Reactivity of damaged pyrimidines: Formation of a schiff base intermediate at the glycosidic bond of saturated dihydrouridine

Article abstract of DOI:10.1021/ja512435j

DNA glycosylases catalyze the first step of the base excision repair (BER) pathway. The chemistry used by these enzymes for deglycosylation has been largely considered as the chemistry of the oxocarbenium ion, e.g., direct rupture of the C1-N1 bond resulting in an oxocarbenium ion intermediate. Here we present mechanistic studies revealing the 2-deoxyribose isomerization and subsequent deglycosylation processes in two pyrimidine lesions: 5,6-dihydro-2-deoxyuridine (dHdU) and 5,6-dihydrothymidine (dHT), formed via ionizing radiation damage to 2-deoxycytidine and thymidine, respectively, under anoxic conditions. Acid or heat treatment of these two lesions leads to the production of two pairs of C1 epimers containing a pyranose and a furanose, respectively, indicating that both lesions favor the rupture of the C1-O4 bond, resulting in a Schiff base intermediate at the N-glycosidic bond. Such a Schiff base intermediate was trapped and characterized by either Pd-catalyzed hydrogenation or thiol-mediated addition reaction. In contrast, in undamaged 2-deoxyuridine and thymidine, reactions at elevated temperatures lead to the release of nucleobases most likely via the traditional oxocarbenium ion pathway. DFT calculations further support the experimental findings, suggesting that the oxocarbenium ion intermediate is responsible for the deglycosylation process if the integrity of the pyrimidine ring is maintained, while the Schiff base intermediate is preferred if the C5=C6 bond is saturated. Currently, the oxocarbenium ion pathway is indicated to be solely responsible for the deglycosylation in BER enzymes, however our results suggest an alternative Schiff base mechanism which may be responsible for the repair of saturated pyrimidine damages.

Full text of DOI:10.1021/ja512435j

Subscriber access provided by McMaster University Library
Article
Reactivity of damaged pyrimidines: formation of a Schiff base
intermediate at the glycosidic bond of saturated dihydrouridine
Yajun Jian, Gengjie Lin, Lidia Chomicz, and Lei Li
J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/ja512435j • Publication Date (Web): 11 Feb 2015
Downloaded from http://pubs.acs.org on February 18, 2015
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of the American Chemical Society is published by the American Chemical
Society. 1155 Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 34
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Reactivity of Damaged Pyrimidines: Formation
of a Schiff Base Intermediate at the Glycosidic
Bond of Saturated Dihydrouridine
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Yajun Jian,†,⊥ Gengjie Lin,†,⊥ Lidia Chomicz,║ and Lei Li †‡*
†Department of Chemistry and Chemical Biology, Indiana UniversityꢀPurdue University Indianapolis
(IUPUI), 402 North Blackford Street, Indianapolis, Indiana, 46202
‡Department of Biochemistry and Molecular Biology and Department of Dermatology, Indiana
University School of Medicine, 635 Barnhill Drive, Indianapolis, Indiana 46202
║Department of Chemistry, University of Gdansk, 80ꢀ308 Gdansk, Poland
lilei@iupui.edu
1
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 2 of 34
ABSTRACT: DNA glycosylases catalyze the first step of the base excision repair (BER) pathway. The
chemistry used by these enzymes for deglycosylation has been largely considered as the chemistry of
the oxocarbenium ion, e.g. direct rupture of the C1ʹꢀN1 bond resulting in an oxocarbenium ion
intermediate. Here we present mechanistic studies revealing the 2ʹꢀdeoxyribose isomerization and
subsequent deglycosylation processes in two pyrimidine lesions: 5,6ꢀdihydroꢀ2ʹꢀdeoxyuridine (dHdU)
and 5,6ꢀdihydrothymidine (dHT), formed via ionizing radiation damage to 2ʹꢀdeoxycytidine and
thymidine respectively under anoxic conditions. Acid or heat treatment of these two lesions leads to the
production of two pairs of C1ʹ epimers containing a pyranose and a furanose respectively, indicating that
both lesions favor the rupture of the C1ʹꢀO4ʹ bond, resulting in a Schiff base intermediate at the Nꢀ
glycosidic bond. Such a Schiff base intermediate was trapped and characterized by either Pd catalyzed
hydrogenation or thiolꢀmediated addition reaction. In contrast, in undamaged 2ʹꢀdeoxyuridine and
thymidine, reactions at elevated temperatures lead to the release of nucleobases most likely via the
traditional oxocarbenium ion pathway. DFT calculations further support the experimental findings,
suggesting that the oxocarbenium ion intermediate is responsible for the deglycosylation process if the
integrity of the pyrimidine ring is maintained; while the Schiff base intermediate is preferred if the
C5=C6 bond is saturated. Currently, the oxocarbenium ion pathway is indicated to be solely responsible
for the deglycosylation in BER enzymes, our results suggest an alternative Schiff base mechanism
which may be responsible for the repair of saturated pyrimidine damages.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2
ACS Paragon Plus Environment

Page 3 of 34
Journal of the American Chemical Society
INTRODUCTION
1
2
3
4
5
6
7
8
9
As the genetic information carrier, DNA is under constant environmental stresses, such as oxidizing
chemicals and UV irradiation. In cells under constant oxidative stress, the deoxyribonucleosides in the
genome can be oxidized to various products. Among the four deoxyribonucleosides, 2ʹꢀdeoxyguanosine
has the lowest redox potential, followed by 2ʹꢀdeoxyadenosine.1 2ʹꢀDeoxyguanosine oxidation leads to
the formations of 8ꢀoxoꢀ7,8ꢀdihydroꢀ2ʹꢀdeoxyguanosine (8ꢀoxoꢀdG) and formamidopyrimidine
deoxyguanosine (Fapy•dG) as the two common products.2 Besides guanine, other nucleobases can also
be oxidized.1 As a matter of fact, although 8ꢀoxoꢀdG is often used as a biomarker for oxidative stress, it
is usually not the most abundant 2ʹꢀdeoxyribonucleoside oxidation product. For instance, in neoplastic
human monocytes exposed to gamma rays, thymidine oxidation products, the four cis and trans
diastereomers of 5,6ꢀdihydroxyꢀ5,6ꢀdihydrothymidine (e.g. thymidine glycol), 5ꢀformylꢀ2ʹꢀdeoxyuridine
and 5ꢀ(hydroxymethyl)ꢀ2ʹꢀdeoxyuridine, were all revealed by a HPLCꢀMS/MS analysis to be more
abundant than 8ꢀoxoꢀdG.3
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The oxidized purine and pyrimidine residues are generally small; their presence usually does not
result in significant helical distortions. The primary repair mechanism for most of these oxidative
lesions is via the base excision repair (BER) pathway, although if a lesion is bulky enough to result in a
severe strand distortion, it can also be repaired by the nucleotide excision repair (NER) pathway.4ꢀ9 As
demonstrated by a recent seminal work by McKibbin et al.,10 further oxidation of 8ꢀoxoꢀdG produces
guanidinohydantoin (Gh) and spiroiminodihydantoin (Sp) lesions, which are prone for amine
modifications. The amineꢀmodified Sps, if bulky, can be efficiently repaired by the NER pathway; while
all Spꢀamine adducts can be repaired by the BER pathway irrespective of size.
A DNA lesion repair by the BER pathway is initiated by DNA glycosylases, which recognize and
remove a specific damaged base by hydrolyzing the Nꢀglycosidic bond to generate an abasic site (AP).9
This deglycosylation chemistry has been largely considered as the chemistry of the oxocarbenium ion,
e.g. rupture of the Nꢀglycosidic bond resulting in an accumulation of positive charge on O4ʹ at the sugar
3
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 4 of 34
ring during the transition state (Route A, Figure 1).4 A Schiff base mechanism (Route B, Figure 1) was
proposed together with the oxocarbenium ion pathway by Zoltewicz et al. in 1970 to offer possible
mechanistic explanation for the acidꢀcatalyzed hydrolysis of some purine nucleosides.11 Cadet and
Teoule observed the isomerization and anomerization of pyrimidic deoxyribonucleosides in 2 M HClO4
via formation of a possible Schiff base intermediate, supporting the mechanism B.12 However, careful
analyses of the acid catalyzed hydrolysis of 2ʹꢀdeoxycytidine and 2ʹꢀdeoxyuridine by Shapiro et al.
found that the reaction kinetics is consistent with a mechanism involving the formation of monoꢀ and
dications of the nucleoside, followed by rupture of the Nꢀglycosidic bond. This finding supports the
route A and rules out the involvement of the Schiff base intermediate.13 This oxocarbenium ion
chemistry shares a common themes with Oꢀglycoside (polysaccharide) hydrolyses and has been
considered as the paradigm for the deglycosylation process for more than forty years.4,8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1. Two possible deglycosylation mechanisms proposed to date.
Analysis of Shapiro’s data indicates that the conclusion was based on experiments using undamaged
deoxyribonucleotides. It is known that nucleobase modifications may reduce their chemical stability,
leading to novel reactions. Recently, our group has demonstrated that saturation of the thymine ring at
the 5ʹꢀend of the 5ꢀthyminylꢀ5,6ꢀdihydrothymine, e.g. the spore photoproduct (SP), activates the C4=O
moiety, supporting a water addition reaction to yield a hemiaminal intermediate.14 In contrast, such a
4
ACS Paragon Plus Environment

Page 5 of 34
Journal of the American Chemical Society
reaction is not observed with an intact thymidine residue.14 A similar reaction was observed with
another damaged pyrimidine: 5,6ꢀdihydroꢀ2′ꢀdeoxyuridine (dHdU),14 suggesting that this reaction may
be ascribed as a general property of a saturated pyrimidine residue. Oxidation, alkylation, and other
electrophilic reagents acting on nucleobases are known to reduce the electron density from the
heterocycle, making the modified nucleobases better leaving groups to form an AP.1 These observations
indicate that the deglycosylation reaction occurs more readily after certain base modifications. We
therefore wonder whether the reaction mechanism with modified nucleobases is truly the same as that
revealed previously with intact deoxyribonucleosides.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
We hereby used dHdU as the general model of a saturated pyrimidine residue to examine its chemical
stability and reactivity, with hope to gain possible mechanistic insights into the deglycosylation process.
It is worth mentioning that dHdU results from ionizing radiation damage to cytosine under anoxic
conditions.15,16 It is the third abundant pyrimidine damages in gamma irradiated genomic DNA, trailing
only by 5ꢀhydroxyꢀ5ꢀmethylhydantoin and 5ꢀhydroxyꢀ5,6ꢀdihydrouracil.17 It is highly mutagenic at both
replication and transcription levels.18,19 Therefore, the dHdU research by itself is also of great
significance. Our experimental results show that the Schiff base formation in route B becomes the
mechanism in the isomerization reaction of dHdU; this route is feasible to result in a subsequent
deglycosylation process. In contrast, in undamaged 2ʹꢀdeoxyuridine, formation of the traditional
oxocarbenium ion via route A is most likely responsible for the deglycosylation process. Similar
reactions were observed with 5,6ꢀhydrothymidine/thymidine residues. Further DFT calculation indicates
that formation of oxocarbenium ion is favorable if the pyrimidine ring remains intact; while Schiff base
is preferred once the C5=C6 bond is saturated. Therefore, both route A and route B are possible; their
involvements in the cleavage of the Nꢀglycosidic bond may be dependent on the nature of the
deoxyribonucleotide lesions. The Schiff base mechanism has been discussed in biꢀfunctional
glycosylases which possess both glycosylase and AP lyase activities,4,7,9 although it is not considered as
a major contributor to the deglycosylation process.4 Our report indicates that a saturated pyrimidine
lesion thermodynamically prefers the Schiff base pathway. The biꢀfunctional DNA glycosylases may
5
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 6 of 34
need to be reꢀexamined to reveal whether the Schiff base pathway is indeed responsible for enzyme
catalysis.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
6
ACS Paragon Plus Environment

Page 7 of 34
Journal of the American Chemical Society
METHODS
1
2
3
4
Materials and General Methods. All solvents and chemicals were of analytical grade and
purchased from SigmaꢀAldrich, Fisher, or VWR and used without further purification. NMR spectra
were obtained via a Bruker 500 MHz Fourier transform NMR spectrometer using deuterium oxide, d4ꢀ
methanol or d6ꢀDMSO as solvent with residual water/MeOH/DMSO acting as an internal standard.
Mass spectrometric (MS) analyses were obtained via electrospray ionization (ESI) employing an ionꢀ
trap mass analyzer. High resolution MS and Tandem mass spectrometry (MS/MS) analyses were
performed using a QꢀTOF LC/MS spectrometer. The MS data were acquired via the “Agilent
MassHunter Workstation Data Acquisition (B.03.00)” software and analyzed via “Qualitative Analysis
of MassHunter Acquisition Data (B.03.00)” software.
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Synthesis of dHdU. dHdU was synthesized via hydrogenation of the 2′ꢀdeoxyuridine as described
in our previous work.14
dHdU isomerization reaction in 0.1 M HCl. dHdU was dissolved in 200 µL 0.1 M HCl to a final
concentration of 85 mM. The resulting solution was maintained at room temperature for ~ 2 days. At
various times, 5 µL of the reaction solution was aliquotted out, immediately frozen in liquid N2, and
saved in −20 °C freezer for future HPLC analysis. Four major products 1, 2, 3, and 4 were generated
from this reaction.
dHdU isomerization reaction at 90 °C in a pH 7.4 buffer. In a separate set of experiments, dHdU
was dissolved in 200 µL pH 7.4 sodium phosphate buffer containing 150 mM NaCl to a final
concentration of 85 mM. The solution was heated to 90 °C using a PCR device with a cap heating
function to minimize water evaporation. At various reaction times, 5 µL solution was aliquotted out,
immediately frozen in liquid N2, and saved in −20 °C freezer for future HPLC analysis. Again, products
1, 2, 3, and 4 were observed during a 3ꢀday reaction.20
Determination of dHdU isomerization rate at 37 °C and pH 7.4. The dHdU solution in the pH
7
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 8 of 34
7.4 buffer as described above was heated to 90 °C, 85 °C, 80 °C, 75 °C and 70 °C respectively in the
PCR device. At various reaction times, 5 µL solution was aliquotted out, immediately frozen in liquid
N2, and stored in −20 °C freezer for HPLC analysis. The formation rates of 1, 2, 3, and 4 were
determined by plotting the HPLC peak integrations against time at a given temperature. The extent of
reaction was carefully controlled so that less than 10% of dHdU was consumed. Under such a situation,
the dHdU concentration change is small and can be neglected in the rate constant calculation. Moreover,
since formations of 2, 3, and 4 from dHdU are reversible, the small extent of reaction determines that
the impact of these reverse processes can also be neglected. The rate constants were therefore calculated
using the starting concentration of dHdU. Arrhenius plots for the dHdU reaction (lnk vs 1/T) were
subsequently constructed and the formation rates for these products at 37 °C and pH 7.4 were deduced.20
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Isomerization of 2′ꢀdeoxyuridine (dU) and thymidine (T) at pH 1. dU and T were dissolved
respectively in 200 µL 0.1 M HCl to a final concentration of 85 mM. The resulting solution was kept at
room temperature for ~ 3 days. At various times, 5 µL of the reaction solution was aliquotted out,
immediately frozen in liquid N2 and saved in −20 °C freezer for future HPLC analysis. No product was
observed in either reaction.
Isomerization of dU and T at 90 °C in a pH 7.4 buffer.
dU and
T
were dissolved
respectively in 200 µL pH 7.4 sodium phosphate buffer with 150 mM NaCl to a final concentration of
85 mM. The solution was heated to 90 ºC using a PCR device with a cap heating function to minimize
water evaporation. At various reaction times, 5 µL solution was aliquotted out, immediately frozen in
liquid N2, and saved in −20 °C freezer for future HPLC analysis. Uracil and thymine were observed in a
3ꢀday reaction, indicating the rupture of the Nꢀglycosidic bond. However, no isomers of dU or T were
observed.20
Preparations of dHdU isomerization products in a large scale.
HPLC analysis reveals that
the acid or heat treatment of dHdU results in four products, including three dHdU isomers 2, 3 and 4. To
8
ACS Paragon Plus Environment

Page 9 of 34
Journal of the American Chemical Society
characterize these products, the reaction was repeated at a bigger scale. Briefly, 200 mg dHdU was
dissolved in 0.1 M HCl and the reaction was allowed to proceed for 8 hours at ambient temperature. The
products were separated via semiꢀpreparative HPLC using procedures described below. The products
were collected, concentrated, and desalted by reinjection into HPLC using H2Oꢀacetonitrile as elution
buffers. The products were then characterized by MS/MS analysis and NMR spectroscopy.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Hydrogenation of the dHdU Schiff base intermediate. To further characterize the Schiff base
intermediate possibly formed during the dHdU isomerization reaction, we attempted to trap this
intermediate via a palladiumꢀcatalyzed hydrogenation reaction. Briefly, dHdU (50 mg, 0.22 mmol) was
dissolved in 0.1 M HCl in MeOH:H2O (1:1, v/v) solution to a final concentration of 5 mM (44 mL), 50
mg PdꢀC was then added to the solution. The resulting suspension was stirred under 1 atm H2 for 8 days
at ambient temperature. At various time points, 5 µl of the reaction mixture was extracted, and
immediately analyzed by HPLC until ~85% dHdU was consumed. After removing the PdꢀC by
filtration, the product was purified by semiꢀpreparative HPLC to yield the hydrogenated dHdU (5) as a
white solid (38 mg, 75%). Formation of the hydrogenated dHdU was confirmed by LCꢀMS/MS analysis
and NMR spectroscopy.
Schiff base intermediateꢀmediated nucleophilic addition reaction. To shed light on the
reactivity of the putative Schiff base intermediate formed in dHdU, we dissolved dHdU (20 mg, 85
ꢁmol) in an aqueous solution at pH 3 to a final concentration of 60 mM. To this solution, 75 fold of
cysteine also dissolved in pH 3 solution was added and the resulting solution allowed to react at 37 ºC.
At various times, 5 µl of the reaction mixture was extracted, and immediately analyzed by HPLC. Two
dHdUꢀcysteine adducts, 6 and 7, were observed, as indicated by the LCꢀMS/MS. However, isolation of
these products from the starting materials by semiꢀpreparative HPLC is problematic, making it difficult
for structural determination via NMR spectroscopy.
To facilitate product purification, we repeated the reaction using 75ꢀfold 3ꢀmercaptopropionic acid
(MPA) as a cysteine analog. MPA retains the thiol moiety as the potential nucleophile, but lacks the αꢀ
9
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 10 of 34
amino moiety in cysteine. Removal of the amino moiety drastically improves the HPLC
chromatography; moreover, it eliminates the possibility for the amino moiety to serve as a potential
competing nucleophile. The reaction solution was allowed to proceed at 37 ºC. At various times, 5 µl of
the reaction mixture was extracted, and immediately analyzed by HPLC. At the end of the 8ꢀday
reaction, the products were purified by semiꢀpreparative HPLC to afford the dHdUꢀMPA adducts 8 (10
mg, 30%) and 9 (18 mg, 56%) as colorless gelꢀlike compounds.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
HPLC Product Analyses. HPLC analyses were performed at room temperature using a Waters
(Milford, MA) HPLC system coupled to a 2489 UV/visible detector at 230 nm. An Agilent ZORBAX
BonusꢀRP column (5 µm particle size, 250 × 4.6 mm i.d.) was equilibrated with buffer A (10 mM diꢀNꢀ
butylammonium acetate or 10 mM ammonium acetate aqueous solution, pH 7.0). Compounds were
eluted with an ascending gradient of acetonitrile (buffer B) in 20 minutes at a flow rate of 1 mL/min.
The semiꢀpreparative HPLC was performed at room temperature with the same Waters HPLC setup.
An XBridgeTM OST C18 column (2.5 µm particle size, 50 × 10 mm i.d.) was equilibrated with buffer A
(10 mM ammonium acetate, pH 7.0), while compounds were eluted with an ascending gradient (0% ~
5%) of acetonitrile (buffer B) in 20 minutes at a flow rate of 4.73 mL/min.
Product Analyses via Tandem Mass Spectrometry (MS/MS). MS/MS analyses of the reaction
products were conducted using an Agilent 6520 Accurate Mass QꢀTOF LC/MS spectrometer. The
column was equilibrated in buffer A (5 mM ammonium acetate in 99% water and 1% acetonitrile, pH
6.5), while compounds were eluted with an ascending gradient (0% ~ 5%) of acetonitrile (buffer B) in
20 minutes at a flow rate of 0.5 mL/min. The mass signals were monitored under both positive and
negative ion modes.
DFT calculations on the deglycosylation reactions of dU/dHdU and T/dHT. All calculations
were conducted using the popular Becke’s threeꢀparameter hybrid functional B3LYP21ꢀ23 and the 6ꢀ
31++G(d,p) basis set,24,25 with the Gaussian09 package.26 Polarizable Continuum Model (PCM) was
10
ACS Paragon Plus Environment

Page 11 of 34
Journal of the American Chemical Society
used to mimic water environment.27ꢀ29 The Gibbs free energy changes (∆Gs) describing considered
reactions were the differences between the electronic energies of reactants and products, with entropy,
zeroꢀpoint vibrational, thermal and pV corrections. Thermodynamic terms were calculated at T = 298 K
and p = 1 atm, in the rigid rotorꢀharmonic oscillator approximation.30,31 ∆G for protonation of a
nucleoside X into XH+ was calculated as ∆G = (GXH+ + GH2O) – (GX + GH3O+), where each of given four
reactants was optimized separately in PCM aqueous environment. Such a procedure allows us to
calculate the thermodynamic effect of protonation, irrespective of the pH or reagents concentrations in
solution.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
We used C3ʹꢀendo conformation for deoxyribose ring and native anti conformation of nucleobase as
the starting geometries of dU/dHdU and T/dHT. No geometrical constraints were used during geometry
optimization and the analysis of harmonic frequencies demonstrated that all resulting structures were at
their energetic minima (all force constants were positive).
11
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 12 of 34
RESULTS
1
2
3
4
5
6
7
8
9
Formation of Four Major Products under Acidic Conditions. Incubation of dHdU in 0.1 M HCl
for 24 hrs at ambient temperature resulted in the formation of four products (1, 2, 3, and 4, Figure 2A).
ESIꢀMS analyses of these products (positive ion mode) revealed that 1 possesses an m/z value of 115.0,
corresponding to the [M + H]+ signal of the dihydrouracil. To confirm this observation, we coꢀinjected
commercially available 5,6ꢀdihydrouracil with 1. HPLC analysis confirms that these two compounds coꢀ
elute under the same peak. Therefore, one of the consequences for the dHdU acid treatment is the
rupture of the Nꢀglycosidic bond.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(2)
100%
(A)
1
(B)
2
80%
60%
dHdU
24 hr
(4)
3
4
dHdU
(3)
(1)
40%
20%
8 hr
0%
12 24 36 48 60 72
hr
4 hr
2 hr
(C)
117.06
99.04
231.10
0 hr
115.05
140
6
8
10
12
14
16
18
20
min
90
190
240
m/z
Figure 2. (A) HPLC chromatogram (monitored at 230 nm) of the dHdU isomerization reaction in the
presence of 0.1 M HCl at ambient temperature for 24 hrs. (B) The formation kinetics for the dHdU
isomerization reaction in 0.1 M HCl. It is obvious that formation of 3 is the kinetically most favorable
reaction, and the formation of 2 is the thermodynamically most favorable reaction. (C) MS/MS analyses
(positive ion mode, [M + H]+) of dHdU. Identical fragmentation patterns were observed for products 2,
12
ACS Paragon Plus Environment

Page 13 of 34
Journal of the American Chemical Society
3, and 4; the mass spectra and the chemical structures of the fragments are shown in the SI.
1
2
3
4
5
6
7
8
ESIꢀMS analysis of products 2, 3, and 4 reveals that they all exhibit a m/z value of 231.1, which is
the same as the [M + H]+ signal of dHdU. The unchanged mass indicates that these three species are
isomers of dHdU. As shown in Figure 2B, among the three products, 3 exhibits the fastest formation
rate, suggesting that its formation is kinetically favorable. As the reaction proceeds, signals
corresponding to 2 and 4 keep growing at the expense of dHdU and 3. After 24 hrs, 2 and 4 become the
major products. This observation suggests that formations of 2 and 4 are thermodynamically favorable.
Moreover, the reaction appears to reach equilibrium in 72 hrs; prolonged incubation at ambient
temperature in 0.1 M HCl does not lead to further change on the ratios among dHdU and its isomers.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Characterization of 2, 3 and 4 via MS/MS spectrometry. Formation of the deglycosylation
product 1 together with dHdU isomers 2, 3 and 4 indicates that the isomerization reaction may result
from activation of the Nꢀglycosidic bond. To shed light on the nature of this reaction, we first used
MS/MS to analyze the product structures. As shown in Figure 2C, the fragmentation of dHdU results in
three major peaks at m/z of 117.0, 115.0 and 99.0 Th respectively. The [M + H]+ signal at 115.0 Th
results from the release of 5,6ꢀdihydouracil moiety; while the [M + H]+ signal of 117.0 Th corresponds
to the resulting 2ʹꢀdeoxyribose moiety after loss of the dihydrouracil. Further elimination of the 3ʹꢀOH
moiety at the deoxyribose by loss of a H2O molecule may lead to the fragment of 99.0 Th. Structures of
these fragments are shown in Figure S8 in the supporting information. The fragmentation patterns of 2,
3 and 4 are indistinguishable from that of dHdU.20 The fact that all these compounds exhibit a fragment
with a [M + H]+ signal at 115.0 Th suggests that the dihydrouracil ring is unchanged; the isomerization
reaction likely occurs at the 2′ꢀdeoxyribose.
Characterization of 2, 3 and 4 via NMR spectroscopy.
The MS/MS analysis supports the
assumption that the dHdU isomerization reactions occur at the Nꢀglycosidic bond. To further reveal the
nature of these reactions, we determined the products’ structures by 1D and 2D NMR spectroscopy. We
13
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 14 of 34
were able to assign all H and 13C NMR signals for dHdU and its isomers, which provide a solid
foundation for structural determination. As shown in the HMBC spectra (Figure 3), H4′ interacts with
C1′ in dHdU and 3, suggesting that C1ʹ and C4ʹ are connected by the ether linkage as commonly found
in a 2ʹꢀdeoxyribofuranose. In contrast, in 2 and 4, interactions between H5′/H5″ and C1′ were observed,
indicating that C1ʹ and C5ʹ are linked. Therefore, 2 and 4 likely possess a 2ʹꢀdeoxyribopyranose.
1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3. HMBC spectra for dHdU and its isomers 2, 3 and 4. Both dHdU and 3 exhibit an interaction
between H4ʹ and C1ʹ, indicating that they possess a deoxyribofuranose. In contrast, new HMBC signals
were found in 2 and 4 for interactions between the two H5ʹ atoms and C1ʹ, suggesting that C5ʹ and C1ʹ
are connected by the ether bonds. Therefore, these two compounds likely possess a deoxyribopyranose.
These observations further indicate that the isomerization reaction occurs at the C1ʹ position. To
reveal the nature of the reaction, we compared the NOE spectra of dHdU and 3 (Figure 4). In dHdU,
both H2ʹ and H3ʹ were found to interact with H6 at the uracil ring, which is indicative of the naturally
14
ACS Paragon Plus Environment

Page 15 of 34
Journal of the American Chemical Society
occurring βꢀglycosidic bond. In contrast, both signals disappear in the NOE spectrum of 3. Instead, 3
exhibits a signal owing to the H1′ꢀH3′ interaction, suggesting that these two atoms are located on the
same side of the furanose. Therefore, the Nꢀglycosidic bond in 3 must adopt an αꢀconfiguration. In the
NOE spectrum of 2, interactions of H1′ꢀH3′ and H2″ꢀH6 were observed, indicating the presence of an
αꢀglycosidic bond. While in 4, H2′ꢀH4′ and H2′ꢀH6 were found to interact, which is consistent with the
existence of a βꢀglycosidic bond. Taken together, the NMR data suggest that the acid treatment of dHdU
produces two pairs of C1ʹ anomers possessing a pyranose and a furanose respectively (Scheme 1).
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 4. NOE spectra for dHdU and its isomers 2, 3 and 4. In dHdU and 4, H2ʹ was found to interact
with H6 at the uracil ring, indicating that both compounds contain a βꢀglycosidic bond. The additional
interactions between H3ʹꢀH6 in dHdU and between H2′ꢀH4′ in 4 further support this assignment. The
H1′ꢀH3′ interaction in 2 and 3 indicates an αꢀglycosidic bond in these complexes.
15
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 16 of 34
1
2
3
Scheme 1
4
5
6
7
8
9
O
O
N
O
OH
OH
N
O
O
NH
HN
H+
NH
HO
O
N+
HO
O
OH
O
HO
O
O
N
N
O
O
OH
+
(2)
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
OH
OH
NH
NH
Schiff base (Sb)
dHdU
O
+
O
H2O
(3)
H2
Pd-C
OH
OH
O
N
(4)
H+
O
HN
O
HN
O
HO
HO
HO
HN
OH
O
OH
OH
O
N
HO
+
(5)
OH
O
O
N
H
OH
OH
OH
OH
abasic site
OH
(1)
Formations of 2, 3 and 4 can be best rationalized by the generation of a Schiff base intermediate
(Sb) at the Nꢀglycosidic bond (Scheme 1). Both the 4ʹꢀOH and the 5ʹꢀOH moieties can undergo
nucleophilic addition to the C1ʹ position of Sb during the ring reꢀclosure process. Moreover, such
addition reactions can happen from either side of the C=N+ bond, resulting in two pairs of C1ʹ epimers
in dHdU/3 and 2/4. Formation of 3 containing an αꢀfuranose requires relatively small conformational
changes from dHdU. Moreover, 5ꢀmembered ring is considered as the most kinetically favorable
configuration in ring closure reactions owing to the lowest entropic cost and transition state strain
energy.32,33 Therefore, it is not surprising that 3 exhibits the fastest formation kinetics among the three
isomers (Figure 2B). On the other hand, 6ꢀmembered pyranose form generally dominates in aqueous
solution over the five membered furanose form due to the enhanced thermal stability,34 explaining the
higher yields for 2 and 4 after a prolonged incubation. Between the two pyranose anomers, the αꢀ
anomer is generally more stable than the β
ꢀform due to the soꢀcalled anomeric effect,35 explaining the
higher yield of 2 than its epimer 4.
The presence of a possible Schiff base intermediate Sb also offers an explanation for the generation
16
ACS Paragon Plus Environment

Page 17 of 34
Journal of the American Chemical Society
of the deglycosylation product 1. In Sb, the C1′ position is electron deficient, and thus prone for the
nucleophilic attack by a water molecule. The resulting water adduct is not stable; it readily undergoes an
elimination reaction, releasing the dihydrouracil and generating an abasic site. Such a mechanism
resembles the reaction route B in Figure 1, indicating that besides the traditional oxocarbenium ion
pathway, the Schiff base mechanism may also be involved in the deglycosylation process under certain
circumstances.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
To trap the Schiff base intermediate via hydrogenation.
To further characterize this Schiff
base formation reaction, we attempted to trap the Sb intermediate. NaBH4 is a common reagent used to
trap Schiff base intermediates.36ꢀ40 The reaction can be conducted in an aqueous buffer at the neutral pH.
It is usually finished within an hour; decomposition of NaBH4 is thus not a severe problem. In contrast,
the dHdU isomerization requires a strong acidic condition and it takes several days for the isomerization
reaction to reach completion. Under such a reaction condition, the NaBH4 readily decomposes by
reacting with H+, releasing H2. Thus, it is unsuitable to be used as a Sb trapping reagent.
O
HN
O
OH
N
HO
(5)
OH
dHdU
(3)
8 d
4 d
2 d
(1)
(2)
(4)
3
6
9
12
min
Figure 5. (A) HPLC chromatograms (monitored at 230 nm) of the dHdU hydrogenation in 0.1 M HCl
for a reaction time of up to 8 days. The dHdU isomerization and deglycosylation reactions still occur
under this acidic environment, as reflected by the formations of products 1−4. The Sb hydrogenation
product (5) is already the dominant species after a 2ꢀday reaction. 5 continues to increase along with the
deglycosylation product 1; while the amounts of dHdU and its isomers 2, 3 and 4 keep decreasing
17
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 18 of 34
during the 8ꢀd reaction. This observation is consistent with the mechanism shown in Scheme 1: dHdU
and its isomers are interꢀconvertible through the Sb intermediate; while the deglycosylation and Sb
hydrogenation reactions yielding 1 and 5 respectively are irreversible processes.
1
2
3
4
5
6
7
8
9
After surveying a number of conditions, we found that 10% Pd/C can catalyze the hydrogenation of
Sb at pH 1 and room temperature, yielding compound 5 in ~ 85% yield after a 8ꢀday reaction (Scheme 1
and Figure 5). ESIꢀMS analysis reveals that 5 exhibits a [M + H]+ signal at 233.1 Th, which is + 2 Th
comparing with that from dHdU.20 5 was isolated by HPLC and characterized by NMR spectroscopy. In
the 1H NMR spectrum taken in MeOD, twelve carbonꢀbased protons were found to be associated with 5,
contrasting to the eleven protons on dHdU. If the spectrum is taken in d6ꢀDMSO to enable the
observation of exchangeable protons, three protons owing to 3ʹ, 4ʹꢀ and 5ʹꢀOH were found in 5.20 All
these spectroscopic data support the chemical structure shown in Figure 5.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
dHdU isomerization reaction at physiological pH.
The results above support the Sb formation
upon acid treatment. Such an acidic condition however does not exist in cells. As shown by Scheme 1,
Sb formation is driven by the protonation of the O4ʹ moiety. The O4ʹ on 2ʹꢀdeoxyribose is a weak base,
whose protonation requires a strong acidic environment. In vivo, such a condition may be fulfilled by
electrostatic and/or Hꢀbonding interactions with amino acid residues on a glycosylase protein.4,7ꢀ9 As a
hydrogen bond may provide 1.8 ~ 3 kcal/mole stablization energy,41,42 and the average energy
contribution from a salt bridge is ~ 3.0 kcal/mole,43 stabilization of the Sb intermediate can be readily
achieved by these interactions within an enzyme. Moreover, in uracil DNA glycosylase, a theoretical
computation indicates that the oxocarbenium ion intermediate is stabilized by the anionic phosphate
groups from four nucleotides on the DNA substrate by 21.9 kcal molꢀ1;44 experimental data suggest that
the phosphodiester interactions with the substrate contribute 6ꢀ8 kcal/mole toward lowering the
activation barrier.45 It is reasonable to assume that the polyanionic DNA backbone may also provide
significant stabilization of the Sb intermediate if it is indeed utilized by DNA glycosylases for catalysis.
Such stabilization networks do not exist in a nonꢀprotein environment. As the consequence, the Sb
18
ACS Paragon Plus Environment

Page 19 of 34
Journal of the American Chemical Society
population is likely to be very small at physiological pH. To shed light on whether the Sb formation can
be achieved at the neutral pH, we accelerated the reaction by heating dHdU dissolved in pH 7.4
phosphate buffer to 90 °C. All three dHdU isomers as well as the cleaved dihydrouracil (1) were
observed.20 Although the reaction yields were lower than those observed at pH 1 and the reaction may
not reach completion in 3 days, the data unambigiously prove that the Sb formation can ocurr at
physicological pH. Via Arrhenius plots, the respective rate constants for the formations of 1, 2, 3 and 4
were calculated to be 0.82, 7.0, 54, and 1.9 × 10−9 s−1 at 37 °C. Again, the αꢀfuranose product 3 is
clearly the kinetically most favorable product.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Thiol addition to the C1ʹ position due to the Sb formation. As rationalized above, the Schiff
base intermediate may be stabilized by the protein network. The C1ʹ position of Sb is electron difficient,
which may support nucleophilic addition reactions by amino acids such as lysine and cysteine.
However, if not located on a protein, the εꢀNH2 moiety in lysine is protonated at pH 7 and the resulting
cationic form is a very poor nucleophile. Moreover, even if the lysine εꢀNH2 group does add to the C1′
position, without the protection from the protein network, the resulting Schiff base is not stable. It is
readily hydrolyzed to produce an abasic site and restore the lysine amino moiety. Such a reaction is
indistinguishable from an oxocarbenium ion mediated deglycosylation reaction and is thus of little
mechanistic significance. In contrast, a thiol group is more difficult to be protonated; the –SH can
function as a reasonably good nucleophile in a slightly acidic environment. More importantly, should
the Schiff base intermediate be invovled, the −SH moiety would be readily added to the C1′ position.
The resulting tetrahedral species is expected to be stable enough to survive the acidic environment
needed for the reaction, enabling a thorough product characterization to shed light on the reaction
mechanism.
We therefore incubated 75ꢀfold cysteine with 60 mM dHdU at pH 3. The acidic condition was
chosen to faciliate the Sb formation. At the same time, the acidity is not too strong to protonate the –SH
moiety, thus enabling the possible trapping of the Schiff base intermediate. Analysis by HPLC reveals
19
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 20 of 34
the gradual increase of two products 6 and 7 in a 3ꢀday reaction. ESIꢀMS analysis under the negative ion
mode reveals that both products possess a molecular mass of 350.1 Th, indicating that they are dHdUꢀ
cysteine adducts. MS/MS analysis reveals three major fragments with m/z of 105.0, 149.0 and 229.1 Th
respectively for both products.20 The 149.0 Th signal can be best explained as the ringꢀopened ribose
fragment with an sulfide attached to the C1′ position.20 The fragmentation patterns are identical for 6
and 7, suggesting that they are likely epimers.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
MPA
(9)
8 d
(8)
4 d
(3)(4)
1 d
dHdU
12 hr
0
(2)
5
(1)
10
min
20
15
Figure 6. HPLC chromatograms (monitored at 230 nm) describing the formation of the dHdUꢀMPA
adducts for a reaction of up to 8 days. The dHdU isomerization and deglycosylation reactions still occur
under this acidic environment, as reflected by the formations of products 1−4. The dHdUꢀMPA adducts
(8 and 9) are already the major species after 12 hours; their yields continue to increase during the 8ꢀd
reaction at the expense of dHdU and its isomers. Please note that 8 and 9 differ at the chiral C1′
position; however we were unable to assign their absolute chiral configuration.
In the HPLC chromatogram, product 6 elutes immediately after the cysteine species, making it
difficult to be purified by HPLC.20 Moreover, although it is unlikely for the αꢀNH2 group of cysteine to
react given its high tendency to protonate resulting in the ammonium form under the strong acidic
20
ACS Paragon Plus Environment

Page 21 of 34
Journal of the American Chemical Society
condition, we would like to firmly exclude its involvement in the reaction. We therefore repeated the
reaction using excess 3ꢀmercaptopropionic acid (MPA). Ionization of MPA in water alters the pH to ~ 3,
facilitating the Schiff base formation. As shown in Figure 6, this acidic condition induces formations of
Sb and dHdU isomers. As the reaction proceeded, dHdU and its isomers became less and less, which
was accompanied by the formation of two new products 8 and 9.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In the 1H NMR spectrum of 8 and 9, all 1H signals from dHdU and MPA remain, which is consistent
with the hypothesis that they are dHdUꢀMPA adducts. We are unable to determine the absolute stereoꢀ
configuration for these two epimers via NOE spectroscopy due to the free rotation of the C1′ꢀC2′ bond.
This observation also agrees with the fact that the deoxyribofuranose ring at dHdU is opened. On the
HMBC spectrum, the hydrogens at the γꢀcarbon of the MPA moiety exhibits a strong coupling with C1′,
indicating that the sulfur from MPA is directly attached to the C1ʹ position.20 Additionally, MS/MS
analysis of 8 and 9 again exhibits three fragments with m/z of 105.0, 149.0 and 229.1 Th, which are
same with those observed in the MS/MS analysis of dHdUꢀcysteine adducts 6 and 7. Such an identical
fragmentation pattern suggests that all thiolꢀadducts possess very similar structures. Taken together, all
the spectroscopic data support our hypothesis that formation of Sb makes the C1ʹ position prone for
nucleophilic addition reactions.
dHT isomerization reaction To ensure that the Schiff base formation is not a unique property of
dHdU, we prepared 5Sꢀdihydrothymidine (dHT) via hydrogenation of thymidine using a literature
protocol.43 We did not use 2ʹꢀdeoxycytidine as the saturated C residue readily deaminates to dHdU.46ꢀ48
This deamination reaction is drastically accelerated under an acidic or heated condition, making the
saturated 2ʹꢀdeoxycytidine unsuitable for the Sb formation studies. Upon treatment with 0.1 M HCl,
three 5SꢀdHT isomers and the corresponding 5Sꢀdihydrothymine residue were indeed generated, as
indicated by the LCꢀMS/MS analysis.20 The reaction closely resembles that obtained with dHdU,
suggesting that the Sb formation is likely a general property of a saturated pyrimidine residue.
21
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 22 of 34
No isomerization was observed with dU or T To further confirm that the Sb formation is due to
the saturated C5=C6 bond, we treated thymidine and 2ʹꢀdeoxyuridine with 0.1 M HCl for 3 days. HPLC
analysis confirms that no reaction occurs under such a condition.20 Heating the thymidine or the 2ʹꢀ
deoxyuridine solution in a pH 7.4 phosphate buffer to 90 ºC for 3 days leads to the formation of thymine
and uracil respectively; however, no corresponding thymidine or 2ʹꢀdeoxyuridine isomers were
observed.20 These observations agree with the previous deglycosylation studies by Shapiro et al. using
2′ꢀdeoxyuridine and 2′ꢀdeoxycytidine,13 indicating that thymidine and 2ʹꢀdeoxyuridine deglycosylate via
the traditional oxocarbenium ion pathway as shown in Figure 1A.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
DFT calculations to understand the Sb formation
Our observations thus suggest that both the
Schiff base and the oxocarbenium ion intermediates shown in Figure 1 can be involved in the
deglycosylation reactions. The oxocarbenium ion intermediate formation may be the dominant pathway
if the aromaticity of the nucleobase is maintained; while the Schiff base route may be involved if the
pyrimidine C5=C6 bond is saturated. To shed light on the possible reaction mechanism with different
pyrimidine damages, we performed DFT calculations to compare the Gibbs free energy changes of these
two reaction pathways.
Scheme 2
The deglycosylation process was suggested to be an acidꢀcatalyzed reaction under physiologically
22
ACS Paragon Plus Environment

Page 23 of 34
Journal of the American Chemical Society
relevant conditions.49,50 Indeed, our calculation found that the crucial step for this process is the
protonation of the reacting nucleosides. If a proton is attached to a “right” position, the system energy is
lower and bond cleavage may subsequently occur, leading to a corresponding product. We considered
protonation at atoms with the lowest calculated Mulliken charges, namely at O4ʹ of the deoxyribose, as
well as at N1, O2, N3, and O4 of the pyrimidine ring. Protonation at the last three positions (O2, N3,
and O4) is omitted in the discussion as these atoms are mainly involved in hydrogen bonding
interactions in duplex DNA and do not directly contribute to the deglycosylation process. Protonation at
O4ʹ and N1 results in putative (O4ʹ)H, (N1ꢀR)H, and (N1ꢀS)H intermediates respectively, as illustrated
using dHdU in Scheme 2. We then compared the stability of these intermediates to reveal their tendency
to further decay into Schiff base or oxocarbenium ion.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 7. (A) Gibbs free energy changes (kcal/mol) for dHdU protonation and decomposition.
Comparing with the general protonated dHdU state, the (O4′)H pathway which leads to the formation of
Schiff base species lowers the free enthalpy by 7~ 8 kcal/mol more than the two putative N1 protonation
pathways (N1ꢀR)H and (N1ꢀS)H. Therefore, in dHdU which contains a saturated C5=C6 bond, the
Schiff base pathway is preferred over the traditional oxocarbenium ion pathway. (B) Geometry
visualization of the optimized dHdU structure. After O4ʹ protonation, the resulting dHdUꢀ(O4ʹ)H is
unstable, which transforms into the Schiff base via a barrierꢀfree C1ʹꢀO4ʹ bond rupture during structure
optimization. (C) Gibbs free energy changes (kcal/mol) for dU protonation and decomposition.
Protonation at N1 produces either (N1ꢀR)H or (N1ꢀS)H, which decomposes into an oxocarbenium ion
23
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 24 of 34
via a barrierꢀfree C1ʹꢀN1 breakage. Both N1ꢀH species are lower in Gibbs free energy than the putative
(O4′)H, suggesting that in dU where the ring aromaticity is maintained, the oxocarbenium ion pathway
is heavily favored. (D) Geometry visualization of the optimized dU structure. dU protonated at N1
readily transforms into the oxocarbenium ion via a barrierꢀfree C1ʹꢀN1 bond rupture. The newly
attached protons in the resulting Schiff base (B) and oxocarbenium ion (D) are indicated by green
circles.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
As expected, the DFT calculations show that for a saturated pyrimidine nucleoside dHdU, protonation
at the O4ʹ position of 2ʹꢀdeoxyribose is preferred. The dHdU molecule with O4ʹ being protonated is
unstable; it reꢀarranges into the Schiff base (Sb) during structural optimization via a barrierꢀfree C1ʹꢀO4ʹ
bond rupture. As we cannot accurately obtain the optimized (O4ʹ)H structure during our calculation, we
made a reasonable hypothesis, using that of Sb to estimate the Gibbs free energy change for the
formation of (O4ʹ)H. Such an approach allows us to obtain a Gibbs free energy change of ꢀ12.93
kcal/mol from the protonated dHdU state for the (O4ʹ)H formation (Figure 7A). In contrast, protonation
at the N1ꢀR and N1ꢀS position only lower the Gibbs free energy by ꢀ5.73 and ꢀ4.81 kcal/mol
respectively. Therefore, the O4ʹ protonation is clearly the favorable pathway, indicating that in dHdU,
the Schiff base formation is the preferred route (Figure 7B).
In contrast, in an undamaged nucleoside such as dU, protonation at the proꢀR or proꢀS position of N1
is clearly favored. Again, the (N1ꢀR)H and (N1ꢀS)H intermediates were found to be unstable; the C1ʹꢀ
N1 bond ruptures during structure optimization, leading to the formation of corresponding
oxocarbenium ion species. The resulting oxocarbenium ions still maintain some conformational feature
of their parent N1ꢀH species, therefore possessing slightly different Gibbs free energy. Again, we used
these oxocarbenium ions to estimate the freeꢀenthalpy change of their parent N1ꢀH species and found
that the formations of (N1ꢀR)H and (N1ꢀS)H lead to Gibbs free energy changes of ꢀ19.17 and ꢀ18.63
kcal/mol respectively from the protonated dU state (Figure 7C). In contrast, the O4ʹ protonation is
clearly unfavorable in dU, which only lowers the Gibbs free energy by ꢀ3.50 kcal/mol. Thus, in dU, the
N1 protonation is the preferred pathway for deglycosylation, leading to the oxocarbenium ion formation
(Figure 7D)
24
ACS Paragon Plus Environment

Page 25 of 34
Journal of the American Chemical Society
Similar trends are observed with the dHT and T deglycosylation processes respectively.20 The Schiff
base pathway is favored by 8.4 ~ 9.5 kcal/mol for dHT (Figure S13). For T, protonation at the N1ꢀR
position is favored; the resulting (N1ꢀR)H subsequently transforms to the oxocarbenium ion via a
barrierꢀfree C1′ꢀN1 bond rupture. In contrast, protonation at the N1ꢀS position is a uphill process
probably due to the steric hindrance; the formed putative (N1ꢀS)H intermediate however is found to be
metastable as a local minimum on the potential energy surface, which can be transformed into the
oxocarbenium ion due to a favorable negative thermodynamic stimulus (~ – 27 kcal/mol, Gibbs free
energy).20 Despite this overall favorable thermodynamics, given the high energy barrier in the (N1ꢀS)H
pathway, we believe that the barrierꢀfree (N1ꢀR)H pathway is responsible for the oxocarbenium ion
formation in T. Although such a stereoꢀselection may not be meaningful in our studies here, it could
have biological indication if that happens in the binding pocket of a glycosylase as only one side of the
pyrimidine ring may be favorable for the proton addition step to trigger the oxocarbenium ion formation.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
25
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 26 of 34
DISCUSSION
1
2
3
4
5
6
7
8
9
Understanding details of the chemical reactivities of damaged DNA nucleobases is important in
efforts aimed at lesion characterization, the chemical syntheses, the elucidation of possible biological
repair routes, or the consequences of their persistence in a genome. The examples discussed here, dHdU
and dHT, represent lesions resulting from ionizing radiation damage to C11,12 and T51 respectively. In
cells, such lesions are removed by DNA glycosylases to initiate the damage repair process by the BER
pathway.4,7,9 The current paradigm governing this removal step is that the damaged base is detached by
rupture of the N1ꢀC1ʹ bond, leaving an oxocarbenium ion intermediate at the deoxyribose (Route A,
Figure 1). Such an intermediate is commonly observed in the hydrolysis of glycans.52ꢀ55 Moreover, using
undamaged 2ʹꢀdeoxyuridine and 2ʹꢀdeoxycytidine, clear evidence was obtained that the Nꢀglycosidic
bond hydrolysis is mediated by the oxocarbenium ion pathway.13 Although there were data supporting
the involvement of a Schiff base intermediate resulting from the rupture of C1ʹꢀO4ʹ bond (Route B,
Figure 1),11,12 such a route is not considered to contribute to the deglycosylation process.4
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Our report here, however, shows that in a saturated pyrimidine lesion such as dHdU and dHT,
formation of the oxocarbenium ion cannot be observed; the reaction instead proceeds via the Schiff base
pathway. In contrast, the oxocarbenium ion pathway is involved if the aromaticity at the pyrimidine ring
is maintained, which is consistent with the previous 2ʹꢀdeoxyuridine and 2ʹꢀdeoxycytidine hydrolysis
studies.13 Our finding highlights the importance to maintain the integrity of the nucleobase ring. As
shown in our previous work, loss of aromaticity in dHdU facilitates the formation of a hemiaminal
intermediate at the C4 position;14 such an intermediate cannot be observed in the undamaged thymine
residue. Now we show that the C5=C6 saturation also surprisingly alters the stability of the remote 2ʹꢀ
deoxyribose, making the O4ʹ more readily to be protonated and promoting the Schiff base intermediate
formation at the Nꢀglycosidic bond. This result emphasizes the complicated impacts of a nucleobase
modification to the overall reactivity of the nucleoside. Although the oxocarbenium ion intermediate is
well established in the glycan chemistry and its formation was observed during the hydrolysis of intact
26
ACS Paragon Plus Environment

Page 27 of 34
Journal of the American Chemical Society
deoxyribonucleosides, the conclusion may still be misleading when applied to the studies of damaged
nucleosides.
1
2
3
4
5
6
7
8
It is worth mentioning that acidꢀcatalyzed ribose ring expansion/anomerization reactions were
observed previously in saturated pyrimidine ribonucleosides or ribonucleoside analogs.56ꢀ60 For instance,
at pH 1, the βꢀfuranose in tetrahydrouridine (THU) was found to quickly expand to a βꢀpyranose via a
similar Schiff base intermediate; the ring expansion reaction, however, was not observed with uridine
and dihydrouridine (dHU).56 These observations partially agree with our dHdU reaction reported here,
although it is puzzling that THU reaction only afforded the βꢀpyranose isomer, in contrast to the three
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
isomers found in our dHdU reaction. Also, it is surprising that dHU did not isomerize. Spontaneous αꢀβ
anomerization was observed with the βꢀfuranose attached to the reduced pyridine ring in nicotinamide
adenine dinucleotide (NADH), but not with that in NAD+ where the pyridine ring maintains its
aromaticity.57,58 These previous observations are largely consistent with our dHdU and dHT reactions,
supporting our conclusion that once the pyrimidine ring aromaticity is lost, the attached (2ʹꢀdeoxy)ribose
is prone for the C1ʹ epimerization reaction via the Schiff base mechanism.
The Schiff base intermediate may also be observed with some purine damages. C1ʹ epimerization
was observed in the synthesis of the phosphoramidites of Fapy•dG (Nꢀ(2ʹꢀdeoxyꢀα,βꢀDꢀ
erythropentofuranosyl)ꢀNꢀ(2,6ꢀdiaminoꢀ4ꢀhydroxyꢀ5ꢀformamidopyrimidine)) and Fapy•dA (N4ꢀ(2ʹꢀ
deoxyꢀα,βꢀDꢀerythropentofuranosyl)ꢀ4,6ꢀdiaminoꢀ5ꢀformamidopyrimidine),61ꢀ65 and posed a significant
challenge for their incorporation into an oligonucleotide via solid phase synthesis. Although a Schiff
base intermediate was proposed to explain the C1ʹ epimerization and facile isomerization reactions,
previous discussions were mainly focused on the interference to the synthetic efforts for Fapy•dG or
Fapy•dA incorporation. Given our data reported here, it is very likely that a similar Schiff base
intermediate may exist in a naturally occurring Fapy•dG or Fapy•dA lesion formed in the genomic
DNA. After facile protonation of the N7ꢀposition, a purine residue can be an excellent leaving group
favoring the oxocarbenium ion formation.4,66 It is unknown whether the Schiff base pathway is involved
27
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 28 of 34
in the depurination process in purine damages other than Fapy•dG and Fapy•dA.
1
2
3
4
DNA-O
O
OH
O-DNA
5
6
AP
7
8
9
mono-functional BER
bi-functional BER
(A)
N
N
DNA-O
DNA-O
DNA-O
H
O+
O
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
O-DNA
DNA-O
O-DNA
OH
O
-elimination
(B)
N
DNA-O
OH
OH
/
elimination
OH
N-enz
O-DNA
O-DNA
NH2-lysine
O
Figure 8. (A) Current mechanism for BER enzymes, which suggests that both monoꢀfunctional and biꢀ
functional BER enzymes utilize the same oxocarbenium ion intermediate. In monoꢀfunctional BER
enzymes, such an oxocarbenium ion formation leads to the direct removal of the damaged nucleobase.
In biꢀfunctional enzymes, the oxocarbenium ion intermediate subsequently reacts with an activeꢀsite
lysine residue, resulting in a nucleoprotein complex connected by a Schiff base intermediate, which
subsequently induces the βꢀelimination reaction and in some cases the β/δꢀelimination reactions to
cleave the DNA strand. (B) An alternative pathway for the biꢀfunctional BER enzymes based on our
Schiff base intermediate formation in this report. The Schiff base in the damaged nucleoside is attacked
by an activeꢀsite lysine, resulting in the nucleoprotein intermediate.
Our data indicate that in a saturated pyrimidine residue, the Schiff base mechanism is favorable over
the oxocarbenium ion pathway. Therefore, at least in the context of free nucleoside, the oxocarbenium
ion may not be responsible for all deglycosylation processes. We wonder whether such a Schiff base
intermediate may be utilized by DNA glycosylases in the BER pathway. BER enzymes can be divided
into two categories: monoꢀfunctional BER enzymes which are DNA glycosylases that hydrolyze the Nꢀ
glycosidic bond to yield an AP, and biꢀfunctional BER enzymes which possess both glycosylase and AP
lyase activities (Figure 8).4 The oxocarbenium ion intermediate has long been suggested in mediating
the deglycosylation process in monoꢀfunctional BER enzymes. Such a hypothesis was strongly
supported by the kinetic isotope effect studies of the reaction catalyzed by uracil DNA glycosylase,
which allowed the elucidation of the transition state configuration, suggesting that the reaction proceeds
through a dissociative transition state, with complete dissociation of the uracil anion followed by
28
ACS Paragon Plus Environment

Page 29 of 34
Journal of the American Chemical Society
addition of water to the resulting oxocarbenium ion intermediate.67
1
2
3
4
5
6
7
8
9
In biꢀfunctional enzymes, the εꢀNH2 of a lysine or the –NH of a proline is suggested to be added to
the C1ʹ of the oxocarbenium ion intermediate, forming a Schiff base intermediate and removing the
damaged nucleobase (Route A, Figure 8).7,9 This Schiff base intermediate subsequently induces a βꢀ
elimination reaction, creating a nick at the 3ʹꢀside of the lesion. Some enzymes can even catalyze an
additional δꢀelimination, cleaving the 5ʹꢀend and generating a gap. In several biꢀfunctional BER
enzymes such as endonuclease III, multiple groups proposed that a Schiff base intermediate may be
formed first at the Nꢀglycosidic bond of the damaged base (Route B, Figure 8) before it reacts with an
active site lysine.9,37,38,68,69 Such a proposal however has not been accepted as the general mechanism
because there is little experimental evidence demonstrating the existence of the Schiff base intermediate
in DNA lesions other than Fapy•dG and Fapy•dA.4 Our dHdU studies here indicate that at least in a
saturated pyrimidine lesion like dHdU and dHT, the Schiff base intermediate is thermodynamically
preferred over the oxocarbenium ion. Such an intermediate readily supports the lysine or proline
mediated substitution reaction, generating the nucleoprotein Schiff base intermediate shown in Route B.
To establish this intriguing hypothesis, more DNA lesions need to be examined. As
aforementioned, the purine N7 can be readily protonated, making it a good leaving group under the
oxocarbenium ion pathway. It is known the 8ꢀoxoꢀdG is repaired by biꢀfunctional BER enzymes. If the
above hypothesis is correct, 8ꢀoxoꢀdG should support the Schiff base chemistry in the nucleoside
context. Moreover, the involvement of the Schiff base intermediate in glycosylase catalysis may need to
be checked by the transition state analysis using kinetic isotope effects similar to that conducted in
uridine DNA deglycosylase.67 To date, the transition state analyses were only performed in monoꢀ
functional BER enzymes.4 Such an analysis should be conducted with biꢀfunctional BER enzymes to
obtain direct evidence on the nature of the reaction intermediate.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The dHdU isomerization results in two pairs of C1ʹ epimers, possessing a furanose and a pyranose
respectively. Such a reaction is expected to occur in the oligonucleotide context as well. However, in an
29
ACS Paragon Plus Environment