Reversal of cystic fibrosis phenotype in a cultured Δ508 cystic fibrosis transmembrane conductance regulator cell line by oligonucleotide insertion

Abstract

Cystic fibrosis (CF) is a lethal genetic disorder that is due to mutations in the gene encoding the cAMP-activated anion CF transmembrane conductance regulator (CFTR) channel. A three-nucleotide base deletion (TTT), encoding phenylalanine in position 508 of the translatable CFTR sequence (accompanied by a C to T replacement immediately 5′ to the deletion), accounts for ≈75% of cases of the disease. In the present study, an oligonucleotide complex (CF4–CF6, 2′-0-methyl RNA–unmodified RNA oligonucleotide duplex, respectively) was used to restore CFTR function by insertion of missing bases in Δ508 CFTR mRNA from a cultured (Δ508) cell line. cAMP-activated whole-cell currents and Cl^– transport were detected in CF4–CF6-treated, but not control Δ508, cells by patch-clamp and 6-methoxy-N-(3-sulfopropyl)quinolinium fluorescence (SPQ) quenching analyses, respectively. Further, the nucleotide addition in the deleted region of Δ508 CFTR was determined after amplification by RT-PCR. Insertion of UGU and replacement of U by C immediately 5′ to the deletion site in Δ508 mRNA appear to have taken place, with phenotypic but not genotypic reversion in tissue culture of treated cells. The mechanism of insertion of nucleotides has yet to be determined.

Cystic fibrosis (CF) is a lethal disorder caused by mutations in the CF transmembrane conductance regulator (CFTR) gene encoding CFTR (1, 2). In this disease, CFTR, a cAMP-activated anion channel (3, 4), is associated with dysfunctional epithelia in several tissues (5), including lungs, pancreas, intestine, sweat glands, and kidneys. The Δ508 CFTR is a misfolded but partially functional channel protein (6, 7) that is unable to translocate perfectly to target plasma membranes (8). Several chemical and pharmacological strategies have restored the CF phenotype partially by bringing Δ508 CFTR to the plasma membrane. Gentamycin was also found to induce a correction of faulty CFTR function in CF, caused by CFTR stop mutations rather than by the TTT deletion (9).

Loss or suppression of deleterious gene function by antisense oligonucleotide technologies involves specific inhibition of DNA and RNA synthesis and protein expression (10, 11). Upon cell entry, exogenous complementary deoxyoligonucleotides hybridize with a target mRNA, inducing an excision endonucleolytic effect on the mRNA, which is due to activation of RNase H (12), resulting in synergism with hybridization inhibition (10, 13). Triplex-forming oligonucleotides have also been used to modify gene function (14, 15). Likewise, RNA editing (16) has been reported in numerous cell systems in which nucleotide sequences can also be modified at the RNA level. Other mechanisms may apply to double-stranded RNA sequences if they are introduced into biological systems. This mechanism is the focus of current attention for their role in RNA interference (RNAi) and gene silencing at the transcriptional level (17). RNAi-mediated gene silencing was discovered in Caenorhabditis elegans (18) and also in plants and mammals (19). Genesilencing results from the successive cleavage of long double-stranded RNA to oligonucleotide small-interfering RNAs (siRNAs) by Dicer enzymes (19). After oligonucleotide hybridization of siRNA with mRNA, a common feature shared with antisense DNA or RNA, the cleavage of target mRNA is catalyzed by an enzyme that is different from RNase H (19). In addition to the above finding, is deoxycytidylate phosphate deoxyguanylate (CpG) stimulation of the host immunomodulatory mechanisms, which has inhibitory implications in both prokaryotic and eukaryotic organisms (20, 21).

Gain of function and/or correction of defective genes present a further challenge for gene therapy. Partial restoration of Δ508 CFTR function may be induced by membrane insertion of the mutated channel without changes in genotype (22, 23). RNA–DNA oligonucleotide hybrids (24, 25) and single-stranded oligonucleotides (26) have also been used for correction of defective genes. Substituted, circular, or single-stranded RNA–DNA chimeras have been used to insert base pairs in deficient genomic DNA (24, 25). Some attempts at replicating this work, however, showed no nucleotide exchange in the targeted loci by cloning of the PCR products (27), and these results were viewed as PCR artifacts created by the RNA–DNA oligonucleotides themselves (27, 28). Transcriptional repair is also the focus of recent attention in the repair of mRNA. Spliceosome-mediated RNA-transplicing (SMaRT) technology has been used to modify Δ508 CFTR transcripts in human CF airway epithelia (29).

It has been observed (12) that mRNA, hybridized to a single-stranded synthetic short piece of DNA, activates RNase H, a ubiquitous enzyme that hydrolyzes specifically the hybridized segment of mRNA. At least a 4- to 5-nt internally located region, hybridized to a phosphorothioate (PS)-modified oligodeoxyribonucleotide, is needed for the excision (12).

Initially, we constructed a modified oligodeoxyribonucleotide with a central segment containing a PS modification and the adjoining segments PS plus 2′-O-methyl modifications. This hybrid oligonucleotide was designed to hybridize in Watson–Crick base complementarity to the region of Δ508 mRNA, where the PS section would be directly opposite the Δ508 mRNA position, flanked 3′ and 5′ by an adjacent few nucleotides complementary to the Δ508 mRNA. In this way, the PS segment would activate endogenous RNase H and cleave the bases opposite the PS complementary oligonucleotide. The 2′-O-methyl PS and segments on both sides of the RNase H-sensitive section would be generally nuclease and RNase H-resistant and, thus, serve as a “genetic Band-Aid.” In theory, this oligonucleotide would hold the 5′ and 3′ segments of Δ508 mRNA in place, in position for an insertion or repair of bases in a second step. For the possible repair step, we constructed two single complementary-strand oligonucleotides: CF4 and CF6, as listed in Table 1. When the CF4–CF6 annealed duplex, however, was added to the tissue culture cells without oligonucleotide treatment, as described above, phenotypic reversion was likewise found. We, therefore, chose to pursue the one-step phenotypic-reversion technique as a simpler model for the sequencing studies described below.

Table 1. Sequences of oligonucleotides used (5′ to 3′).

Oligonucleotide	Sequence
CF4	AUC AUA GGA AAC ACC AAA GAU GAU AUU UUC UUU
CF6	pC AUC UUU GGU Gp
F1	GGG AGA ACT GGA GCC TTC A
N1	GTA TCT ATA TTC ATC ATA GGA AAC ACC ACA
M1	GTA TCT ATA TTC ATC ATA GGA AAC ACC ATT
NF1	GCC TGG CAC CAT TAA AGA AAA TAT CAT CTT
MF2	GCC TGG CAC CAT TAA AGA AAA TAT CAT TGG
CFR	GTT GGC ATG CTT TGA TGA CGC TTC
CFFW	GGC ACC ATT AAA GAA AAT ATC ATC TT
CFFM	GGC ACC ATT AAA GAA AAT ATC ATT GG
SCFR	GTT GGC ATG CTT TGA TGA CGC TTC
SNF1	GCC TGG CAC CAT TAA AGA AAA TAT CAT CTT
SMF2	GCC TGG CAC CAT TAA AGA AAA TAT CAT TGG

Primers and constructs used to repair CFTR mRNA are given. CF4 sequence corresponds to bases of 2′-O-methylribosyl oligonucleotides, with internucleoside phosphate bonds. Lightface type in CF6 corresponds to natural RNA sequences, and lowercase p corresponds to the terminal phosphate groups. Bold type corresponds to the natural DNA sequences. Shaded region in SCFR, SNF1, and SMF2 corresponds to internucleoside phosphorothioate bonds.

In this article, single complementary-strand oligonucleotides were constructed and annealed to form a duplex, as shown in Fig. 1. Duplexes were added to growing Δ508 CFTR cell cultures, and patch-clamping studies (30) showed evidence of phenotypic reversion of cells to a normal phenotype, namely cAMP-activated anion currents. Phenotypic recovery was confirmed by PKA-activated single-channel currents and 6-methoxy-N-(3-sulfopropyl)quinolinium (SPQ) Cl^– fluorescence. Nucleotide-treated cells contained an insertion of missing bases in the CFTR mRNA. The mechanism of insertion into mRNA is still under investigation.

Fig. 1.

Hypothetical schematic representation of repair of CFTR Δ508 mRNA by CF4–CF6 duplex (A) and sequences found in restored mRNA (B). Deletion in Δ508 mRNA, possible configuration of CF4/CF6 in the hybridization step, and the various replacements found are depicted. Bold and italicized type indicates bases of 2′-O-methyl ribosyl oligonucleotides with normal internucleo-side phosphate bonds (CF4 oligonucleotide). Lightface and roman type corresponds to natural RNA. Bold and underlined type corresponds to inserted bases of ribonucleotides. Shaded bases in the CF4 and Δ508 mRNA indicate noncomplementary Watson–Crick base pairs. Sequence analysis of RT-PCR products obtained by using different primers is also shown. (B) Half-boxed “A” was present in 7 of 10 sequencings but was absent in 3 other sequencings, leaving three deletions (a and b). Dashes indicate deletions, possibly induced by PCR (a and b). There are no deletions in c. Half-boxed “U” indicates that, in this position, U sometimes is replaced by another nucleotide (c).

Materials and Methods

Synthesis of Oligonucleotides. Standard DNA and RNA oligonucleotides (Table 1) were synthesized on a 394 DNA/RNA synthesizer (Applied Biosystems) with phosphoramidite chemistry and standard phosphoramidite monomers (Glen Research, Sterling, VA). For the introduction of 3′ and 5′ phosphate groups on the CF4 oligonucleotide (Table 1), 5′ and 3′ phosphorylation reagents were used accordingly (Glen Research). PS bonds were introduced by sulfurization with Beaucage thiolation reagent (31). We synthesized 2′-0-methyl modifications, oligoribonucleotides, and PS oligonucleotides, and HPLC was purified as described (32, 33).

Annealing of CF4–CF6 Oligonucleotide Duplex. Stock solutions of CF4 (1 mM) and CF6 (1 mM) were prepared by dissolving compounds in high-ionic-strength buffer (0.2 M NaCl/20 mM MgCl₂/20 mM Tris·HCl, pH 7.0). Duplex formation was prepared with a 1:1 (vol/vol) mixture of compounds by heating to 75–80°C and cooling down gradually to room temperature. All duplexes and compounds were sterilized by passage through 0.45-μm cellulose acetate centrifuge filters (Costar).

Cell Culture and Incubation Procedures. Mouse mammary carcinoma cells (c127i) transfected with human epithelial CFTR wild-type (WT1) or Δ508 CFTR cells (34, 35) were used for these studies. Cells were grown and maintained in DMEM supplemented with 10% FBS and 1% l-glutamine, as reported (35).

Electrophysiology and Chemicals. Whole-cell and excised inside-out patches were obtained to assess cAMP–PKA-dependent anion currents in treated Δ508 cells. Currents and command voltages were obtained and driven, respectively, with a 3900 amplifier (Dagan Instruments, Minneapolis) by using a 1 GΩ headstage. The excised patch-clamp configuration was carried out as described (30). Single-channel data were obtained between ± 100 mV in symmetrical Cl^–. Data were further analyzed as described (30). The pipette and bathing solution contained either 140 mmol/liter NaCl/1.0 mmol/liter MgCl₂/KCl 5/10 mmol/liter N-2-hydroxyethylpiperazine N′-2-ethanesulfonic acid (Hepes), pH 7.4, or 70 mmol/liter MgCl₂/10 mmol/liter Hepes, pH 7.4. The bathing solution also contained 1.0 mmol/liter CaCl₂. Where indicated, the patch-pipette was filled up to at least one-third of its height with either Mg·ATP or Tris·ATP (100 mmol/liter, pH 7.4, adjusted with N-methyl-glucamine), as reported in ref. 30. Experiments were conducted at room temperature. The cAMP stimulatory mixture contained 8-Br-cAMP, isobutyl methyl xanthine (IBMX), and forskolin at final concentrations of 500 μmol/liter, 200 mol/liter, and 10 μmol/liter, respectively. The catalytic subunit of PKA was used at a final concentration of 20 μg/ml. The Cl^– channel blocker, diphenylamine-2-carboxylate (DPC) was kept in a 100-fold stock solution (20 mmol/liter) in 50% water/50% ethanol. DPC was used at a final concentration of 500 μM.

SPQ Fluorescence Technique. Cyclic-AMP-stimulated Cl^– transport was also followed by fluorescence changes of cells loaded with the Cl^–-sensitive dye SPQ (36). Briefly, cells were grown to partial confluence on glass coverslips. SPQ cell loading was conducted by a 15-min incubation in a Cl^– free diluted salt solution containing SPQ (5 mM). The saline solution contained 135 mM Na⁺-gluconate, 2.0 mM KH₂PO₄, 1.0 mM MgSO₄, 1.0 mM Ca²⁺-gluconate, and 10 mM Hepes (pH 7.4). In some experiments, gluconate was replaced by either isethionate or aspartate with similar results. SPQ fluorescence was determined under an oil immersion in an E800 fluorescence microscope (×20 objective; Nikon) with a UV filter (96101C UV-2E/C). Images were captured with a C4742–95 digital camera (Hamamatsu Photonics, Hamamatsu City, Japan) and stored as TIFF files on a Macintosh computer with iplab spectrum software (Signal Analysis, Vienna, VA). Pictures were analyzed digitally with nih image 1.62b7 software. SPQ cytoplasmic fluorescence values were normalized between an intranuclear (considered as zero quenching, i.e., low Cl^–) and extracellular background. Data were expressed as percentage of fluorescence with respect to time 0, when cells were placed in an isotonic saline solution. The solution contained 135 mM NaCl, 2.0 mM KH₂PO₄, 1.0 mM MgSO₄, 1.0 mM Ca²⁺-gluconate, and 10 mM Hepes (pH 7.4), with or without the cAMP stimulatory combination mixture.

Primers and RT-PCR of Δ508 CFTR mRNA. Total RNA isolated by using TRIzol reagent according to the manufacturer's protocol (Invitrogen) from either WT1 or Δ508 cells treated with or without CF6–CF4 annealed duplex oligonucleotides (final concentration of duplex in reaction mixture 10 μM) was used to perform the RT-PCR assay (37) in two steps by using a ThermoScript RT-PCR system (GIBCO/BRL). Several sets of primers were used (see Table 1). In the first step, total RNA (≈2 μg) was incubated for 60 min at 55°C by using an amplification refractory mutation system (ARMS) (38) reverse mutant (M1) and normal (N1) primers (0.5 μg each) separately for the first-strand synthesis. In a second step, after heating at 94°C for 5 min, 35 cycles of PCR (38) were carried out on the samples. PCR cycles were performed first by denaturation at 94°C for 2 min, annealing at 62°C for 1 min, and extension at 72°C for 2 min, followed by a final extension for 10 min at 72°C. Two ARMS reverse primers (N1 and M1) (38) and a forward (F1) primer were used (see Table 1). Similarly, RT-PCR was performed with ARMS PS reverse primer (SCFR) and PS forward primer (SNF1; see Table 1). RT-PCR products were separated in 3% agarose gel and subjected to automated DNA-sequence analysis.

Similarly, RT-PCR was conducted by using allele-specific primers (Fig. 1B and Table 1) as follows. In the first step, total RNA (≈2 μg) was incubated for 60 min at 55°C with reverse (CFR) primer (0.5 μg each) for the first-strand synthesis. In a second step, samples were heated at 94°C for 1 min. PCR was performed for 30 cycles at 60°C, including denaturation at 94°C (45 sec), annealing at 60°C (45 sec), and extension at 72°C (1 min). This procedure was ended with a 7-min final extension at 72°C. Two forward primers (CFW and CFM) and one reverse primer (CFR) were used for allele-specific RT-PCR (Table 1). The RT-PCR assay was performed by using the MasterAmp RT-PCR system (Epicentre Technologies, Madison, WI). In this procedure, combined reverse transcription and PCR were performed in the presence of forward normal (NF1), forward mutant (MF2), and reverse (CFR) primers (0.5 μg each) separately. First, samples were incubated for 20 min at 60°C. PCR was performed for 40 cycles at 94°C, and denaturation (30 sec), annealing (62°C for 30 sec), and extension (72°C for 1 min) were performed. The reaction ended by a final extension for 6 min at 72°C. Two forward primers (NF1 and MF2) and a reverse primer (CFR) were used for this procedure. RT-PCR products were separated in a 3% agarose gel and were then subjected to automated DNA-sequence analysis. ThermoScript RT-PCR products obtained by ARMS primers were subcloned for further separation and purification of possible heterogeneity of oligomers obtained in pCR-Blunt vector, according to the manufacturer's protocol (Invitrogen). The resulting clones were subjected to automated DNA-sequence analysis by using a T7 promoter primer.

Results and Discussion

Restoration of Δ508 CFTR Phenotype. To restore the normal (wild-type) phenotype in cells expressing Δ508 CFTR, modified oligonucleotides were constructed and their effect on ion transport was assessed in Δ508 cells treated with these constructs. Initially, a modified oligodeoxyribonucleotide, containing PS and PS plus 2′-O-methyl modifications for the central and flanking segments, respectively, was synthesized. This oligonucleotide was designed to hybridize in Watson–Crick base complementarity to the region of Δ508 mRNA where the PS section would be directly opposite the Δ508 mRNA position, flanked 3′ and 5′ by adjacent nucleotides complementary to the Δ508 mRNA. Thus, the PS segment would activate endogenous RNase H and cleave the bases opposite the PS complementary oligonucleotide (10 μM). The insertion of bases might then occur in a second step. For the possible insertion step, two complementary single-strand oligonucleotides, CF4 and CF6 (Table 1 and Fig. 1A), were constructed. The annealed CF4–CF6 duplex (final concentration, 10 μM) was added to the Δ508 CFTR cells after a 2-h incubation period and subsequent washout of the PS compound. Next, after incubation overnight or after up to 72 h of incubation in the second (insertion) step, patch-clamp examination was conducted on the treated cells. Whole-cell currents of two-step-treated Δ508 cells (Fig. 2) showed a 243% increase after cAMP stimulation (2.98 ± 0.68 ns per cell vs. 0.87 ± 0.16 ns per cell; n = 24, P < 0.01), which was absent in the control Δ508 cells (Fig. 2). The cAMP-activated currents were largely (>84%) inhibited by the CFTR inhibitor DPC (500 μM), as expected for wild-type CFTR (Fig. 2). Similar results were obtained with control WT1 cells, overexpressing wild-type CFTR in the same cellular background (data not shown). The treated Δ508 cells (10–12 pS) displayed PKA- and ATP-activated 10–12 pS Cl^– channels (Fig. 3) not observed in the control Δ508 cells (0 of 24).

Fig. 2.

Whole-cell currents of Δ508 CFTR expressing cells. (Upper) Representative tracings of Δ508 CFTR-expressing cells after treatment with CF4–CF6. Addition of cAMP activated currents that were inhibited by DPC (500 μM). (Lower Left) Untreated Δ508 cells lacked a cAMP activated whole-cell conductance (n = 21). (Lower Center) CF4–CF6-treated Δ508 cells had a robust cAMP response. (Lower Right) cAMP activation of treated cells was inhibited by DPC (500 μM).

Fig. 3.

Single-channel currents of Δ508 CFTR expressing cells. (A) Representative tracings observed after (Lower) PKA (100 nM) and MgATP (1 mM) addition of CF4–CF6 treated Δ508 CFTR cells. Currents were obtained in excised inside-out patches. Data are representative of n = 12. (B Upper) The PKA-activated Cl^– single-channel currents had a single-channel conductance of 12 pS (n = 6). (B Lower) Single-channel conductances are shown in the all-point histogram.

CF4–CF6 annealed duplex treatment alone (second step) also showed, however, phenotypic reversion, as determined by the patch-clamp technique. We, therefore, chose to pursue this one-step phenotypic reversion technique as a simpler model for the sequencing studies described below.

Interestingly, CF4–CF6 cells sporadically showed large, DPC-inhibitable whole-cell currents in the absence of cAMP stimulation (data not shown). Both WT1 and CF4–CF6-treated Δ508 cells responded with a similar change in SPQ fluorescence in response to cAMP stimulation (Fig. 4) (39). Control Δ508 cells showed no cAMP-induced change in SPQ fluorescence (Fig. 4). The data indicate that CF4–CF6 treatment of Δ508 cells restores a normal tissue-culture phenotype, consistent with the presence of functional CFTR.

Fig. 4.

SPQ fluorescence assay. Cl^–-induced changes in fluorescence were followed in cells loaded with the Cl^–-sensitive dye SPQ. SPQ fluorescence was tested in WT1 cells expressing wild-type CFTR, as well as control and treated Δ508 cells in a custom-made chamber, under UV-fluorescence microscopy. A cAMP-induced response (5–15 min) was observed only in WT1 and treated Δ508 cells (P < 0.05). Numbers of individual cells analyzed are given in parentheses.

Sequencing Analysis of PCR Products. To determine the extent of restored phenotype in the CF4–CF6-treated Δ508 cells, total RNA was isolated and CFTR specific primers were used to amplify the predicted region. CFTR wild-type specific primers efficiently amplified single bands of the expected size in the CF4–CF6 oligonucleotide-treated cells. The sequence region after this treatment did not show changes in the PCR-amplified oligonucleotides when wild-type primers were used. Because the number of Δ508 mRNA copies that are potentially repaired may be low, allele-specific primers for either wild-type or mutated CFTR were used next (CFFW and CFFM; see Table 1). Allele-specific primers detected mRNA from serial dilutions of wild and Δ508 total RNA by means of a shift of the amplified band (Fig. 5). Wild-type RT-PCR product could be detected in 1:10,000 dilutions in Δ508 mRNA background (Fig. 5). Initially, RT-PCR product from the allele-specific wild-type primer (CFFW; see Table 1) in the oligonucleotide-treated Δ508 cells also failed to detect insertion in the amplified band. Therefore, total RNA of CF4–CF6-treated Δ508 cells was tested further by PCR analysis with ARMS-specific primers (Fig. 6) (40). Wild-type and mutated ARMS primer-amplified PCR bands were examined by DNA-sequence analysis. The sequence of the amplified mutated cDNA showed a variety of one-codon insertions, rich in G residues. Whether a GGG codon (glycine) is an acceptable substitute for phenylalanine remains to be determined because systematic study of such base insertion has not been done to our knowledge. To analyze this observation further, the PCR products were subcloned in pCR-Blunt vector and subjected to DNA-sequence analysis. The fraction of control RNAs isolated from untreated Δ508 cells showed no oligonucleotide insertion. However, mRNA isolated from CF4–CF6-treated Δ508 cells showed 20–30% UGU insertion, based on analysis of the percentage of subclones showing TGT insertion into the RT-PCR-generated Δ508 DNA. This percentage of insertion is apparently sufficient for phenotypic reversion in the tissue-culture system. None of the subcloned untreated Δ508 cells displayed false positives (three-base insertion) in the region flanking the initial deletion (Fig. 7). Combined reverse transcription and PCR were performed with forward wild-type (NF1), forward mutant (MF2), and reverse (CFR) primers (0.5 μg each) separately. Insertion of bases in the proper position (Fig. 1B) has been found in subcloned and sequenced RT-PCR products, from treated but not control Δ508 cells.

Fig. 5.

Allele-specific RT-PCR analysis of CFTR mRNA. Samples from WT1 mRNA were serially diluted into Δ508 mRNA (lanes 2–14, from 0 to 10⁶ order with 10-fold serial dilution) to test efficiency of allele-specific primers (CFFW and CFFM; Table 1). Lanes 2 and 3 show WT1 and Δ508 mRNA amplified with respective primers. The wild-type primers recognized WT1 mRNA in dilutions up to 1:10,000 (lanes 11 and 12). Gel electrophoresis was conducted in agarose gels (3%), and a 25-bp ladder is shown on the left.

Fig. 6.

Specific reverse-primer analysis of mRNA. Samples of RT-PCR material obtained by amplification with wild-type (N1)- and mutant (M1)-specific reverse primers (see Table 1). Lanes 1 and 2 are Δ508-untreated and RT-PCR with N1 and M1 primers, respectively. Similarly, lanes 3 and 4 are Δ508-treated with CF4–CF6 and after RT-PCR with N1 and M1 primers, respectively. Gel electrophoresis was performed in 3% agarose gel.

Fig. 7.

DNA sequencing of RT-PCR products. Sequence analysis of the CF4–CF6-treated Δ508 CFTR was conducted by RT-PCR analysis of total RNA. The PCR materials observed in the gel were subcloned in pCR-Blunt vector and sequenced at the Massachusetts General Hospital Sequencing Facility. Minor peaks represent background. Variations in heights of peaks are due to instrumental sensitivity.

The question arises, then, whether base insertion in the treated Δ508 cells may occur only at the transcriptional level. It was possible that the repair mechanism was carried out instead of, or in addition to, the replication level, by way of reverse transcription extending back to the DNA genome. This hypothesis was tested by RT-PCR analysis of DNA from subcloned Δ508 cells that were originally treated with CF4–CF6. We have, thus far, found no evidence that the restored phenotype is carried back to the inheritable DNA genome level (data not shown). Western blot analysis was conducted with antibodies targeted to epitopes upstream and down-stream of the Δ508 deletion, respectively. The data indicate that the full-length protein is made in the presence of the oligonucleotides (data not shown). This finding eliminates the possibility that the CF4–CF6 treatment causes the translation of a truncated form of CFTR. Further, the data also rule out the possibility that partially degraded protein was translated after treatment with the oligonucleotides because of stop-codon missing signals in the treated mRNA (9). Both antibodies showed the same level of protein without shorter, truncated, or degraded peptides. Thus, treatment with CF4–CF6 does not act as an inhibitor of protein synthesis.

Mechanism of Insertion. As an initial step, we sought restoration of the most common CF phenotype by antisense oligodeoxynucleotide hybridization to the region immediately adjacent to the trinucleotide deletion on both sides of the Δ508 CFTR mRNA. The double-stranded synthetic 2′-O-methyl-RNA–unmodified-RNA oligonucleotide chimera (CF4–CF6) was constructed and used to anneal, selectively cut, and repair the missing region (Fig. 1B). This duplex, as shown, contains a single-stranded 2′-O-methyl-substituted 33-mer oligoribonucleotide (CF4) hybridized to an unmodified 11-mer oligoribonucleotide with 5′ and 3′ monophosphate termini (CF6). The role of the 5′- and 3′-phosphate termini of CF6, if any, remains to be determined. The CF4–CF6 chimera may, theoretically, hybridize with the mRNA bearing the deletion (Fig. 1B), followed by an mRNA cleavage step, as a result of which CF6 might be spliced into this region or, alternatively, serve as a triplex backbone sequence for a one-by-one insertion mechanism. Although insertion of missing bases in the deficient mRNA has been found, as mentioned previously, the specific molecular mechanism is uncertain. The mismatch that we created immediately proximal in the 5′ direction to the Δ508 mRNA UUU deletion may induce a deletion mechanism that is as yet unknown. One possibility is that the deletion step may be induced by either a new enzyme or the Dicer enzyme. Another possibility is that RNase H, in the presence of a triplex, may induce the deletion. There may conceivably be a deletion in RNA analogous to the MutS DNA mismatch mechanism (41). The process by which CF4–CF6 restores the phenotype is consistent with specific docking and cleavage of selective nucleotide sequences hybridized to an mRNA sequence. A TTT insertion into the PCR-amplified deoxyoligonucleotide would be expected if the UUU from CF6 had been inserted into the Δ508 mRNA. However, a TGT (UGU in the Δ508 mRNA) insertion was observed consistently in clones obtained by ARMS-primer amplification RT-PCR products (Figs. 1B and 7). This finding suggests a one-by-one insertion mechanism with a G in place of a U.

Initial sequencing of RT-PCR oligodeoxynucleotides obtained with ARMS primers specific for the CFTR wild-type sequence (N1; see Table 1) revealed the presence of a mixture of residues rich in Gs in the Δ508 site. Subcloning of this oligodeoxynucleotide band resulted in the finding that some but not all of the sequenced cDNA material expressed a TGT at this Δ508 site. Because it was mRNA that was subjected to RT-PCR amplification, the corresponding bases in the Δ508 region were actually UGU. The G residue in the UGU could be accounted for by a restoration mechanism with other than Watson–Crick complementarity in insertion (42). The 3′ hybridization initial steps enabled by the ARMS primers may be flawed also by exonucleotide-induced primer degradation in the reverse-transcriptase step of the RT-PCR amplification procedure (40, 43). Other possibilities of error introduction (44) may explain the apparently artifactual deletions shown in Fig. 1B, which accompany insertions. This consideration was partially confirmed by wild-type ARMS primer-RT-PCR product amplification of the sequence in WT1 cells, which carries only wild-type CFTR. In this reaction, the expected TTT was found by sequence analysis of the wild-type RT-PCR material, in contrast to the TGT (actually UGU in mRNA) when the repaired Δ508 mRNA was sequenced.

For further clarification of the results described above, new sets of primers were constructed with an initial PS-substituted nucleotide in the 3′ end, followed by several (PO) standard nucleotides. This improved primer selectively amplified CF4–CF6-treated but not control (untreated) RT-PCR product in Δ508 cells. The ARMS forward PS primer (SNF1) inserts UGU without a concomitant new deletion (Fig. 1B and Table 1), thus presenting the best case for phenotypic reversion to wild type. A C residue is located 5′ to the UUU in the wild-type CFTR, whereas in the mutant Δ508 CFTR, this residue is a U. This base change may conceivably result in replacement failure 5′ to the inserted UGU using the ARMS reverse primer (N1). The purpose of using CF4–CF6 complex, which has a G residue rather than U proximal to the 5′ end of the Δ508 deletion, was to make this residue complementary to that in the wild-type sequence rather than that in the deleted Δ508 sequence. Both AUC and AUU code for isoleucine. This base mismatch for the Δ508 mRNA (see Fig. 1) may, however, induce a single-strand break analogous to that found for single DNA mismatches as mentioned above (41), necessary for a subsequent repair mechanism to be initiated. This finding eliminates the possibility that the change of C to a U, immediately 5′ to the TTT in the same position in the Δ508 gene, may contribute, in addition to the Δ508 TTT deletion, to the phenotypic change in CFTR.

DNA polymerase has a high degree of Watson–Crick fidelity in synthesizing complementary strands. Reverse transcriptase, which starts at the 3′ end of the PCR amplification, however, has a lower level of this specific type of fidelity. This property of reverse transcriptase may be a possible explanation for the UGU (the equivalent of TGT) found in the amplified and restored Δ508 mRNA. In relation to phenotypic restoration, benign mutations of the TTT present in the wild-type gene do exist. TGT is one of these changes, coding for cysteine, which appears to be an acceptable substitute for phenylalanine in the Δ508 region (44).

The 2′-O-methyl group, plus the 5-methyl of thymidine, which uridine does not have, may also alter the tertiary structure of the CF4 chimera. Such factors may influence the nucleophilicity, electrophilicity, and polarizability of the bases, which make up the mRNA–CF4–CF6 triple-stranded structure. Tinoco and colleagues (45) have described numerous double-stranded DNA/RNA base complementarities, which conformational and other experimental molecular conditions may cause to favor over the standard Watson–Crick AT and GC ones. The highly sensitive nanosphere–gold procedure of Letsinger and colleagues (46, 47) may, in the future, be applicable as an alternative to the PCR technique, avoiding possible artifacts introduced by PCR amplification, or else it may be used after subcloning.

In summary, the above data indicate that specific base insertion in Δ508 mRNA has been made. Certain PCR-introduced artifacts have been avoided by PS modification of the 3′-terminal residues of ARMS specific primers. Under our best conditions, thus far, insertion of UGU has taken place in a sufficient fraction of Δ508 mRNA to induce phenotypic but not genotypic reversion in a tissue-culture cell line. Present results suggest a general approach to modulation of genetic diseases.