|Home | About | Journals | Submit | Contact Us | Français|
The collision-induced dissociation pathways of isomeric cytosine-guanine and cytosine-adenine intrastrand cross-link-containing dinucleoside monophosphates were investigated with the stable isotope-labeled compounds to gain insights into the effects of chemical structure on the fragmentation pathways of these DNA modifications. Dimroth-like rearrangement, which was reported for protonated 2′-deoxycytidine and involved the switching of the exocyclic N4 with the ring N3 nitrogen atom, was also observed for the cytosine component in the protonated ions of C[5–8]G, C[5-2]A and C[5 – 8]A, but not C[5-N2]G or C[5-N6]A. In these two sets of cross-links, the C5 of cytosine is covalently bonded with its neighboring purine base via a carbon atom on the aromatic ring and an exocyclic nitrogen atom, respectively. On the contrary, the rearrangement could occur for the deprotonated ions of C[5-N2]G, C[5-N6]A and unmodified cytosine, but not C[5 – 8]G, C[5-2]A or C[5 – 8]A. In addition, ammonia could be lost more readily from C[5-N2]G and C[5-N6]A than from C[5 – 8]G, C[5-2]A and C[5 – 8]A. The results from the present study afforded important guidance for the application of mass spectrometry for the structure elucidation of other intrastrand/interstrand cross-link lesions.
Endogenous and exogenous sources can lead to the formation of a variety of DNA damage products including abasic sites, single-nucleobase lesions, cross-link lesions and strand breaks [1, 2]. The formation of these lesions is implicated in the natural processes of aging  and many pathological conditions including cancer  and neurodegeneration . Among those lesions, intrastrand and interstrand cross-link products can form in DNA upon the attack of reactive oxygen species (ROS) induced from the γ-radiolysis of water or Fenton reaction [6–13]. Recently, the structures of a number of ROS-induced intrastrand cross-link lesions of DNA have been identified [6–11, 14–20]. Biochemical studies revealed that this type of lesion could be subjected to repair by nucleotide excision repair enzymes [21, 22]. In addition, these intrastrand cross-link lesions could block replicative DNA polymerases and induce errors during replication by translesion synthesis DNA polymerases [12, 13, 17, 23], underscoring the biological significances of this group of DNA lesions.
Mass spectrometry has evolved into a powerful technique for the structure elucidation and quantification of nucleic acid components and their derivatives that play important roles in numerous fundamental biological processes [24, 25]. To quantify these intrastrand cross-link lesions formed in vitro or in vivo by LC-MS/MS with the isotope-dilution method [12, 13, 19, 26], we synthesized the stable isotope-labeled analogs of a number of intrastrand cross-link-bearing dinucleoside monophosphates (Scheme 1). Since the stable isotopes, that is, 15N and 13C, are selectively incorporated into the pyrimidine nucleoside of the cross-linked dinucleoside monophosphates, this offers us an opportunity to examine the fragmentation chemistry of these cross-links. The outcome of the study may help us to understand better the cleavage pathways of this type of lesions so that it may provide a basis for the structure elucidation of new intrastrand cross-link lesions by mass spectrometry.
[2-13C, 1,3-15N]-5-bromo-2′-deoxyuridine ([2-13C, 1,3-15N]-5-BrdU) was obtained from Cambridge Isotope Laboratories (Andover, MA). All other chemicals were obtained from either Sigma-Aldrich or VWR International (West Chester, PA). Compound [1,3-15N, 2′-D]-2′-deoxycytidine was synthesized following previously described procedures [8, 19]. The isotopically labeled d(G[8-5]C), d(C[5 – 8]G), d(C[5-N2]G), d(C[5-N6]A), d(C[5-2]A) and d(C[5 – 8]A) (Scheme 1) were synthesized starting from [2-13C, 1,3-15N]-5-BrdC-containing dinucleoside monophosphates and oligodeoxynucleotides according to the published procedures for the preparation of the relevant unlabeled compounds [9, 10, 20]. In this respect, the [2-13C, 1,3-15N]-5-BrdC-bearing substrates were synthesized from [2-13C, 1,3-15N]-5-BrdU by using phosphoramidite chemistry . The purities and identities of the synthesized isotope-labeled compounds were verified by HPLC and mass spectrometry.
ESI-MS experiments were carried out on an LCQ Deca XP ion-trap mass spectrometer (ThermoFinnigan, San Jose, CA). For ESI-MS measurements, the cross-link-bearing dinucleoside monophosphates were prepared in a 2-μM aqueous solution containing 50% methanol, and 0.2% formic acid was also added to the sample solutions for experiments in the positive ion-mode. The solution was infused directly into the mass spectrometer with a syringe pump at a flow rate of 2.0 μL/min. The spray voltages were 4.5 and 3.0 kV for experiments in the positive- and negative-ion modes, respectively, and the temperature for the heated capillary was maintained at 225 °C. The automated gain control (AGC) feature was employed, and the maximum numbers of ions were set at 5×107 and 4×107 in MS and MSn modes, respectively. The normalized collisional energy was optimized (15–35%) so that the precursor ion was clearly present (10–40%), but not the most abundant in the product-ion spectrum. Isolation width was set at 3 and 2 m/z units for MS/MS and MSn, respectively. To obtain product-ion spectra for low-mass precursor ions, the activation Q was increased slightly (in a range of 0.25–0.29) to minimize the loss of fragment ions of low m/z values.
The product-ion spectra of many intrastrand cross-link-bearing dinucleoside monophosphates discussed in this paper have been reported previously [8–10, 14, 20]. Some dissociation pathways, however, remain elusive owing largely to the fact that the origins of neutral losses cannot be revealed without the isotope-labeled samples. We took advantage of the availability of dinucleoside monophosphates carrying a stable isotope-incorporated pyrimidine nucleoside (Scheme 1) and the multi-stage MS (MSn) capability of an ion-trap mass spectrometer to elucidate the fragmentation pathways of the cross-link lesions. These data, together with the previously reported MS2 and MSn results of the corresponding unlabeled compounds [8–10, 14, 20], allowed for the unambiguous assignments of the origins of major neutral-loss fragments and the revelation of new fragmentation pathways.
All cross-link lesions studied here harbor a covalent bond between the C5 of cytosine and a ring carbon or exocyclic nitrogen atom of its neighboring purine base. The protonated and deprotonated ions of the covalently linked nucleobase portion can often be produced from the fragmentation of the cross-link-bearing dinucleoside monophosphates. Thus, we reason that the elucidation of the dissociation pathways of those covalently joined nucleobases may build up a basis for employing MSn for the structure determination of other cross-link lesions of DNA. In this context, it is worth noting that any pair of sequence isomers, e.g., d(C[5 – 8]A)/d(A[8-5]C) and d(C[5 – 8]G)/d(G[8-5]C), are cleaved to the identical protonated or deprotonated ions of the cross-linked nucleobase portion because the sequence difference is revoked once the backbone 2-deoxyribose/phosphate is eliminated. In light of this, we refer, in the following discussion, to the cross-linked nucleobase portion of d(C[5 – 8]G) and d(G[8-5]C) as C[5 – 8]G, and similar annotations are used for the nucleobase portions of other intrastrand cross-links.
It was recently reported that, similar as adenine , the protonated ions of cytosine and some of its oxidatively modified derivatives could undergo Dimroth-like rearrangement in the gas phase [29, 30]. This rearrangement allows for the switching of the exocyclic NH2 nitrogen with the N3 nitrogen of cytosine. We began by asking whether the same cleavage chemistry occurs for the two isomeric cross-links formed between cytosine and its neighboring guanine, i.e., C[5 – 8]G and C[5-N2]G. In the viewpoint that HNCO is commonly lost from the cytosine component, we would expect to observe the loss of HN13CO, in addition to H15N13CO, if the same rearrangement occurs for C[5 – 8]G or C[5-N2]G. It turned out that the ions arising from the expulsion of both H15N13CO and HN13CO (of m/z 219 and 220) are of high abundance in the production spectrum (MS3) of the [M + H]+ ion of C[5 – 8]G (Figure 1b), supporting the occurrence of Dimroth-like rearrangement.
Other than the elimination of labeled HNCO, we also observed the losses of both NH3 and 15NH3, with the latter being less facile . In this context, our previous study on the fragmentation of the protonated ion of unmodified cytosine revealed that ammonia can be eliminated mainly from the N3 nitrogen atom, and a small fraction of ammonia can also be lost from the exocyclic N4 nitrogen . We speculate that two factors may account for the preferential loss of unlabeled NH3 over 15NH3 from the protonated C[5 – 8]G. First, the unlabeled NH3 can also be eliminated from the guanine portion of the cross-link. Along this line, previous work by Gregson et al.  showed that the protonation at the N1 of guanine and the subsequent cleavage of the N1-C2 bond can result in the loss of an ammonia from the N1 and N2 of guanine. Second, the attachment of the C5 of cytosine with the C8 of guanine may also alter the fragmentation of the cytosine portion by leading to the facile loss of ammonia from the exocyclic N4 nitrogen atom of cytosine.
In sharp contrast, the loss of H15N13CO, but not HN13CO, was evident in the corresponding product-ion spectrum resulting from the fragmentation of the [M − H]− ion of C[5 – 8]G (Figure 1d), underscoring the prohibition of the Dimroth-like rearrangement for the [M − H]− ion of the conjugate where the C5 of cytosine is covalently bonded with the C8 of guanine. Moreover, we found the loss of an unlabeled NH3 to give the ion of m/z 245 in low abundance (Figure 1d).
We next examined the fragmentation of C[5-N2]G, a cross-link lesion where the C5 of cytosine is coupled to the exocyclic NH2 group of its neighboring guanine. It turned out that, upon the collisional activation of the protonated ion, the fragment ion arising from the loss of unlabeled NH3 dominates the product-ion spectrum. Additionally, while the ion formed from the loss of H15N13CO was of very low abundance, the ion emanating from the elimination of an HN13CO was not detectable (Figure 1a). These results showed that the Dimroth-like rearrangement does not occur for the [M + H]+ ion of C[5-N2]G. In this context, it is worth noting that the ion resulting from the loss of H15N13CO is of low abundance; nonetheless, the results were reproducible. By contrast, the losses of HNCO, H15N13CO and HN13CO, in decreasing preference, were among the most abundant neutral losses found in negative-ion MS3 (Figure 1c). The unlabeled HNCO can only be eliminated from guanine. The HNCO loss resulting from the Dimroth-like rearrangement of the cytosine component is estimated to be approximately 30% by comparing the relative abundances of the ions of m/z 218 and m/z 217 (Figure 1c).
Intrastrand cross-link products formed between cytosine and its neighboring adenine are of special interest because we identified three isomers, in which the C5 of cytosine is coupled to the N6, C2 or C8 of adenine to yield C[5-N6]A, C[5-2]A and C[5 – 8]A, respectively (Scheme 1) . In the product-ion spectra of the [M + H]+ ions of the nucleobase portions of the three cross-links, we found that the neutral loss of NH3 (m/z 231) occurs much more readily for C[5-N6]A than for C[5 – 8]A and C[5-2]A (Figure 2a-c). The fragment ion arising from the loss of 15NH3, i.e., the ion of m/z 230, was also found for C[5-N6]A, though at an abundance of about 1/3 of that for the ion emanating from the loss of unlabeled NH3 (the m/z 231 ion). As discussed above, protonated cytosine can undergo the neutral loss of ammonia from both the N3 and N4 nitrogen atoms, with the former loss being more facile . In addition, ammonia can be eliminated from the N1 or N6 nitrogen of protonated adenine, with the loss from the former nitrogen being slightly preferred . Thus, similar as the preferential loss of unlabeled NH3 over 15 NH3 from C[5-N2]G, the more favorable loss of unlabeled NH3 from C[5-N6]A might be due to the elimination of NH3 from the adenine component of the cross-link or attributed to the more facile loss of NH3 from the N4 of cytosine emanating from the covalent linkage of the C5 of cytosine with the adenine.
Similar as what we found for the isomeric cross-links involving cytosine and guanine, the three cross-links formed between cytosine and adenine also exhibit difference in their susceptibility toward Dimroth-like rearrangement. The ion of m/z 204, arising from the neutral loss of an HN13CO after the rearrangement, has relative abundances of approximately 35% and 20% for C[5 – 8]A and C[5-2]A, respectively, but this ion is almost absent for C[5-N6]A (Figure 2). The above results, therefore, demonstrate that the Dimroth-like rearrangement is prohibited for the [M + H]+ ion of C[5-N6]A, but allowed for the [M + H]+ ions of C[5 – 8]A and C[5-2]A. This finding is in keeping with the inhibition of the Dimroth-like rearrangement for the [M + H]+ ion of C[5-N2]G, but not C[5 – 8]G (vide supra). Therefore, we may conclude that the covalent linkage of the C5 of cytosine with the exocyclic amino nitrogen of a purine base, i.e., the N6 of adenine or the N2 of guanine, results in the inhibition of the Dimroth-like rearrangement for the protonated ion of the cross-linked nucleobase portion. The underlying cause for this inhibition remains unclear.
The product-ion spectra acquired in the negative-ion mode revealed that all three cytosine-adenine cross-links undergo a very facile loss of the labeled HNCO moiety, though the fragment ion arising from the loss of 15NH3 was also found in the product-ion spectrum of C[5-N6]A. In addition, the negative-ion spectra for the three isomers exhibited completely opposite trend in Dimroth-like rearrangement for the three isomers, i.e., the rearrangement was found for C[5-N6]A, but not for the other two isomers. In this context, the product-ion spectrum of C[5-N6]A showed two major fragment ions of m/z 201 and 202 (Figure 2d), which are attributed to the losses of H15N13CO and HN13CO, respectively. The latter neutral loss can only form from the Dimroth-like rearrangement product of C[5-N6]A. On the contrary, the ion resulting from the loss of H15N13CO (m/z 201) dominates the product-ion spectra of C[5 – 8]A and C[5-2]A (Figure 2e-f), whereas the elimination of HN13CO is barely detectable, supporting that the Dimroth-like rearrangement does not occur for the [M − H]− ions of these two isomers. This result is consistent with what we observed for the two isomeric cytosine-guanine cross-links, where the Dimroth rearrangement was found for the [M − H]− ion of C[5-N2]G, but not C[5 – 8]G.
The unique structural features of the cross-links may account for the difference in their susceptibilities toward Dimroth-like rearrangement in the negative-ion mode. For deprotonated ions, the Dimroth-like rearrangement of the cytosine portion necessitates the cleavage of the C2-N3 bond of cytosine. Therefore, the deprotonation of the N1 of cytosine and the localization of the negative charge on the carbonyl oxygen (O2) of cytosine are expected to promote the cleavage (Scheme 2). In C[5-N6]A and C[5-N2]G, the two nucleobases remain as two isolated π systems, which may favor the localization of the negative charge on the O2 of cytosine. In the other three lesions, i.e., C[5 – 8]G, C[5 – 8]A and C[5-2]A, the conjugation of the π systems of the two nucleobases renders the negative charge delocalized between the two nucleobases. The decreased negative charge density on the O2 of cytosine may disfavor the cleavage of the C2-N3 bond of cytosine thereby inhibiting the subsequent Dimroth-like rearrangement.
We also examined the fragmentation of the [M − H]− ion of [1,3-15N]cytosine, and it turned out that approximately 50% of the HNCO loss can arise from the Dimroth-like rearrangement product of the deprotonated cytosine (Figure 3). In this context, the Dimroth-like rearrangement occurred at somewhat lower frequency (ca. 30%) for the protonated cytosine . It is also worth noting that we did not find fragment ion emanating from the loss of ammonia in Figure 3, whereas this loss occurs readily during the collisional activation of the protonated cytosine . This difference might be attributed to the fact that there are only two acidic protons in deprotonated cytosine and the loss of ammonia would require the transfer of a non-exchangeable proton to the nitrogen atom to be eliminated. However, no such transfer is needed for the loss of ammonia from protonated cytosine.
Several general conclusions can be drawn from the results presented above. First, upon collisional activation of the protonated or deprotonated ions of the cross-linked nucleobases, the pyrimidine base is more susceptible to cleavage than the purine base possibly due to the lower level of conjugation for the pyrimidine bases than for the purine bases. Second, we found that, in the positive-ion mode, the Dimroth-like rearrangement can occur for the cross-links where cytosine is conjugated with its neighboring adenine or guanine base to yield an extended π-system, i.e., C[5 – 8]G, C[5 – 8]A, and C[5-2]A. This rearrangement, however, is not found for C[5-N6]A or C[5-N2]G, where the C5 of cytosine is coupled to an exocyclic amino nitrogen of the neighboring purine base. By contrast, such rearrangement occurs upon the fragmentation of the deprotonated ions of C[5-N6]A and d[5-N2]G, but not for C[5 – 8]G, C[5 – 8]A, or C[5-2]A. Furthermore, when the C5 of cytosine is conjugated with the exocyclic amino group of a purine base, i.e., the N2 of guanine or the N6 of adenine, the loss of ammonia from the cross-linked nucleobase portion is much more facile than the same loss from the cross-links where the C5 of cytosine is linked with the ring carbon of its neighboring purine base, i.e., the C2 or C8 of adenine and the C8 of guanine.
The fragmentation chemistry reported here may be useful for the structure elucidation of other intrastrand cross-link lesions. In addition, it may be applied for gaining insights into the structures of interstrand cross-link lesions if the cross-linked nucleobase component can be produced upon the fragmentation of the interstrand cross-link-bearing nucleosides/nucleotides.
The authors want to thank Haizheng Hong and Yu Zeng for providing cross-link products for this study. This work was supported by the National Institutes of Health (R01 CA101864).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.