Short Antisense Oligonucleotide-mediated Inhibition Is Strongly ...

Vol. 269,No.23,Issue of June 10,pp. 16187-16194, 1994 Printed in U.S.A.

THEJOURNALOF BIOLOGICAL CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc

Short AntisenseOligonucleotide-mediated Inhibition Is Strongly Dependent on Oligo Length and Concentration but Almost Independent of Location of the Target Sequence* (Received forpublication, January 26, 1994, and in revised form, March 30, 1994)

Bernd FaklerSO, Stefan HerlitzeO, Beate AmthorO, Hans-Peter ZennerS, and Johann-Peter Ruppersberg+§ll From the SSektion fur Sensorische Biophysik der HNO-Universitatsklinik Tubingen, 0-72076Tubingen, Germany and the §Mar-Planck-Znstitutfur Medizinische Forschung, Abteilung Zellphysiologie, 0-69120Heidelberg, Germany

5’-untranslated region or the initiation site on the mRNA (6, Theinhibitoryeffectofshortantisenseoligodeoxynucleotides (aODNs) on cRNA expression in Xenopus 14). In both mechanisms protein expression is suppressed at oocytes was measured using an electrophysiological as-the RNA level. say based on subunit-specific block of cloned cu-amino- The main scientific application of aODNs is expression inhi3-hydroxy-5-methyl-4-isoxazole-propionatereceptors. bition of mRNAs with known nucleotide sequence in order to role of proteins by selectively inhibThe effect of both phosphorothioate-modified (PS)and characterize the functional iting theirexpression. By aODN-mediated inhibition of expreslengthdephosphodiester (PO) aODNswasstrongly pendent with a half-maximal inhibition calculated for sion of integrin a1 and PI, Lallier and Bronner-Fraser demonrole of these proteins in the 7.6 nucleotides (nt) and 9.9 nt, respec- stratedthefundamental an oligo length of in tively. More than 96%inhibition was mediated by a PS attachment of neural crestcells to extracellular matrices and aODNof 12 nt and by PO aODNs 2 15 nt. At a given cellular migration (15). Reis and co-workers inhibited expression of NMDA receptor channels withaODNs to elucidate their length PS and PO aODNsshoweddifferentialdependence of their inhibitory effect on the injected aODN role in Ca2+-dependent focal ischaemic infarction (16). concentration (half-maximal inhibition at 18 ng/pl for a On the other hand, mRNAs with unknown nucleotide sePO 12-mer and at0.19 ng/pl fora PS 12-mer) and differ- quence can be inhibited by aODNs or by antisense cRNAs or entialsaturationbehavior.Theinhibitoryeffectof cDNAs. Gundersen and Umbach (17) introduced a so-called suppression cloning technique in which positive cDNA clones aODNs, even as short as 8 ntfor PS oligomers,was highly sequence specific, but almost independent of the were selected accordingto theircapability to inhibit the expresPS sion of the gene of interest. Instead of antisense cDNAs or position of the respective target site on the cRNA (for 8-mers, 270% expressioninhibitionthroughoutthe cRNAs it might be possible to use an oligodeoxynucleotide litested target sitesfrom the translation initiation to the brary to obtain sequence information of a n unknown proteinby 3’-untranslated region). successively testing the inhibitorypotency of many individual Thus, shortPS aODNs can be reliably used in order to oligomers of such a library. To realize this strategy, however, specifically inhibit protein expression in experiments oligodeoxynucleotides must fulfill at least thefollowing condiaddressing physiological, molecular biological, and pertions: oligos should be short in order t o cover a high percentage haps even therapeutical issues. of all possible sequence variations in a library of reasonable size. Nevertheless, the inhibitorypotency of these short oligos must be strong enough to guarantee significant results in the Since theirintroductionin1978 (1) antisense oligodebioassay used to check suppression of the mRNA expression. oxynucleotides (aODNs)’ have been used more and more as Moreover, the inhibitory potency of short aODNs should be powerful tools for specific inhibition of protein expression in independent of the position of their target sequence on the vitro (2-6) andin vivo (7-11). The specificity of aODNs is due to mRNA. ahighly specific hybridization of these oligomers totheir Since short unmodified aODNs were found to have rather complementary target sequence on the mRNA, which causes weaksuppressing effects (5, 10) modified aODNs withininhibition of protein expression by at least two now widely creasedstabilitytoenzymaticdegradationand/orwithinaccepted mechanisms.Thefirst is degradation of RNA by creased hybrid stability were usedto increase their inhibitory RNase H, which selectively cleavesthe RNA at DNA-RNA heteffect. Modification of the phosphodiester bonds to phosphoroeroduplexes (7,10,12, 13).The second mechanism is the arrest thioates or covalent linkage of an intercalating agent to the 3’of translation initiation caused by aODN hybridization to the and/or 5”terminal residue were found to be most successful. The shortest modified oligo for which an inhibitory effect has * This work was supported by the DFG (KliFo Horforschung). The costs of publication of this article were defrayedin part by the payment been described was a nonamer, towhich an acridine derivative of page charges. This article must thereforebe hereby marked “adver- and a dodecanol chain were covalently linked (18). tisement” in accordance with 18 U.S.C. Section 1734 solely to indicate Short oligomers have several potential advantages.Specificthis fact. ity is higher for short oligos as pointed out by Woolf and coll To whom correspondence should be addressed: Sektion fur Sensorische Biophysik derHNO-Universitatsklinik Tiibingen, Wntgenweg workers (191, who found that the ratio of intended to uninis reduced for longer oligos. 11, 72076 Tiibingen,Germany. Tel.: 07071-29-4162;Fax: 07071-29- tended RNA degradation 3311. Furthermore, short antisense oligomers seem t o be less often The abbreviations used are: aODNs, antisense oligodeoxynucleoti- prevented from hybridization by secondary structures of the des; PS aODNs, phosphorothioate-modifiedaODNs; PO aODNs, phosphodiester aODNs; nt, nucleotides; ATX, argiotoxin,,; AMPA, a-amino- RNA (20). Another advantageis a faster cellular uptake,which 3-hydroxy-5-methyl-4-isoxazole-propionate;GluR-B,,GluR-D,, flip was found to be size dependent, showing higher rates of transsplice variants of the glutamate receptors B and D; UT, untranslated. port for a septamer than for 15- or 20-mers (21). Although the

16187

16188

Short Antisense Oligonucleotide-mediated Inhibition

advantages of short aODNs are evidentfrom the previous work of their inhibiof many groups,a detailed quantitative analysis tory effect under various conditions is still lacking. In this study we investigate the effect of short aODNs in dependence of length, concentration, and localization of their target site on the RNA. To provide exact quantitative data about the efficiency of inhibition of RNA expression, we introduce an electrophysiological assay based on subunit-specific Xenopus oocytes. block of cloned AMPA receptors expressed in This assayallows relatively precise estimates of aODN effects.

aODN a t a concentrationof 1pg/pl. The final concentrationof aODN in the injection solution was thus 0.5 pg/pl. For the controls the aODNcontaining volume was replaced by 1 pl of water. aODN concentration to the final concentration in refers to the injected solution rather than the oocytes, since the latter is strongly influenced by the volume injected and by the volume of the free diffusion compartment of the oocyte. If the latter volume is assumed to be 0.5 pl, the final concentration of a phosphodiester octamer is about 19p ~ . To obtain the calibration curve (Fig. 2) the ratio of GluR-B,-/GluRDl-cRNAs was varied in the cRNA mixture in the following way (volumes in111): for the ratioof 10:1(10 GluR-B,-cRNA,1GluR-Dl-cRNA,1.5 H,O), for the ratioof 5:1(5,1,6.5), for the ratioof 2.5:1(2.5,1,9), for the ratio of 1:l (1,1, 10.5), and for the ratio of 1:5 (1,5, 6.5). MATERIALSANDMETHODS Current Recording and Data Analysis-All experiments were perRNA Synthesis-Complementary DNAs encoding the AMPA receptor formed in a recording chamber continuously perfused with bathing subunits GluR-B, and GluR-D, (i refers to the flip splice variants of solution (Mg-NFR) containing (in mM) 115 NaCl, 2.5 KCl, 1.8 MgCl,, 10 GluR-B and GluR-D (22)) were subcloned into pSP64T- (23) derived HEPES (pH 7.2). Both the agonist, kainate (Sigma, 100 PM), and the vectors with most of the genuine 5'- and 3"untranslated regions reantagonist, ATX (5 J~M),were dissolved in Mg-NFR. moved (24). Plasmids were linearized 3' of the poly(A) stretch, and two-microelecsubunit-specific cRNAs were synthesized in vitro using SP6 polymerase Currents in response to kainate were measured with (23). cRNAs were then stored at -80 "C in stock solutions a t a concen- trode voltage clamp a t a clamp potential of -70 mV and recorded on a Graphtec chart recorder. Current and voltage electrodes were filled tration of about 1 pg/pl. Oligodeoxynucleotide and ATX Synthesis-Unmodified and phospho- with 3 M KC1 and had resistancesof 0.5-5 MR. ) Each recording started with the application of kainate (100 p ~ Mgrothioate oligodeoxynucleotides were synthesizedon a n Applied Biosystems 380B automated synthesizer using standard cyanoethyl phos- NFR to obtain the control current mediated by the AMPA receptor ofATX. Next we switched toATX (5 PM), kainate phoramiditechemistry.Forthepreparation of phosphorothioate channels in the absence linkages tetraethylthiuram disulfide (TETD) in acetonitrile was toused(100 J~M),Mg-NFR to induceATX block. When the ATX block had almost convert the internucleotide cyanoethyl phosphite to phosphorothioate reached steady state, stability of the base line was checked by switching triester. In allPS aODNs, the internucleoside linkages werecompletely back to agonist-free Mg-NFR. Following another agonist application, phosphorothioate-modified. Coupling, oxidation, or sulfurization and ATX was applied for a second time until the current no longer decleavage from the column were performed as usual. After deprotection creased, finally reaching steadystate of the ATX block. oligomers were dissolved in water, precipitated with ethanol, and reThe ATX block was determinedfrom these recordingsas percentage dissolved in water. Concentrations were determined by assuming 1A,,, reduction of the current in thepresence of ATX relative to the control unit = 33 pg (19). current. The oligomer length purity was evaluatedby an analytical run on a The dependence of ATX block ( B ) on the ratio of GluR-B,-/GluR-D,20% polyacrylamide gel stained with ethidium bromide. For all oli- cRNAs (r)(Fig. 2) was fitted with the logistic function: gomers 518 nt, only the correct band could be detected on the gel. Oligos 5 3 0 n twere checked to be about 90% full-length material. B(r) = I(B,- - B,,,,)41 + W 0 5)n)1 + B,,, (Eq. 1) The synthesizedoligodeoxynucleotides, which wereall designed with the help of commercial software (OLIGO), were antisense to GluR-B,- with the asymptotic value B,,, (0.03) being the relative ATX block cRNA and had thefollowing nomenclature and sequences (the adenine measured for homomeric GluR-B, receptor channels and B , (0.95) of the initiation codon of the GluR-B,-cRNA is counted as +l;all sebeing the respective block determined for GluR-D, homomers. ro (0.99) quences are written in the 5' to 3' direction): (i) phosphodiester oli- is the subunit ratio for half-maximal ATX block and n the Hill coeffigomers: PO8-1 (8-mer, sequence complementary to nucleotides 1-8 of cient (1.61). GluR-B,), TTTTGCAT; P010-1, TCTTTTGCAT; PO12-1, AATCTTTTThe relative activityof GluR-B,-cRNA was calculated asfollows: in a GCAT PO15-1,CATAATCTTTTGCAT;PO18-1,ATGCATAATCTTI" first step the mean ? S.D. of the ATX block was determined in the TGCAT; P O l W 8 3 , GAATCCAGAACAGC'ITGC; PO18-983, GGTTG- absence (controls,n 2 10 for each injection series) andin thepresence GCCAAACAATCCC; ~018-1483, TAATGGAGCAATTGCAAT; polsof aODNs (usually7 5 n 5 10). With helpof the calibration curvein Fig. 1983, GCAATTTCTGTTTGCTTA,PO18-2483,CAATCAAAGCCACC2, the meansof both block values were then correlated to certain cRNA AGCA, PO21-1, AATATGCATAATCTTTTGCAT; PO30-1, GAGGACAratios. In a second step the remainingcRNA activity in the presenceof GAAATATGCATAATC'ITITGCAT; (ii) Phosphorothioate oligomers: aODNs was estimated by dividing the two cRNA ratios determined in PSG1511, TCACGA; PS8-1, T"GCAT; PS8-736, AGCAGGTC; PS8the presence and in the absence of aODNs. The (mean + S.D.) and 1511, TCTCACGA; PS8-2134, ACTCTGGC; PS8-2557, ATATTCTG, PS8-3'UT (8-mer, sequence complementary to a stretch in the 3'-un- (mean - S.D.) values of ATX block were also correlated tocRNA ratios and divided by the control cRNA ratio; the resulting deviations were translatedregion), ATCCCTCG; PS10-1, TCTTTTGCAT PS12-1, then referred to as S.D.*.Therefore, the control values of the relative AATCTTTTGCAT. All aODNs listed under (i) and (ii) are specific for their target sequence; that is, for oligo an length of 8 nt they have 5 6 activity of GluR-B,-cRNAare also presented withan errorbar. To fit the relative activity of GluR-B,-cRNA in dependence on the consecutive matching base pairs in common with other positionson the cRNAs of GluR-Bi and GluR-D,for oligos 2 10 nt therespective number length of the aODNs, we calculated the hybridization free energies(AG) according to nearest neighbor analysis (26). Then a Boltzmann curve is 57, and for PS€L1511 the respective number is5 5. (iii) PhosphoroGluR-B,-cRNAactivity thioate oligomer targeting theGluR-B,- and GluR-D,-cRNA(nomencla- could be fitted to the experimentally determined ture refers to the target position on GluR-B,-cRNA):PS8-204, AACT- (A) with respect to the calculatedAG values: G'ITG. (iv) Phosphorothioate oligomer targeting neither the GluR-B,A(AG) = ((1- c)/(l + exp [(AG - AG,,)/k])J + c (Eq. 2) nor the GluR-Dl-cRNA PS8-0, GGCGGTGG. (v) Phosphorothioate oligomer targeting theGluR-Dl-cRNA:PS12-lD,, AATAATCCTCAT. is the midpoint of the curve, c is the relativecRNA activity which ATX was a gift from Drs. W. Jahn, M. Raditsch, and V. Witzemann could not be inhibited, and k determines the amount of free energy and was synthesized as described previously (25). required for an e-fold change of cRNA activity. Preparation a n d Injection of Oocytes-Xenopus oocytes were surgiEquation 2 represents a simplified form of a two-state hybridization cally removed from adult females and manually dissected. About 50 nl of a solution containing the respectivecRNAs and antisenseoligomers model, in which the concentration of aODNs, differing for the various was injected into Dumont stage VI oocytes. Oocytes were then incu- oligo lengths, is neglected. We therefore used an extended fit formula bated at 19 "C and treated with collagenase typeI1 (Sigma, 0.5 mg/ml) which takes changes in aODNs concentration into account. Nevertheabout 12 h after injection. After collagenase treatment thefollicle cell less, the values for the slope factor ( k ) resulting from this fit differed 2 less than2.5% for PS and layer was manuallyremoved. Electrophysiological measurements were from the values obtained with Equation by PO aODNs. done 45-60 h afterinjection. The dose-response curves of Fig. 5 were fitted as usual. Unless otherwise stated the injection solution wascomposed of 1 p1 All computational work was done on a n Apple Macintosh computer of cRNA-mixture (volumes in pl: 5 GluR-B,-cRNA, 1 GluR-Dl-cRNA,6.5 commercial software (Igor, wave metrics; CANVAS). H,O) containing GluR-B,-/GluR-D,-cRNAsat a ratio of 5:l and 1 pl of (Quadra 800) using

16189

Short Antisense Oligonucleotide-mediated Inhibition KAINATE A

..

KAINATE

ATX

1.0

1

ATX

KAINATE B

KAINATE

..

ATX

ATX

0. I

I .o

IO

ratio of injected GluR-Bi- / GluR-Di-cRNA 1 mm

C

- -

KAINATE

ATX

KAINATE

ATX

_ 1

1

FIG.2. Dependence of relative ATX block on the injected ratio of GluR-B,-/GluR-D,-cR.GluR-B,- and GluR-D,-cRNAs were injected into oocytes in ratios of lO:l, 51, 2.5:1, 1:1, and 1:5 (see “Materials andMethods”),and theATX block was determined from measurements as in Fig. 1. Each data point represents the mean 2 S.D. of 10 measurements. The line represents the fit of a logistic function to the mean values (see “Materials and Methods”).The ratio for half-maximal block ( F ~ , and ~ ) the Hill coefficient obtained from the fit were 0.99 and 1.61, respectively.

located at the so called Q/R site of the putative pore aligning region of the receptor channel (24). The GluR-A, -C, and -D subunits contain an uncharged glutaminethis at site insteadof I min FIG.1. ATX block of currents mediated by heteromeric GluR-B, arginine. Coexpression of GluR-Bi and GluR-D, subunits yields /Gl&D,A”Areceptor channels expressed in the presence and heteromeric receptor channels, whose sensitivity toATX is deabsence of octameric aODNs. A, controlcurrent from an oocyte pre- termined by the GluR-B, subunit; that is, heteromeric ionoviously injected with GluR-B,- and GluR-D,-cRNh at a ratio of 5:l but tropic receptors containing the GluR-B, subunit are far less without addition of aODNs. Current responses were evoked by 100 PM sensitive toATX than homomeric GluR-D, receptors. As a conkainate and blocked by 5 p~ ATX at a clamp potential of -70 mV. The sequence an unmodifiedaODN targeting the first 30 nt of duration of kainate andATX application is shown by horizontal bars.E and C, recordings as inAmade from oocytesexpressingAMPAreceptors GluR-B,-cRNA was found to increase the percentage ofATX in the presence of the phosphorothioate octamer PS8-1 ( E ) and the block of receptors forming upon coinjection of GluR-B,- and phosphodiester octamer PO&1 (C). Both oligomers targeted the first GluR-Di-cRNAs to more than 80% (27). eight nucleotides of the coding regionof GluR-B,. Note the difference in The percentageofATX blockobtained from measurements as ATX block at its steady state a t the end of the second application of kainate andATX. In E and C, aODNs wereadded to the cRNAs to yield in Fig. 1is, however, only an indirect measure of the inhibition a final concentration of 0.5 pg/d in the injection solution. of cRNA expression. To quantify expression inhibition we used the following calibration experiment: the current component RESULTS blocked by 5 p~ ATX was measured in oocytes which were The Effect of Antisense Octamers Determined with the AMPA coinjected with GluR-B,- and GluR-D,-cRNAs in ratios varying Receptor-ATX System-For a detailed analysisof the inhibitory from 1O:l to 1:5. Fig. 2 shows the dependence of the ATX block effect of short phosphodiester and phosphorothioate-modified on this ratio. The results, which could be well fitted with a aODNs on protein expression, we used cRNA expression of logistic function, show aclear correlationbetween the injected AMPA receptor subunits with different sensitivities to spithe ratio of the two cRNAs and ATX block. Although the absolute der venom Argiotoxin 636 (ATX). cRNAs of the ATX-insensitive concentration of cRNAs in thecytoplasm varies with thesize of GluR-B, and theATX-sensitive GluR-Di subunit werecoinjected the oocyte and with the injected volume (judged by eye), the into Xenopus oocytes in a ratio of 5:l (see “Materials and Meth-standard deviations of block values in Fig. 2 were remarkably ods”). The resulting AMPA receptor channels were almost in- small. This strongly suggests that the ratio of cRNAs rather the of ATX block. sensitive to ATX at a clamp potential of -70 mV (Fig. lA), so than their absolute amounts determines size that 5 p~ ATX blocked only 10% * 2% (mean t S.D., n = 10) of When we use the calibrationcurve in Fig. 2 to quantify the 100 p~ kainate. When oc- effect of aODNs we suppose that, on the RNAlevel, the the total current in response to aODNtameric aODNs targeting the first8 nt of the coding region of mediated inhibition of cRNA expression is equivalent to a reGluR-B,-cRNA(PO8-1, a phosphodiester oligodeoxynucleotide, duction of the active concentration of GluR-B,-cRNA at a conand PS8-1, the corresponding phosphorothioate) were coin- stant active concentration of GluR-D,-cRNA. In other words, jected (final aODN concentration, 0.5 pg/pl) the ATX block was the ratio of the activities of the two cRNAs is varied by the increased to42% 2 4% ( n = 11)in the PS8-1- (Fig. 1B)injected aODNs. The terms “active concentration” and “cRNA activity” for PO8-1 (Fig. E ) . refer to the cRNA ratio yielding the same ATX block of the oocytes while controlvalues were obtained This suggests that PS8-1 but not PO8-1 inhibited expression expressed receptorchannels, as measured for channels exof the ATX-insensitive GluR-B, subunit. pressed in the presence of aODNs. Thus, theATX block of 45% The determinant of the insensitivity of GluR-B receptor found for PS8-1 in the experimentof Fig. 1B can be related to channels t o ATX is a positively charged arginine residue (27) a GluR-B,-/GluR-Di-cRNAratio of 1:l while the ratio of the

16190 A


- -

KAINATE

KAINATE

B

ATX

-

KAINATE

"

KAINATE

"

ATX

ATX

. ATX

-

- . I min

"

.

.-

"

"

0

6

I

I

I

I

I

0

50

100

150

200

free energy [kJ/moll FIG. 4. Dependence of aODN-mediated expression inhibition on the lengthof PO and PS aODNs. The relative activity of GluRB,-cRNA was quantified from ATX block measurements (Fig. 3) by ap" plication of the calibration curve in Fig. 2 (see "Materials and MethATX ATX 0 ods"). Each data point represents the mean -c S.D.' of 7-10 measurements. Open circles correspondto PS aODNs and filled circles to PO aODNs. Control values are indicated by the triangle. With the exception of the PS hexamer, the sequences of the PO and PS aODNs, which are indicated from 3' to 5' on the abscissa, are complementary to the first8-30 nucleotides of the GluR-Bi-cRNA(second lane). Since the sequence of the first 6 bases of the GluR-Bi-cRNA alsoexists on the GluR-D,-cRNA(nucleotides 1819-18241, PS6-1511 which has a unique target site on GluR-Bi-cRNAwas used instead (first lane). The inhibitory effect of the corresponding PS octamer (PS8-1511) targeting the same site was not significantly different from that of PS8-1 (see also FIG.3. Length dependence of the effect of PO and PS aODNs Fig. 6A).PS oligomers longerthan 12 nt were not used because of their on ATX block. Experimental conditions and concentrations were as in potential inhibition of RNase H (45). In order to fit a reasonable enerFig. 1. A, the effect of phosphodiester aODNs increases with oligo getic model to the experimental data, the sequence of the aODNs has length. The uppermost current trace was a control trace obtained in the been correlated with the free energy of hybridization according to the same batch of oocytes in the absence of aODNs. The coinjected aODNs nearest neighbor analysis for DNA-DNA-hybrids (26) and the z values were from top to bottomP010-1, PO12-1, PO15-1, and PO18-1. B, at of the data points were adjusted to the estimated free energy indicated a given length phosphorothioate aODNs showed a stronger effect on on the lowest lane (all energy values refer to the disruption of the ATX block. Daces from top to bottom are: control, PS10-1, and PS12-1. interaction in an existing hybrid). Since in up to three nucleotides no C , ATX block of GluR-D, homomericAMPA receptors. Note the similar- spontaneous hybridization occurs, free energy changes sign (-20.9 kJ/ ity to ATX blockof GluR-BjGluR-D, receptors inthe presence of mol is assumed for the hybridization initiation energy (26)). The boxPO15-1, PO18-1, and PS12-1 in A and B . width around a particular nucleotide in lanes 1 and 2 represents the increase in hybridization free energy caused by addition of this nucleotide to the aODN sequence according to Breslauer et al. Boltzmann injected cRNAs in the same experiment was 4.6. The remaining curves (see "Materials and Methods"), fitted to the values of relative activity of GluR-Bi-cRNA withrespect to the estimated hybridization relative activityof GluR-B,-cRNAin thepresence of PS8-1 was free energy of the tested aODNs, yieldedthe following parameters: the thus 24%. amount of free energy required for an e-fold change in relative activity Dependence of the Effect of Antisense Oligodeoxynucleotides of GluR-B,-cRNA was 3.9 kJ/mol for PS aODNs and 4.2 kJ/mol for PO on Oligo Length a n d Internucleoside Linkage-As shown in aODNs, the length of aODNs mediating a half-maximal effect was 7.6 and 9.9 nt, respectively. Fig. 1, we did not observe any inhibition with the octamer

C

.

KAINATE

KAINATE

PO8-1, while PS8-1 reduced the activity of GluR-Bi-cRNAby between 70 and 77% (mean, 74%; n = 11).This difference may be caused by rapid enzymatic degradation which has been reported for phosphodiester oligos in Xenopus oocytes (5, 10, 28, 29) and which phosphorothioates lack. Since two to three times were indeed longer phosphodiester oligos of 15-30 nt in length found to mediate inhibition of RNA expression (7, 301, we extended the aODNs according to their target sequence on the GluR-B,-cRNAin the 3' direction, in each case beginning with the initiation codon of GluR-B, (see "Materials andMethods"). Fig. 3 shows representative current traces evoked by 100 V M kainate at a clamp potentialof -70 mV, where ATX was applied a t a concentration of 5 1.1~as in Fig. 1. For direct comparison controls and current traces recorded from oocytes coinjected with phosphodiester (Fig. 3 A ) or phosphorothioate (Fig. 3B),

oligomers of the same length and sequence were arranged side by side. For both types of aODNs, the increase in length apparently caused an increase in ATX block almost reaching the value measured for the GluR-D, homomeric receptor channels (Fig. 3C). This indicates that expression of the GluR-Bi subunit is indeed more potently suppressed by longer aODNs. The two types of oligomers, however, clearly differ in the minimal length required to attain the extent of block of GluR-D, homomeric receptor channels. Whereas in the presence of the 12-mer phosphorothioate (PS12-1) the ATX block had already reached this maximal value, 15- or 18-mers of the phosphodiester oligos were necessary. Fig. 4 shows the quantified inhibitory effects obtained as


16191

described in the previous section. (i) For both types of aODNs A tested (phosphodiester oligos represented by filled circles and phosphorothioates by open circles), the remaining activity of 4 z GluR-B,-cRNAstrongly depended on the lengthof the antisense g.- 1.2 oligodeoxynucleotides. (ii) Therelativeactivity of GluR-B,m % 1.0 cRNA was reduced by more than 95% by a phosphorothioate 12-mer, and by phosphodiester oligos 215 nt. fiii)No inhibitory x 0.8 effect could be detected for phosphorothioate aODNs5 6 nt, and ..> for phosphodiester aODNs 5 8 nt. Since a n aODN targeting the 2 0.6 first 6 n t of GluR-B,-cRNA would also target a hexamer on .m GI~-Di-cRNA(nucleotides 1819-18241, PS6-1511 which has a 2 0.4 unique target site on GluR-Bi-cRNA was used instead. (iv) A Boltzmann curve (see "Materials and Methods") fitted to the effect of phospho~thioateaODNs is shifted to theleft by about three nucleotides relative to thecorresponding curve for phosphodiester oligos. As mentioned above the difference in antisense effects between phosphodiester and phosphorothioate oligomers are generally assumed to be due to the considerably longer half-livesof phosphorothioates. One would therefore expect that thisdifference in inhibitory efficiencies of aODNs could be compensated by changing theirconcentration in theoocytes while keepingall other experimental conditions constant. Dose Dependence of the Effect of Antisense Oligodeoxynucleotides-Dose-response curves were determined for four different aODNs by injecting dilution series covering the concentration range from 5.10-1 pg/p1 to 5.10-6 pg/p1 in the injection solution. In a first experiment, a phosphodiester12-mer (PO12-1) was compared with thecorresponding phosphorothioate 12-mer (PS12-1). Both oligomers were identical in length and sequence but differed significantly in their effects at a 0'2 , ,,,,, , , (,,, , , / / , / , , / , , concentration of 0.5 pg/p1 (Fig. 4). Fig. 5A shows the experi0.0 mentally determined dose response curvesfor which a logistic 10-4 10-5 10-4 10-3 10-2 10-1 IO* fit yielded a half-maximal effect at 1.9-W4pg/$ for PS12-1 concentration of injected aODNs [pglpll and at 1.8-10-2pg/d for PO12-1. This suggests that under our experimental conditions the effective concentration of a phosFIG.5. Dependence of aODN-mediatedexpression inhibition phodiester aODN was reduced by about a factor of 100 relative on oligo concentration. aODNs were injected into oocytes in concento thecorresponding phosphorothioate oligomer. Furthermore, trations of 5.10" pg/pl to 5.10-6&pl together with GluR-Bi-/GluR-DicRNAs a t a ratio of 5:l. The relative activity of G l ~ - B i - c R N A was then i t is evident from Fig. 5A that the effect of PS12-1 is already quantified as in Fig. 4. Each data point represents the mean 2 S.D.' of saturated at a concentration of 0.5 pg/pl, whereas PO12-1 is 7-10 measurements. A, dose-response curves of an unmodified 12-mer not yet in the saturation range. A further increase of the con- (PO12-1, filled circles) and a phosphorothioate 12-mer (PS12-1, open centration of PO12-1 to 5 pg/p1 (corresponding to 120-200 p~ circles). The concentrations at which aODNs mediated a half-maximal pg/pl for PO12-1 inside theoocyte), however, was found to havetoxic side effects effect (IC6o)were obtained by a logistic fit to be 1.8.10-* and 1.9-10-* pglpl for PS12-1.B, dose-response curves of PS8-1 (open destroying the cells, as also reported by others (9, 10, 31). circles) and PO30-1 (filled circles) yielded IC,, values of 1.9.10-3and In a second experiment, we compared a phosphodiester 30- 8.6.10-4pg/pl, respectively. mer (PO30-1) with a phosphorothioate octamer (PS8-1) which was supposed t o have a similar dose dependence of inhibitory effects of sixphosphorothioate octamers (PS8-1,PS8-736, effect. As expected the dose-response curveof PO30-1 is shifted PS8-1511, PS8-2134, PS8-2557, and PS8-3'UT) targeting difto theleft compared with PO12-1 by about a factor of 20 (Fig. ferent positions on the GluR-B,-cRNAfrom the initiationcodon 5 B ) , and the half-maximal effect was at 8.6.W4pg/pl. On the to the 3"untranslatedregion were compared (localizationof the other hand, the half-maximal effect of PS8-1 (1.9.10-3pg/pl) target sequences indicated in Fig. 6B). Each aODN was dewas shifted toward higher concentrations by about a factor of signed to have an unique target on sitethe GluR-B,-cRNA,that 10 compared with PS12-1. is, it shared 5 6 consecutive matching base pairs with other Surprisingly, the maximal effect of PS8-1 was already positions on the GluR-B,- or on the GluR-D,-cRNA. reached at a concentration of 5 . W pg/pl (Fig.5B);the remainAs shown in Fig. 6.4, the six octamers, all testeda t a conceningrelative activity of GluR-B,-cRNA was, however, much tration of 0.5 pg/pl, reduced the relative activity of the GluRhigher than for the longeraODNs. This indicates that the B,-cRNA by between 69% (found for PS8-736) and 80% (found length of phosphorothioate aODNs determines their maximal for PS8-21341, indicating that the maximal aODN effect is only effect independently of concentration. slightly dependent on the position of the targetsequence on the Since all experiments up to thispoint have been made with cRNA. The small variations of the maximal aODN effect may aODNs targeting the same region of the cRNA, the experimen- reflect structural featuresof the cRNA molecule such as loops tal findings might be sequence specific. Sequence specificity of or hairpins, which are known to reduce hybridization affinity aODN effects has indeed been reported(8,9, 14). We therefore (20,321. Similarly also forsix PO 18-mers (POl8-1, P018-483, designed another set of experiments inwhich the targetsite of PO18-983, PO18-1483, PO18-1983, and P018-2483), we conthe aODNs was varied. sistently found an inhibitory effect of more than 95%, indeDependence of the Effect of Antisense OZ~godeox~nucZeotides pendent of their target site on the GluR-B,-cRNA (data not on the Position of Their Target Sequence on the cRNA-The shown).

2

j,

I

I

I I

16192

Oligonucleotide-mediated Antisense Short Inhibition

In Xenopus oocytes, Cazenave and co-workers reported efficient expression inhibition by an acridine-linked 11-mer(5) and Shuttleworth et aZ. (34) found that a phosphodiester 10-mer induced a just detectable cleavage of RNA by RNase H. Furthermore, Helene andco-workers (18) showed that a modified T .9-mer (5' substitution with an acridine derivative and3' sub1.0 stitution by a dodecanol chain) directed RNase H cleavage in vitro and inhibited T24 cell proliferation t o some extent. $!0 0.8 The methodsmost frequently used quantify aODN effects by .-x .-> probing RNA and/or protein which have been extracted and 5 0.6 0 fractionated. Compared with these methods theAMPA recep.tor-ATX system used in this study may have the following 0.4 advantages. First, the reliability of aODN effects tested witha single pair of oocytes (control and aODN application) is com0.2 For example, in parable t o conventionaldetectionmethods. 0.0 PS12-1 we found an ATX block of 88-95% in 12 oocytes, while in the 10 correspondingcontrol oocytes the ATX block was between 8 and 13%. Second, since the oocyte method is faster than conventional methods and up to100 oocytes can be studied in parallel, this opened up the possibility to study many different aODN species or concentrations simultaneously with B sample sizes of up to 20 oocytes. The large sample size increases accuracy to a level considerably higher than in conventional methods which allow only single experiments. The results of ATX block obtained with our system are not a direct measure of the aODN effects. cRNA activities aswell as aODN effects could, however, be quantified from ATX block measurements by simple application of a calibration curve. This cali5' UT coding region of GluR-Bi 3' UT bration curve shown in Fig. 2 reflects the basic finding that the FIG.6. Site specificity of octameric PS aODNs. Effects of oc- ATX block of heteromeric GluR-BjGluR-D, AMPA receptor tameric PS aODNs targeting various sites on the GluR-Bi-cRNA were channels is directly correlated to the injected ratio of the two determined as in Figs. 4 and 5. A, PS8-1, PS8-736, PS8-1511, PS8cRNAs. The absolute amount of the two cRNAs was found to 2134, and PS8-2557 targeted sites in the coding region of GluR-B,, PS83"UT targeted a site inthe 3'-untranslated region, PS8-0 had no have no effect on ATX block, so that for constant cRNA ratios of 5:1,2.5:1, and 1:lthe toxinblock did not change,even when the target on either GluR-B, or GluR-D,, while PS8-204 targeted both of the two cRNAs. Only the first six octamers differed significantly from the absolute cRNA amounts were reducedby factors of 10-50 (data control values (right bar). B , schematic representation of the various not shown). target sites on the GluR-B,-cRNA. The shape of the calibration curve is characteristic for the Three additional shortphosphorothioate aODNs were tested AMPA receptor-ATX system rather than for translation effias negative controls: the first was PS8-0 which had lessthan 6 ciency of a particular oocyte. As a consequence, the aODN efconsecutive matching base pairs incommon with any position fects determined in different batches of oocytes or with slightly on the two cRNAs, the second was PS8-204, an octamer tar- different compositions of the cRNA solution due t o pipetting corrected by geting both the GluR-B,- (nucleotides 204-2111, and GluR-D,- errors aredirectly comparable to each other, when the calibration curve. Thus, the AMPA receptor-ATX system cRNA (nucleotides 222-229). No difference in the remaining allows quantitative estimates of aODN effects with relatively relativeactivity of GluR-B,-cRNA wasdetectablebetween PS8-0, PS8-204, and the values measured in the absence of high precision. Moreover, the rangeof maximal sensitivity of the AMPA reaODNs (Fig. 6A).As expected, both PS octamers wereobviously not able tomodify the ratioof activities of the two cRNAs. The ceptor-ATX system is presetby the ratioof the injected cRNAs absolute amplitude of the kainate-evoked current was, how- which definesthe position of the respective controlvalue on the calibration curve.If a quantification of extremely smallaODN ever, reduced in the presence of PS8-204 (data not shown). Finally, we tested PS12-lD, targeting the first 12nt of the effects with a threshold sensitivity of about 5% is needed, the GluR-D,-cRNA.The mean S.D. of the observed ATX block was control value on the calibration curve should be chosen in the 3 2 2% ( n = 7) which is significantly smaller than thecontrol linear range,i.e. at a GluR-B,-/GluR-D,-cRNAratio of about 1.0. valuesbutidenticaltothevalue recently determined for This, however, restricts the range of detection and aODN efGluR-B, homomeric receptor channels (27). This indicatesthat fects of more than one potency can no longer bediscriminated. PS12-1Di indeed specifically suppressed the expression of the Since we were mainly interested in aODNs with inhibitory efficiencies of more than 30% up totwo potencies, acRNA ratio GluR-D, subunit. of 5 (ATX block of 10%) was chosen. This is just besides the DISCUSSION linear range and corresponds t o a detection limit of about 10In the present study we have shown that completely phos- 15%. Because this cRNA ratio is in the flat range of the caliphorothioate-modified aODNsmust havea minimal lengthof 8 bration curve, the control values as well as the results of exeffects yielded relatively large nt to mediate a specific reduction of more than 70% of the periments with small inhibitory activity of a cRNAencoding aglutamate receptor subunit when error bars. For phosphorothioate octamers, we found an inhibitory effect coinjected into Xenopus oocytes. Although this is twice the length of the minimal deoxyoli- of about 70%, while unmodified aODNs mediating the same gomer-RNA hybrid actingas a substrate for RNase H under in degree of expression inhibition were about 11nt in length(Fig. vitro conditions (33), it is shorter than the minimal aODNs 4).Furthermore, while the inhibitory effect of PS oligomers already saturated at a concentration of5.10-' pg/pl, for PO which were found to be effective in vivo. A

2 -

I


16193

oligomers 5 1 5 n no t saturating concentrationcould be reached whereas PS8-0, an octamer having no target site on either of (Fig. 5 A ) . Therefore, the inhibitory efficiencies of unmodified the two cRNAs, was found t o have no inhibitory effect (Fig. 6A). aODNs 5 15 nt shown in Fig. 4 are characteristic valuesfor a n Moreover, a phosphorothioate octamertargetingboththe oligo concentration of 0.5 pg/pl rather than maximal aODN GluR-Bi- and GluR-D,-cRNA did not cause any shift in the ratio of the two cRNAs, which may indicate that PS8-204 indeed effects. Both the differential saturation behavior and length PS aODN effects might be explained inhibited both RNAs to the same extent (Fig. 6A). A finding dependence of the PO and by the reported stability of PS aODNs t o enzymatic degrada- questioning the assumption that longer aODNs have higher tion inside the oocytes (9, 10, 28). It has recently been sug- specificity was reported by Woolf et al. These authors demonstrated that partially matchingaODNs mediated partialcleavgested (35, 36) toconvert only the 3‘-terminal phosphodiester age of the targetRNA by RNase H (19). Since longer oligomers bonds to thioate linkages since the numberof phosphorothioate specificity is desubstitutions were found to correlate negatively with the hy- contain more potentialtargetsites,their brid stability especially when the sulfuris in the R configura- creased rather than increased.Based on our finding that spetion (37-40). This point should be further investigated. Prelim- cific aODN effects of at least 70% can already be mediated by inary experiments, however, in which the number of aODNs of 8 nt in length, phosphorothioate 14- and 21-mers, phosphorothioate-modified linkages of PS8-1 was reduced to containing 7 and 14 internal octamers, mayrecognize 2136 or two or four (one or two 5’- and 3’4erminal PS modifications) 4272 target sitesin a RNApool of a typical eukaryotic cell (19). revealed an insignificantdecrease of theinhibitory effect As a consequence, the activities of the RNAs carrying these target sites mightbe reduced t o a significant extent. For phosrather than a further increase (data notshown). The maximal inhibitory efficiency of PS oligomers was shown phodiester aODNs, which we found to mediate an aODN effect to strongly dependon the oligo length under saturating condi- comparable tothat of PS octamers only at a length of 11nt, the tions (Figs. 4 and 5) which suggests that the maximaleffect of resulting number of putative recognition sites of PO 14- and aODNs is determined by their hybridizationaffinity (14,411.To 21-mers would be 19 and 53, respectively. Thus, the ratio of test this hypothesis we estimated the hybridization free ener- intended to unintendedexpression inhibition might be signifiby application of nearest neighbor cantly higher for phosphodiester aODNs compared with the gies (AG) of the used aODNs widely used phosphorothioates. analysis, based on AG values for DNA-DNA hybrids, to the aODNs sequences (26). These estimates for AG were used to fit These estimates are,however, based on the assumptionthat of nucleotides actuallymediatingthe Boltzmann curves (see “Materials and Methods”) to the data in theinternalstretch Fig. 4. For the phosphorothioates, the fitting procedure re- aODN effect may be located anywhere on the longer aODN. sulted in a slope factor of 3.9 kJ/mol which is about 1.6 times Any reduction in hybridizationaffinity, which is caused by the the valueexpected from theory for a temperature of 19 “ C , and non-matching base pairs andwould be expected to reduce the for the phosphodiester aODNs a respective value of 4.2 kJ/mol inhibitory aODN effect, was neglected. Whether inhibitory efwas obtained. In otherwords larger changes inoligo length or fects of internal octamers on longer phosphorothioate aODNs hybridization free energy are necessary to mediate an e-fold can be comparable to those of pure octameric aODNs must be change in relative activity of GluR-B,-cRNAthan theywould be further investigated.A first hint is already given by a PS octaexpected if the aODN effect is exclusively determined by hy- mer which did not have an octameric target site butcontained bridization. This discrepancy between hypothesis and experi- two internal 7-mers which both targeted theGluR-B,-cRNA a t different positions. The resulting reduction in GluR-B,-cRNA mental data might be due to either errors in our theoretical assumptions or the fact that the maximalaODN effect is not activity was 85% which is even slightly more than found for exclusively determined by hybridization. Errors in our theoreti-octameric PS aODNs. On the other hand, PS8-1 has 7 consecucal assumptions mightbe the approximation of the hybridiza- tive matching base pairs in common with the GluR-Di-cRNA. tion process by a simplified two-state reaction or the estimates However, the potency of PS8-1 to shift the ratiobetween the in AG values which were madeon the basisof nearest neighbor two cRNAs is similar to those of the other octameric aODNs interactions originally obtained for DNA-DNA hybrids (26) in- exclusively specific for GluR-Bi-cRNA. stead of DNA-RNA interactions, and for phosphodiester oligos As pointed out in the Introduction, itwould be desirable to rather thanfor phosphorothioates. On the other handmaxithe develop an alternativecloning strategy which is based on sepamal aODN effects might be influenced by properties of the rately probing the potency of the members of an oligomer litarget cRNA as secondary structures of RNAs have been rebrary to inhibit m/cRNA expression in Xenopus oocytes in order ported to interfere with hybridization and thus reduce the ef- to obtain sequence information about a certain m/cRNA. The ficiency of aODNs (20).This is further supported by a finding of feasibility of such a strategy would, however, strongly depend the present study that slight variations in maximal aODN ef- on the properties of the individual oligomers. The findings of fects were obtained for a series of phosphorothioate octamers this study on short aODNs may be a first step toward such a all having about the same theoretical value for hybridization strategy. free energy but targeting different positions on the GluR-BiAcknowledgments-We thank Drs. R. Schoepfer and P. H. Seeburg cRNA (Fig. 6). The values for half-maximal aODN effect on the for providingthe AMPA receptor clones, Drs. W. Jahn, M. Raditsch and abscissa in Fig. 4 cannot be interpreted as AG values for the V.Witzemann for the gift ofATX, Dr. U.Brandle and U. Rexhausen for respective hybridization reaction since the effective concentra- help with oligo design and data evaluation, Dr. M. Monem for helpful tion of the aODNs in the oocyte is not known, and may be discussions, Drs. A. Gummer, M. Hausser, and J. Mosbacher for critical reading of the manuscript, and Dr. B. Sakmann for continuous support different because of differential stability of the aODNs. In many studies applying aODNs as a tool to selectively of this work. inhibit proteinexpression oligomers of 14-21 nt in length were REFERENCES used (16, 4244). The main reasonfor choosing aODNs of this 1. Zamecnik, P. C., and Stephenson, M. L. (1978)Proc. Natl. Acad. Sci. U. S. A. size is that longeroligos exhibit a stronger effect and a higher 75,280-284 2. Kawasaki, E. S. (1985)Nucleic Acids Res. 13, 4991-5004 specificity than shorterones. In thisstudy, however, we showed 3. Minshull, J., and Hunt, T.(1986)Nucleic Acids Res. 14, 64334451 that even aODNs as short as8 nt mediated inhibitionof cRNA 4. Maher, L. J., 111, and Dolnick, B. J. (1988)Nucleic Acids Res. 16,3341-3358 5. Cazenave, C., Loreau, N., Thuong, N. T., ToulmB, J. J., and HBlBne, C. (1987) expression with a high sequence specificity. Thus, all the PS Nucleic Acids Res. 15,47174736 octamers designed as sequence specific for GluR-B, reduced the 6. Boiziau, C., Thuong, N. T., and Toulmk, J. J. (1992)Proc. Natl. Acad. Sci. activity of the corresponding cRNA by between 69 and BO%, U. S. A. 89, 768-772

16194

Oligonucleotide-mediated Antisense Short Inhibition

7. Dash, P., Lotan, I., Knapp, M., Kandel, E. R., and Goelet, P. (1987) Proc. Natl. Acad. Sci. U. S. A . 8 4 , 7896-7900 8. Shuttleworth, J., and Colman, A. (1988) EMBO J. 7, 4 2 7 4 3 4 9. Baker, C., Holland,D., Edge, M., and Colman,A. (1990) Nucleic Acids Res. 18, 35374543 10. Woolf,T.M., Jennings, C. G. B., Rebagliati, M., and Melton, D.A. (1990) Nucleic Acids Res. 18, 1763-1769 11. Kleuss, C., Scheriibl, H., Hescheler,J.,Schultz, G., and Wittig, B. (1992) Nature 358, 424-426 12. Walder, R. Y., and Walder, J. A. (1988) Proc. Natl. Acad. Sci. U. S. A . 85, 5011-5015 13. Agrawal, S., Mayrand, S . H., Zamecnik, P. C., and Pedersen, T. (1990) Proc. Natl. Acad. Sci. U. S. A. 87, 1401-1405 14. Chiang, M. Y., Chan, H.,Zounes, M. A,, Freier, S. M., Lima, W. F., and Bennett, C. E (1991) J. Biol. Chem. 266, 18162-18171 15. Lallier, T., and Bronner-Fraser, M. (1993) Science 259, 6 9 2 4 9 5 16. Wahlestedt, C., Golanov, E., Yamamoto, S., Yee, E , Ericson, H., Yoo, H., Inturrisi, C. E., and Reis,D.J. (1993) Nature 363, 260-263 17. Gundersen, C. B., and Umbach, J. A. (1992) Neuron 9, 527-537 18. Saison-Behmoaras, T., Tocque, B., Rey, I., Chassignol, M., Thuong, N. T., and Helene, C. (1991) EMBO J. 10, 1111-1118 19. Woolf, T. M., Melton, D. A,, and Jennings,C. G. B. (1992) Proc. Natl. Acad. Sci. U. S. A. 89,7305-7309 20. Lima, W. E , Monia. B. P., Ecker. D. J.,and Freier,S. M.(1992)Biochemistry . 31. . 12055-12061 21. Loke, S . L., Stein, C. A,, Zhang, X. H., Mori, K, Nakanishi, M., Subashinge, C., Cohen, J. S., and Neckers, L. M. (1989) Proc. Natl. Acad. Sci. U. S. A . 86, 3474-3478 22. Sommer, B., Keinanen, K., Verdoorn, T., Wisden, W., Burnashev, N., Herb, A., Kohler, M., Takagi, T., Sakmann, B., and Seeburg, P. H. (1990) Science 252, 1715-1718 23. Krieg, P. A,, and Melton, D. A. (1984) Nucleic Acids Res. 12, 7057-7070 24. Burnashev, N., Monyer, H., Seeburg, P. H., and Sakmann,B. (1992) Neuron 8, 189-198 25. Draguhn, A., Jahn, W., and Witzemann, V. (1991) Neurosci.Lett. 132, 187-190

26. Breslauer, K. J., Frank, R., Blocker, H., and Marky, L. A. (1986) Proc. Natl. Acad. Sci. U. S. A . 83, 37463750 27. Herlitze, S., Raditsch, M., Ruppersberg, J. P., Jahn, W., Monyer, H., Schoepfer, R., and Witzemann, V. (1993) Neuron 10, 1131-1140 28. Spitzer, S., and Eckstein, F. (1988) Nucleic Acids Res. 16, 11691-11704 29. Dagle, J. M., Walder, J. A,, and Weeks, D. L. (1990) Nucleic Acids Res. 18, 47514757 30. Lotan, I., Volterra, A,, Dash, P., Siegelhaum, S. A,, and Goelet, P. (1988)Neuron 1,963-971 31. 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. 1 7 , 4 2 5 5 4 2 7 3 32. Stull, R. A,, Taylor, L. A,, and Szoka, F. C. (1992) Nucleic Acids Res. 20, 3501-3508 33. Donis-Keller, H. (1979) Nucleic Acids Res. 7,179-192 34. Shuttleworth, J., Matthews, G., Dale, L., Baker, C., and Colman, A. (1988) Gene (Amst.) 72, 267-275 35. Shaw, J.-P., Kent, K., Bird, J., Fishback, J., and Froehler, B. (1991) Nucleic Acids Res. 19, 747-750 36. Kitajiama, I., Shinohara, T., Minor, T., Bibbs, L., Bilakovics, J., and Nerenberg, M. (1992) J . Biol. Chem. 267,25881-25888 37. Cosstick, R., and Eckstein, F. (1985) Biochemistry 24, 3630-3638 38. LaPlanche, L.A., James, T. L., Powell, C., Wilson, W. D., Uznanski, B., Stec, W. J., Summers, M. E, and Zon, G. (1986) Nucleic Acids Res. 14, 9081-9093 39. Stein, C. A,, Suhasinghe, C., Shinozuka, K., and Cohen, J . S. (1988) Nucleic Acids Res. 16,3209-3221 40. Stein, C. A,, and Chen, Y.C. (1993) Science 261, 1004-1012 41. Monia, B. P., Johnston, J . E , Ecker, D. J., Zounes, M. A,, Lima, W. E, and Freier, S. M. (1992) J . Biol. Chem. 267, 19954-19962 42. El-Baradi, T., Bouwmeester, T., Giltay, R., and Pieler, T. (1991) EMBO J . 10, 1407-1413 43. Wang, H., Watkins, D. C., and Malbon, C. C. (1992) Nature 358,334-337 44. Simons, M., Edelman, E. R., DeKeyser, J . L., Langer, R., and Rosenherg, R. D. (1992) Nature 359,67-70 45. Gao, W. Y.,Han, F. S., Storm, C., Egan, W., and Cheng, Y. C. (1992) Mol. Pharmacol. 41,223-229