubated for 0, 0.5, 1, 2, 5 and 23 h in the culture medium. The oligonucleotides were separated on a 20% polyacrylamide sequencing gel. The acridine-substituted oligonucleotide migrates more slowly than the unsubstituted one as expected due to the size and charge of the acridine substituent. In this experiment the acridine-substituted oligonucleotide was contaminated with a small amount of the unsubstituted one which is slowly degraded in the culture medium. B. Penetration of the ras anti-sense 9-mer linked to the dodecanol substituent into the cytoplasm and nuclei of T24 cells. T24 cells were incubated in the presence of 1 /tM of unlabelled oligomer with 5 x 106 cpm of 5' 32P-labelled oligomer. At the indicated time periods, cells were collected and fractionated into nuclei (N) and cytoplasm (C) fractions. The nucleic acids were extracted from the two fractions and subjected to electrophoresis on 20% polyacrylamide gel containing 7 M urea. The lane labelled 0 corresponds to the control oligonucleotide. The bands migrating above the oligonucleotide in the nucleus after 6 h and in both the cytoplasm and nucleus after 24 h are due to radioactivity incorporated in cellular macromolecules (RNA and DNA) after removal of the 5' [32P]phosphate from the 5'-end of the oligonucleotide by phosphatases.

Figure 5B shows the uptake of the 9-mer linked to the dodecanol substituent by T24 cells. One hour after addition to the culture medium the labelled oligonucleotide was detected in both cytosolic and nuclear fractions. After 24 h of incubation radioactivity was incorporated into nucleic acid macromolecules both in cytoplasm and nucleus. This arises from a dephosphorylation of the oligomer occurring in the culture medium, free phosphate being taken up by cells and incorporated into nucleic acids. However even after 24 h incubation there was still undegraded oligomer both in cytosolic and nuclear fractions. It is noteworthy that for each incubation time there was a two-fold higher incorporation in nucleus than in cytoplasm. The distribution of the cytosolic and nuclear fraction of the unmodified 9-mer was the same as that of the 9-mer linked to the dodecanol tail but its uptake efficiency was four times less (data not shown). These results indicate that the dodecanol tail increases uptake of the 9-mer by T24 cells. After lh of incubation about 9% of the initial radioactivity was found inside cells. A steady-state value of the intracellular oligonucleotide concentration was reached within 1 h since the radioactivity associated with the band corresponding to the labelled oligonucleotide extracted from cells after 3,

6 and 24 h remained almost unchanged. Using rough estimates of cell volume we calculated that the intracellular concentration of the oligomer was around 4 zM. Recently it was reported that oligonucleotides are taken up by cells in a saturable, size-dependent manner compatible with receptor-mediated endocytosis (Loke et al., 1989). Our results show that a short oligodeoxynucleotide (9-mer) linked to a dodecanol tail was taken up very efficiently by T24 cells. When the oligonucleotide was further substituted by the (fluorescent) acridine derivative its localization inside cells could be followed by fluorescence microscopy. A punctate distribution of fluorescence was observed in the cytoplasm suggesting that a large part of the oligonucleotide was trapped in vesicles (endosomes) but not available at the site of mRNA translation. The oligomer Acr-m5-CCACACCGA-m120H complementary to human activated Ha-ras inhibits proliferation of human T24 bladder carcinoma cells It was reported that NIH 3T3 cells induced to divide by adding serum to the culture medium were unable to enter the S phase of the cell cycle after microinjection of anti-ras antibody showing that the protein product of the ras

1 1 15

T.Saison-Behmoaras et al.

60000

0 os 0

=

40000

co -

0

.0 E

20000

i

4

3

2

5

day Fig. 6. Effect of Acr-m5-5-CCACACCGA-3'mI20H antisense 9-mer on the proliferation of T24 cells (3 x 103 cells) growing in DMEM medium in the presence of 7% foetal calf serum. Cells were exposed to antisense oligomer complementary to activated Ha-ras or to a non-specific (NS) oligomer. Each oligonucleotide was added at the beginning of the experiment at a concentration of 1 /iM. The cell culture medium with the oligomer was changed on the third day. The sequence of non-specific 9-mer (NS) is Acr-m5-5'-ACGCAACCC-3'-m120H.

protooncogene is required for initiation of replication (Mulcahy et al., 1985). T24 human bladder carcinoma cells express activated Ha-ras with the G - T mutation described in Figure 1. In order to check the effects of antisense oligomers on T24 cell growth we added 1 ,uM of 9-mer linked to acridine and dodecanol substituents to the culture medium and counted the cells daily over a 5 day period. The medium containing the oligomer was changed once on the third day and all assays were carried out in triplicate.

~60O 0 C

0

Figure 6 shows that addition of 1 MtM of antisense oligomer to T24 cells produced 60% decrease in the growth rate over a 5 day period. Cell growth was not influenced by the

addition of 1 MM of a non-specific oligomer carrying the same substituents as the antisense oligomer. It should be noted that the non-specific oligonucleotide had the same base composition as the specific one but a different sequence. T24 cells grown 24 h with 5 MM of modified 9-mer contained about 50% less total ras protein (as measured by immunoprecipitation) than did either control T24 cells or cells grown with 5 /sM of non specific oligomer (results not shown). Inhibition of cell growth associated with antisense oligomer was dose-dependent over the range of 0.1-2 MtM (Figure 7). It is noteworthy that growth inhibition of T24 cells treated with the 9-mer linked to the acridine and dedecanol substituents reached a plateau at 60% inhibition. The modified 9-mer specifically recognizes activated Ha-ras and is not expected to alter the expression of other ras species (the 9-mer has only 5 bases complementary to Ki-ras and 6 bases complementary to N-ras). The p21 proteins corresponding to Ki-ras and N-ras should still be able to support cell division. The proliferation of the nontumorigenic human mammary cell line HBL 100 (which contain only normal ras) was not affected by the addition of the antisense 9-mer (Figure 7) suggesting, as in studies in vitro, that the effects of the modified 9-mer are due to 1 116

20 C

1

2

3

[oligonucleotide] PM Fig. 7. Dose-response curve for T24 growth inhibition by antisense oligomer. Percent inhibition was calculated by comparison with the number of cells present in the control cultures grown in the absence of oligomers. T24 (0,A) or HBL100 (,X) cells were grown for 5 days in the medium supplemented with 1 AM anti-ras 9-mer substituted by acridine (5') and dodecanol (3') (oligo (4)) (0,U) or with 1 AM of non-specific 9-mer substituted by acridine and dodecanol (oligo (8)) (A,X). Each point represents the mean of triplicate cultures with error bars as indicated.

its inhibitory effect on activated ras. NIH 3T3 cell growth also unaffected by addition of mutated Ha-ras-specific and non-specific 9-mers (data not shown). was

Discussion Although several groups have reported the successful use of oligodeoxyribonucleotides to inhibit the expression of a single gene inside cultured cells (Heikkila et al., 1987; Harel-Bellan et al., 1988; Holt et al., 1988; Jaskulski et al.,

Oncogene inhibition by antisense oligonucleotides

1988), Tidd et al., have reported the failure to inhibit expression of the N-ras oncogene in T15 cells (a line of NIH 3T3 cells transfected with multiple copies of the human N-ras oncogene under control of the glucocorticoid-inducible MMTV promoter) with either a 9-mer oligomethylphosphonate or a 20-mer consisting of a phosphodiester sequence flanked by two methylphosphonate linkages at each end (Tidd et al., 1988). Moreover it was reported that an 1 1-mer antisense methylphosphonate directed to the initiation codon region and an 8-mer targeted to the 12th codon region of human c-Ha-ras inhibited p21 translation only at very high concentrations (100-200 /tM) in a rabbit reticulocyte lysate only after preannealing of mRNA and oligomers by heating to 90°C and then cooling gradually to room temperature (Yu et al., 1989). Since then it has been shown that phosphodiester oligomer-RNA duplexes but not methylphosphonate oligomer-RNA duplexes are substrates for RNase-H (Walder and Walder, 1988; Cazenave et al., 1989; Furdon et al., 1989). The three most likely mechanisms of inhibition of gene expresson by antisense oligomers appear to be (i) degradation of the RNA transcript by RNase H at the site of oligonucleotide binding independently of whether the oligonucleotide is targeted to a coding or non-coding region (Cazenave et al., 1989); (ii) physical block of pre-mRNA splicing when the oligonucleotide is targeted to an intron-exon junction (Kulka et al., 1989); (iii) physical inhibition of mRNA translation initiation when the oligonucleotide is targeted to a sequence close to the CAP site in the 5'-untranslated region (Goodchild et al., 1988). In this report, the inhibitory effect of a series of 9-mers complementary to the 12th amino acid codon in activated Ha-ras from T24 bladder carcinoma cells was tested in a cell-free translation system. We demonstrated that RNase H activity was required for the sequence-specific inhibition of p21 synthesis by antisense 9-mers. Efficient inhibition was obtained with the 9-mer linked to the acridine at its 5' end and to a dodecanol substituent at the 3' end: 0.5 ,LM of this oligomer inhibited 50% of p21 synthesis in cell free translation systems. The sequence specificity was further tested by replacing the mutated Ha-ras mRNA being translated in the rabbit reticulocyte lysate by the mRNA which encodes the normal Ha-ras. The single nucleotide change, from G to T at the 12th amino acid codon in the ras gene, was sufficient for discrimination between the two mRNAs by the 9-mer oligonucleotide. Indeed the modified 9-mer at 1 gM and 37°C induced preferential RNase-H cleavage of the messenger RNA coding for mutated ras. These results demonstrate that it is possible to design an anti-ras oligomer which will preferentially inhibit the synthesis of the mutated ras p21 protein. Furthermore we demonstrated that at 1 ,LM the 9-mer linked to the acridine and dodecanol substituents inhibited the proliferation of T24 cells containing only the activated Ha-ras whereas the proliferation of human mammary cells containing normal Ha-ras was unaffected. The doubly substituted 9-mer oligonucleotide is efficient at 1 ,tM: (i) acridine protects the oligomer against exonucleases (Figure 5); (ii) the hydrophobic tail increases oligonucleotide uptake by cells; (iii) the acridine and hydrophobic tail stabilize 9-mer interaction with the mRNA, enough to induce RNase H cleavage only with a perfectly hybridized target since one mismatch destabilizes the oligomer. Since point mutation in the 12th or 61st codons has been found in numerous chemical and radiation induced as well as naturally occurring malignant mammalian tumors

the design of such modified oligonucleotides might provide new tools to explore the function of ras proteins and their role in cell transformation.

Methods and materials Synthesis and purification of oligomers Unmodified oligodeoxyribonucleotides were synthesized on an automated solid-phase synthesizer (Applied Biosystems Inc) by using standard phosphoramidite chemistry and were purified by high pressure liquid chromatography. Oligonucleotides linked at their 3' end to 2-methoxy, 6-chloro, 9-amino acridine via a pentamethylene bridge, were synthesized according to a previously published procedure (Asseline et al., 1986). Oligonucleotides linked at their 5' end to an acridine were synthesized on an automatic solid phase DNA synthesizer as previously described (Thuong and Chassignol, 1988). The synthesis of oligonucleotides bearing an c-hydroxydodecamethylene group at the 3' end was performed via the solid support 3 which was obtained in a four-step process from 1,12 dodecanediol (scheme 1) according to a described procedure (Atkinson and Smith, 1984) by replacing the 5'-dimethoxy-trityl-nucleoside with the hydroxyl compound 1. From the support 3 the assembly of the oligonucleotide chain, together with the incorporation of the acridine at the 5' end, and the unblocking of the oligomer were carried out following the procedure previously described (Thuong and Chassignol, 1988). Oligodeoxynucleotides were labelled at the 5' end using T4 polynucleotide kinase and [-y-32P]ATP (Amersham) and were analyzed on a 20% polyacrylamide-7 M urea sequencing gel.

RNA isolation, in vitro translation and immunoprecipitation Total cytoplasmic RNA was purified from confluent cultures of Chinese hamster lung fibroblast line (CCL39) transfected with T24 Ha-ras as described (Maniatis et al., 1982). These cells (39 THac) express very high levels (> 10-fold over wild type cells) of activated Ha-ras mRNA and p21 protein (Seuwen et al., 1988). Wheat germ extract and rabbit reticulocyte lysate were purchased from Genofit (Geneva) or Amersham. Oligomers were added to a translation mixture containing 20 Mg of total RNA, 20 zM amino acid mixture (minus methionine) and 3.7 MBq of [35S]methionine. The reaction mixture with wheat germ extract (60 itl) was incubated 30 min at 25°C. The reaction mixture with rabbit reticulocyte lysate (50 ML) complemented or not with RNase H from E.coli (Genofit) was incubated at 30°C for 45 min. After incubation the translation mixture was diluted in 500 M1 of extraction buffer (1% Triton X-100/0.1% NaDoSO4/0.5% sodium deoxycholate/0. 1 M NaCl/ 10 mM sodium phosphate pH 7.4/1 mM phenylmethane sulfonyl fluoride/aprotinin (100 Kallikrein inactivator units per ml) and incubated with monoclonal antibody Y13-259 (Cliniscience) overnight at 4°C. The immunocomplexes were collected 2 h at 4°C with protein A - Sepharose (Pharmacia) coated with rabbit anti-rat IgG (Cliniscience). The Sepharose was washed four times in extraction buffer; the proteins were then dissolved in electrophoresis buffer (0.25 M Tris-HCI, pH 6.8, 0.1 M dithiothreitol, 10% glycerol, 2 % SDS) treated at 90°C for 5 min, and separated on 0.5% SDS, 13.5% polyacrylamide gels. After fixing, the gels were soaked in Amplify (Amersham) dried and fluorographed at -70°C. Autoradiographs were then scanned with a densitometer (LKB, ultrascan XL) and the area under the peak corresponding to the p21 protein was determined. In vitro transcription and RNase H cleavage A BamHI -Sall fragment of 760 bp containing the entire coding region of normal and activated Ha-ras (mutation on the 12th codon; GGC - GTC; Gly - Val) was introduced into the BamHI site of the plasmid pBT+ using a BamHI linker attached to the Sail site of the fragment. BamHI inserts were ligated into the BamHI site of plasmid pGEM-l (Promega-Biotec). The orientation (determined with SnaI) is SP6 3'........ 5' T7 for both constructs. The base sequence at the 12th codon (GTC) was confirmed by NaeI digestion. Uniformly labelled capped SP6 transcripts were synthesized in a 50 M1 transcription reaction containing 40 mM Tris (pH 7.5), 6 mM MgCl2, 2 mM spermidine, 10 mM DTT, 0.5 mM ATP, CTP, 0.1 mM GTP, 0.2 mM RNA cap analogue [m7G(5')ppp(5')G]; 0.5 mM UTP, 3.7 MBq of [a-32p] UTP (30 TBq/mmol), 80 U RNasin (Promega), 10 U SP6 polymerase (Boehringer) and 2 MLg of EcoRI linearized plasmid. The reaction mixture was incubated for 1 h at 40°C. After phenol -chloroform extraction and ethanol precipitation the synthesized RNA was suspended in 20 ul of water and 2M was used for in vitro assays with rabbit reticulocyte lysate (Amersham) supplemented with 25 mM methionine. Samples were incubated at 37°C for 1 h and subjected to electrophoresis on 6% polyacrylamide gels containing 7 M urea. 0.8 units of RNase H from E.coli

1117

T.Saison-Behmoaras et al. (Promega, Biotec) and oligomers were added when indicated to the reaction mixture. Artificial chain termination reaction using 3'-0-methyl guanosine triphosphate Chain termination of nascent RNA transcripts by incorporation of 3'-Omethyl guanosine triphosphate was utilized to generate RNA chains of known length and sequence from SP6 promoter. RNA synthesis reactions with modified chain-terminating substrate was performed as in the standard in vitro synthesis assays except for the addition of 0.1 ,AM 3'-0-methyl guanosine (Pharmacia).

Cell lines and proliferation assays Human urinary bladder cancer cell line T24, rat NIH 3T3 and human HBL 100 cell lines were obtained from the American cell culture collection (Rockville, USA). Cells were grown in Dulbecco's modified Eagle's medium (DMEM, Flow) supplemented with 7% heat inactivated (30 min at 65°C) foetal calf serum, antibiotics (50 U/mI of penicillin and 50 U/mi streptomycine) and 4 mM glutamine. Cells were seeded on microtitre plates at a density of 2 x 103/well. Oligomers were added at various concentrations, each well contained 200 ,u medium. After 3 days the microcultures were fed by replacing 200 Id medium with fresh medium containing oligomers until the fifth day. Cell numbers were determined daily using a counting chamber for quantification of cells or colourimetric assay using (3-(4,5-dimethyl thiazol-2-yl) 2,5-diphenyltetrazolium bromide) as described (Denisot and Lang, 1986). Treated and untreated cells showed 98-100% viability after 5 days of growth. In 5 days the number of untreated cells was increased 30-fold on the average. Transport and stability of oligonucleotides in T24 cells For each time point 5 x 106 c.p.m. of 5' 32P-labelled oligodeoxynucleotide (2 pmole of labelled oligonucleotide and 1 /LM carrier unlabelled oligonucleotide) was added to 5 x I05 T24 cells in 2 ml of cell culture medium. Following incubation for the specified period, the cells were collected and oligomer extracted from cytoplasm and nuclei as previously described (Teichman-Weinberg et al., 1988). Oligonucleotide extracted from 50 41 culture medium, cytoplasm and nucleus was subjected to electrophoresis on 20% polyacrylamide gels containing 7 M urea.

Acknowledgements We thank Dr J.Pouyssegur and M.Goubin for providing 39 THac and HBL100 cells, respectively. Drs F.Dautry, H.Neel and M.Boidot-Forget for construction of plasmid vectors and I.Duroux for technical assistance. This work was supported in part by the Ligue Nationale Francaise contre le Cancer.

References Anfossi,G., Gewirtz,A. and Calabretta,B. (1989) Proc. Natl. Acad. Sci. USA, 86, 3379-3383. Asseline,U., Delarue,M., Lancelot,G., Toulme,F., Thuong,N.T., Montenay-Garestier,T. and He&ne,C. (1984) Proc. Natl. Acad. Sci. USA, 81, 3297-3301. Asseline,U., Thuong,N.T and Helene,C. (1986) Nucleosides Nucleotides, 5, 45-63. Atkinson,T. and Smith,M. (1984) In Gait,M.J. (ed.), Oligonucleotide Synthesis-A Practical Approach. IRL Press, Oxford, pp. 35-81. Bos,J.L. (1989) Cancer Res., 49, 4682-4689. Cazenave,C., Loreau,N., Toulme,J.J. and Helene,C. (1987) Nucleic Acids Res., 15, 4717-4736. Cazenave,C., Stein,C.A., Loreau,N., Thuong,N.T., Neckers,L.M., Subasinghe,C., Helene,C., Cohen,J.S. and Toulme,J.J. (1989) Nucleic Acids Res., 17, 4255-4273. Denisot,F. and Lang,R. (1986) J. Immunol. Meth., 89, 271-277. Furdon,P.J., Dominski,Z. and Kole,R. (1989) Nucleic Acids Res., 17, 9193-9204.

Goodchild,J., Caroll,E.,III and Greerberg,J.R. (1988) Arch. Biochem. Biophys., 263, 401 -409. Harel-Bellan,A, Ferris,D.K., Vinocour,M., Holt,J.T. and Farrar,W.L. (1988) J. Immunol., 140, 2431-2435. Heikkila,R., Schwab,G., Wickstrom,E., Loke,S.L., Pluznik,D.H., Watt,R. and Neckers,L.M. (1987) Nature, 328, 445-449. Helene,C. and Toulme,J.J. (1989) In Cohen,J.S. (ed.), Oligodeoxynucleotides: Antisense Inhibitors of Gene Expression. Macmillan Press, London, pp. 137- 172.

1118

Holt,J.T., Redner,R.L. and Nienhuis,A.W. (1988) Mol. Cell. Biol., 8, 963-973. Jager,A., Levy,M.J. and Hecht,S.M. (1988) Biochemistry, 27, 7237-7246. Jaskulski,D., de Riel,J.L., Mercer,W.E., Calabreta,B. and Baserga,R. (1988) Science, 240, 1544-1546. Kulka,M., Smith,C.C., Aurelian,L., Fishelevich,R., Meade,K., Miller,P. and Ts'o,P.O.P. (1989) Proc. Natl. Acad. Sci. USA, 86, 6868-6872. Letsinger,R.L., Bach,S.A. and Eadie,J.S. (1986) Nucleic Acids Res., 14, 3487-3499. Loke,S.L., Stein,C.A., Zhang,X.H., Mori,K., Nakanishi,M., Subasinghe,C., Cohen,J.S. and Neckers,L.M. (1989) Proc. Natl. Acad. Sci. USA, 86, 3474-3478. Maniatis,T., Fritsch,E.F. and Sambrook,J. (1982) Molecular Cloning. A Laboratory Manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. McCormick,F. (1989) Cell,56, 5-8. Minschull,J. and Hunt,T. (1986) Nucleic Acids Res., 14, 6433-6451. Mulcahy,L.S., Smith,M.R. and Stacey,D.W. (1985) Nature, 313, 241-243. Reddy,E.P., Reynolds,R.K., Santos,E. and Barbacid,M. (1982) Nature, 300, 149-152. Seuwen,K., Lagarde,A. and Pouyssegur,J. (1988) EMBO J., 7, 161-168. Teichman-Weinberg,A., Littauer,U.Z. and Ginzburg,I. (1988) Gene, 72, 297-307. Thuong,N.T. and Chassignol,M. (1988) Tetrahedron Lett., 29, 5905-5908. Tidd,D.M., Hawley,P., Warenius,H.M. and Gibson,I. (1988) Anti-cancer Drug Design, 3, 117-127. Ulsh,L.S. and Shih,T.Y. (1984) Mol. Cell. Biol., 4, 1647-1652. Walder,R.Y. and Walder,J.A. (1988) Proc. Natl. Acad. Sci. USA, 85, 5011-5015. Yu,Z.P., Chen,D.F., Black,R.J., Blake,K., Ts'o,O.P., Miller,P. and Chang,E.H. (1989) J. Exp. Pathol., 4, 97-108. Received on December 21, 1991; revised February 7, 1991