An assessment of the antisense properties of RNase H-competent and steric-blocking oligomers.

The antisense activity and gene specificity of two classes of oligonucleotides (ONs) were directly compared in a highly controlled assay. One class of ONs has been proposed to act by targeting the degradation of specific RNAs through an RNase H-mediated mechanism and consists of C-5 propynyl pyrimidine phosphorothioate ONs (propyne-S-ON). The second class of antisense agents has been proposed to function by sterically blocking target RNA formation, transport or translation and includes sugar modified (2'-O-allyl) ONs and peptide nucleic acids (PNAs). Using a CV-1 cell based microinjection assay, we targeted antisense agents representing both classes to various cloned sequences localized within the SV40 large T antigen RNA. We determined the propyne-S-ON was the most potent and gene-specific agent of the two classes which likely reflected its ability to allow RNase H cleavage of its target. The PNA oligomer inhibited T Ag expression via an antisense mechanism, but was less effective than the propyne-S-ON; the lack of potency may have been due in part to the PNAs slow kinetics of RNA association. Interestingly, unlike the 2'-O-allyl ON, the antisense activity of the PNA was not restricted to the 5' untranslated region of the T Ag RNA. Based on these findings we conclude that PNAs could be effective antisense agents with additional chemical modification that will lead to more rapid association with their RNA target.

Strand-invasion of duplex DNA by peptide nucleic acid oligomers.

Polyamide oligomers, termed peptide nucleic acids (PNAs), bind with high affinity to both DNA and RNA and offer both antisense and antigene approaches for regulating gene expression. When a PNA binds to a complementary sequence in a double-stranded DNA, one strand of the duplex is displaced, and a stable D-loop is formed. Unlike oligodeoxynucleotides for which binding polarity is determined by the deoxyribose sugar, the unrestrained polyamide backbone of the PNA could permit binding to a DNA target in an orientation-independent manner. We now provide evidence that PNAs can, in fact, bind to their complementary sequence in DNA independent of the DNA-strand polarity--that is, a PNA binds to DNA in both "parallel" and "antiparallel" fashion. With a mixed-sequence 15-mer PNA, kinetic studies of PNA.DNA interactions revealed that D-loop formation was rapid and the complex was stable for several hours. However, when measured either by gel-mobility-shift analysis or RNA polymerase II-elongation termination, D-loop formation was salt dependent, but PNA-strand dissociation was not salt dependent. We observed that D-loop-containing DNA fragments had anomalous gel mobilities that varied as a function of the position of the D-loop relative to the DNA termini. On the basis of permutation analysis, the decreased mobility of the PNA.DNA complex was attributed to a bend in the DNA at or near the D-loop.

Antisense and antigene properties of peptide nucleic acids.

Peptide nucleic acids (PNAs) are polyamide oligomers that can strand invade duplex DNA, causing displacement of one DNA strand and formation of a D-loop. Binding of either a T10 PNA or a mixed sequence 15-mer PNA to the transcribed strand of a G-free transcription cassette caused 90 to 100 percent site-specific termination of pol II transcription elongation. When a T10 PNA was bound on the nontranscribed strand, site-specific inhibition never exceeded 50 percent. Binding of PNAs to RNA resulted in site-specific termination of both reverse transcription and in vitro translation, precisely at the position of the PNA.RNA heteroduplex. Nuclear microinjection of cells constitutively expressing SV40 large T antigen (T Ag) with either a 15-mer or 20-mer PNA targeted to the T Ag messenger RNA suppressed T Ag expression. This effect was specific in that there was no reduction in beta-galactosidase expression from a coinjected expression vector and no inhibition of T Ag expression after microinjection of a 10-mer PNA.

DNA triple-helix formation at physiologic pH and temperature.

Oligonucleotides that form a triple helix with duplex DNA offer a novel way to site specifically regulate gene expression in vivo. Triple helices formed by homopyrimidine oligomers containing both cytosine and thymine are stabilized by acid pH and low temperature, and there is little information about triplex formation with these oligomers at both pH 7.5 and 37 degrees C. Therefore, we examined the effect of changing various conditions on triplex formation at pH 7.5. A 30-mer oligonucleotide (composed of T and 5-methyl C) at submicromolar concentrations formed a triplex with its target duplex at pH 7.5 and 37 degrees C. Association of the 30-mer oligomer with the duplex was slow, with complete association requiring about 1 h. At 37 degrees C, a 21-mer oligomer bound weakly to the target duplex but both a 25-mer and the 30-mer readily formed a triplex. This relationship of triplex formation with length was temperature dependent, as at 25 degrees C the 21-mer behaved similarly to the longer oligomers. Increasing spermine concentrations (from 0.2 to 1 mM) increased the amount of triplex formed. Spermine may be important only for the association of the oligomer to the duplex, since decreasing the spermine concentration after the triplex formed did not reduce the amount of triplex detected. At 1 mM spermine, formation of the triple-helical complex was very dependent on the concentration of KCl; increasing the KCl from 50 to 100 mM prevented triplex formation. However, the inhibitory effect of KCl could be abrogated by raising the spermine concentration to 2 mM. Our observations indicate that a triple helix can form under physiologic conditions but its formation is affected by several competing interactions.

Multiple non-B-DNA conformations of polypurine.polypyrimidine sequences in plasmids.

A polypurine.polypyrimidine (Pur.Pyr) sequence with a central interruption in a plasmid can adopt multiple non-B-DNA conformations depending on the conditions as revealed by specific chemical probes (OsO4, diethyl pyrocarbonate, and dimethyl sulfate) and two-dimensional electrophoresis. The relatively long mirror repeat Pur.Pyr sequences (GAA)9TTC(GAA)8 and (GGA)9TCC(GGA)8 form single canonical intramolecular triplexes at pH 7.0-6.0 in negatively supercoiled plasmids as isolated from Escherichia coli. With a lowering of the pH and/or an increase in the degree of negative supercoiling, these sequences undergo a novel conformational change as revealed by diethyl pyrocarbonate hypermodification of adenines in the middle of the polypurine strand and OsO4 reaction with thymines in the center and the quarter points of the polypyrimidine strand. To evaluate this structure, a family of related Pur.Pyr sequences were cloned and studied. The non mirror repeat sequence (GGA)9TCC(GAA)8 forms a non-B conformation only under acidic pH conditions, but the structural properties are different from those of the mirror repeat sequences. Furthermore, when the central interruptions of a mirror repeat sequence were increased from 3 to 9 bp, two canonical triplexes formed independently at pH 5.0 [at the (GAA)9 and (GAA)8 regions in the sequence (GAA)9TTAATTCGC(GAA)8]. Thus, if an interruption is sufficiently long, the two halves of the Pur.Pyr sequence do not interact with each other. Novel types of folded DNA geometries which explain these results are described.

Site-specific inhibition of EcoRI restriction/modification enzymes by a DNA triple helix.

The ability of oligopyrimidines to inhibit, through triple helix formation, the specific protein-DNA interactions of the EcoRI restriction and modification enzymes (EcoRI and MEcoRI) with their recognition sequence (GAATTC) was studied. The oligonucleotides (CTT)4 and (CTT)8 formed triplexes in plasmids at (GAA)n repeats containing EcoRI sites. Cleavage and methylation of EcoRI sites within these sequences were specifically inhibited by the oligonucleotides, whereas an EcoRI site adjacent to a (GAA)n sequence was inhibited much less. Also, other EcoRI sites within the plasmid, or in exogenously added lambda DNA, were not inhibited. These results demonstrate the potential of using triplex-forming oligonucleotides to block protein-DNA interactions at specific sites, and thus this technique may be useful in chromosome mapping and in the modulation of gene expression.

Intramolecular DNA triplexes in supercoiled plasmids. I. Effect of loop size on formation and stability.

The capacity of four oligopurine.oligopyrimidine (pur.pyr) sequences with different lengths of interruptions in the center [GAA)4(N)n(GAA)4G) (n = 3, 5, 7, and 9) to adopt intramolecular DNA triplexes was evaluated in recombinant plasmids. The hyperreactive patterns of the pur.pyr inserts to specific chemical probes (OsO4, diethyl pyrocarbonate, and dimethyl sulfate) at the base pair level demonstrate that intramolecular triplexes with identical 12-base triads in the stem but with different loop sizes (4, 6, 8, and 10 bases) can form in supercoiled plasmids. Furthermore, the extent of OsO4 modification was measured as a function of temperature and of average negative supercoil density. In addition, the transition free energy of B-DNA to triplexes at pH 4.5 was determined by two-dimensional electrophoresis. These comparative studies show that longer loops require more supercoil energy for triplex formation and are less thermostable than triplexes with shorter loops. Also, it may be that not only the loop size but the base composition of the loop region affects the structural transition and triplex stability. Thus, these results significantly broaden the range of natural pur.pyr sequences that may adopt triplexes.

Intramolecular DNA triplexes in supercoiled plasmids. II. Effect of base composition and noncentral interruptions on formation and stability.

The effects of interruptions in the homopurine bias and the G+C content of the homopurine.homopyrimidine (pur.pyr) sequences on intramolecular triplex formation and stability in supercoiled plasmids were evaluated. In addition, the interconversion of triplex and duplex, after altering the stabilizing factors (low pH or supercoiling), was studied. We conclude: (a) a 42-base pair pur.pyr sequence with three consecutive interruptions does not form a large triplex with three unpaired nucleotides in the stem. Instead, a mixture of two smaller (27- and 28-nucleotide) triplexes forms. (b) A 28-nucleotide sequence with a single interruption forms a triplex with one unpaired nucleotide in the stem. This interruption causes the triplex to be 7 degrees C less thermostable and requires more superhelical energy for formation than the control triplex. (c) As the G+C content of a pur.pyr sequence increases, the thermostability of the triplex increases and the triplex requires less supercoiling for formation. (d) The interconversion between duplex and triplex is fast. After negative supercoiling is removed, all triplex becomes duplex in about 3 min. When the pH is shifted from 8.0 to 5.2, the conversion of duplex to triplex in a negatively supercoiled plasmid is complete in less than 2 min. Hence, these kinetic properties are consistent with important biological roles for triplexes. In summary, the results from both this and the accompanying paper show that a substantial amount of sequence imperfections is tolerated for triplex formation and stability.

The chemistry and biology of unusual DNA structures adopted by oligopurine.oligopyrimidine sequences.

A family of unusual DNA structures has been discovered in segments with predominantly purines in one strand (pur.pyr sequences). These sequences are overrepresented in eukaryotic DNA and have been mapped near genes and recombination hot spots. When cloned into recombinant plasmids, many pur.pyr sequences are reactive to chemical and enzymic probes that are generally specific for single-stranded DNA. An intramolecular triplex is adopted by mirror repeats of G's and A's. Other non-B DNA structures adopted by similar sequences remain to be fully clarified but may be a family of related conformations. It is likely that these unorthodox structures play an important role in the function of the eukaryotic genome.

Intramolecular DNA triplexes in supercoiled plasmids.

A series of inserts with oligopurine.oligopyrimidine mirror repeat sequences was investigated at the base pair level with specific chemical probes (OsO4 and diethylpyrocarbonate) to evaluate the in vitro existence of intramolecular triplexes. Two parent inserts in recombinant plasmids with (GAA)9 and (AG)12 sequences and three mutant inserts (containing transitions or transversions) revealed that base pair changes at one location affected the chemical reactivity 13 base pairs away. The specificity and nature of these reactions, as well as the thermal stability of the complexes, provide direct evidence for the existence of a triplex with a portion of the pyrimidine-rich strand folded back and Hoogsteen-paired in the major groove of the Watson-Crick duplex. The biological implications of this unorthodox DNA structure are discussed.

8-Ketodeoxycoformycin and 8-ketocoformycin as intermediates in the biosynthesis of 2'-deoxycoformycin and coformycin.

An enzyme has been isolated from cell-free extracts of Streptomyces antibioticus that can catalyze the reduction of 8-ketodeoxycoformycin (8-KetodCF) and 8-ketocoformycin (8-ketoCoF) to the naturally occurring nucleoside analogues 2'-deoxycoformycin (dCF) and coformycin (CoF), respectively. The partially purified reductase requires NADPH as the cofactor and stereospecifically reduces the 8-keto group of both ketonucleoside substrates to a hydroxyl group with the R configuration at C-8. This is the same configuration of the hydroxyl group as that of the dCF and CoF isolated from S. antibioticus. The reduction proceeds at the nucleoside level, and ATP is not required. The reductase is stereospecific for the NADPH cofactor in that it transfers the pro-S but not the pro-R hydrogen from C-4 of NADPH to the 8-keto group. The apparent Km for 8-ketodCF and 8-ketoCoF were 250 and 150 microM, respectively. These in vitro results, which show that 8-ketodCF and 8-ketoCoF may be intermediates in the biosynthesis of dCF and CoF, support and extend our earlier results from in vivo studies which established that adenosine and C-1 of D-ribose are the carbon-nitrogen precursors of dCF. A possible mechanism for the formation of dCF is presented.

Influence of DNA sequence on the formation of non-B right-handed helices in oligopurine.oligopyrimidine inserts in plasmids.

A systematic study was conducted on seven recombinant plasmids harboring synthetic inserts which had all purines on one strand and all pyrimidines on the complementary strand (Pur.Pyr). The inserts ranged in G+C content from 100% [G19.C19] to 0% [A20.T20] with intermediate contents at 66% [(TCC)8.(GGA)8], 50% [(CT)12.(AG)12 and (TTCC)6.(GGAA)6], 33% [(TTC)8.(GAA)8], and 25% [(GAAA)6.(TTTC)6]. The specific reactions at the base pair level of these inserts with enzymatic (S1 and P1 nucleases) and chemical (bromoacetaldehyde, OsO4, diethyl pyrocarbonate, and dimethyl sulfate) probes were evaluated as influenced by pH, negative supercoiling, and ionic strength (NaCl). Supercoil-induced relaxation studies using two-dimensional gels also provided important conformational information. We conclude that the five inserts with 66-25% G+C adopt a non-B right-handed conformation which is stabilized by negative supercoiling. Low pH (pH values 4.5-5.0) tends to stabilize this structure but is not essential for its formation. Surprisingly, an end bias of reactivity from the center toward the 5'-end of the purine strand of these inserts was generally found for the enzymatic and chemical probes which was irrespective of the orientation of the insert in the pRW790 vector. An intramolecular triple-stranded model for the unusual structure of the insert accounts most favorably for these observations. Unexpectedly, the A20.T20 insert seems to remain in an orthodox right-handed B-conformation under all conditions tested. The G19.C19 insert does adopt a non-B right-handed structure as for the five inserts with 66-25% G+C, but the pattern of reactivities and hence its conformation is different.

Biosynthesis of 2'-deoxycoformycin: evidence for ring expansion of the adenine moiety of adenosine to a tetrahydroimidazo[4,5-d][1,3]diazepine system.

2'-Deoxycoformycin (2'-dCF), a nucleoside antitumor agent produced in trace quantities by Streptomyces antibioticus, has been shown in earlier work to originate from the intact carbon-nitrogen framework of adenosine. Additional experiments using 13C and two-dimensional Fourier transform NMR techniques, together with radiolabeling studies, identify the C-1 of D-ribose, and not the tetrahydrofolate "C-1 pool", as the source of the C-7 carbon in the aglycon of 2'-dCF. These results show that the adenine portion of adenosine (or a nucleotide thereof) undergoes a unique ring expansion, by insertion of a -CH2- unit between the N-1 and C-6 of the adenine ring, to furnish the 1,3-diazepine portion of 2'-dCF.

Evidence for the conversion of adenosine to 2'-deoxycoformycin by Streptomyces antibioticus.

The incorporation and distribution of 14C in 2'-deoxycoformycin, elaborated by Streptomyces antibioticus, were studied with [U-14C]glycine, [U-14C]adenosine and [U-14C]adenine. Similar ratios of 14C in the aglycon and carbohydrate portions of 2'-deoxycoformycin, ara-A, and adenosine isolated from the RNA indicated that [U-14C]adenosine was incorporated into 2'-deoxycoformycin without cleavage of the N-glycosylic bond. Following the addition of [U-14C]adenine, 98% of the 14C isolated from [14C]-2'-deoxycoformycin resided in the aglycon. 2'-Deoxycoformycin biosynthesis may not require the de novo purine biosynthetic pathway as evidenced by the failure to detect incorporation of [U-14C]glycine into 2'-deoxycoformycin. These data suggest that the biosynthesis of 2'-deoxycoformycin involves the incorporation of the carbon-nitrogen skeleton of an intact purine nucleoside or nucleotide, thereby implying that a purine ring is opened enzymatically between C-6 and N-1 and a one-carbon unit is added to form the 1,3-diazepine ring of 2'-deoxycoformycin.