Synthetic and oxidative studies on 8-(arylamino)-2'-deoxyguanosine and -guanosine derivatives.

Facile aerial oxidation is a general feature of guanine ribo- and 2'-deoxyribonucleosides that are substituted at the 8-position by an aminoaryl group. In previous work, it had been suggested that two of the major oxidation products are a pair of diastereomers having a spiro structure. These were presumed to be related by a chiral difference at the spiro carbon atom. The pattern of the oxidative process involves a contraction of the pyrimidine ring. It was thought to be analogous to that suggested by other investigators for the oxidation of uric acid, but for which no really definitive evidence had been presented. We have been able now to isolate in a crystalline state one of the diastereomers produced by the aerial oxidation of 8-phenylaminoguanosine under alkaline conditions. Analysis by X-ray diffraction has now confirmed the type of spiro structure promulgated previously. These findings also imply that spiro compounds are likely to be produced during the aerial oxidation of any 8-arylaminoguanine nucleoside or 2'-deoxynucleoside. In addition, this work adds considerable weight to the results of Poje and Sokolic-Maravic who proposed that a spiro intermediate is produced during the aerial oxidation of uric acid (12,13). However, they found this compound to be unstable to base, in contrast to the arylaminoguanine oxidation products. In the course of the above work we showed that the 8-arylamino derivatives of guanosine can be converted by the Barton deoxygenation method to the corresponding 2'-deoxyribonucleosides. This makes available a number of the latter compounds, which are not easily prepared by other methods.


Introduction
It is now well established that the treatment of DNA with the ultimate carcinogen 2-(N-acetoxy-N-acetyl)aminofluorene (AAAF) under solvolytic conditions leads principally to the formation of adducts in which aminoarylation has occurred at the C-8 position of deoxyguanosine residues.
The major adduct formed in DNA in vivo can be viewed as being derived from the 2'-deoxynucleoside (Structure 1) deoxyguanosine-aminofluorene (dGuo-AF), whereas under in vitro conditions the major adduct is derived from the acetyl This paper was presented at the Fifth International Conference on Carcinogenic and Mutagenic N-Substituted Aryl Compounds held 18-21 October 1992 in Wurzburg, Germany. derivative (Structure 2) deoxyguanosineacetylaminofluorene (dGuo-AAF). These reactions of AAAF with DNA may be regarded as prototypical of the way that most carcinogenic amines behave towards DNA after they have been metabolically activated.
What seems to have escaped general attention, however, is that all of these arylamino adducts, either as the free deoxynucleosides or as residues in DNA, are not stable in solution above pH 7. Nevertheless, the instability associated with AFmodified RNA and DNA, N-(guanosine-8-yl)-2-aminofluorene (Guo-AF) and dGuo-AF had been reported (1-3) more then 20 years ago, but its origin was unrec-ognized at that time. Two groups reported further studies around 1980. Spodheim-Maurizot et al. (4), conducted kinetic studies on the action of alkali on the AAF adducts of guanosine (Guo), deoxyguanosine (dGuo), deoxyguanosine 5'-monophosphate and denatured DNA, whereas Kriek and Westra (5,6)  directly of products obtained when dGuo-AAF was hydrolyzed by alkali. Both groups observed the formation of two products and these were identified (5) tentatively by the latter group as 2'-diastereomers of the 7,8-seco-imidazole derivative (Structures 3a and 3b). Around the same time, Kadlubar et al. (7) presented data on the reaction of Nhydroxy-2-naphthylamine with DNA. In addition to two other adducts, they also found a degradation product in the alkaline enzymatic digests of the DNA. Although this was assigned as the alternate ringopened structure, 8,9-seco-imidazole, in neither case was the spectroscopic evidence sufficient to make the identifications unambiguous (Structures 4 and 5).
the level of the deoxynucleoside (dGuo-AF) and the oligomers containing this residue, it was found that the decomposition could be completely suppressed even under highly alkaline conditions, provided either an antioxidant was present or the reaction was run under anaerobic conditions. Thus, the earlier views of Stohrer et al. (8) were confirmed. However, in the presence of air at 75°C [Kriek and Westra conditions (5)], the acetylated nucleoside 2 was found to disappear rapidly in a 0.2 N sodium hydroxide solution, being replaced by three new compounds (as determined by HPLC analysis), which were simply designated for convenience as ring-opened products (ROP-1, -2, and -3) (Structures 6a, 6b, and 7, respectively). Under these Later reports by Stohrer and his associates (8) described the alkali instability of synthetic DNA containing a dGuo-AAF residue. They found, nevertheless, that the addition of a thiol at the ammonia deblocking step protected the DNA from degradation; this indicated, for the first time, that the degradation might be oxidative. They further suggested that the nature of the oxidation might involve an allantoin type of residue, but did not pursue the matter further.
Recent work by the author and his associates (9,10) has suggested that the degradation might be considerably more complicated than first imagined. Both at Structure 7 (ROP- 3) conditions, ROP-3 appeared only in the early stages of the reaction, as did a fourth peak representing the intermediate deacetylated nucleoside 2 ( Figure 1). Treatment of ROP-1 under identical conditions also gave rise to the same ringopened products. Although the 1H-NMR spectra of ROP-1 and ROP-2 indicated that they were identical to the two substances 3a and 3b isolated previously (5), careful scrutiny of both the positive-ion and negative-ion fast atom bombardment mass spectroscopy (FAB-MS) showed that each has a molecular ion at m/z 462 rather than at 464 as reported earlier (5). This left no doubt that these substances arise via an oxidation mechanism rather than by a simple alkaline hydrolysis. Based on the IH-NMR and FAB-MS analyses, it was concluded (9)(10)(11)) that instead of the diaminopyrimidine structures (3a,3b), most probably these substances are the spirodiastereomers 6a and 6b.
In assigning these structures we noted that there was a parallel with the case of uric acid whose oxidation in alkali had been studied most recently by Poje and Sokolic-Maravic (12,13). The authors concluded through an indirect proof that one of the degradation products was 8, but they noted that it underwent further rapid degradation in the alkaline medium to give noncyclic products. This stands in contrast to Structure 6a and 6b, both of which are stable to further oxidation by oxygen in alkali. Despite a wealth of spectroscopic evidence, Structures 6a, 6b, and 8 still remain speculative. Major  conducting further analytical studies were the unavailability of quantities of Structure 6a or 6b in greater than milligram amounts, and the lack of a general synthetic method for the preparation of 8-arylamino derivatives of deoxyguanosine derivatives. Thus, the latter problem had to be tackled before further structure studies were possible. former reputedly generates a nitrenium ion, the intermediate that attacks the guanine base, principally at C-8 to give the derived product (Structure 11) (14)(15)(16).
This approach works well (yields up to 35%) with compounds such as AAAF (14) that solvolyse easily. However, in cases where hydrolysis predominates (17,18), such as with phenylhydroxylamine and the corresponding biphenyl derivatives, or where the hydroxylamine is very unstable (19), as in the case of the food mutagen (20,21) have described an elegant, intramolecular version of the solvolytic method for the elaboration of the dG-C8-AF; however, the reaction is quite pH-dependent, and the yield is not satisfactory for large-scale production. From the chemical point of view then, the solvolytic method is neither general nor reliable, especially for a multigram scale synthesis. In addition, it is hazardous, as it involves the ultimate carcinogen. Therefore, we decided to investigate an alternative route to the synthesis of arylamine-modified deoxyguanosines, which focuses on a nucleophilic rather than an electrophilic approach.
Although aliphatic amines are known to react with 8-haloguanosines to give the expected 8-alkylamino derivatives (22,23) only one report has described the use of an aromatic amine, and this was in the deoxyribonucleoside series (24). This involved an attempt to induce 2-aminofluorene to react directly with 8-bromo-2'-deoxyguanosine. However, because of the lability of the glycosidic linkage at the high temperatures used, the reaction was accompanied by complete depurination. We attempted to improve the nucleophilicity of the amino group by conversion to the cyanamide (Structure 12). However, reaction of the compound represented by Structure 12 with 13, even under forcing basic conditions, did not lead to the compound shown in Structure 14.
In the more stable ribonucleoside series, it has been reported by Jacobson et al. Despite the fact that the substitution reaction involves an amine nucleophile, the mechanism of the reaction is driven by acid catalysis (20). In the absence of added amine hydrobromide, the reaction is extremely slow, whereas the addition of the proton sponge (1,8-diaminonaphthalene) completely inhibits both the substitution and the depurination reactions (26). Under the conditions, even after 60 hr at 1 10°C, both starting materials were the only substances present.
Although this substitution reaction appears to work well with the amines of aromatic hydrocarbons, unfortunately we could not obtain any reaction with the heterocyclic amine IQ. Neither substitution nor depurination was observed. The lack of substitution probably is related to the poor nucleophilicity associated with 2-aminoimidazoles, a property that likely is diminished even further by electron delocalization into the pyridine ring.
To obtain the corresponding deoxyribonucleosides, compounds 16  Although the overall yields from guanosine are only in the region of 10 to 15%, the reactions are easy to carry out on a large scale. Now, with substantial amounts of both the nucleosides 18 and 19, and the corresponding 2'-deoxy compounds 1 and 23 on hand, we were able to examine in detail their aerial oxidation reactions. However, in this article we present only the work that confirms the structure of the spiro derivatives.

Further Oxidation and Structure Studies
The major objective at this point in the study was to obtain for X-ray analysis a crystal of one of the spiro-derivatives 6a or 6b. However, all attempts to obtain satisfactory crystals from these oxidation products of dGuo-AAF were unsuccessful. Nevertheless, because it had already been demonstrated (4) that the alkaline (aerobic) degradation of Guo-AF and dGuo-AF led to the formation of what appeared to be parallel products (5), the data indicated that the substitution pattern of the sugar residue probably has little influence on the oxidative pathway. In order to confirm these results, we treated Guo-AF under the aerobic alkaline conditions (700C), that we had used previously (9-11) with dGuo-AF. The results were quite analogous ( Figure 2). Three major products can be observed by HPLC analysis (which we trivially designate GOP-1, -2 and -3), and their behavior is very similar to what was observed in the dGuo-AF series. Levels of GOP-1 and GOP-2 rise steadily with time, arriving at roughly a ratio of 2:3. On the other hand, GOP-3 is produced only in low concentration and rapidly vanishes from the system. It likely is the guanosine analog of the compound represented by Structure 7, but structure proof is lacking. 100 Scheme 1. General approach for the removal of 2 -OH of arylamine-modified guanosines. modified Barton deoxygenation method (27,28) to remove the 2'-hydroxyl groups (Scheme 1). The protection of the 3' and 5' hydroxyl groups as the tetraisopropyl disiloxane derivatives (Structure 20) was accomplished by using 1,1',3,3'-tetraisopropyl-dichlorodisiloxane. In both cases, this reaction afforded a clean product. The protected nucleosides were then converted to their phenoxythioacyl derivatives (28) (Structure 21), which could be cleanly deoxygenated by heating with tributyltin hydride/azobis(isobutyronitrile) (AIBN).

F
The product 23, when deprotected by tetrabutylammonium fluoride, afforded the desired 2'-deoxynucleoside 1 in the case where Ar = fluorenyl, and product 22 where Ar = phenyl.
In the former case, the spectral data of the product were identical to those previously published (14). Analytical methods including IH-NMR, 13C-NMR, and FAB-MS confirmed that the compounds obtained in the phenylamine series had the designated corresponding structures. The positive FAB mass spectrum of GOP-I shows a parent ion at m/z 479 [M+H]+, whereas that of 6a (ROP-1) is known to be at m/z 463 [M+H]+, a mass difference of 16 corresponding exactly to one oxygen atom (9,11). Thus, it appears likely that the oxidation products from Guo-AF and dGuo-AF are quite homologous.
Unfortunately, neither GOP-1 nor GOP-2 could be induced to give crystals large enough for X-ray analysis. We turned, therefore, to an examination of the aniline derivative of guanosine-namely compound 19 (Guo-Anil). Surprisingly, when 19 is stirred under air in 1 N sodium hydroxide solution at room temperature, only two products are observed in about equal quantity (designated AOP-1 and AOP-2). At 70°C (Figure 3), AOP-2 vanishes after about 2 hr, while the concentration of AOP-1 rises slowly as that of 19 falls. In the presence of marcaptoethanol, the aerial oxidation of Guo-Anil is completely suppressed, in keeping with all previous studies on analogous compounds.  In contrast to the work with corresponding alkali stable derivatives of dGuo-AF and Guo-AF, AOP-1 derived from Guo-Anil (Structure 19) gave crystals from an aqueous solution that were sufficiently stable in the X-ray beam to provide sufficient data for structure determination. Positional and equivalent isotropic thermal parameters were obtained and give rise to the structure (a trihydrate) shown in Figure 6.
A conventional three-dimensional view of AOP-1 is presented in Figure 7. These figures clearly show the spiro structure of the oxidation product, the spiro center having the S-configuration. The bond lengths of N(2)-C(3) and C(5)-N(6) are 1.29 (M)A, 1.31 (1)A, respectively, and show that the guanidino groups on both of the 5-membered rings exist in the exocyclic imino, rather than the more normal amino tautomeric form that is found in a wide variety of guanidino derivatives (29,30). This may be related to the ring strain energy associated with 5-membered rings in which an exocyclic imine is more stable than one that is endocyclic. In Figure 6, the torsion angle 0(3)-C(12)-N(5)-C(5)X= -54(1)o indicates that AOP-1 adopts the anti conformation about the glycosidic bond in the crystalline state, and that the ribose adopts the C' 2 endo conformation (torsion angle 0(6)-C(16)-C(15)-C(14): X=-58°). Both the imine C(5)-N(6) and the 5'-0(6)H of the ribose bisect the sugar ring, and the distance between N(6) to 0(6) is 2.75A, which suggests that there is a hydrogen bond between them, thus perhaps stabilizing the anti-conformation of  Figure 7. Conventional chemical structure representation of AOP-1.
the glycosidic bond. Also in Figure 6 the torsion angle N(5)-C(5)-(N(6)-C(6): X= 179.1(9)0, indicates that the N-phenyl substituted imine prefers the E-configuration, which is obviously related to the steric hindrance that would exist between the ribose and the phenyl group were they in the Z-related positions.
In the light of the determination of the structure of AOP-1, it would now appear that all of the other derivatives (ROP-1, ROP-2, GOP-1, and GOP-2) likely have the same general structure, given the close chemical relationships of their production. In addition, ROP-1 and ROP-2 almost certainly are related in having opposite chiralities at the spiro-center, a kinship that probably also is shared by GOP-1 and GOP-2. On the other hand, the lack of observation of two diastereosmers in the case of ROP-1 is a mystery that still awaits resolution. Finally, the current work strongly supports the postulated spironature of the alkali labile intermediate [8] observed in the oxidation of uric acid under alkaline conditions (12,13).
Further investigation of the oxidation chemistry of the 8-arylaminoguanosine derivatives is being pursued.