Search

DNA interstrand crosslinks (ICLs) represent a severe form of damage that blocks DNA metabolic processes and can lead to cell death or carcinogenesis. The repair of DNA ICLs in mammals is not well characterized. We have reported previously that a key protein complex of nucleotide excision repair (NER), XPA-RPA, recognizes DNA ICLs. We now report the use of triplex technology to direct a site-specific psoralen ICL to a target DNA substrate to determine whether the human global genome NER damage recognition complex, XPC-hHR23B, recognizes this lesion. Our results demonstrate that XPC-hHR23B recognizes psoralen ICLs, which have a structure fundamentally different from other lesions that XPC-hHR23B is known to bind, with high affinity and specificity. XPC-hHR23B and XPA-RPA protein complexes were also observed to bind psoralen ICLs simultaneously, demonstrating not only that psoralen ICLs are recognized by XPC-hHR23B alone, but also that XPA-RPA may interact cooperatively with XPC-hHR23B on damaged DNA, forming a multimeric complex. Since XPC-hHR23B and XPA-RPA participate in the recognition and verification of DNA damage, these results support the hypothesis that interplay between components of the global genome repair sub-pathway of NER is critical for the recognition of psoralen DNA ICLs in the mammalian genome.

Human high mobility group box (HMGB) 1 and -2 proteins are highly conserved and abundant chromosomal proteins that regulate chromatin structure and DNA metabolism. HMGB proteins bind preferentially to DNA that is bent or underwound and to DNA damaged by agents such as cisplatin, UVC radiation, and benzo[a]pyrenediol epoxide (BPDE). Binding of HMGB1 to DNA adducts is thought to inhibit nucleotide excision repair (NER), leading to cell death, but the biological roles of these proteins remain obscure. We have used psoralen-modified triplex-forming oligonucleotides (TFOs) to direct a psoralen-DNA interstrand cross-link (ICL) to a specific site to determine the effect of HMGB proteins on recognition of these lesions. Our results reveal that human HMGB1 (but not HMGB2) binds with high affinity and specificity to psoralen ICLs, and interacts with the essential NER protein, replication protein A (RPA), at these lesions. RPA, shown previously to bind tightly to these lesions, also binds in the presence of HMGB1, without displacing HMGB1. A discrete ternary complex is formed, containing HMGB1, RPA, and psoralen-damaged DNA. Thus, HMGB1 has the ability to recognize ICLs, can cooperate with RPA in doing so, and likely modulates their repair by the NER machinery. The abundance of HMGB1 suggests that it may play an important role in determining the sensitivity of cells to DNA damage under physiological, experimental, and therapeutic conditions.

Naturally occurring DNA sequences can form noncanonical structures such as H-DNA, which are abundant and regulate the expression of several disease-linked genes. Here, we show that H-DNA-forming sequences are intrinsically mutagenic in mammalian cells. This finding suggests that DNA is a causative factor in mutagenesis and not just the end product. By using the endogenous H-DNA-forming sequence found in the human c-myc promoter, mutation frequencies in a reporter gene were increased approximately 20-fold over background in COS-7 cells. H-DNA-induced double-strand breaks (DSBs) were detected near the H-DNA locus. The structures of the mutants revealed microhomologies at the breakpoints, consistent with a nonhomologous end-joining repair of the DSBs. These results implicate H-DNA-induced DSBs in c-myc gene translocations in diseases such as Burkitt's lymphoma and t(12;15) BALB/c plasmacytomas, where most breakpoints are found near the H-DNA-forming site. Thus, our findings suggest that H-DNA is a source of genetic instability resulting from DSBs and demonstrate that naturally occurring DNA sequences are mutagenic in mammals, perhaps contributing to genetic evolution and disease.

A link between genetic abnormalities and carcinogenesis is well established. It follows that a correlation exists between mutation frequency and malignant progression. We have determined the spontaneous and DNA damage-induced mutation frequencies for a series of cell lines derived from SENCAR mouse keratinocytes at various stages of malignant progression. Nontumorigenic mouse keratinocytes (3PC), papillomas (MT1/2), squamous-cell carcinomas (CH72), and spindle-cell carcinomas (CH72T4) were transfected with damaged or undamaged shuttle vectors containing a supF mutation reporter gene. The plasmid mutation frequencies were determined by blue/white screening. The spontaneous plasmid mutation frequency of the squamous-cell carcinoma line was slightly higher than the mutation frequencies of the other cell lines tested. The DNA damage induced by triplex-directed psoralen crosslinks increased the mutation frequencies sixfold to eighteenfold in all cell lines tested, with no significant differences among the cell lines. Sequence analyses revealed that the spindle-cell carcinoma line had a different spontaneous mutation spectrum from the other cell lines. DNA damage-induced mutations were predominantly point mutations at the triplex-duplex junction in all of the cell lines tested, as expected. These data suggested that a strong mutator phenotype was not required for progression to an advanced malignant phenotype in our model system., (Copyright 2004 Wiley-Liss, Inc.)

It has been reported that 80-90% of human cancers may result, in part, from DNA damage. Cell survival depends critically on the stability of our DNA and exquisitely sensitive DNA repair mechanisms have developed as a result. In humans, nucleotide excision repair (NER) protects the DNA against the mutagenic effects of carcinogens and ultraviolet (UV) radiation from sun exposure. By preventing mutations from forming in the DNA, the repair machinery ultimately protects us from developing cancers. DNA damage recognition is the rate-limiting step in repair, and although many details of NER have been elucidated, the mechanisms by which DNA damage is recognized remain to be fully determined. Two primary protein complexes have been proposed as the damaged DNA recognition factor in NER: xeroderma pigmentosum protein A-replication protein A (XPA-RPA) and xeroderma pigmentosum protein C-human homolog of RAD23B (XPC-hHR23B). Here we compare the evidence that supports damage detection by these protein complexes and propose a model for DNA damage recognition in NER based on the current understanding of the roles these proteins may play in the processing of DNA lesions., (Copyright 2003 Wiley-Liss, Inc.)

Site-specific DNA binding molecules offer the potential for genetic manipulation of mammalian cells. Peptide nucleic acids (PNAs) are a DNA mimic in which the purine and pyrimidine bases are attached to a polyamide backbone. PNAs bind with high affinity to single-stranded DNA via Watson-Crick base pairing and can form triple helices via Hoogsteen binding to DNAPNA duplexes. Dimeric bis-PNAs capable of both strand invasion and triplex formation can form clamp structures on target DNAs. As a strategy to promote site-directed recombination, a bis-PNA was coupled to a 40-nt donor DNA fragment homologous to an adjacent region in the target gene. The PNA-DNA conjugate was found to mediate site-directed recombination with a plasmid substrate in human cell-free extracts, resulting in correction of a mutation in a reporter gene at a frequency at least 60-fold above background. Induced site-specific recombination was also seen when the bis-PNA and the donor DNA were co-mixed without covalent linkage. In addition, the bis-PNA and the bis-PNA-DNA conjugate were found to induce DNA repair specifically in the target plasmid. Both the PNA-induced recombination and the PNA-induced repair were found to be dependent on the nucleotide excision repair factor, XPA (xeroderma pigmentosum complementation group A protein). These results suggest that the formation of a PNA clamp on duplex DNA creates a helical distortion that strongly provokes DNA repair and thereby sensitizes the target site to recombination. The ability to promote recombination in a site-directed manner using PNA-DNA conjugates may provide a useful strategy to achieve targeted correction of defective genes.

Nucleotide excision repair (NER) plays a central role in maintaining genomic integrity by detecting and repairing a wide variety of DNA lesions. Xeroderma pigmentosum complementation group A protein (XPA) is an essential component of the repair machinery, and it is thought to be involved in the initial step as a DNA damage recognition and/or confirmation factor. Human replication protein A (RPA) and XPA have been reported to interact to form a DNA damage recognition complex with greater specificity for damaged DNA than XPA alone. The mechanism by which these two proteins recognize such a wide array of structures resulting from different types of DNA damage is not known. One possibility is that they recognize a common feature of the lesions, such as distortions of the helical backbone. We have tested this idea by determining whether human XPA and RPA proteins can recognize the helical distortions induced by a DNA triple helix, a noncanonical DNA structure that has been shown to induce DNA repair, mutagenesis, and recombination. We measured binding of XPA and RPA, together or separately, to substrates containing triplexes with three, two, or no strands covalently linked by psoralen conjugation and photoaddition. We found that RPA alone recognizes all covalent triplex structures, but also forms multivalent nonspecific DNA aggregates at higher concentrations. XPA by itself does not recognize the substrates, but it binds them in the presence of RPA. Addition of XPA decreases the nonspecific DNA aggregate formation. These results support the hypothesis that the NER machinery is targeted to helical distortions and demonstrate that RPA can recognize damaged DNA even without XPA.

Triplex-forming oligonucleotides (TFOs) bind specifically to duplex DNA and provide a strategy for site-directed modification of genomic DNA. Recently we demonstrated TFO-mediated targeted gene knockout following systemic administration in animals. However, a limitation to this approach is the requirement for a polypurine tract (typically 15-30 base pairs (bp)) in the target DNA to afford high affinity third strand binding, thus restricting the number of sites available for effective targeting. To overcome this limitation, we have investigated the ability of chemically modified TFOs to target a short (10 bp) site in a chromosomal locus in mouse cells and induce site-specific mutations. We report that replacement of the phosphodiester backbone with cationic phosphoramidate linkages, either N,N-diethylethylenediamine or N,N-dimethylaminopropylamine, in a 10-nucleotide, psoralen-conjugated TFO confers substantial increases in binding affinity in vitro and is required to achieve targeted modification of a chromosomal reporter gene in mammalian cells. The triplex-directed, site-specific induction of mutagenesis in the chromosomal target was charge- and modification-dependent, with the activity of N,N-diethylethylenediamine > N,N-dimethylaminopropylamine phosphodiester, resulting in 10-, 6-, and <2-fold induction of target gene mutagenesis, respectively. Similarly, N,N-diethylethylenediamine and N,N-dimethylaminopropylamine TFOs were found to enhance targeting at a 16-bp G:C bp-rich target site in a chromatinized episomal target in monkey COS cells, although this longer site was also targetable by a phosphodiester TFO. These results indicate that replacement of phosphodiester bonds with positively charged N,N-diethylethylenediamine linkages enhances intracellular activity and allows targeting of relatively short polypurine sites, thereby substantially expanding the number of potential triplex target sites in the genome.

Gene targeting in mammalian cells has proven invaluable in biotechnology, in studies of gene structure and function, and in understanding chromosome dynamics. It also offers a potential tool for gene-therapeutic applications. Two limitations constrain the current technology: the low rate of homologous recombination in mammalian cells and the high rate of random (nontargeted) integration of the vector DNA. Here we consider possible ways to overcome these limitations within the framework of our present understanding of recombination mechanisms and machinery. Several studies suggest that transient alteration of the levels of recombination proteins, by overexpression or interference with expression, may be able to increase homologous recombination or decrease random integration, and we present a list of candidate genes. We consider potentially beneficial modifications to the vector DNA and discuss the effects of methods of DNA delivery on targeting efficiency. Finally, we present work showing that gene-specific DNA damage can stimulate local homologous recombination, and we discuss recent results with two general methodologies--chimeric nucleases and triplex-forming oligonucleotides--for stimulating recombination in cells.

Triple helix-forming oligonucleotides (TFOs) can bind to polypurine/polypyrimidine regions in DNA in a sequence-specific manner. Triple helix formation has been shown to stimulate recombination in mammalian cells in both episomal and chromosomal targets containing direct repeat sequences. Bifunctional oligonucleotides consisting of a recombination donor domain tethered to a TFO domain were found to mediate site-specific recombination in an intracellular SV40 vector target. To elucidate the mechanism of triplex-induced recombination, we have examined the ability of intermolecular triplexes to provoke recombination within plasmid substrates in human cell-free extracts. An assay for reversion of a point mutation in the supFG1 gene in the plasmid pSupFG1/G144C was established in which recombination in the extracts was detected upon transformation into indicator bacteria. A bifunctional oligonucleotide containing a 30-nucleotide TFO domain linked to a 40-nucleotide donor domain was found to mediate gene correction in vitro at a frequency of 46 x 10(-)5, at least 20-fold above background and over 4-fold greater than the donor segment alone. Physical linkage of the TFO to the donor was unnecessary, as co-mixture of separate TFO and donor segments also yielded elevated gene correction frequencies. When the recombination and repair proteins HsRad51 and XPA were depleted from the extracts using specific antibodies, the triplex-induced recombination was diminished, but was either partially or completely restored upon supplementation with the purified HsRad51 or XPA proteins, respectively. These results establish that triplex-induced, intermolecular recombination between plasmid targets and short fragments of homologous DNA can be detected in human cell extracts and that this process is dependent on both XPA and HsRad51.

Triplex-forming oligonucleotides (TFOs) recognize and bind to specific duplex DNA sequences and have been used extensively to modify gene function in cells. Although germ line mutations can be incorporated by means of embryonic stem cell technology, little progress has been made toward introducing mutations in somatic cells of living organisms. Here we demonstrate that TFOs can induce mutations at specific genomic sites in somatic cells of adult mice. Mutation detection was facilitated by the use of transgenic mice bearing chromosomal copies of the supF and cII reporter genes. Mice treated with a supF-targeted TFO displayed about fivefold greater mutation frequencies in the supF gene compared with mice treated with a scrambled sequence control oligomer. No mutagenesis was detected in the control gene (cII) with either oligonucleotide. These results demonstrate that site-specific, TFO-directed genome modification can be accomplished in intact animals.

Previous studies from our laboratories have shown that (a) Triapine() is a potent inhibitor of ribonucleotide reductase activity and (b) hydroxyurea-resistant L1210 leukemia cells are fully sensitive to Triapine. In an analogous manner, Triapine was similarly active against the wild-type and a hydroxyurea-resistant subline of the human KB nasopharyngeal carcinoma. Triapine was active in vivo against the L1210 leukemia over a broad range of dosages and was curative for some mice. This agent also caused pronounced inhibition of the growth of the murine M109 lung carcinoma and human A2780 ovarian carcinoma xenografts in mice. Optimum anticancer activity required twice daily dosing due to the duration of inhibition of DNA synthesis which lasted about 10 hr in L1210 cells treated with Triapine in vivo. DNA synthesis in normal mouse tissues (i.e. duodenum and bone marrow) uniformly recovered faster than that in L1210 leukemia cells, demonstrating a pharmacological basis for the therapeutic index of this agent. Triapine was more potent than hydroxyurea in inhibiting DNA synthesis in L1210 cells in vivo, and the effects of Triapine were more pronounced. In addition, the duration of the inhibition of DNA synthesis in leukemia cells from mice treated with Triapine was considerably longer than in those from animals treated with hydroxyurea. Combination of Triapine with various classes of agents that damage DNA (e.g. etoposide, cisplatin, doxorubicin, and 1-acetyl-1,2-bis(methylsulfonyl)-2-(2-chloroethyl)hydrazine) resulted in synergistic inhibition of the L1210 leukemia, producing long-term survivors of tumor-bearing mice treated with several dosage levels of the combinations, whereas no enhancement of survival was found when Triapine was combined with gemcitabine or cytosine arabinoside. The findings demonstrate the superiority of Triapine over hydroxyurea as an anticancer agent and further suggest that prevention by Triapine of repair of DNA lesions created by agents that damage DNA may result in efficacious drug combinations for the treatment of cancer.

Targeting DNA damage by triplex-forming oligonucleotides (TFOs) represents a way of modifying gene expression and structure and a possible approach to gene therapy. We have determined that this approach can deliver damage with great specificity to sites in the human gene for the G-protein-linked receptor rhodopsin, mutations of which can lead to the genetic disorder autosomal dominant retinitis pigmentosa. We have introduced DNA monoadducts and interstrand cross-links at multiple target sites within the gene using TFOs with a photoactivatable psoralen group at the 5'-end. The extent of formation of photoadducts (i.e., monoadducts and cross-links) was measured at target sites with a 5'-ApT sequence at the triplex-duplex junction and at a target site with 5'-ApT and 5'-TpA sequences located four and seven nucleotides away, respectively. To improve psoralen reactivity at more distant sites, psoralen moieties were attached to TFOs with nucleotide "linkers" from two to nine nucleotides in length. High-affinity binding was maintained with linkers of up to 10 nucleotides, but affinities tended to decrease somewhat with increasing linker length due to faster dissociation kinetics. DNase I footprinting indicated little, if any, interaction between linkers and the duplex. Psoralen-TFO conjugates formed DNA cross-links with high efficiency (56-65%) at 5'-ApT sequences located at triplex junctions. At a 5'-ApT site four nucleotides away, the efficiency varied with linker length; a four-nucleotide linker gave the highest efficiency. Duplexes with 5'-TpA and 5'-ApT sites two nucleotides away, in otherwise identical sequences, were cross-linked with efficiencies of 56 and 38%, respectively. These results indicate that TFO-linker-psoralen conjugates allow simultaneous, efficient targeting of multiple sites in the human rhodopsin gene.

Specific recognition of a region of duplex DNA by triplex-forming oligonucleotides (TFOs) provides an attractive strategy for genetic manipulation. Based on this, we have investigated the ability of the triplex-directed approach to induce mutations at a chromosomal locus in living cells. A mouse fibroblast cell line was constructed containing multiple chromosomal copies of the lambdasupFG1 vector carrying the supFG1 mutation-reporter gene. Cells were treated with specific (psoAG30) or control (psoSCR30) psoralen-conjugated TFOs in the presence and absence of UVA irradiation. The results demonstrated a 6- to 10-fold induction of supFG1 mutations in the psoAG30-treated cells as compared with psoSCR30-treated or untreated control cells. Interestingly, UVA irradiation had no effect onthe mutation frequencies induced by the psoralen-conjugated TFOs, suggesting a triplex-mediated but photoproduct-independent process of mutagenesis. Sequencing data were consistent with this finding since the expected T.A-->A.T transversions at the predicted psoralen crosslinking site were not detected. However, insertions and deletions were detected within the triplex binding site, indicating a TFO-specific induction of mutagenesis. This result demonstrates the ability of triplex-forming oligonucleotides to influence mutation frequencies at a specific site in a mammalian chromosome.

We have explored the application of triplex technology to the human rhodopsin gene, which encodes a G-protein-linked receptor involved in the genetic disorder autosomal dominant retinitis pigmentosa (ADRP). Our results support the hypothesis that most human genes contain high-affinity triplex sites and further refine the rules governing identification and successful targeting of triplex-forming oligonucleotides (TFOs) to these sites. Using a computer search for sites 15 nucleotides in length and greater than 80% purine, we found 143 distinct sites in the rhodopsin gene and comparable numbers of sites in several other human genes. By applying more stringent criteria, we selected 17 potential target sites in the rhodopsin gene, screened them with a plasmid binding assay, and found 8 that bound TFOs with submicromolar affinity (Kd = 10(-)9-10(-)7 M). We compared purine (GA) and mixed (GT) TFOs at each site, and found that GA-TFOs consistently bound with higher affinity, and were less sensitive to pyrimidine interruptions in the target strand. High G-content favored high-affinity binding; only sites with >54% G-content bound TFOs with Kd

Oligonucleotides offer enormous potential for manipulating gene function in cells and, as such, constitute a promising new class of pharmaceutical agents. Oligonucleotides that form triple helices (triplexes) at specific DNA sequences in defined genes can be used to reduce transcription selectively, to introduce site-specific mutations or to stimulate gene-specific targeted recombination.

Targeting damage to specific sites in the genome represents an attractive approach to manipulating gene function in mammalian cells. To test the applicability of triple-helix formation as a means for achieving precisely timed site-specific damage within a mammalian gene, a triplex-forming oligodeoxyribonucleotide (TFO) that binds with high affinity to a specific site within the hamster adenine phosphoribosyltransferase (APRT) gene was modified with the photochemically reactive psoralen derivative 4'-(hydroxymethyl)-4,5',8-trimethylpsoralen (HMT). The modified TFO, psorTFO1, bound with high affinity to a target site within intron 1 of the APRT gene. Upon irradiation, photomonoadducts (i.e., covalent adducts of psorTFO1 to one strand of the target duplex) were formed with high efficiency (approximately 50%). Introduction of 5'-TpA sequences (the preferred site for psoralen-induced photo-cross-links) at or near the triplex junction leads to increased efficiency of total photoadduct formation and to efficient formation of products that had the electrophoretic mobility on denaturing PAGE expected for three-stranded photo-cross-links (i.e., products containing psorTFO1 covalently linked to both strands of the duplex). Their identities as cross-links were verified by (1) identical electrophoretic mobility of products formed with either duplex strand radiolabeled and (2) coprecipitation of the radiolabeled duplex strand with its complementary biotinylated strand following denaturation. In addition, the cross-links were completely reversible upon irradiation at 254 nm, as expected for psoralen-mediated cross-links. The yield and distribution of photoadducts depended on 5'-TpA position. The most efficient photoadduct formation (approximately 90%) and photo-cross-link formation (approximately 90% of total photoadducts) were observed for a 5'-TpA adjacent to the triplex junction, with significant, but lower, cross-linking efficiency within three base pairs of the junction. Molecular models of the psoralen-conjugated triplex with its six-carbon linker suggested a simple explanation for this distance dependence: facile intercalation near the triplex/duplex junction, with increasing strain required for intercalation at more distant sites. These results indicate that psorTFO1 allows for DNA damage with high precision and high efficiency, and the likely proportion of monoadducts and cross-links can be estimated from the target sequence.

Specific recognition of duplex DNA by a single-stranded oligonucleotide via the formation of triplex DNA is a rational approach for targeting specific regions of a genome. By screening a number of potential target sites for triple helix formation within mammalian genes that allow genetic selection in cell culture, we have identified a site within intron 1 of the hamster adenine phosphoribosyltransferase (APRT) gene that specifically binds a triplex-forming oligodeoxyribonucleotide (TFO) with high affinity. Under optimal conditions for triplex formation, the equilibrium dissociation constant is in the nanomolar range (Kd = 7 x 10(-10) M). This high-affinity binding is very specific, as a 10(5)-fold excess of genomic DNA reduced triplex formation less than 10-fold, and within a 6928-bp plasmid bearing the APRT gene, only restriction fragments containing the intron 1 site were found to bind the TFO. Results of DNase I protection assays were consistent with the TFO binding in an antiparallel orientation via reverse Hoogsteen hydrogen bonds in the major groove of the duplex. We have examined the kinetics of triplex formation as well as the effects of ionic composition and chemical modifications of the TFO on triplex formation. While divalent cations were not required for triplex formation, Mg2+ stabilized the triplex apparently through inhibition of TFO dissociation, with a mean bound lifetime of > 17 h for the triplex at Mg2+ concentrations above 5 mM.(ABSTRACT TRUNCATED AT 250 WORDS)

Development of oligonucleotide derivatives as therapeutic agents requires an understanding of their pharmacokinetic behavior. The in vivo disposition and stability of a prototype of such compounds are reported here. The compound studied, a relatively G-rich 38 base 3' phosphopropyl amine oligonucleotide (TFO-1), was cleared from the circulation with a half-life of approximately 10 minutes, displaying distribution kinetics consistent with a two compartment model. TFO-1 was also readily absorbed into circulation from the peritoneal cavity. All tissues examined except brain accumulated the compound reaching concentrations calculated to be in the micromolar range. TFO-1 was found to be stable in circulation and in tissues in that a large fraction of intact material was detected 8 hours after injection, as assessed by gel electrophoresis. Approximately 20-30% of the injected dose was excreted in the urine over an 8 hour period. These results suggest that G-rich oligonucleotides, minimally modified at the 3' end, are relatively stable in vivo and have distribution kinetics favorable to use as therapeutic agents.

In preclinical investigations performed in mice, 2-amino-9-beta-D-ribofuranosyl purine-6-sulfinamide (sulfinosine), a novel derivative of 6-thioguanosine (6TGR), was active against six solid tumors and four strains of experimental leukemia. Sulfinosine penetrated the central nervous system more readily than did 6TGR and, when given repeatedly, was much more effective in the treatment of L1210 leukemia, being curative for some mice. Other findings of major interest to us were the different dosing characteristics of sulfinosine and 6TGR, the divergent efficiencies of the two drugs in generating cellular resistance, and the activity of sulfinosine against experimental leukemias refractory to 6TGR and other experimental or clinically used chemotherapeutic agents. The chemotherapeutic characterization of sulfinosine that evolved from these studies suggests that this agent may have unique properties that deserve clinical consideration. Both the dosing characteristics of the drug and its pronounced activity against thiopurine-resistant experimental leukemia favor the possibility that sulfinosine could be used to advantage in the treatment of human leukemia unresponsive to 6-mercaptopurine or 6-thioguanine.

Preclinical investigations in vivo revealed unexpected differences in the biological characteristics of 2-amino-9-beta-D-ribofuranosylpurine-6-sulfenamide (sulfenosine, 1) and 2-amino-9-beta-D-ribofuranosylpurine-6-sulfonamide (sulfonosine, 2), two novel but structurally related derivatives of 6-thioguanosine (6TGR). Strikingly, the addition of a fully oxidized sulfur atom at the 6 position of sulfenosine produced a purine derivative (sulfonosine) that was remarkably active against experimental leukemia resistant to treatment with either sulfenosine or 6TGR. This slight structural modification also appeared to influence solubility, scheduling capability, and oral activity as well as penetration of the central nervous system (CNS) and the onset of cellular resistance. These findings underscore the dramatic changes in biologic activity that can be produced by subtle modifications in molecular structure. We trust they may also contribute to the development of improved clinical therapy.