Recognition of Artificial Nucleobases by E. coli Purine Nucleoside Phosphorylase versus its Ser90Ala Mutant in the Synthesis of Base-Modified Nucleosides

DOI: 10.1002/chem.201501334

Full Paper

&

Enzymes

Recognition of Artificial Nucleobases by E. coli Purine Nucleoside

Phosphorylase versus its Ser90Ala Mutant in the Synthesis of

Base-Modified Nucleosides

Ilja V. Fateev,^[a]Maria I. Kharitonova,^[a]Konstantin V. Antonov,^[a]Irina D. Konstantinova,^[a]

Vasily N. Stepanenko,^[a]Roman S. Esipov,^[a]Frank Seela,^[b]Kartik W. Temburnikar,^[c]

Katherine L. Seley-Radtke,^[c]Vladimir A. Stepchenko,^[d]Yuri A. Sokolov,^[d]

Anatoly I. Miroshnikov,^[a]and Igor A. Mikhailopulo*^[d]

Abstract: A wide range of natural purine analogues was

used as probe to assess the mechanism of recognition by

the wild-type (WT) E. coli purine nucleoside phosphorylase

(PNP) versus its Ser90Ala mutant. The results were analyzed

from viewpoint of the role of the Ser90 residue and the

structural features of the bases. It was found that the Ser90

residue of the PNP 1) plays an important role in the binding

and activation of 8-aza-7-deazapurines in the synthesis of

their nucleosides, 2) participates in the binding of a-d-pen-

tofuranose-1-phosphates at the catalytic site of the PNP, and

3) catalyzes the dephosphorylation of intermediary formed

2-deoxy-a-d-ribofuranose-1-phosphate in the trans-2-deoxy-

ribosylation reaction. 5-Aza-7-deazaguanine manifested ex-

cellent substrate activity for both enzymes, 8-amino-7-thia-

guanine and 2-aminobenzothiazole showed no substrate ac-

tivity for both enzymes. On the contrary, the 2-amino deriva-

tives of benzimidazole and benzoxazole are substrates and

are converted into the N1- and unusual N2-glycosides, re-

spectively. 9-Deaza-5-iodoxanthine showed moderate inhibi-

tory activity of the WT E. coli PNP, whereas 9-deazaxanthine

and its 2’-deoxyriboside are weak inhibitors.

Introduction

action mixture in a 25% yield. A careful analysis of the pub-

lished crystallographic data of the catalytic site of the E. coli

PNP allowed us to hypothesize that the key role of the side

chain of Ser90-OgH involves the initial recognition followed by

the formation of the glycosyl bond in the synthesis of 8-aza-7-

deazapurine nucleosides.^[1]In that regard, the close proximity

of the Ser90 hydroxyl group to the imidazole ring of the ade-

nine bound at the catalytic site of the E. coli PNP containing

also a-d-ribofuranose-1-phosphate (Rib-1P)^[2,3]was a clue for

such suggestion. Note that this proximity may be responsible

for the absence of substrate activity of the C8-substituted pu-

rines (except for 8-aminoadenine, see below) for the E. coli

PNP. Ab initio analyses of the possible tautomeric structures of

8-aza-7-deazapurine in a complex with the Ser90 hydroxyl

group showed that the spatial arrangements of the N9-amino

sp³-hybridized (=N8-N9(H)-C4, purine numbering throughout

this paper) tautomer and the sp²-hybridized N9-imino (-(H)N8-

N9=C4) tautomer are compatible with the suggested equilibri-

um implying the nucleophilic attack of the N9-imino nitrogen

atom on the electrophilic C1 atom of Rib-1P in the synthesis

reaction.^[1,4]

Recently, we have disclosed satisfactory substrate properties

for a number of representatives of the pyrazolo[3,4-d]pyrimi-

dine (8-aza-7-deazapurine) scaffolds 1–4 for the recombinant

E. coli purine nucleoside phosphorylase (PNP) (Figure 1).^[1]The

most surprising finding was the observation of substrate activi-

ty of 2-amino-8-aza-6-chloro-7-deazapurine (4) that was enzy-

matically converted into its 2’-deoxyriboside 4a as the main

product (40–55% according to HPLC) and isolated from the re-

[a] Dr. I. V. Fateev, M. I. Kharitonova, K. V. Antonov, Dr. I. D. Konstantinova,

Dr. V. N. Stepanenko, Dr. R. S. Esipov, Prof. A. I. Miroshnikov

Department of Biotechnology

Shemyakin and Ovchinnikov Institute of Bioorganic Chemistry

RAS, Miklukho-Maklaya 16/10, 117997 GSP

Moscow B-437 (Russian Federation)

[b] Prof. F. Seela

Laboratory of Bioorganic Chemistry and Chemical Biology

Center for Nanotechnology

Heisenbergstr. 11, 48149 Münster (Germany)

[c] Dr. K. W. Temburnikar, Prof. K. L. Seley-Radtke

University of Maryland, Baltimore County, 1000 Hilltop Circle

Baltimore, Maryland 21250 (USA)

Remarkably, a PNP preparation from human erythrocytes

was shown to be capable to catalyze the ribosylation of allo-

purinol (3) that was converted into 8-aza-7-deaza-9-(b-d-ribo-

furanosyl)-purine-6-one (allopurinol ribosides, Allo-Rib) by

using diverse ribosyl donors (i.e., uridine, inosine, xanthosine,

or Rib-1P) and the biocatalysts.^[5,6]The mechanism of the

[d] V. A. Stepchenko, Dr. Y. A. Sokolov, Prof. I. A. Mikhailopulo

Institute of Bioorganic Chemistry, National Academy of Sciences

220141 Minsk, Acad. Kuprevicha 5/2 (Belarus)

E-mail: imikhailopulo@gmail.com

Supporting information for this article is available on the WWW under

htt p ://dx.doi.org/10.1002/chem.201501334.

Chem. Eur. J. 2015, 21, 13401 – 13419

13401

ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper

Taking into account that E. coli

PNP catalyzes the reversible

phosphorolysis of purine nucleo-

sides, one can expect that the

Asp204 side chain makes the

main contribution in the binding

of the heterocyclic substrate in

the synthetic reaction. The

Asp204 interacts with the N7

atom and/or substituent at the

C6 atom (NH₂, NHR, carbonyl,

OR) of the purine base giving

rise, in all likelihood, to the

proper base positioning at the

catalytic site and to an enhance-

ment of the nucleophilic proper-

ties of the N9 atom, that is, acti-

vation of the substrate.^[1]It was,

therefore, surprising that a re-

placement of the N7 atom with

a CH group did not abolish the

Figure 1. Structures of the 8-aza-7-deazapurines 1–7 and their 2’-deoxy-b-d-ribofuranosides 1a–7a, 5-aza-7-deaza-

guanine (8, ^5A:7DAGua) and its glycosides 9a and 9b (^5A:7DAG and ^5A:7DAdG), the 7-thiapurines 10–14, the 9-deazapur-

ines 15 and 16, the 2’-deoxy-b-d-ribofuranosides of 9-deazaxanthine (17) and -guanine (18), as well as 1,3-benzi-

midazole (BI, 19), 1,3-benzthiazole (BT, 20), and 1,3-benzoxazole (BO, 20) and their respective 2-amino derivatives

19a–21a. Purine numbering is used throughout the text.

heterocyclic substrate binding and its activation was not inves-

tigated.^[6]However, it is noteworthy that an analysis of the ter-

nary complex of the E. coli PNP, sulfate or phosphate, and for-

mycin B (Figure 2), which is a strong inhibitor of the E. coli

substrate activity of bases 1–3 pointing to the rather efficient

contribution of the Asp204 bC(=O)OÀH···XC6 (X=NH₂, OMe, or

carbonyl) interaction in the correct binding and activation of

the substrates. However, 2-amino-8-aza-7-deaza-6-chloropurine

(4) does not contain a N7 atom and moreover contains a chlor-

ine atom at the C6 atom that is unable to take part in the hy-

drogen bonding and none the less showed satisfactory sub-

strate activity for E. coli PNP.

To gain further insight into the role of the Ser90 side chain

of the catalytic site of E. coli PNP in the recognition of the 8-

aza-7-deazapurines 1–4 as well as a number of other purine

analogues in the synthesis reaction, we 1) prepared the

Ser90Ala mutant of the E. coli PNP [see the Supporting Infor-

mation, with the preparation of the wild-type E. coli PNP^[16]],

2) analyzed kinetic parameters of the transribosylation of hypo-

xanthine, allopurinol (3), and 5-aza-7-deazaguanine (8,

^5A:7DAGua) by using the wild-type (WT) E. coli PNP versus the

mutant enzyme, 3) investigated the substrate activity of 8-aza-

7-deazapurines toward the WT E. coli PNP, (iv) tested the sub-

strate properties of a number of challenging purine analogues,

that is, the 7-thiaguanines 10–14, the 9-deazapurines 15–18,

and 1,3-benzimidazole (19), 1,3-benzthiazole (20), and 1,3-ben-

zoxazole (21), and their 2-amino derivatives 19a–21a towards

both enzymes. The results have been analyzed from viewpoint

of the electronic structures of the bases and suggested possi-

ble modes of substrate recognition aiming at the implication

for the enzymatic synthesis of new nucleosides.

Figure 2. Main tautomers of formycin B.

PNP,^[7a,b]led the authors to conclude that the N2-H tautomer of

formycin B may form a hydrogen bond to the Ser90-(H)Og

oxygen atom.^[7a,8]Notably, based on the crystallographic data,

the Ser90 residue of the catalytic site of the E. coli PNP was

suggested to participate in the binding of inorganic phos-

phate^[2,8–10]and the ribose moiety^[8,10]of a purine nucleoside in

the phosphorolysis reaction.^[2,3,8–11]

It is presently broadly accepted that the b-carboxyl group of

the Asp204 residue of the catalytic site of E. coli PNP plays an

important role in the binding of natural purine nucleosides by

means of the protonation of the N7 nitrogen atom followed

by activation leading to phosphorolytic scission of the glycosyl

bond^[2,3](for recent reviews, see reference [4]). Indeed, the re-

placement of the Asp204 residue by l-alanine gave rise to the

Asp204Ala mutant of E. coli PNP, the phosphorolytic activity of

which was dramatically reduced.^[3,12]There are only few excep-

tions from this postulate and some of them are exemplified by

the very efficient phosphorolysis of N7-methylguanosine (21a)

and its 2’-deoxy-counterpart 21b (see below), the positively

charged N7 atom of which mimics an enzymatic protona-

tion.^[13–15]

It should be emphasized that the motivation of this work is

that the E. coli PNP is very valuable in biotechnology of modi-

fied nucleosides, including a number of antiviral and antican-

cer drugs.^[2–4]Analysis of the biocatalytic potential of this

enzyme, as well as an understanding of the mechanism of its

functioning are of considerable interest for the design of new

substrates and/or inhibitors as well as for the synthesis of new

analogues of natural purine nucleosides that are of potential

importance for molecular biology and medicine.

Chem. Eur. J. 2015, 21, 13401 – 13419

www.chemeurj.org

13402

ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper

with the earlier data on the crucial importance of the N7 atom

of purine nucleosides in an analogous reaction (reviewed in

reference [4]).^[17,18]

Results and Discussion

Substrate specificity of the WT E. coli PNP versus that of the

The similar substrate properties of hypoxanthine and allo-

purinol for the WT E. coli PNP indicate a non-critical role of the

N7 atom of the heterocyclic substrate in the enzymatic synthe-

sis of nucleosides. As discussed above, two hydrogen bonds

between the b-carboxyl group of the Asp204 residue (side

chain) and the N7 atom and the C6 carbonyl group^[2,11]are re-

sponsible for the correct binding/activation of hypoxanthine in

the productive complex of the WT E. coli PNP. In the case of al-

lopurinol (3), the binding and activation are realized by the

side chain of Asp204 and the C6=O function as well as by the

Ser90À(H)Og···HÀN8 hydrogen bond that makes a compensa-

tion for the loss of the Asp204/N7 hydrogen bond

(Figure 3).^[1,4]

Ser90Ala mutant

The transribosylation of hypoxanthine and 8-aza-7-deazahy-

poxanthine (3, allopurinol, Allo) by using uridine and the re-

combinant E. coli uridine phosphorylase (UP)^[16]for the in situ

generation of intermediary Rib-1P revealed the differences in

the reaction rates at the starting phase of the synthesis cata-

lyzed by the WT E. coli PNP versus the Ser90Ala mutant

(Figure 3). In the presence of the former, the formation of 1) in-

osine reached approximately 80% in 30 min and after the next

30 min established an equilibrium of hypoxanthine+Rib-1PQi-

nosine+inorganic phosphate (Pi) at a base/nucleoside ratio of

approximately 2:8 in the reaction mixture and 2) allopurinol ri-

boside (Allo-Rib) proceeded somewhat slower achieving a 95%

conversion of allopurinol (3) to its N9 riboside after 24 h. Thus,

the reversible condensation of Rib-1P and hypoxanthine is dis-

placed to the product formation, whereas the formation of

Allo-Rib proceeded irreversibly as it was proven by the experi-

ments with 8-aza-7-deaza-9-(2-deoxy-b-d-ribofuranosyl)-purin-

6-one (3a, Allo-dRib) and the WT E. coli PNP. The Allo-dRib was

not phosphorolyzed by E. coli PNP and this result is in consent

Replacement of the WT E. coli PNP with the mutant enzyme

in the synthesis reactions resulted in 1) the reduction of the in-

itial rate of the inosine formation (35% yield of inosine in one

hour; an equilibrated mixture with a base/nucleoside ratio of

ca. 2:8 after 24 h) and 2) a significant decrease of the rate of

the Allo-Rib formation attaining a approximately 20% yield

after 48 h. A slight reduction of the rate of the inosine forma-

tion without loss of the catalytic efficiency points to a moderate

contribution of the Ser90 side

chain of the catalytic center of

the WT E. coli PNP in the binding

of hypoxanthine^[2,9]and/or Rib-

1P^[10]in the presence of all the

natural factors, responsible for

the catalytic efficiency. On the

contrary, the Ser90 residue of

the WT E. coli PNP is of unique

importance in the binding and

activation of allopurinol as well

as other 8-aza-7-deazapurines in

the enzymatic synthesis of their

nucleosides. It is obvious that

a

hydrogen bond between

Asp204 and the C6 carbonyl

function of allopurinol is an im-

portant factor in the observed

formation of Allo-Rib in the reac-

tion catalyzed by the mutant

PNP. Indeed, 2-amino-8-aza-6-

chloro-7-deazapurine (4) was

transformed into its riboside in

a yield of less than 1% after 48 h

(HPLC) under similar reaction

conditions.

Replacement of Ser90 with l-

alanine eliminates the possible

interactions with hypoxanthine

and/or Rib-1P resulting in the re-

tardation of the velocity of the

inosine formation catalyzed by

the mutant enzyme; however,

Figure 3. Progress of the synthesis of inosine versus allopurinol riboside (HPLC ratio [%]) by using the WT E. coli

PNP and the Ser90Ala mutant of the PNP. Reaction conditions: reactions (1.0 mL) were performed in the presence

of hypoxanthine or allopurinol (2 mm) and uridine (2 mm) in buffer solution (pH 7.0, 2 mm KH₂PO₄) and the re-

combinant E. coli enzymes (15 mg of the WT E. coli PNP or the Ser90Ala mutant and 9 mg of the recombinant E.

coli UP^[16]) at 508C for 48 h.

Chem. Eur. J. 2015, 21, 13401 – 13419

www.chemeurj.org

13403

ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper

when the equilibrium was reached after 24 h, a similar 2:8 stoi-

chiometric mixture of hypoxanthine/inosine was obtained. On

the contrary, in the case of the synthesis of Allo-Rib catalyzed

by the WT E. coli PNP the main function of the Ser90 residue

appears to be the initial binding followed by an activation of

allopurinol (3) and dramatic loss of efficiency of the product

formation by going from the wild type to the mutant enzyme

is in harmony with this suggestion.

faster than Allo-Rib and 5-aza-7-deazaguanosine (9a), respec-

tively, catalyzed by the WT E. coli PNP (Table 1). As expected, in

the case of the mutant E. coli PNP, the synthesis of Allo-Rib

proceeded with substantially reduced (463 times slower) effi-

ciency as compared with that of inosine, pointing to the im-

portance of the Ser90 residue in this reaction. On the contrary,

5-aza-7-deazaguanosine (9a) was synthesized surprisingly 1.8

times faster than inosine. On the other hand, the efficiency of

the syntheses of inosine, Allo-Rib, and 5-aza-7-deazaguanosine

(9a) is decreased by 93, 1438, and 28 times, respectively, in the

transition from the wild-type enzyme to the mutant enzyme

(Table 1).

In order to gain further insight in the mechanism of the gly-

cosyl bond formation, the formation of inosine, Allo-Rib, and 5-

aza-7-deazaguanosine (9a) (Figure 4) catalyzed by PNP of both

A significant reduction in the catalytic power of the mutant

enzyme versus that of the WT E. coli PNP in the synthesis of al-

lopurinol riboside agrees with our hypothesis^[1](see above).

However, the reason for the differences in the drop of the sub-

strate activity of hypoxanthine (93 times) versus that of 5-aza-

7-deazaguanine (28 times) is not obvious. We hypothesize that

the characteristics of the electronic structures of hypoxanthine

and ^5A:7DAGua (8) may provide a clue to the understanding of

the observed differences in the velocity of the enzymatic gly-

cosylation. With this aim in view, we analyzed the electronic

properties of the compounds under study by the ab initio

method (restricted Polak–Ribiere (RPR) method by using basis

set of the 6-31G** level; the files of the PM3 geometry optimi-

zation were used as starting approximation for the ab initio

calculations; HyperChem 8.1 (Hypercube, Inc., 1115 NW 4^thSt.

Gainesville, FL 32601, USA; www.hyper.com) and compared

them to structurally similar natural substrates (see the Support-

ing Information, Table S1).

Figure 4. Structures of the tautomers 8a and 8b of 5-aza-7-deazaguanine

(8) and its nucleosides 9.

types was investigated in detail. The selection of this set of

compounds was based on one common structural motif com-

prising the presence of C6=O carbonyl groups with rather simi-

lar electronic structure (Table S1 in the Supporting Informa-

tion). Thus, hypoxanthine may participate in the formation of

two hydrogen bonds with the Asp204 side chain (compare ref-

erences [2,11]); allopurinol (3) may form a hydrogen bond with

Asp204 and the other with the Ser90-OgH hydroxyl group,

whereas 5-aza-7-deazaguanine (8) has only one possibility to

form a hydrogen bond between the C6=O functionality and

the Asp204 side chain (Table 1).

According to the quantum chemical parameters, hypoxan-

thine populates in an approximately 1:1 mixture of the sp³-hy-

bridized N9 and the sp²-hybridized tautomers, whereas the

equilibrium of allopurinol is imperceptibly displaced to the sp³-

hybridized N9 structure. Remarkably, the partial negative

charge of the sp²-hybridized N9 tautomer of hypoxanthine

(À0.563 e) is higher (the absolute value!) compared with that

of allopurinol (À0.385 e) implying the higher nucleophilicity of

the former. These data are in consent with the observed

higher reactivity of hypoxanthine in the transribosylation reac-

tion catalyzed by the WT E. coli PNP (see above). The sp³-hy-

bridized N9 (8a)Qsp²-hybridized (8b) equilibrium of 5-aza-7-

deazaguanine (8) is slightly biased toward the former structure

(Figure 4). The partial negative charges of the sp²-hybridized

N9 atoms of hypoxanthine, allopurinol, and the guanine ana-

logue 8b are in satisfactory agreement with the observed reac-

tivity of these bases toward both PNPs.

First of all, we studied the formation of ribosides and found

that inosine is synthesized approximately 30 and 1.9 times

Table 1. Rates of the inosine, Allo-Rib, and 5-aza-7-deazaguanosine (9a,

^5A:7DAG) formation catalyzed by the WT and Ser90Ala E. coli PNPs.^[a]

Product of the

synthetic reaction

Formation rate

[mmolmin^À1per mg of enzyme]

wild-type PNP

Ser90Ala PNP

inosine

69

2.3

37

0.74

0.0016

1.3

allopurinol riboside (Allo-Rib)

5-aza-7-deazaguanosine (9a; ^5A:7DAG)

Comparative kinetics of the reactions catalyzed by the wild-

type E. coli PNP and its mutant

[a] Reaction conditions: reaction mixtures (1 mL) contained uridine

(4 mmol), hypoxanthine, allopurinol, or 5-aza-7-deazaguanine (2 mmol),

and potassium phosphate (20 mmol). The reactions were initiated by addi-

tion of the recombinant E. coli enzymes UP (6.6 mg) and WT PNP (from

0.34 to 0.68 mg) or the Ser90Ala mutant PNP (from 2.4 to 18 mg). The re-

actions were performed at room temperature and pH 7.0 and the prog-

ress of the reaction was monitored by HPLC. Each reaction was conduct-

ed in triplicate and average values are given.

The aforementioned considerations prompted us to study the

kinetics of the synthesis of inosine, Allo-Rib, and 5-aza-7-deaza-

guanosine (9a) by the condensation of the corresponding

bases with a-d-ribofuranose-1-phosphate (Rib-1P) catalyzed by

the WT E. coli PNP and the mutant enzyme (Table 2) (for de-

tails, see the Experimental Section).

Chem. Eur. J. 2015, 21, 13401 – 13419

www.chemeurj.org

13404

ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper

In the transition from the

wild-type E. coli PNP to a mutant

enzyme, the catalytic constant

k_catof the ribosylation of hypo-

xanthine, allopurinol, and 5-aza-

7-deazaguanine (8) decreased by

approximately 10, 770, and 346

times, respectively. As expected,

among the compounds studied,

allopurinol showed the lowest

substrate efficiency. These data

undoubtedly point to an impor-

tant role of the Ser90À

OgH···N8(9) interaction of allo-

purinol in the synthesis of its ri-

boside.

Table 2. Kinetic parameters for the synthesis of the ribosides of hypoxanthine, allopurinol, and 5-aza-7-deaza-

guanine (8) and for the phosphorolysis of inosine catalyzed by the WT E. coli PNP versus those catalyzed by

the Ser90Ala mutant PNP (for details, see the Experimental Section).

Substrate

K_m

[mm]

V_max

k_cat

[s^À1

]

Relative catalytic

constant^[a]

[mmolmin^À1mg^À1

]

wild-type E. coli PNP

hypoxanthine

allopurinol (3)

5-aza-7-deaza-guanine (8)

inosine

(0.041Æ0.006)

(0.16Æ0.02)

(59Æ12)

(1.0Æ0.1)

(34Æ6)

(153Æ31)

(2.7Æ0.2)

(90Æ15)

(66Æ5)

57

1

33

5.5

(0.15Æ0.01)

(0.052Æ0.008)

(25Æ2)

Ser90Ala mutant

hypoxanthine

allopurinol (3)

5-aza-7-deaza-guanine (8)

inosine

(20Æ3)

(5.3Æ0.4)

(0.0014Æ0.0003)

(0.10Æ0.03)

(4.6Æ1.6)

(14Æ1)

4000

1

74

1

(0.17Æ0.01)

(0.14Æ0.02)

(9.2Æ1.2)

(0.0035Æ0.0009)

(0.26Æ0.09)

(12.0Æ4.0)

[a] The lowest value of the catalytic constant k_catfor the WT and the mutant E. coli enzyme was taken for one

(1) in the relevant calculations.

The similarity of the kinetic

parameters of two analogues

has no straightforward explana-

Binding of hypoxanthine at the catalytic site of the WT E.

coli PNP is realized mainly through the hydrogen bonding be-

tween the Asp204 side chain and the N7 and the C6 carbonyl

group of the base (see references [2,11]). Replacement of

Ser90 with l-alanine does not disturb this interaction; however,

essential loss of the hypoxanthine affinity (increase of the Mi-

chaelis–Menten constant K_mvalue 490 times) is likely associat-

ed with the Ser90/base interaction, the origin of which is un-

known (see, e.g., references [2,9]). The decrease of the maxi-

mum initial velocity (V_max) and the catalytic constant (turnover

number, k_cat) both by approximately one order of magnitude

are likely related to the absence of the side chain of Ser90/Rib-

1P interaction (see, references [3,10]).

tion owing to, first of all, unique substrate properties of 5-aza-

7-deazaguanine (8). Indeed, it was shown that 5-aza-7-deaza-

guanine (8) is a substrate of mammalian^[4a,19]and bacterial^[20]

PNPs, despite the known differences in the binding and activa-

tion of natural substrates, namely, 1) Asn243 of the human

PNP (hPNP) participates in the substrate binding instead of

Asp204 of the WT E. coli PNP; note that the C6=O···HÀN(H)À

Cb(=O)ÀAsn243 hydrogen bond is weaker than that of C6=

O···HOÀC(=O)ÀR of Asp204 owing to a lower “donor strength”

of the amide group of Asn243,^[21]and nonetheless the glycosyl

bond formation takes place; 2) hPNP catalyzes the transforma-

tion of allopurinol (3) to its N9 riboside (Allo-Rib)^[5]and here-

with does not contain l-serine in the catalytic center; indeed,

the corresponding b5-sheets of the hPNP TLVVT¹¹⁴N¹¹⁵AAGG

and E. coli PNP KIIRVGS⁹CGA^[9]allow suggesting the threonine

T¹¹⁴or asparagine N¹¹⁵residues of hPNP as key amino acid resi-

dues in the synthesis of Allo-Rib; 3) formycin B (Figure 2) was

shown to be a good inhibitor of the WT E. coli PNP^[7a,b]owing

likely to the strong hydrogen bonding of Asp204/N1H and

C7=O as well as Ser90/N2H,^[2,10]but formycin B is inert against

hPNP^[7a,c,d]implying an inability of the T¹¹⁴or N¹¹⁵residues of

the catalytic site of hPNP to assume the Ser90 function of E.

coli PNP and a low binding efficiency between Asn243/N1H

and C7=O as well.

The affinity of allopurinol and 5-aza-7-deazaguanine for the

WT E. coli PNP is only four times lower compared to hypoxan-

thine. The K_mvalues for the ribosylation of both analogues are

very similar and remain practically unchanged at transition

from the wild-type enzyme to a mutant enzyme. On the

whole, the changes of the K_mvalues in experiments with the

wild-type and mutant enzymes, as well as in the transition

from the wild-type to the mutant enzyme can be explained by

the interplay of different factors, such as the residual charges

of the C6 carbonyl groups of all the studied bases, the efficien-

cy of the C6=O/Asp204 side chain, and the side chain of

Ser90/bases interactions, herewith the contribution of the

latter interaction on the K_mvalues for hypoxanthine is more es-

sential compared to the purine analogues.

The kinetic parameters obtained and the aforementioned

discussion led us to suggest an additional role of the Ser90

residue of the catalytic site of the WT E. coli PNP in the

binding and/or activation of hypoxanthine. To shed light on

the strong influence of the Ser90 on the affinity of hypoxan-

thine for the WT E. coli PNP, we studied the kinetics of the re-

verse reaction of phosphorolysis of inosine (Table 2). Similar to

The breakdown of the substrate–WT E. coli PNP complex

and the release of the ribosides of the analogues proceed ap-

proximately 60 times slower in the case of Allo-Rib and about

1.7 times slower in the case of 5-aza-7-deazaguanosine (9a)

versus that of inosine. It can be assumed that the Ser90À the synthesis reaction, the inosine affinity (K_m), as well as the

OgH···N8 hydrogen bond reduces the release rate of the Allo-

Rib molecule formed from the catalytic site of the enzyme,

which leads to the lowest V_maxand k_catvalues of the ribosyla-

tion reaction in comparison with the same kinetic parameters

for two other nucleoside syntheses in which such a hydrogen

bond is absent.

maximum initial velocity (V_max) and the turnover number (k_cat)

decreased by 177, 5.4, and 5.5 times, respectively, at transition

from the wild-type PNP to the mutant enzyme. One can hy-

pothesize that in the case of natural substrates the Ser90

residue participates in the substrate binding at the catalytic

site of the WT E. coli PNP, whereas this function is reduces or is

Chem. Eur. J. 2015, 21, 13401 – 13419

www.chemeurj.org

13405

ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper

not operative upon interaction with some analogues (see

below).

chemical parameters (e.g., the partial charge of À0.636 e)

(Table S1 in the Supporting Information) are in favor of its

higher substrate activity versus that of 5-aza-7-deazaguanine

(8) (see, Table S2 in the Supporting Information). These contra-

dictory data allow us to suggest the formation of the wrong

binding complex 25 between the base and the Asp204 side

chain of the bacterial PNP catalytic site (see reference [11]),

which consists of two strong hydrogen bonds dominating over

one correct C6ÀN(H₂)···HOÀCb(=O)ÀAsp204 hydrogen bond.

However, similar structures with wrong hydrogen binding can

apparently exist in the case of 5-aza-7-deazaguanine (8) and

the natural purine bases with the most populated tautomers

characterized by the sp³-hybridized N9 and sp²hybridized N3

atoms, respectively. To assess such a probability of the exis-

tence of such erroneous complexes, the structures of 5-aza-7-

deazaguanine (8a, ^5A:7DAGua) and its isomer 22b (^5A:7DAiGua)

with acetic acid as a mimic of the wrong Asp204/base interac-

tion, as well as guanine (Gua) and 8-aza-7-deazaguanine

As distinct from 5-aza-7-deazaguanine (8), its structural

isomer 5-aza-7-deazaisoguanine (22) displayed very low sub-

strate activity as it was observed by Voegel et al. in the study

on the enzymatic synthesis of its b-d-ribo- and 2’-deoxy-b-d-

ribo-nucleosides.^[15]A bacterial PNP of unknown origin (Sigma)

was used as biocatalyst and the relevant N7-methyl nucleo-

sides 23 as the pentofuranose donors. It was found that the re-

actions proceed very slowly and 5-aza-7-deazaisoguanosine

(24a) and its 2’-deoxy counterpart 24b were obtained in 22

(10 d incubation!) and 54% (in five consecutive reaction

cycles!) yield, respectively (Figure 5).

The reason for such a low substrate activity of 5-aza-7-de-

azaisoguanine (22, ^5A:7DAiGua) is difficult to understand. Indeed,

an equilibrium of the tautomers 22a and 22b is biased to-

wards the sp²-hybridized N9 tautomer 22b and its quantum

(

^8A:7DAGua) (both bases populated predominantly in the sp³-hy-

bridized N9 tautomers) were analyzed by the restricted Har-

tree–Fock (RHF) ab initio method by using the basis set of the

6-31** FIREFLY QC package,^[22]which is partially based on the

GAMESS (US)^[23]source code (see, also reference [1]) and the

selected data are presented in Table 3. Note that all the four

analyzed bases have analogous empirical formula.

The hydrogen-bonded structure ^5A:7DAGua-N9H/AcOH was

found to be thermodynamically most stable among the com-

plexes analyzed taking into account the total energy values.

This result contrasts with a unique substrate activity of 5-aza-7-

deazaguanine, which surpasses all the studied compounds, in-

cluding guanine. Notably, the probability of substrate binding

in the active site of the enzyme as shown for guanine was pre-

viously proven by us on an example of 2’-deoxyglycosylation

of N²-acetylguanine, wherein initially the N7 glycoside was

formed, which rearranges into the thermodynamically more

stable N9 glycoside.^[20]Such a course of the enzymatic glycosy-

Figure 5. Structures of the tautomers 22a and 22b of 5-aza-7-deazaisogua-

nine (22, ^5A:7DAiGua) and its nucleosides 24. Hypothetical structure of the het-

erocyclic base binding with Asp204 (25, N9 sp²character) inhibiting the en-

Table 3. Geometry-optimized structures of the most populated tautomers of 5-aza-7-deazaguanine (8, ^5A:7DAGua, sp³-hybridized N9 atom) and 5-aza-7-dea-

zaisoguanine (22, ^5A:7DAiGua, sp²-hybridized N9 atom) in double wrong hydrogen bonding with acetic acid as a mimic of the Asp204 binding at the catalytic

site of the E coli PNP; similar hypothetic structures of guanine and 8-aza-7-deazaguanine (^8A:7DAGua) (both in the most populated sp³-hybridized N9 form)

are calculated for comparison reason.

^5A:7DAGua-N⁹H(sp³)/AcOH

^5A:7DAiGua-N⁹(sp²)/AcOH

Gua-N⁹H(sp³)/AcOH

^8A:7DAGua-N⁹H(sp³)/AcOH

E_T[kcalmol^À1

E_HOMO[eV]

E_LUMO[eV]

]

À481585.3

À8.8846

3.5674

À481570.1

À8.9825

2.5824

À481579.3

À8.0383

3.3851

À481570.3

À8.5091

3.2572

Chem. Eur. J. 2015, 21, 13401 – 13419

www.chemeurj.org

13406

ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

zymatic glycosylation.

Full Paper

lation is excluded in the case of the other three bases, and

their glycosylation depends on several factors, particularly the

ability of the Asp204 side chain to bind the C6 carbonyl group

of the substrate. Surprisingly, the values of E_LUMOsatisfactorily

correlate with the substrate properties of the analyzed set of

bases, wherein ^5A:7DAiGua-N9/AcOH with the lowest substrate

activity is characterized by the lowest value of E_LUMO(see

below).

and the corresponding nucleosides 6a and 7a were detected

by HPLC/mass spectrometry analysis in the reaction mixtures

in 95–98 yields. The synthesis of the nucleoside of 8-aza-7-

deaza-2,6-diaminopurine (6) that is an analogue of 2,6-diami-

nopurine (DAP) is of interest for further study of its chemistry

and enzymatic transformations, which are developed for DAP

and its nucleosides, to obtain potentially biologically important

nucleosides, for example, nucleosides of 6-amino-8-aza-7-

deaza-2-fluoropurine (7), and oligonucleotides. It is noteworthy

that the chemical glycosylation of 8-aza-7-deazapurines result-

ed in the formation of the N9 and N8 regioisomeric nucleo-

sides, which is a significant drawback of chemical protocols

(see e.g., reference [24]).

Substrate activities of the new 8-aza-7-deazapurines to-

wards the WT E. coli PNP

To validate the area of the enzymatic synthesis of 8-aza-7-de-

azapurine nucleosides, the substrate activity of 8-aza-7-deaza-

2-fluoro-6-methylthiopurine (5), 8-aza-7-deaza-2,6-diaminopur-

ine (6),^[24a]and 6-amino-8-aza-7-deaza-2-fluoropurine (7) to-

wards the WT E. coli PNP was tested and the data for the com-

plete set of the analogues were analyzed from viewpoint of

their electronic structure (Tables S1 and S2 in the Supporting

Information). The fluoride 7 was synthesized from the diamino

The transglycosylation of the base 6 by using a combination

of thymidine/thymidine phosphorylase (TP)^[16]for an in situ

generation of 2-deoxy-a-d-ribofuranose-1-phosphate (dRib-1P)

as well as E. coli PNP^[16]gave, after workup and silica gel

column chromatography, the desired nucleoside 6a in 71%

yield. The formed thymine formed and nucleoside 6a were co-

precipitated from the reaction mixture and isolation of the

pure nucleoside 6a from this mixture needed careful and labo-

rious chromatography. Keeping this in mind, the synthesis of

nucleoside 7a was realized in two steps: the preparation of

dRib-1P from thymidine in the presence of E. coli TP that was

isolated as stable barium salt in 37% yield (see reference [25])

and then condensation with the base 7 affording nucleoside

7a in 95% yield after conventional silica gel column chroma-

tography.

heterobase

(Scheme 1).

6 by the Schiemann reaction in 21% yield

Unexpectedly, the base 5 showed no substrate activity for

the WT E. coli PNP. On the contrary, the enzymatic 2-deoxyribo-

sylation of the bases 6 and 7 proceeds with high efficiency

The pure individual a anomer of dRib-1P (Ba²⁺, purity>

1

97%) was found to be stable at room temperature and its H,

¹³C, and ³¹P NMR spectra are in fair agreement with the expect-

ed structure, albeit some coupling constants differ from the

corresponding constants obtained for the cyclohexylammoni-

um salt.^[26]The structure of the synthesized nucleosides 6a and

1

7a were proven by the integrity of spectral methods (UV, H,

and ¹³C NMR spectroscopy as well as mass spectrometry).

Treatment of the 2,6-diamino nucleoside 6a with adenosine

deaminase (ADA from calf intestine, Sigma) resulted in a gradu-

al transformation into 8-aza-7-deaza-2’-deoxyguanosine as it

was monitored by the changes of the UV spectra characterized

by two isobestic points at l=235 and 267 nm (see refer-

ence [24] as well as Figure S1 in the Supporting Information).

It is obvious that the trans-2-deoxyribosylation studied de-

pends on the electronic structure of the heterocyclic bases

that was analyzed for the complete set of 8-aza-7-deazapurines

studied as described above in comparison with the corre-

sponding purine bases (Table S1 in the Supporting Informa-

tion). Noteworthy that the sp³-hybridized N9 tautomers prevail

over the sp²-hybridized N9 tautomers in the case of 8-aza-7-

deazapurines (except allopurinol) and the partial negative

charges of the sp²-hybridized N9 tautomers of the purines are

significantly higher (absolute values!) than the one of the cor-

responding nitrogen atoms of 8-aza-7-deazapurines, which is

generally consistent with a higher substrate activity towards

the WT E. coli PNP.

Scheme 1. Reaction conditions: a) Synthesis of dRib-1P: incubation of thymi-

dine in K/phosphate buffer (0.5m) at 508C for 12 h in the presence of E. coli

TP gave rise to the phosphorolysis of the starting nucleoside in approxi-

mately 80% yield (HPLC). Workup followed by the final treatment with

Ba(OAc)₂afforded the desired phosphate in 37% yield; the latter was used

in the synthesis of compound 7a (2 mm Tris·HCl (pH 7.2) 238C, 2 h) (95%).

b) In situ dRib-1P was obtained from thymidine (0.2m K/phosphate buffer

(pH 7.4) at 408C for 1 h) in the presence of E. coli TP to give approximately

95% conversion of thymidine to thymine and dRib-1P; the base 6 and E. coli

PNP were added to this reaction mixture; 408C, 18 h (71%); c) 70% HF/pyri-

dine, À11 8C and an aqueous solution of KNO₂for 1 h; À11 8C!68C for 2 h

(21%).

Chem. Eur. J. 2015, 21, 13401 – 13419

www.chemeurj.org

13407

ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper

Electronic structure of the 8-aza-7-deazapurines and their

complexes with the Ser90 side chain of E. coli PNP

An ab initio analysis of the electronic structure of the bases 2,

4, and 5 showed that the C6 oxygen atom of the former is

negatively charged, whereas the corresponding chlorine and

sulfur atoms of the latter ones are positively charged, preclud-

ing the hydrogen bonding with Asp204. Previously, it has been

shown that purine heterocycles having thioalkyl or mercapto

substituents at the C6 atom are substrates similar to the corre-

sponding oxygen-containing purines^[17,27,28]apparently due to

the Asp204/N7 hydrogen bonding at the catalytic site of the

PNP. Moreover, we have recently shown that the WT E. coli

PNP accepts a number of purine bases with bulky C6 substitu-

ents (e.g., kinetin, N⁶-benzoyl- and N⁶-benzyladenines, N²-

acetyl-O⁶-methyl- and N²-acetyl-O⁶-benzylguanines, or N²-

acetyl-O⁶-[2-(4-nitrophenyl)ethyl]guanine) as substrates in the

transglycosylation reactions pointing to a large hydrophobic

space close to the C6-N1-C2 segment of the pyrimidine ring of

the purine derivatives in the active site that can accommodate

these groups.^[28–30]In addition, we have shown that 1-deaza-

purines, 3-deazapurines, and 1,3-dideazapurines (benz-

imidazole) are good substrates of the WT E. coli PNP.^[31–35]

These data point to a modest contribution of the C6 amino- or

oxygen-containing functions of the purines in the substrate

binding and presumably its activation.^[32,33]However, in the

case of 8-aza-7-deazapurines the Asp204···O(N)ÀC6 interactions

were assumed to become operational for the binding of the

substrate in the catalytic site of E. coli PNP,^[1]the value of

which can be assessed indirectly by testing such compounds

as the C6ÀCl and ÀSMe derivatives 4 and 5.

Figure 6. Structures of the heterocyclic base binding with Ser90 (N9 has sp³

character) and transformation to the productive complex (N9 has sp²charac-

ter).

the bases, that is, the base 5 with the lowest E_LUMOvalue of

2.119 eV (the binding) and 1.640 eV (the productive complex)

showed no substrate activity for the WT E. coli PNP (see, e.g.,

with the HOMO and the LUMO analysis by Pearson^[36]). Note-

worthy, 1) that the hydrogen-bonded complex of ^5A:7DAiGua-N9/

AcOH showed the lowest E_LUMOvalue (see above), which is in

harmony with the lowest substrate activity of ^5A:7DAiGua versus

the related bases (Table 3), and 2) that the LUMO energy level

was shown to be one of the most informative quantum chemi-

cal parameters in the correlation of biochemical properties

(see, e.g., references [37–39]). A similar dependence of the sub-

strate activity of 8-aza-7-deazapurines on the DE value of the

hardness of binding (2hB) and productive complex (2hPC)^[40]

was observed, that is, the calculated DE value of 0.320 eV for

compound 5 is the lowest value among the analogues stud-

ied.

The value of DE=2hBÀ2hPC of À0.238 eV for the prevailing

C6=O keto tautomer of allopurinol (Table S2 in the Supporting

Information) is entirely different from a number of the corre-

sponding values calculated for all other heterocycles studied.

Unexpectedly, the DE value of 0.452 eV for the less populated

C6ÀOH tautomer is in agreement with similar data for other

bases. Presently, we have no arguments in favor of this struc-

Comparison of the tautomeric structure of the bases 4 and

5 resulting from the geometry optimization (Table S1 in the

Supporting Information) shows a close similarity as for the pre-

vailing population of the sp³-hybridized =N8ÀN9HÀC4 tauto-

mer (DE_T=À11.5 and 10.3 kcalmol^À1, respectively) as well as

the partial charges of the sp²-hybridized N9 (ÀN8(H)ÀN9=C4,

À0.422 and À0.398 e, respectively) and the N8 (=N8ÀN9(H)À ture as the real hypoxanthine tautomer at the catalytic site of

C4, À0.279 and À0.273 e, respectively) nitrogen atoms. Taking

into account that 1) neither the chlorine nor the sulfur atom of

the respective bases 4 and 5 is able to interact with the b-car-

boxyl group of Asp204, and 2) the steric hindrances created by

the SMe group for the substrate binding at the catalytic site

appear to be minimal (see above), these data do not allow to

make reasonable conclusion regarding the absence of sub-

strate activity of the base 5 as compared with the satisfactory

substrate activity of the chloride-containing base 4.

E. coli PNP.

Recognition of the challenging heterocyclic bases by the

wild-type E. coli PNP versus its Ser90Ala mutant—7-thia

motif and 7-thia/a-amino group combination

Among the unnatural heterocyclic compounds showing sub-

strate activity for E. coli PNP, N-(1,3,4-thiadiazol-2-yl)-cyanamide

(LY217896, 26) (Figure 7) occupies an exclusive position be-

cause it has no structural elements reminiscent of any natural

heterocyclic bases, and nevertheless, is an excellent substrate

for both the mammalian and the bacterial (Sigma) PNP, and

moreover, the formation of two regioisomeric ribosides, that is,

the N4- and N3-b-d-ribofuranosides 26a (reversible synthesis!)

and 26b (irreversible ribosylation) was observed in the pres-

ence of a-d-ribofuranose-1-phosphate (Rib-1P)^[41,42](reviewed

in [4a]).

Analysis of the quantum chemical parameters of the studied

bases in a complex with the Ser90 side chain (Figure 6, binding

vs. productive complex) by the restricted Hartree–Fock ab

initio method^[22,23](see, also reference [1]) led to interesting ob-

servations (Table S2 in the Supporting Information). Among

the energy values of the highest occupied molecular orbital

(HOMO), the lowest unoccupied molecular orbital (LUMO), and

the HOMO–LUMO gap describing the electronic structures of

the 8-aza-7-deaza-purines 1–7, only the E_LUMOparameters mani-

fest a qualitative correlation with the substrate properties of

The quite puzzling substrate activity of the thiadiazole 26

for the PNPs of diverse origin prompted us to test the sub-

Chem. Eur. J. 2015, 21, 13401 – 13419

www.chemeurj.org

13408

ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper

according to the ab initio analysis (6-31G**, the basis set of pa-

rameters) its most populated tautomer is characterized by the

sp²-hybridized structure of the N9 atom (two thermodynami-

cally equivalent C6 oxo/hydroxy forms) with a partial charge of

À0.612 e as distinct from 8-oxo-7-thiaguanine (10) that pref-

erably populates in the C8=O oxo form, thereby precluding

the nucleophilic attack on the C1 atom of an a-d-pentofura-

nose-1-phosphate (Table 4) (Table S3 in the Supporting Infor-

mation).

Doskocil and Holy found that 8-aminoguanosine is a good

substrate towards the E. coli PNP, comparable to the corre-

sponding natural substrate guanosine,^[17]and a poor inhibitor

of guanosine phosphorolysis by the same enzyme. It was rea-

sonably to expect the similar situation in the case of an enzy-

matic glycosylation of 8-amino-7-thiaguanine (11), that is, the

C8 amino function will not create any additional hindrances to

the formation of the productive substrate–enzyme complex.

On the other hand, the formation of a Ser90/C8ÀNH₂hydrogen

bond is a highly probable event, however, the impact of this

event on the formation of glycosyl linkages is difficult to pre-

dict (see below).

Figure 7. Structures of N-(1,3,4-thiadiazol-2-yl)-cyanamide (LY217896, 26) and

its regioisomeric ribosides 26a and 26b, as well as the supposed hydrogen

bonding of the purine analogues with the Asp204 residue in the catalytic

site of the E. coli PNP.

strate properties of 8-oxo-7-thiaguanine (10), 8-amino-7-thia-

guanine (11), and 8-chloro-7-thiaguanine (12) as well as 7-

thiaxanthine (13), 9-deaza-7-thiaxanthine (14), and its 2’-deoxy-

riboside 14a (Figure 1), keeping in mind that these structures

vaguely resemble the five-membered ring of the thiadiazole

26. Unfortunately, all the 7-thiaguanine derivatives 10–12

showed extremely low substrate activity in the transribosyla-

tion and trans-2-deoxyribosylation reactions catalyzed by the

wild-type and Ser90Ala mutant E. coli PNPs. In experiments

with the 8-oxo derivative 10, careful HPLC analysis by using

chemically synthesized nucleosides as standards, allowed us to

identify the nucleoside peaks; however, the yield of the nu-

cleosides was on the level of 1–2%. By analogy with 5-aza-7-

deaza-iso-guanine (22) (Figure 5), miserable substrate activity

of the 7-thiaguanine derivatives 10 and 11 along with modest

inhibition of the PNP-catalyzed inosine synthesis (see below)

are presumably due to the formation of wrong complexes of

the bases with the Asp204 residue (see structures 27 and 28 in

Figure 7).

The chloride 12 and 7-thiaxanthine (13) (see reference [43])

showed no substrate and inhibitory properties for the E. coli

PNP despite the fact that they exist in the N9 imino tautomers

(partial charges of À0.470 and 0.512 e, respectively) (Table S3

in the Supporting Information) demonstrating thereby a very

low affinity to the catalytic site of the E. coli PNP. 8-Oxo-7-thia-

guanine (10) and 8-amino-7-thiaguanine (11) showed a moder-

ate affinity to the catalytic site of the WT E. coli PNP in experi-

ments on the synthesis of inosine from Rib-1P (barium salt)

and hypoxanthine in the presence of equimolar concentrations

of both bases. The velocity of the inosine formation was re-

duced by 7 and 16%, respectively, pointing to the modest

competition of the 7-thiaguanines 10 and 11 for the catalytic

site of the enzyme (data not shown). Moreover, we found that

uric acid, existing in multiple hydroxy/oxo tautomeric forms,^[44]

is the modest substrate for the E. coli PNP, which catalyzes the

synthesis of N9 riboside^[45]in the reaction with Rib-1P in 12%

yield; it is likely that one of tautomeric forms with a sp²-hybrid-

ized N9 atom is populated in the aqueous reaction buffer ena-

bling the formation of a glycosyl bond (this work is in prog-

ress).

The lack of substrate activity of 8-amino-7-thiaguanine (11)

for the E. coli PNP was the most unexpected finding, because

Table 4. Geometry-optimized (ab initio method, 6-31G**, basis set of parameters, HyperChem 8.1) structures of the two tautomers of 8-oxo-7-thiaguanine

(10) and 8-amino-7-thiaguanine (11).

Most stable tautomer

Less stable tautomer

E_T[kcalmol^À1

]

À600386.3^[a]

À587933.5

À600368.1

À587933.4

[a] DE=À18.5 kcalmol^À1

.

Chem. Eur. J. 2015, 21, 13401 – 13419

www.chemeurj.org

13409

ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper

The differences in the substrate properties of uric acid

versus the 7-thiaguanines 10 and 11 can be attributed to 1) a

more efficient binding and correct docking at the catalytic

center of the PNP of the former through two Asp204/base hy-

drogen bonds, and/or 2) a stringent control of the space di-

mensions of the substrate molecule. Regarding the latter, the

discussed differences of the substrate properties can be attrib-

uted to the known differences in the van der Waals radii of the

N7 nitrogen atom of uric acid versus a sulfur atom of com-

pounds 10 and 11.

Comparative ab initio analysis of the electronic structures of

8-amino-7-thiaguanine (11) and N-(1,3,4-thiadiazol-2-yl)-cyana-

mide (26) and two modes of an overlay of the geometry-opti-

mized structures show a close similarity of both five-mem-

bered rings. The lack of substrate activity of the former re-

mains obscure. Indeed, by assuming for the nucleoside 26 sim-

ilar hydrogen bonding as proposed for the base 11 (Figure 7,

structure 29), one would expect a similar inhibitory activity

(Table 5, right overlay), but this is contrary to the known sub-

atoms, that is, N3 and N4, on the anomeric C1 atom. Anyway,

testing of the substrate properties of heterocyclic compounds

having cyano or acetylene group for E. coli PNP is of obvious

interest.

9-Deaza-7-thiaxanthine (14) and its 2’-deoxyriboside 14a did

not inhibit the phosphorolysis of inosine (1:1 (mol) inosine/in-

hibitor ratio, 0.1m phosphate buffer (pH 7.3), 208C, time of re-

action 10 min). The lack of activity of the heterocyclic base 14

is consistent with data obtained for the related 7-thiapurines

10–13 and its glycosylation does not result in an increased af-

finity for the enzyme.

9-Deazapurines and their nucleosides as valuable probes of

the Ser90/N9 inosine co-operation

To evaluate the efficiency of the Asp204 and N7H/C6=O

double hydrogen binding, we studied the phosphorolysis of in-

osine (see references [46,47]^]) under standard reactions condi-

tions in the presence of 9-deazaxanthine (15) and 9-deaza-9-io-

doxanthine (16) as well as the 2’-deoxy-b-d-ribonucleosides of

9-deazaxanthine (17) and 9-deazaguanine (18), the syntheses

of which were published recently.^[48,49]Compounds of this

group do not contain a substituent at the C8 atom and this

allows eliminating of the consideration of interactions with the

Ser90 side chain. It was found that 9-deazaxanthine (15) and

its 2’-deoxy-b-d-riboside 17 exhibited weak inhibitory activity

(13–15% inhibition) and 9-deaza-9-iodoxanthine (16) remark-

able activity (77% inhibition) in the phosphorolysis of inosine;

9-deaza-2’-deoxyguanosine (18) as opposed to 9-deaza-2’-de-

oxyxanthosine (17) did not inhibit the inosine phosphorolysis

under standard reaction conditions. Differences in the inhibito-

ry properties of the nucleosides 17 and 18 reflect apparently

different electronic properties of the H7NÀC5ÀC6=O fragments

of the corresponding aglycones, which determine the efficien-

cy of their interaction with Asp204 by the formation of two hy-

drogen bonds and the ability to compete with similar inosine

hydrogen bonds.

Table 5. Geometry optimization by the ab initio method (6-31G**) of 8-

amino-7-thiaguanine (11) and N-(1,3,4-thiadiazol-2-yl)cyanamide (26) and

overlay of the five-membered rings of both structures.

8-Amino-7-thiaguanine

N-(1,3,4-Thiadiazol-2-yl)cyanamide

partial charges of selected atoms [e]

N9=À0.612

N3=À0.351

N4=À0.226

S1= +0.311

S7= +0.303

O6=À0.625

two kinds of overlay of both structures

At first glance, a significantly higher inhibitory activity of the

iodide 16 in comparison with the parent base 15 is unexpect-

ed. However, it was previously shown that 5’-deoxy-5’-iodo-9-

deazainosine is a significantly more effective competitive inhib-

itor of hPNP from human erythrocytes than the parent com-

pound 9-deazainosine,^[46]which corresponds to the behavior of

the bases 15 and 16, and reveals a certain similarity of the two

PNPs. Kinetics of the inosine phosphorolysis by the E. coli PNP

was studied by the conventional initial velocity method as well

as with the use of three concentrations of the iodide 16 and

the data are displayed by Lineweaver–Burk plots (Figure 8).

The Michaelis–Menten constant, K_M=23.3 mm, the maximal

velocity V_max=0.70 mmmin^À1mg^À1, and k_cat=67 sec^À1were cal-

culated for the inosine phosphorolysis. Unlike 5’-deoxy-5’-iodo-

9-deazainosine,^[46]a non-competitive or rather a mixed type of

inhibition of the inosine phosphorolysis was established in the

case of iodide 16 with respect to the nucleoside substrate. It

was found that the phosphorolysis in the presence of the in-

hibitor is accompanied by a decrease in the V_max/K_Mratio and

therefore the inhibition constant (K_i=(10Æ1) mm) was calculat-

strate properties of the thiadiazole 26. Apparently, the correct

positioning of the thiadiazole 26 in the catalytic center of the

PNP, which enables the attack of both nitrogen atoms at the

C1 atom of the Rib-1P, is implemented through an interaction

of the exocyclic ÀN(H)ÀCꢀN group with the Asp204 residue

(Table 5, left overlay). It is logical to suggest that the binding

of the base 11 through the C6 carbonyl group/Asp204 hydro-

gen bond, if it is realized at all, leads to a spatial arrangement

of the substrate at the catalytic center that is not compatible

with the PNP requirements, whereas the unique binding

through the cyanamide group in the catalytic site of the PNP

enables a nucleophilic attack of both sp²-hybridized nitrogen

Chem. Eur. J. 2015, 21, 13401 – 13419

www.chemeurj.org

13410

ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper

mutant, the substrate properties of benzimidazole (19), 1,3-

benzothiazole (20), and 1,3-benzoxazole (21), as well as their

NH₂

respective 2-amino derivatives 19a (^NHBI), 20a ( BS), and 21a

(

2

NH₂

BO) (Figure 1) towards the wild-type and mutant E. coli

PNPs were studied. Selection of these bases was made to

obtain complementary data with regard to the exocyclic

amino motif in a combination with the endocyclic amino,

sulfur, and oxygen substitution in the five-membered ring in

the recognition of the bases by the E. coli PNP. The substrate

properties of these compounds (except BI) towards the E. coli

PNP have not, to our knowledge, been studied before and are

of interest for the reason that they differ only by one substitu-

ent in the five-membered ring and enable to eliminate the

contribution of the pyrimidine ring substituents in the enzy-

matic reactions under consideration.

Figure 8. Double reciprocal plot (1/[V] vs. 1/[S]) for the inosine phosphoroly-

sis in the presence of 0, 11, 14, and 18 mm of 9-deaza-9-iodoxanthine (16).

All the reactions were run in 0.1m K/phosphate buffer (pH 7.3) at 208C for

10 min, 93 mgL^À1of the PNP with an activity of 27 Umg^À1, 35–105 mm of ino-

sine. The progress of the phosphorolysis was monitored at l=248 nm

(De=3160m^À1cm^À1).

Benzimidazole was studied for comparison reason and taken

as a substrate, the high activity of which for E. coli PNP was

shown previously.^[31–33]The standard reaction conditions for the

transglycosylation of the complete set of bases were employed

based on the BI transribosylation data. A combination of the E.

coli uridine phosphorylase (UP)^[16]and uridine or 2’-deoxyuri-

dine was used for an intermediary generation of a-d-ribo-fura-

nose-1-phosphate (Rib-1P) or a-d-2-deoxyribo-furanose-1-

phosphate (dRib-1P) in the synthesis of the respective ribo-

and 2’-deoxyribonucleosides.

ed as it was suggested by Shugar et al.^[47]One can hypothesize

that a non-specific binding of the 5’-iodo substituent within

the amino acid residue(s) of the exocyclic 5’-OH binding site of

the hPNP contributes to the strong anchoring of the rest of

the molecule at the catalytic site. On the contrary, a similar

non-specific binding of the base 16 prevails over the specific

double hydrogen bonding of the E. coli Asp204 side chain with

the HNÀCÀC=O fragment of the base.

The formation of the ribonucleosides (31a, Figure 9, A) and

the 2’-deoxyribonucleosides (31b,Figure 9, C) of BI catalyzed

by the WT E. coli PNP proceeded very quickly and the main dif-

The most surprising finding was the weak inhibitory activity

of 9-deaza-2’-deoxyxanthosine (17) and furthermore, the lack

of such activity of the 9-deazaguanine analogue 18. Indeed,

the nucleosides 17 and 18 have all the functional groups nec-

essary for an efficient binding in the active site of the E. coli

PNP. One can assume that the E. coli PNP requirements to the

spatial arrangement of the substrate at the catalytic center are

significantly different from those of the human enzyme, and,

apparently, the nucleosides 17 and 18 are unable to adopt the

necessary stereochemistry for the docking in the active center

(for a detailed discussion of peculiar stereochemistry of the N-

vs. C-nucleosides, see Chattopadhyaya et al.^[50]). An alternative

explanation is that the Ser90 residue of the E. coli PNP plays an

essential role in linking the nucleoside substrate by a hydrogen

bond between the Ser90ÀOgH hydroxyl group and the N9

atom of natural purine nucleosides. The latter assumption is

consistent with the data presented above (Table 2) for the

phosphorolysis of inosine by the wild-type E. coli PNP and

a mutant enzyme, wherein a significant decrease of the PNP af-

finity was found by replacing the Ser90 residue by l-alanine.

These suggestions are of undoubted interest and deserve

more detailed study.

Figure 9. Progress of the synthesis of benzimidazole nucleosides PNP Rib-BI

and dRib-BI (in [%], ordinate) by using the WT E. coli PNP (A=Rib-BI,

C=dRib-BI) versus the mutant PNP (B=Rib-BI, D=dRib-BI) (HPLC analysis).

Reaction conditions: a) Synthesis of the ribosides: reaction mixture (1 mL)

contained the base (2 mm) and uridine (2 mm) in K/phosphate buffer (2 mm,

pH 7.0). b) Synthesis of the 2-deoxyribosides: reaction mixture (1 mL) con-

tained the base (2 mm) and 2’-deoxyuridine (4 mm) in K/phosphate buffer

(4 mm, pH 7.0). The reactions were initiated by addition of the recombinant

E. coli enzymes: UP (0.036 mgmL^À1, 3.6 units) and the E. coli PNP

(0.06 mgmL^À1, 3.24 units) or the mutant PNP (0.06 mgmL^À1, 0.072 units) and

the reaction mixtures were incubated at 508C.

ference between both reactions consisted in the changes of

the base/nucleoside ratio that occurred during long-term stor-

age of the reaction mixtures. Thus, the quantitative transforma-

tion of BI into 1-(2-deoxy-b-d-ribofuranosyl)benzimidazole

(dRib-BI) (Figure 9, C) occurred in a few minutes and then the

concentration of dRib-BI in the reaction mixture decreases

gradually until approximately 15% after 140 h. On the contrary,

Benzimidazole and its 2-amino derivative as well as their 3-

thia- and 3-oxa derivatives as probes of the role of the PNP

Ser90 residue in the enzymatic glycosylation

To minimize the contribution of the pyrimidine ring of the pu-

rines in the reactions catalyzed by the E. coli PNP and its

Chem. Eur. J. 2015, 21, 13401 – 13419

www.chemeurj.org

13411

ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper

the formation of Rib-BI (Figure 9, A) quickly attained around

55% and the equilibrium BI+Rib-1PQRib-BI+Pi remained un-

changed in the reaction mixture during 144 h (Figure 9).

In reactions catalyzed by the mutant enzyme, the formation

of dRib-BI (Figure 9, D) attained approximately 95% yield after

24 h and further maintaining the reaction is accompanied by

the very slow decrease of the product concentration to around

83%. Ribosylation of BI (Figure 9, B) was running very slowly

and yet reached approximately 50% output after 144 h of in-

cubation.

The high efficiency of the BI deoxyribosylation is fully consis-

tent with the results obtained previously in similar experi-

ments.^[31]The enzymatic synthesis has an equilibrium nature

and the decline of the dRib-BI concentration obtained by

using the E. coli PNP (Figure 9, C) versus the use of the mutant

PNP (Figure 9, D) indicates complementary WT PNP activity

(see references [51,52]), which catalyze 2-deoxy-a-d-ribofur-

anse-1-phosphate dephosphorylation. The intermediary a-d-ri-

bofuranse-1-phosphate is more resistant against this side activ-

ity of the WT E. coli PNP under reaction conditions used. It was

supposed that the Ser90 residue of the E. coli PNP takes a part

in the hydrogen bonding with (d)Rib-1P^[8,10]and the side activi-

ty of the wild-type enzyme is likely connected with this inter-

action.

Figure 10. Progress of the synthesis of the ribonucleosides (A and B) versus

the 2’-deoxyribonucleosides (C and D) nucleosides (in [%], ordinate) of 2-

aminobenz-imidazole by using the WT E. coli PNP (A and C) or the mutant

PNP (B and D) (HPLC analysis). Reaction conditions: see Figure 9.

NH₂

C) was observed only in the synthesis of BI-dRib catalyzed

by the wild-type enzyme. Participation of the Asp204 residue

of the catalytic site of the E. coli PNP in the binding and activa-

tion of ^NHBI as well as BI appears to be similar to that in the re-

2

actions of natural heterocyclic bases.

To validate the possibility of the practical enzymatic synthe-

NH₂

sis of

BI nucleosides, the reaction conditions of the

BI-

dRib (32b) synthesis were optimized and the desired com-

pound was obtained in 91% yield (see the Experimental Sec-

tion).

Comparison of the reactions A and B (Figure 9) unequivocal-

ly points to the participation of the Ser90 residue of the cata-

lytic site of the WT E. coli PNP in the binding/activation of the

substrate, just (the same way) as it was mentioned above for

the synthesis of inosine. The nature of this positive impact is

difficult to imagine, and the replacement of Ser90 with l-ala-

nine leads to a marked slowdown in the synthesis of the

riboside BI-Rib (Figure 9, B) cata-

It is obvious that the binding and activation of BI at the cat-

alytic site of the wild-type and mutant PNPs is realized mainly

by the interaction of the Asp204 side chain with one of the en-

docyclic nitrogen atoms (Scheme 2). It is likely that in the case

of the BI glycosylation catalyzed by the E. coli PNP a positive

contribution of the side chain of Ser90 in the reaction effec-

lyzed by the mutant enzyme

(Figure 10).

Characteristic features of the

synthesis of the ribosides

(Figure 9, A and B) and the 2’-de-

oxyribosides (Figure 9, C and D)

of BI catalyzed by the two en-

zymes are manifested also in the

synthesis of the relevant glyco-

sides of 2-aminobenzimidazole,

but the efficiency of the synthe-

sis is much lower (Figure 10). It

is obvious that the Ser90 residue

has a positive effect on the syn-

thetic reactions catalyzed by the

wild-type enzyme (Figure 10, A

and C), and, like the glycosyla-

tion of benzimidazole, 2’-deoxy-

ribosylation (Figure 10, C and D)

occurred with greater efficiency

than ribosylation (Figure 10, A

and B). Again, the decrease of

the product concentration in the

reaction mixture after attaining

Scheme 2. Suggested structures of hydrogen-bonded heterocyclic bases at the catalytic site of E. coli PNP, as well

NH₂

the maximal content (Figure 10, as products of the glycosylation of BI (31) and BI (32), and BO 35a and 35b and 36a and 36b.

Chem. Eur. J. 2015, 21, 13401 – 13419 www.chemeurj.org ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

13412

Full Paper

tiveness is similar to that of hypoxanthine glycosylation (see

and ¹³C NMR spectra [including (¹H,¹H) and (¹H,¹³C) 2D COSY,

HSQC, HMBC, and NOE spectra—see the Supporting Informa-

tion, NMR data for compounds 36a (pages 8–10) and 36b

(pages 11–13)] allowed us to suggest the unexpected structure

36a, the b-d-ribopyranose fragment of which is attached to

the exocyclic amino group. This assignment is based on 1) the

presence of one hydrogen atom attached to the exocyclic

amino group manifesting strong coupling with the anomeric

proton (9.3 Hz, trans arrangement), 2) the H1’,H2’ coupling

constant of J=9.14 Hz, which is characteristic for the trans ar-

rangement of the vicinal protons of the pyranose ring (the

H1’,H2’ torsion angle of the pyranose ring of compound 36a

NH₂

above). Lower efficiency of the synthesis of

BI-dRib

(Figure 10, C) versus that of dRib-BI (Figure 9, C) is apparently

associated with the cooperation or competition of the two

types of substrate binding in the catalytic center of the

enzyme, that is, 1) the main binding through protonation of an

endocyclic nitrogen atom by the Asp204 side chain (e.g., struc-

tures 30 and 19, Scheme 2) and 2) additional hydrogen-bind-

ing C2ÀNH₂function with the Ser90 side chain (see structures

33 and 34, Scheme 2). The latter mode of interaction can also

takes place in the case of 2-aminoadenine and thereby increas-

es the affinity to the enzyme as it was stated by Doskocil and

Holy:^[17]the K_iconstants of the phosphorolysis of guanosine in

the presence of adenosine and 8-aminoadenosine were found

to be 350 and 5.5 mm, respectively. This type of hydrogen

bond may also be present in the binding of the 7-thiapurines

10 and 11 at the catalytic site of the E. coli PNP, however, if in

the case of good substrates such as BI it reduces the reaction

rate, in the case of poor substrates the formation of such hy-

drogen bonding further reduces the probability of the forma-

tion of the glycosyl linkage.

4’

in the most populated C_1’conformation (see the Experimental

Section) correlates well with this coupling constant (compare

with data for similar b-d-ribopyranosyl amines^[53,54])], 3) the C4’

hydroxyl proton resonance, which appears as a doublet point-

ing to the coupling with the C4’ vicinal pyranose proton rather

than with the C5’ protons of the furanose isomer 35a, 4) cross

peaks of the 4’-OH proton (d=4.70 ppm) with the C3’ (d=

71.11 ppm), C4’ (d=67.38 ppm), and C5’ (d=64.61 ppm)

atoms (see the Supporting Information, Figure 5-E). The ¹³C

HMBC spectrum of compound 36a (pentopyranose fragment,

Supporting Information, page 10) gives further support for the

pyranose structure.

Unlike benzimidazole, 1,3-benzthiazole (20) and 1,3-benzox-

azole (21) revealed no substrate activity towards the wild-type

enzyme and its mutant as well. In contrast to 2-aminobenzimi-

dazole (19a), 2-amino-1,3-benzothiazole (20a) showed neither

substrate nor inhibitory activities; the latter was tested in the

synthesis of inosine pointing to inefficient hydrogen binding at

the catalytic site of the wild-type PNP.

NH₂

The 2’-deoxyribonucleoside 36b of BO was formed in 8%

yield (HPLC) and was characterized by high-resolution mass

spectrometry (see the Experimental Section). Chromatographic

purification by using reversed-phase silica gel gave an amor-

phous product of approximately 90% purity (NMR data, com-

pound is not stable) The detailed analysis of this compound by

NMR spectroscopy showed close structural similarity with the

riboside 36a (see the Experimental Section, as well as the

The base 21a showed modest substrate activity in the trans-

ribosylation and trans-2’-deoxyribosylation catalyzed by the

WT E. coli PNP, only giving rise to the formation of the respec-

tive hypothetical furanosides 35a and 35b that spontaneously

isomerize into the b-d-ribopyranosides 36a and 36b in 27 and

8% yield, respectively, according to HPLC analysis (Figure 11)

(for the HPLC and mass spectrometry data, see the Experimen-

tal Section).

1

spectra in the Supporting Information). The H NMR spectrum

of the 2’-deoxyribopyranoside 36b agrees satisfactorily with

those of the ribopyranoside 36a within a range of the reso-

nance signals of aglycone as well as the resonance signals

for H1’ [i.e., d=5.0 (dt, ³J(H1’,NH)=³J(H1’,H2’a)=9.8,

³J(H1’,H2’b)=2.2 Hz)] and H3’ [i.e., d=3.73 (dd, ³J(H3’,OH)=

The amorphous N2-(b-d-ribopyranosyl)-2-aminobenzoxazole

(36a) (IUPAC name: (2R,3S,4R,5R)-2-(benzo[d]oxazol-2-ylamino)-

tetrahydro-2H-pyran-3,4,5-triol) was obtained after column

chromatography on reversed-phase silica gel in 25% yield of

3

5.6, J(H3’,H2’)=10, J(H3’,H2’b)=4.1, J(H3’,H4’)<1 Hz)].

It appears to be reasonable that an enzymatic transforma-

1

NH₂

approximately 93% purity (HPLC). A careful analysis of its H

tion of BI and BO into the corresponding nucleosides is

predominantly realized through the interaction of the Asp204

at the catalytic site of the wild-type PNP with the endocyclic

NH₂

nitrogen atom. In the case of the BI ribosylation, an involve-

ment of the Ser90 side chain in the binding resulted in percep-

tible enhancement of the reaction rate versus the reaction cat-

alyzed by the Ser90Ala mutant enzyme (Figure 10). The princi-

NH₂

pal difference between the BI and BO glycosylation rates

and the structures of the glycosides formed consists in the

spatial arrangement of the productive complexes that is suita-

ble electronically for the glycosyl bond formation between the

NH₂

endocyclic nitrogen atom of BI (19 where X=ÀNH···HO of

the Ser90 side chain) and the C1 atom of pentofuranose-1-

phosphate on the one hand, whereas the involvement of the

exocyclic C2 nitrogen atom (33Q34) in the glycosyl bond for-

Figure 11. Progress of the synthesis of the ribonucleoside 36a (A) versus the

2’-deoxyribo 36b (B) (in [%], ordinate) of 2-amino-1,3-benzoxazole by using

the WT E. coli PNP (HPLC analysis) (no glycoside formation was observed in

the reactions catalyzed by the Ser90Ala mutant enzyme). Reaction condi-

tions: see the Experimental Section.

NH₂

mation is the only possibility in the BO glycosylation on the

other. It is noteworthy that from the viewpoint of the electron-

Chem. Eur. J. 2015, 21, 13401 – 13419

www.chemeurj.org

13413

ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper

NH₂

ic structures of the hydrogen-bonded molecules of

BO

substrates in the enzymatic trans-2’-deoxyribosylation (thymi-

dine and E. coli TP + E. coli PNP) and 2’-deoxyribosylation

(dRib-1P and E. coli PNP). A good correlation of the E_LUMOvalue

and the substrate activity of 8-aza-7-deazapurines was found.

The five-membered rings of 8-amino-7-thiaadenine (11) and

N-(1,3,4-thiadiazol-2-yl)cyanamide (26) revealed close similarity,

however, the former devoid of substrate activity for both PNPs

studied, whereas the latter was found to be a substrate to-

wards mammalian and bacterial PNPs.^[41,42]Apparently, the cy-

anamide function is a major cause of the unusual substrate ac-

tivity of the base 26 and, therefore, purine analogues (and

more structurally distant heterocycles!) containing a cyanamide

group are of interest to study their substrate properties.

NH₂

(33Q34) and

BS (33Q34 with a sulphur atom instead an

oxygen atom) the glycosylation of the exocyclic nitrogen atom

NH₂

of the latter is not forbidden. However, the binding of BS at

the catalytic site of the wild-type PNP apparently gives rise to

an architecture that is not compatible with the glycosyl bond

formation owing to the remarkable dimensional differences of

NH₂

BS and BO (Figure 12).

The reduction of the inosine affinity on going from the WT

E. coli PNP to the Ser90Ala mutant (Table 2), on the one hand,

and a similar weak inhibition of the inosine phosphorolysis by

9-deazaxanthine (15) and its 2’-deoxyriboside 17 points to the

perceptible role of the Ser90 side chain—N9 interaction in the

substrate/inhibitor binding in the E. coli PNP catalytic site.

The study of an enzymatic glycosylation of 1,3-benzimida-

zole (19), 1,3-benzothiazole (20), and 1,3-benzoxazole (21) as

well as their 2-amino derivatives 19a–21a revealed that 1) BI is

Figure 12. Geometry optimization by ab initio calculations (6-31G**, basis

set of parameters, HyperChem 8.1) of 2-aminobenzimidazole (19a, left), 2-

aminobenzoxazole (21a, right), and 2-aminobenzthiazole (20a) and overlay

of six-membered rings of selected pairs of bases.

an excellent substrate for both enzymes, 2) ^NHBI, unlike 8-

2

It is obvious that the b-d-furanosides 35a and 35b are the

primary products of the PNP-catalyzed reactions, which isomer-

ize stereospecifically into the b-d-pyranosides 36a and 36b.

The non-enzymatic spontaneous isomerization of the furano-

sides 35 into the pyranosides 36 may occur either similar to

those described by Montgomery and Thomas^[53]and Marquez

et al.^[54]for relevant b-d-ribofuranosyl amines, or through the

tautomeric structure with the intermediary exocyclic double

bond (RÀN=C2ÀN1H) and a sp3-hybridized N1 atom followed

by the Amadori-like rearrangement reaction accompanied by

the enlargement of the ribose ring (see reference [55], for

a review see reference [56]).

amino-7-thiapurine (11), shows satisfactory and poor substrate

activities for the wild-type and mutant PNP, respectively, 3) BS,

NH₂

BS (like 8-amino-7-thiapurine (11)), and BO are no substrates

of both PNPs, 4) ^NHBO is a very poor substrate only for the

2

wild-type PNP in the reactions of trans-ribosylation and trans-

2’-deoxyribosylation, giving rise initially to the respective exo-

cyclic nitrogen atom glycofuranosides 35a and 35b that spon-

taneously isomerize into the b-d-ribopyranosides 36a and

36b. Noteworthy, after reaching the maximum conversion of

NH₂

BI and BI into their N1-2’-deoxyribosides, a gradual decrease

of the concentration of compounds 31b and 32b under a fur-

ther maintaining the reaction mixture was observed only in

the case of the wild-type enzyme pointing to dephosphoryla-

tion of the intermediate dRib-1P that is likely caused by the

participation of the Ser90 residue.

Conclusion

The substrate properties of a wide range of challenging heter-

ocyclic bases for the recombinant wild-type E. coli PNP versus

the Ser90Ala mutant were studied to evaluate the mechanism

of substrate recognition by PNP and the role of various elec-

tronic and structural features in this process. The hypothesis^[1]

regarding an important role of the Ser90 residue in the recog-

nition of 8-aza-7-deazapurines was confirmed by studying the

transribosylation rate (uridine and E. coli UP for an in situ Rib-

1P generation) and kinetics (co-substrate Rib-1P) of allopurinol

(3), hypoxanthine, and 5-aza-7-deazaguanine (8) as well as by

the electronic structure analysis (ab initio calculations and

HOMO, LUMO, and HOMO–LUMO gap) of the bases. Replace-

ment of Ser90 with l-alanine gives rise to a dramatic loss of

the catalytic activity of the Ser90Ala mutant PNP in the Allo-

Rib synthesis.

In general, substrate activity of natural purines and their an-

alogues for PNPs is determined by the interplay of various

electronic and structural features of heterocyclic bases that

affect the interactions with different amino acid residues at the

catalytic site. One of the features, for example, the nature of

five-membered ring substituent(s), can play a decisive role in

the manifestation of substrate properties. Understanding the

role of separate feature in the enzymatic synthesis is very im-

portant for the design of new heterocycles with the aim of

turning them into potentially biologically important nucleo-

sides.

Experimental Section

The substrate activity of 8-aza-7-deaza-2-fluoro-6-methylthio-

purine (5), 8-aza-7-deaza-2,6-diaminopurine (6), and 6-amino-8-

aza-7-deaza-2-fluoropurine (7) was studied and it was unex-

pectedly found that the former base 5 is not recognized by

the E. coli PNP, whereas the two other analogues are excellent

General methods and material: All chemicals and solvents were

of laboratory grade as obtained from commercial suppliers and

were used without further purification. TLC was performed on TLC

aluminum sheets covered with silica gel 60 F254. UV/Vis spectra

were recorded with a Carry 100 spectrometer (Varian, USA). NMR

Chem. Eur. J. 2015, 21, 13401 – 13419

www.chemeurj.org

13414

ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper

spectra were recorded at Brucker Avance 500-DRX and 700-DRX

spectrometers (Bruker, Germany). Chemical shifts (d) are given in

[ppm]. J values are given in [Hz]. For NMR spectra recorded in

[D₆]DMSO, the chemical shift of the solvent peak was set to

then filtrated. The reaction mixture was evaporated to 300 mL and

concentrated ammonia (30 mL), a water solution of barium acetate

(20 mL, 4.3 g, 16.8 mmol), and ethanol (300 mL) were added. After

one day at 48C, precipitated a-d-ribofuranose-1-phoshate (barium

salt) was filtered off, washed with ethanol, and dried in vacuum

1

2.50 ppm for H NMR and to 40.03 ppm for ¹³C NMR. Mass spectra

1

under P₂O₅. Yield: 1.86 g (5.1 mmol, 31%). H NMR (700 MHz, D₂O):

were measured on Agilent 6224, ESI-TOF, LC/MS (USA) in positive

ion mode (ESI), LCQ Fleet ion trap mass spectrometer (Thermo

Electron, USA) in 80% acetonitrile. For the high-resolution mass

spectrometry results of synthesized benzimidazole nucleosides see

Table 6. HPLC system: A) HPLC COMPACT Pump 2050 with

3

d=5.60 (dd, J(H₁,H₂)=6.5, ³J(H₁,³¹P)=4.1 Hz, 1H; H₁), 4.20–4.27 (m,

3

1H; H₄), 4.09 (dd, J(H₂,H₃)=3.9, J(H₂,H₁)=6.5 Hz, 1H; H₂), 4.02 (dd,

³J(H₃,H₄)=6.1, ³J(H₃,H₂)=3.9 Hz, 1H; H₃), 3.74 (dd, ³J(H₄,H₅)=3.3,

3

²J(H_5a,H_5b)=12.4 Hz, 1H; H_5a), 3.62 ppm (dd, J(H_5a,H₄)=5.2 Hz, 1H;

H_5b); ¹³C NMR (176 MHz, D₂O): d=97.26 (C1), 84.34 (C4), 71.64 (C2),

69.84 (C3), 61.56 ppm (C5).

2-Deoxy-a-d-ribofuranose-1-phosphate (barium salt): Thymidine

(1.3 g, 5.4 mmol) was dissolved in potassium/phosphate buffer

(0.5m, 15 mL, pH 7.4). TP (200 mL, 39 U) was added and the reac-

tion mixture was kept at 508C overnight. After that time HPLC

analysis (method A) indicated that about 80% of the thymidine

were phosphorolyzed. The reaction mixture was cooled to room

temperature, precipitated thymine was filtered off, and concentrat-

ed ammonia (25% by weight, 5 mL) was added to the filtrate fol-

lowed by an aqueous solution (50 mL) containing ammonia chlo-

ride (4 g, 74.8 mmol) and magnesium chloride (1.7 g of hexahy-

drate, 8.4 mmol). The mixture was kept in a freezer at 48C for 3 h

and was then filtrated. To the filtrate, barium acetate (1.4 g,

5.5 mmol) in water (20 mL) and ethanol (270 mL) were added.

After 24 h in the freezer at 48C, precipitated 2-deoxy-a-d-ribofura-

nose-1-phoshate (barium salt) was filtered off, washed with etha-

nol, and dried in stream of warm air (about 408C). Yield: 0.70 g

Table 6. Data of the HR mass spectrometry experiments of the benzimi-

dazole nucleosides.

Starting

base

[M+H]⁺

(calcd)

Nucleoside

synthesized

[M+H]⁺

(calcd)

19

119.0595 (119.0609)

134.0502 (134.0718)

135.0543 (135.0540)

Rib-BI

251.0998 (251.1032)

235.1074 (235.1074)

266.0716 (266.1141)

250.1180 (250.1192)

267.0946 (267.0945)

251.1023 (251.1032)

dRib-BI

2

19a

21a

Rib-^NHBI

dRib-^NHBI

2

Rib-^NHBO

2

dRib-^NHBO

2

a Lambda 1010 UV detector (BISCHOFF Chromatography, Germa-

ny), NucleosilC₁₈column, 1504.6 mm², 5 mm, isocratic elution,

1 mLmin^À1, 5% acetonitrile in 0.1% trifluoroacetic acid TFA/H₂O, v/

v; detection at l0260 nm; B) Waters system (Waters 1525, detector

2487, Breeze 2; USA), MZ PerfectSil 100 ODS-3, 5 mm, 150

4.6 mm²column, by using 5% aqueous methanol as the eluent at

the wavelength of l=254 nm, see also in the description of the

experiments; C) Waters system by using column Nova Pack C18,

4.6150 mm², 4 mm, detection at l=280 nm, gradient elution,

1 mLmin^À1: eluent A: 0.2% TFA/water, eluent B: 70% CH₃CN in

0.1% TFA/water, gradient 0–70% B over 20 min. Flash column

chromatography was carry out on silica gel 60, 35–70 mm (Merck,

USA).

The following recombinant E. coli enzymes^[16]were used in the

present study: UP, specific activity 100 units per mg, 9 mg per mL,

and TP, a solution in 5 mm potassium phosphate buffer (pH 7.0)

with an activity of 195 UmL^À1, PNP (the product of the deoD gene,

EC 2.4.2.1) specific activity 54 units per mg, 17 mg per mL. A solu-

tion in 20 mm Tris HCl buffer (pH 7.5) with an activity of

918 UmL^À1was used in the kinetic experiments and the synthesis

of nucleosides 36a and 36b.

1

(37%). H NMR (500 MHz, D₂O with methanol as an internal stan-

dard): d=5.77 (t, ³J(H₁,H₂)=5.2, ³J(H₁,³¹P)=5.2 Hz, 1H; H₁), 4.26–

4.19 (brm, 2H; H₃, H₄), 3.72 (dd, ³J(H_5a,H₄)=2.7, ²J(H_5a,H_5b)=

3

2

12.0 Hz, 1H; H_5a); 3.58 (dd, J(H_5b,H₄)=5.5, J(H_5b,H_5a)=12.0 Hz, 1H;

H_5b), 2.36 (dt, ³J(H_2’a,H_1’)=³J(H_2’a,H_3’)=5.30, ²J(H_2’’a,H_2b)=14.75 Hz,

1H; H_2’a), 2.06ppm (d, ³J(H_2’b,H_1’)=³J(H_2’b,H_3’)<0.7, ²J(H_2’a,H_2b)=

14.2 Hz, 1H; H_2’b); ¹³C NMR (126 MHz, D₂O): d=99.55 (C1), 86.17

(C4), 71.19 (C3), 61.83 (C5), 41.67 ppm (C2); ³¹P NMR (202 MHz,

D₂O): d=19.03 ppm (d, J(³¹P,H₁)=5.5 Hz); MS (ESI, negative ion

mode): m/z calcd for C₅H₁₀O₇P₁: 213.02; found: 213.2 [M⁺ÀBa²⁺

+H]

Kinetic analysis of the ribosylation of hypoxanthine, allopurinol

(3), and 5-aza-7-deazaguanine (8) catalyzed by the WT and

Ser90Ala mutant E. coli PNPs: Each reaction mixture (0.5 mL,

pH 7.0) contained a) hypoxanthine [from 0.005 to 1.2 mm for the

wild-type enzyme (final concentration in the reaction mixture

0.255 mgL^À1) or from 0.2 to 20 mm for the Ser90Ala mutant

(2.4 mgL^À1)], b) allopurinol [from 0.02 to 4 mm for both enzymes,

the WT E. coli PNP (2.4 mgL^À1) or the mutant PNP (6 mgL^À1)], or

c) 5-aza-7-deazaguanine [from 0.01 to 1.2 mm for both enzymes,

the WT E. coli PNP (0.255 mgL^À1) or the mutant (6 mgL^À1)], d) ino-

The preparation of the Ser90Ala E. coli PNP (specific activity

1.2 mmolmin^À1mg^À1of protein) is described in the Supporting In-

formation.

sine [from 0.005 to 1.2 mm for the wild-type enzyme (0.255 mgL^À1

)

A-d-Ribofuranose-1-phosphate (barium salt): In the present syn-

thesis, uridine and E. coli UP were used instead of inosine and PNP

from rat liver and xanthine oxidase originally described by Kalck-

ar.^[57.58]

or from 0.5 to 50 mm for the Ser90Ala mutant (2.4 mgL^À1)],

e) 2 mm a-d-ribofuranose-1-phosphate (barium salt) or potassium

phosphate.

Uridine (4 g, 16.4 mmol) was dissolved in potassium/phosphate

buffer (0.1m, 330 mL, pH 7.0). UP (25 mL, 56 U) was added and the

reaction mixture was kept at 508C for three days. After that time

HPLC analysis (method A) indicated that about 34% of the uridine

were phosphorolyzed. The reaction mixture was cooled to room

temperature, precipitated uracil was filtered off, and concentrated

ammonia (25% by weight, 110 mL) was added to the filtrate fol-

lowed by an aqueous solution (80 mL) containing ammonia chlo-

ride (17.5 g, 325 mmol) and magnesium chloride (10.1 g of decahy-

drate, 36.7 mmol). The mixture was kept at 48C overnight and was

For each experiment, sixteen reaction mixtures were prepared. The

reaction was performed at room temperature for various times

(from one minute to one day), according to the rate of nucleoside

formation. The concentrations of the substrate and the product

were determined by isocratic high-performance liquid chromatog-

raphy (method B). The kinetic parameters were determined by

nonlinear regression analysis by using the SciDAVis software v0.2.4

(Data Visualization Platform, SciDAVis; http://sourceforge.net/proj-

ects/scidavis/) Then the average value of the constants for three

experiments was calculated. The k_catvalues (turnover number)

Chem. Eur. J. 2015, 21, 13401 – 13419

www.chemeurj.org

13415

ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper

+

were calculated per whole enzyme by using a PNP mass of

156 kDa.^[16]

(10) [M⁺+2H], 267.1 (100) [M⁺+H], 151.2 (23), [C₅H₇N₆(base+H⁺

)], 113.2 (2), 60.1 (6).

Kinetic analysis of the inhibitory activities of the 9-deazapurines

15 and 16 and their nucleosides 17 and 18: In preliminary experi-

ments the rate of inosine phosphorolysis in the presence or ab-

sence of the tested compounds was compared. Reaction condi-

tions: 100 mm potassium/phosphate buffer (pH 7.3), 0.75 mm ino-

sine, approximately 0.75 mm inhibitor, 0.005 U of PNP per mmol of

inosine. After being kept at 208C for 10 min a drop of 1m HCl was

added and the reaction mixture was analyzed by HPLC (method A).

6-Amino-8-aza-7-deaza-2-fluoropurine (7)

Based on preliminary data, phoshorolysis of inosine in the range of

concentrations of 35–105 mm was analyzed in the presence of 0,

11, 14, and 18 mm of 9-deaza-9-iodoxanthine (16), 100 mm potassi-

um/phosphate buffer (pH 7.3), 2.5 UL^À1(0.93·10–3 mgL^À1) of PNP,

progress of the phosphorolysis was monitored by UV at l=

248 nm (De=3160m^À1cm^À1).

To a suspension of 4,6-diaminopyrazolo[3,4-d]pyrimidine (0.55 g,

4.23 mmol) in HF/pyridine (70%, 5 mL) cooled at À11 8C with an

ice/salt bath was added an aqueous solution of KNO₂(0.49 g,

5.76 mmol) in water (0.2 mL) over one hour. After the addition, the

mixture was stirred for another two hours with the temperature in-

creasing from À11 to 68C. The mixture was poured into an ice-

cooled calcium carbonate mixture and stored overnight. Methanol

(500 mL) was used to extract the product from the mixture. The

product was further purified by flash chromatography and a color-

less solid (120 mg, 21%) was received. R_f(CH₂Cl₂/CH₃OH, 9:1): 0.31;

elemental analysis calcd for C₅H₄FN₅(153): C 39.22, H 2.63, N 45.74;

found: C 38.96, H 2.50, N 45.30.

8-Aza-7-deaza-9-(2-deoxy-b-d-ribofuranosyl)-2,6-diaminopurine

(6a)

6-Amino-8-aza-7-deaza-9-(2-deoxy-b-d-ribofuranosyl)-2-fluoro-

purine (7a)

To a solution of thymidine (200 mg, 0.83 mmol) in potassium/phos-

phate buffer (0.2m, pH 7.4, 10 mL) was added TP (40 mL, 7.8 U) and

the reaction mixture was stirred at 408C for 1 h resulting in almost

complete phosphorolysis of thymidine according to HPLC analysis

(method A). PNP (120 mL, 13 U) was added to the reaction mixture

and then the heterocyclic base 6^[22a][solution of 70 mg

(0.47 mmol) in DMSO (2 mL)] was added dropwise to the mixture,

which was stirred at 408C and the formation of the nucleoside 6a

was monitored by HPLC (method A): R_tvalues: thymidine=4.6,

thymine=3.2, compound 6=6.1, compound 6a=3.7 min. After

18 h, very small quantity of the starting base remained in the reac-

tion mixture and the precipitate consisting of thymine and the de-

sired nucleoside was formed. It was filtered off, dissolved in etha-

nol, absorbed on silica gel (3 mL), and put on the top of the silica

gel column (315 cm²), which was then eluted with chloroform/

methanol (8:1, v/v) to afford the nucleoside 6a as a white powder

(90 mg, 71%). ¹H NMR (500 MHz, [D₆]DMSO): d=7.83 (s, 1H; H₃),

7.20 (brs, 2H; NH₂), 6.31 (t, ³J(H_1’,H _2’a)=³J(H_1’,H _2’b)=6.6 Hz, 1H;

H_1’), 6.14 (brs, 2H; NH₂), 5.20 (d, ³J(OH,H_3’)=4.3 Hz; 1H; 3’-OH),

4.98–4.72 (brs, 1H; 5’-OH), 4.34 (d of brt, ³J(H_3’,H_2’a)=6.3,

2-Deoxy-a-d-ribofuranose-1-phoshate

(barium salt) (14 mg,

40.1 mmol) was dissolved in Tris·HCl buffer (20 mm, 5 mL, pH 7.2).

PNP (20 mL, 2.2 U) was added and 2-fluoro-6-amino-8-aza-7-deaza-

purine (7) (3 mg, 19.6 mmol) in DMSO (0.5 mL) was added dropwise

with stirring. The progress of the reaction was monitored by HPLC

(method A). R_tvalues: base 7=6.2, nucleoside 7a=12.0 min). After

2 h, complete transformation of the base 7 into the nucleoside 7a

was observed according to HPLC. The reaction mixture was evapo-

rated to dryness, the residue was dissolved in ethanol, absorbed

on silica gel (0.5 mL), and put on the top of the silica gel column

(1.010 cm²), which was then eluted with chloroform/methanol

(10:1, v/v) to afford the nucleoside 7a as a white powder (5 mg,

1

95%). H NMR (500 MHz, D₂O with CH₃CN as an internal standard):

3

³J(H_3’,H_2’b)=3.6, J(H_3’,H_4’)=3.6 Hz, 1H; H_3’), 3.74 (dt, J(H_4’,H_5’)=5.4,

d=7.44 (s, 1H; H₃), 5.84 (t, J(H_1’,H_2’a)=³J(H_1’,H_2’b)=6.6 Hz, 1H; H_1’),

2

3

³J(H_4’,H_3’)=3.1 Hz, 1H; H_4’), 3.46 (dd, J(H_5’a,H_5’b)=11.5, J(H_5’a,H_4’)=

3.99 (dt, J(H_3’,H_2’a)=6.5, J(H_3’,H_2’b)=3.9, J(H_3’,H_4’)=3.9 Hz, 1H; H_3’),

3.43 (dt, ³J(H_4’,H_3’)=³J(H_4’,H_5’a)=3.9, ³J(H_4’,H_5’b)=6.9 Hz, 1H; H_4’),

3.11 (dd, ³J(H_5’a,4’)=4.1, ²J(H_5’a,H_5’b)=12.2 Hz, 1H; H_5’a), 3.00 (dd,

³J(H_5’b,4’)=6.0, ²J(H_5’b,H_5’a)=12.3 Hz, 1H; H_5’b), 2.29 (dt, ³J(H_2’a,1’)=

³J(H_2’a,3’)=6.4, ²J(H_2’a,H_2’b)=13.9 Hz; 1H, H_2’a), 1.83 ppm (ddd,

³J(H_2’b,1’)=6.8, ³J(H_2’b,3’)=4.1, ²J(H_2’b,H_2’a)=14.1 Hz, 1H; 2’-H_b);

¹³C NMR (126 MHz) (purine numbering is italicized): d=162.43 (d,

¹J(¹³C₆,¹⁹F)=212.88 Hz; C6/C2), 160.47 (d, ³J(¹³C_7a,¹⁹F)=20.80 Hz,

C7a/C4), 155.45 (d, ³J(¹³C₄,¹⁹F)=18.96 Hz; C4/C6), 135.11 (C3/C7),

119.50 (Me-C=N), 99.90 (C3a/C5), 87.23 (C4’), 84.86 (C1’), 71.64 (C3’),

5.5 Hz, 1H; H_5’a), 3.36 (dd, ³J(H_5’b,H_4’)=5.5 Hz; 1H; H_5’b), 2.71 (dt ,

²J(H_2’a,H_2’b)=12.9,

³J(H_1’,H_2’a)=³J(H_2’a,H_3’)=6.3 Hz,

1H;

H_2’a),

2.12 ppm (ddd, ²J(H_2’b,H_2’a)=13.0, ³J(H_2’b,H_1’)=6.6, ³J(H_2’b,H_3’)=

3.6 Hz, 1H; H_2’b); the resonance signal of H_5’bis overlapped by the

HOD signal; ¹³C NMR (126 MHz, D₂O) (purine numbering is itali-

cized): d=162.61 (C4/C6), 158.53 (C7a/C4), 156.88 (C6/C2), 133.94

(C3/C7), 95.93 (C3a/C5), 87.56 (C4’), 84.01 (C1’), 71.74 (C3’), 63.00

(C5’), 38.24 ppm (C2’), the NMR spectrum is in a fair agreement

with data published by He and Seela;^[59]UV/Vis (H₂O): l_max(e)=274

(7100), 257 (6200) (pH 7), 274 (7400), 257 (6900) (pH>10), 256

(8100 mol^À1m³cm^À1), and shoulder at 280 nm (pH<4); l_min(e)=

263 (6200), 239 (3500) at pH 7, 263 (6500), 239 (6900) at pH>10,

239 nm (4900 mol^À1m³cm^À1) (pH<4); MS (Thermo electron; posi-

tive ion mode): m/z (%) calcd for C₁₀H₁₄O₃N₆: 266.11; found: 268.1

63.50 (C5’), 39.11 (C2’), 1.42 ppm (CH₃-C=N); UV/Vis (H₂O): l_max

=

264 (pH 7), 263 (pH<4), 265 nm (pH>10); l_min=237 (pH 7, pH<

4), 241 nm (pH>10); MS (ESI, positive ion mode): m/z (%) calcd for

C₁₀H₁₂FO₃N₅: 269.09; found: 270.0 (100) [M+H⁺], 251.2 (40), [M⁺

+

ÀH₂O], 154.1 (48) [C₅H₅FN₅(base+H⁺)], 104.1 (59).

Chem. Eur. J. 2015, 21, 13401 – 13419

www.chemeurj.org

13416

ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper

2-Amino-1-(2-deoxy-b-d-ribofuranosyl)-1,3-benzimidazole (32b)

54 Umg^À1) in potassium phosphate buffer (20 mm, 75 mL, pH 7.0)

was incubated at 558C for 744 h, monitoring the formation of the

nucleoside 36a by HPLC (method C, R_t: Ura=2.1, Urd=2.5, 21a=

7.0, 36a=7.6 min). The reaction mixture was evaporated to 4 mL;

and compound 36a was obtained after preparative HPLC on re-

versed phase C18 silica gel (20250 mm²; MZ-preparative, Germa-

ny). Compound 36a was eluted with a linear MeOH gradient 0!

20% (v). The fractions containing the nucleoside 36a were pooled

and evaporated to give the desired nucleoside 36a (0.01 g, purity

93% by HPLC) in 25% yield. For a geometry-optimized structure of

nucleoside 36a by using the PM3 method, see below. ¹H NMR

(700 MHz, [D₆]DMSO, 308C): d=8.49 (d, ³J(NH,H_1’)=9.3 Hz, 1H;

NH), 7.38 (d, ³J(H₄,H₅)=7.9 Hz, 1H; H₄), 7.28 (d, ³J(H₇,H₆)=7.7 Hz,

A mixture of 2-amino-1,3-benzimidazole (19a) (0.13 g, 1 mmol), 2’-

deoxyguanosine (dG, 0.4 g, 1.5 mmol), and PNP (54 IU) in potassi-

um/phosphate buffer (20 mm, 15 mL, pH 7.0) was gentle stirred at

508C for 30 h, monitoring the formation of the nucleoside 32b by

TLC (UV detection, elution with the CH₂Cl₂/EtOH/AcOH, 85:10:5 (v/

v/v), R_f: 15a=0.25; dG=0.02, 15=0.15 min). The reaction mixture

was evaporated to dryness; the residue was dissolved in dry EtOH,

one tee spoon of SiO₂was added to the solution, and the mixture

was evaporated to dryness. The SiO₂with the product was placed

on the top of the SiO₂column [20180 mm², prepared in

CH₂Cl₂(MeOH (95:5 v/v)] that was eluted with a linear MeOH gradi-

ent 5!20% (v) in CH₂Cl₂(21000 mL). The fractions containing

the nucleoside 32b were pooled and evaporated to give the TLC-

and HPLC-pure product (0.22 g, 91%), a portion of which was crys-

tallized from a EtOH/ether mixture. M.p. 179–1848C [measured on

3

4

1H; H₇), 7.14 (dt, J(H₆,H₅)=³J(H₆,H₇)=7.7, J(H₆,H₄)ꢁ1 Hz, 1H; H₆),

3

4

7.01 (dt, J(H₅,H₄)=³J(H₅,H₆)=7.8, J(H₅,H₇)ꢁ1 Hz, 1H; H₅), 5.035 (t,

³J(H_1’,NH)=³J(H_1’,H_2’)=9.14 Hz, 1H; H_1’), 4.85 (d, J(OH,H_3’)=3.6 Hz,

1H; 3’-OH), 4.77 (d, ³J(OH,H_2’)=7.0 Hz, 1H; 2’-OH), 4.69 (d,

³J(OH,H_4’)=5.8 Hz, 1H; 4’-OH), 3.93–3.95 (m, ³J(H_3’,OH)=3.6,

3

³J(H_3’,H_2’)=2.8, J(H_3’,H_4’)ꢁ2 Hz, 1H; H_3’), 3.53–3.55 (m, J(H_4’,OH)=

5.8, ³J(H_4’,H_5’)ꢁ5, ³J(H_4’,H_3’)ꢁ2 Hz, 1H; H_4’), 3.53 (dd, ²J(H_5’a,H_5’b)=

11.0, ³J(H_5’a,H_4’)ꢁ3 Hz, 1H; H_5’a), 3.47 (dd, ²J(H_5’b,H_5’a)=11.0,

³J(H_5’b,H_4’)ꢁ5 Hz, 1H; H_5’b), 3.40 ppm (td, ³J(H_2’,H_1’)=9.1,

³J(H_2’,OH)=7, ³J(H_2’,H_3’)=2.8 Hz, 1H; H_2’); ¹³C NMR (176 MHz,

[D₆]DMSO, 308C): d=162.42 (C2), 148.35 (C4a), 143.22 (C7a), 124.21

(C6), 121.02 (C5), 116.29 (C7), 109.19 (C4), 81.35 (C1’), 71.27 (C3’),

69.99 (C2’), 67.53 (C4’), 64.73 ppm (C5’); ¹⁵N NMR (71 MHz,

[D₆]DMSO, 308C): d=84.3 ppm (NH); UV/Vis (H₂O): l_max: 240,

278 nm (pH 7); MS (ESI, positive ion mode): m/z (%) calcd for

C₁₂H₁₄O₅N₂: 267.0945; found: 267.09679 [M⁺+H]; calcd for

C₇H₆N₂O⁺: 135.0540; found: 135.05484 [base+H].

1

the WRS-2 (BestScope, China)]; H NMR (500 MHz, [D₆]DMSO): d=

7.28 (d, ³J(H₄,H₅)/(H₇,H₆)=7.77 Hz, 1H; H₄/H₇), 7.12 (d, ³J(H₇,H₆)/

(H₄,H₅)=7.72 Hz, 1H; H₄/H₇), 6.96 (dt, ³J(H₅,H₄)/(H₆,H₅)=³J(H₅,H₆)/

(H₆,H₇)=7.60, ⁴J(H₅,H₇)/(H₆,H₄)=0.90 Hz, 1H; H₆/H₇), 6.87 (dt,

³J(H₆,H₅)/(H₅,H₄)=³J(H₆,H₇)/(H₅,H₆)=7.56, ⁴J(H₅,H₇)/(H₆,H₄)=1.02 Hz,

1H; H₆/H₇), 6.51 (s, 2H; NH₂), 6.23 (dd, ³J(H_1’,H_2’a)=8.80,

³J(H_1’,H_2’b)=6.15 Hz, 1H; H_1’), 5.30 (d, ³J(OH,H_3’)=4.17 Hz, 1H; 3’-

OH) 5.29 (t, ³J(OH,H_5’a)=³J(OH,H_5’b)=4.88 Hz, 1H; 5’-OH), ca. 2.5

N2-(2-Deoxy-b-d-ribopyranosyl)-2-amino-1,3-benzoxazole (36b)

[IUPAC name (2R,3R,4R,5R)-2-(benzo[d]oxazol-2-ylamino)tetrahy-

dro-2H-pyran-4,5-diol] by transglycosylation of 2-amino-1,3-

benzoxazole (17a)

(H_2’a, overlapped by the [D₆]DMSO resonance), 2.02 (ddd,

³J(H_2’b,H_1’)=6.12, ³J(H_2’b,H_3’)=2.25, ²J(H_2’b,H_2’a)=13.21 Hz, 1H; H_2’b),

4.39 (brdt, ³J(H_3’,OH)=4.20, ³J(H_3’,H_4’)=3.0, ³J(H_3’,H_2’b’)=2.50 Hz,

1H; H_3’), 3.82 (dt, ³J(H_4’,H_3’)=3.00, ³J(H_4’,H_5’a)=³J(H_4’,H_5’b)=3.16 Hz,

1H; H_4’), 3.67 ppm (dt, ³J(H_5’a,OH)=4.58, ³J(H_5’a,H_4’)=³J(H_5’b,H_4’)=

3.16 Hz, 2H; H_5’a,b); ¹³C NMR (126 MHz, [D₆]DMSO): d=154 (C2),

142.83 (C7a), 132.74 (C4a), 120.68 (C7), 118.19 and 114.89 (C4, C5),

108.74 (C6), 86.67 (C4’), 83.53 (C1’), 70.41 (C3’), 61.01 (C5’),

37.86 ppm (C2’); UV/Vis (H₂O): l_max(e)=280 (6350), 243 (2000)

(pH 7), 282 (6250), 244 (7650) (pH>10), 280 (6450), 274 nm

(7050 mol^À1m³cm^À1) (pH<4); l_min(e)=259 (2000), 235 (5650)

(pH 7), 260 (1850) 230 (5500) (pH>10), 278 (6250), 246 nm

(750 mol^À1m³cm^À1) (pH<4); MS (ESI, positive ion mode): m/z (%)

calcd for C₁₂H₁₅O₃N₃: 249.11; found: 251.2 (9), [M+2H]⁺, 250.1

(100), [M+H]⁺, 135.1 (3), [base+2H]⁺, 134.2 (56) [base+H]⁺; MS

HR (ESI, positive ion mode): m/z (%) calcd for C₁₂H₁₅O₃N₃: 250.1180

[M⁺+H]; found: 250.1192.

A mixture of 2-amino-1,3-benzoxazazole (21a) (0.02 g, 2 mmol), 2’-

deoxyuridine (dU, 0.068 g, 4 mmol), UP (246 IU, preparation:

9 mgmL^À1, 100 Umg^À1), and PNP (149 IU, preparation: 17 mgmL^À1

,

54 Umg^À1

) in potassium phosphate buffer (20 mmol, 75 mL,

pH 7.0) was incubated at 558C for 744 h monitoring the formation

of the nucleoside 36b by HPLC (detection at l=280 nm, elution

with the CH₃CN/H₂O/TFA 70:30:0.1 (v/v/v), R_t: Ura=2.1, dU=2.5,

17a=7.0; 33b=8.9 min). The reaction mixture was evaporated to

3 mL and compound 36b was obtained after preparative HPLC on

reversed phase C18 silica gel (20250 mm², MZ-preparative). Nu-

cleoside 36b was eluted with a linear MeOH gradient 0!20% (v).

The fractions containing the nucleoside 36b were pooled and

evaporated to give the HPLC-pure product (0.003 g, 93%) in 8%

N2-(b-d-Ribopyranosyl)-2-amino-1,3-benzoxazole (36a) [IUPAC

name:

(2R,3R,4R,5R)-2-(benzo[d]oxazol-2-ylamino)tetrahydro-

2H-pyran-3,4,5-triol] by transglycosylation of 2-amino-1,3-ben-

zoxazole (21a)

1

3

yield. H NMR (700 MHz, [D₆]DMSO, 308C): d=8.62 (d, J(NH,H_1’)=

9.0 Hz, 1H; NH), 7.39 (d, ³J(H₄,H₅)=7.9 Hz, 1H; H₄), 7.29 (d,

³J(H₇,H₆)=7.8 Hz, 1H; H₇), 7.15 (dt, J(H₆,H₅)=³J(H₆,H₇)=7.9 Hz, 1H;

3

H₆), 7.04 (dt, ³J(H₅,H₄)=³J(H₅,H₆)=7.7 Hz, 1H; H₅), 5.0 (dt,

³J(H_1’,NH)=³J(H_1’,H_2’a)=9.8, ³J(H_1’,H_2’b)=2.2 Hz, 1H; H_1’), 4.81 (d,

³J(OH,H_3’)=5.6 Hz, 1H; 3’-OH), 4.48 (d, ³J(OH,H_4’)=3.2 Hz, 1H; 4’-

OH), ca. 3.73 (overlapped with low field doublet of the H_5areso-

nance; the following coupling constants are obtained from the res-

A mixture of 2-amino-1,3-benzoxazazole (21a) (0.02 g, 2 mmol), uri-

3

dine (U, 0.366 g, 20 mmol), UP (540 IU, preparation: 9 mgmL^À1

100 Umg^À1), and PNP (550.8 IU, preparation: 17 mgmL^À1

,

onance of vicinal protons and 3’-OH: J(H_3’,OH)=5.6, J(H_3’,H_2’)=10,

3

³J(H_3’,H_2’b)=4.1, J(H_3’,H_4’)<1 Hz, 1H; H_3’), 3.3-3.7 (brm, the follow-

Chem. Eur. J. 2015, 21, 13401 – 13419

www.chemeurj.org

13417

ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper

[13] a) E. Kulikowska, A. Bzowska, J. Wierzchowski, D. Shugar, Biochim. Bio-

phys. Acta 1986 , 874 , 35 5 –3 66; b) A. Bzowska, E. Kulikowska, D. Shugar,

Z. Naturforsch. C 1990, 45, 59–70.

[14] W. J. Hennen, C.-H. Wong, J. Org. Chem. 1989 , 54 , 469 2 –4 695.

[15] J. J. Voegel, M. M. Altorfer, S. A. Benner, Helv. Chim. Acta 1993 , 76 ,

206 1 –2 069.

[16] R. S. Esipov, A. I. Gurevich, D. V. Chuvikovsky, L. A. Chupova, T. I. Mura-

vyova, A. I. Miroshnikov, Protein Expression Purif. 2002 , 24 , 5 6 –6 0.

[17] J. Doskocil, A. Holy, Collect. Czech. Chem. Commun. 1977, 42, 370–383,

and references therein.

[18] I. Votruba, A. Holy, H. Dvorakova, J. Gunther, D. Hockova, H. Hrebabecky,

T. Cihlar, M. Masojidkova, Collect. Czech. Chem. Commun. 1994 , 59 ,

230 3 –2 330.

[19] a) H. Rosemeyer, F. Seela, J. Org. Chem. 1987 , 52 , 513 6 –5 14 3; F. Seela,

H. Rosemeyer in Recent Advances in Nucleosides: Chemistry and Chemo-

therapy (Ed.: C. K. Chu), Elsevier, Amsterdam, 2002, pp. 505–533.

[20] I. V. Fateev, K. V. Antonov, I. D. Konstantinova, T. I. Muravyova, F. Seela,

R. S. Esipov, A. I. Miroshnikov, I. A. Mikhailopulo, Beilstein J. Org. Chem.

2014 , 10 , 165 7 –1 669.

ing coupling constants are obtained from the resonance of vicinal

3

protons and 4’-OH: J(H_4’,OH)=3.2, J(H_4’,H_5’a)=2.8, J(H_4’,H_5’b)=1.1,

³J(H_4’,H_3’)<1 Hz, 1H; H_4’), 3.75 (dd, ²J(H_5’a,H_5’b)=12.0, ³J(H_5’a,H_4’)=

2

3

2.8 Hz, 1H; H_5’a), 3.48 (dd, J(H_5’b,H_5’a)=12, J(H_5’b,H_4’)=1.8 Hz, 1H;

H_5’b), 1.92 (td, ²J(H_2’a,H_2’b)=12.4, ³J(H_2’a,H_1’)=9.8, ³J(H_2’a,H_3’)=

10.0 Hz, 1H; H_2’a), 1.76 ppm (td, ²J(H_2’b,H_2’a)=12, ³J(H_2’b,H_1’)=2.2,

³J(H_2’b,H_3’)=4.1 Hz, 1H; H_2’b); ¹³C NMR (176 MHz, [D₆]DMSO, 308C):

d=161.50 (C2), 148.24 (C4a), 143.0 (C7a), 124.18 (C6), 121.05 (C5),

116.31 (C7), 109.15 (C4), 80.33 (C1’), 68.14 (C3’), 66.86 (C4’), 67.66

(C5’), 34.82 ppm (C2’); ¹⁵N NMR (71 MHz, [D₆][D₆]DMSO, 308C): d=

88.93 ppm (NH); UV/Vis (H₂O): l_max=270, 227 nm (pH 7); MS (ESI,

positive ion mode): m/z (%) calcd for C₁₂H₁₄O₄N₂: 251.1032; found:

251.10006 [M⁺+H], 273.08183 [M⁺+Na]; calcd for: 135.0540;

found: 135.05366 [base+H]⁺.

Acknowledgements

[21] T. Steiner, Angew. Chem. Int. Ed. 2002 , 41 , 4 8 –7 6; Angew. Chem. 2002 ,

114 , 5 0 –8 0.

The authors are thankful to the International Science and Tech-

nology Centre (ISIC, http://www.istc.ru, project #B-1640) for

the financial support of this work. I.A.M. is deeply thankful to

the Alexander von Humboldt-Stiftung (Bonn–Bad-Godesberg,

Germany) and the Byelorussian Republican Foundation for

Fundamental Research (http://www.fond.bas-net.by, project

No. X13MC-027) for partial financial support. We are indebted

to Dr. T. A. Balashova (Institute of Bioorganic Chemistry,

Moscow) for assistance in the analysis of the NMR spectra of

the new nucleosides.

[22] A. A. Granovsky, Firefly version 7.1.G, http://classic.chem.msu.su/gran/

firefly/index.html.

[23] M. W. Schmidt, K. K. Baldridge, J. A. Boatz, S. T. Elbert, M. S. Gordon, J. H.

Jensen, S. Koseki, M. Matsunaga, K. A. Nguyen, S. Su, T. L. Windus, M.

Dupuis, J. Comput. Chem. 1993 , 14 , 134 7 –1 363.

[24] a) F. Seela, K. Xu, Org. Biomol. Chem. 2007 , 5 , 303 4 –3 04 5; b) F. Seela, H.

Driller, Helv. Chim. Acta 1988 , 71 , 75 7 –7 61.

[25] M. Friedkin, J. Biol. Chem. 1950, 184, 449–460.

[26] H. Komatsu, H. Awano, J. Org. Chem. 2002 , 67 , 541 9 –5 421.

[27] I. A. Mikhailopulo, S. Lapinjoki, Biocatalytic Synthesis of Pharmaceutical-

ly Important Nucleosides: Proceedings of the FinMed 2006 (Eds.: S. Sor-

vari, O. Toldi), Saariselkä, Lapland (Finland), 27–31 March, 2006, ISBN-

13: 978-952-99302-4-1, pp. 42–51.

[28] J. Roivainen, T. Elizarova, S. Lapinjoki, I. A. Mikhailopulo, R. S. Esipov, A. I.

Miroshnikov, Nucleosides Nucleotides Nucleic Acids 2007 , 26 , 90 5 –9 09.

[29] I. D. Konstantinova, K. V. Antonov, I. V. Fateev, A. I. Miroshnikov, V. A.

Stepchenko, A. V. Baranovsky, I. A. Mikhailopulo, Synthesis 2011, 1555–

1560.

Keywords: ab initio calculations

glycosylation · nucleobases · purine analogues · substrate

properties

·

escherichia coli

·

[30] A. I. Kulikovskiy, R. S. Esipov, A. I. Miroshnikov, I. A. Mikhailopulo, Collect.

Symp. Ser. 2011, 12, 397–240.

[1] V. A. Stepchenko, F. Seela, R. S. Esipov, A. I. Miroshnikov, Y. A. Sokolov,

I. A. Mikhailopulo, Syn. Lett. 2012, 23, 1541–1545.

[2] E. M. Bennett, C. Li, P. W. Allan, W. B. Parker, S. E. Ealick, J. Biol. Chem.

2003 , 278 , 4711 0 –4 7118.

[3] Y. Zhang, W. B. Parker, E. J. Sorscher, S. E. Ealick, Curr. Topics Med. Chem.

2005, 5, 1259–1274.

[4] a) I. A. Mikhailopulo, A. I. Miroshnikov, Acta Naturae 2010, 2, 38–61;

b) I. A. Mikhailopulo, A. I. Miroshnikov, Mendeleev Commun. 2011 , 21 ,

5 7 –6 8.

[5] T. A. Krenitsky, G. B. Elion, R. A. Strelitz, G. H. Hitchings, J. Biol. Chem.

1967, 242, 2675–2682.

[6] M. D. Erion, J. D. Stoeckler, W. C. Guida, R. L. Walter, S. E. Ealick, Biochem-

istry 1997 , 36 , 1173 5 –1 1748.

[7] a) A. Bzowska, E. Kulikowska, D. Shugar, J. Clin. Pharm. Ther. 2000, 88,

349–425; b) A. Bzowska, E. Kulikowska, D. Shugar, Biochim. Biophys.

Acta 1992 , 1120 , 23 9 –2 47; c) M. R. Sheen, B. K. Kim, R. E. Parks Jr., Mol.

Pharmacol. 1968, 4, 293–299; d) B. Wielgus-Kutrowska, E. Kulikowska, J.

Wierzchowski, A. Bzowska, D. Shugar, Eur. J. Biochem. 1997 , 243 , 40 8 –

417.

[31] I. D. Konstantinova, O. M. Selezneva, I. V. Fateev, T. A. Balashova, S. K. Ko-

tovskaya, Z. M. Baskakova, V. N. Charushin, A. V. Baranovsky, A. I. Mirosh-

nikov, J. Balzarini, I. A. Mikhailopulo, Synthesis 2013, 272–280.

[32] I. A. Mikhailopulo, Curr. Org. Chem. 2007, 11, 317–335.

[33] I. A. Mikhailopulo, A. I. Zinchenko, S. B. Bokut, N. V. Dudchik, V. N. Barai,

E. N. Kalinichenko, H. Rosemeyer, F. Seela, Biotechnol. Lett. 1992 , 14 ,

88 5 –8 90.

[34] I. A. Mikhailopulo, Z. Kazimierczuk, A. I. Zinchenko, V. N. Barai, V. V. Ro-

manova, L. A. Eroshevskaya, Nucleic Acids Symp. Ser. 1994, 31, 83.

[35] I. A. Mikhailopulo, Z. Kazimierczuk, A. I. Zinchenko, V. N. Barai, V. V. Ro-

manova, L. A. Eroshevskaya, Nucleosides Nucleotides 1995 , 14 , 47 7 –4 80.

[36] R. G. Pearson, Proc. Natl. Acad. Sci. USA 1986 , 83 , 844 0 –8 441.

[37] G. Desimoni, P. P. Righetti, E. Selva, G. Tacconi, V. Riganti, M. Specchiarel-

lo, Tetrahedron 1977 , 33 , 282 9 –2 836.

[38] S. J. Gho, Bull. Korean Chem. Soc. 2003, 24, 731–732, and references

therein.

[39] D. L. Heyl, S. Fernandes, L. Khullar, J. Stephens, E. Blaney, H. Opang-

Owusu, B. Stahelin, T. Pasko, J. Jacobs, D. Bailey, D. Brown, M. C. Milletti,

Bioorg. Med. Chem. 2005 , 13 , 426 2 –4 268.

[40] Z. Zhou, R. G. Parr, J. Am. Chem. Soc. 1990 , 112 , 572 0 –5 725.

[41] W. J. Ehlhardt, W. J. Wheeler, A. P. Breau, S. H. Chay, G. M. Birch, Drug

Metab. Dispos. 1993, 21, 162–170.

[8] G. Koellner, M. Luic, D. Shugar, W. Saenger, A. Bzowska, J. Mol. Biol.

1998 , 280 , 15 3 –1 66.

[9] C. Mao, W. J. Cook, M. Zhou, G. W. Koszalka, T. A. Krenitsky, S. E. Ealick,

Structure 1997 , 5 , 137 3 –1 383.

[42] F. G. Hayden, A. R. Tunkel, J. J. Treanor, R. F. Betts, S. Allerheiligen, J.

Harris, Antimicrob. Agents Chemother. 1994 , 38 , 117 8 –1 181.

[43] a) G. Stoychev, B. Kierdaszuk, D. Shugar, Eur. J. Biochem. 2002 , 269 ,

404 8 –4 057; b) E. Kulikowska, B. Kierdaszuk, D. Shugar, Acta Biochim.

Pol. 2004, 51, 493–531.

[10] G. Koellner, A. Bzowska, B. Wielgus-Kutrowska, M. Luic, T. Steiner, W.

Saenger, J. Stepinski, J. Mol. Biol. 2002 , 315 , 35 1 –3 71.

[11] W. Ye, D. Paul, L. Gao, J. Seckute, R. Sangaiah, K. Jayaraj, Z. Zhang, P. A.

Kaminski, S. E. Ealick, A. Gold, L. M. Ball, PLoS ONE 2014, 9, DOI:

10.1371/journal.pone.0115082.

´

[44] E. D. Raczynska, M. Makowski, M. Szela˛g, B. Kaminska, K. Zientara, J. Mol.

[12] G. Mikleusevic, Z. Stefanic, M. Narczyk, B. Wielgus-Kutrowska, A. Bzow-

ska, M. Luic, Biochimie 2011, 93, 1610–1622.

Structure (THEOCHEM) 2010, 947, 83–91.

Chem. Eur. J. 2015, 21, 13401 – 13419

www.chemeurj.org

13418

ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper

ˇ

[45] J. A. Maurins, R. A. Paegle, M. J. L¯ıdaks, E. I. Kvasyuk, P. V. Kuzmichkin,

[53] J. A. Montgomery, J. H. Thomas, J. Org. Chem. 1971 , 36 , 196 2 –1 967.

[54] J. A. Kelley, J. S. Driscoll, J. J. McCormack, J. S. Roth, V. E. Marquez, J.

Med. Chem. 1986 , 29 , 235 1 –2 358.

[55] V. V. Mossine, C. L. Barnes, M. S. Feather, T. P. Mawhinney, J. Am. Chem.

Soc. 2002 , 124 , 1517 8 –1 5179.

[56] V. V. Mossine, T. P. Mawhinney, Adv. Carbohydr. Chem. Biochem. 2010 , 64 ,

29 1 –4 02.

[57] H. M. Kalckar, J. Biol. Chem. 1947, 167, 477–486.

[58] P. E. Plesner, H. Klenow, Preparation of ribose-1-phosphate in Methods in

Enzymology, Vol. 3 (Eds.: S. P. Colowick. N. O. Kaplan), Academic Press,

New York, 1957, pp. 181–182.

I. A. Mikhailopulo, Nucleosides Nucleotides 1986, 5, 79–93.

[46] J. D. Stoeckler, J. B. Ryden, R. E. Parks, Jr., M.-Y. Chu, M.-I. Lim, W.-Y. Ren,

R. S. Klein, Cancer Res. 1986, 46, 1774–1778.

[47] A. Bzowska, E. Kulikowska, N. E. Poopeiko, D. Shugar, Eur. J. Biochem.

1996 , 239 , 22 9 –2 34.

[48] K. Temburnikar, K. Brace, K. L. Seley-Radtke, J. Org. Chem. 2013 , 78 ,

730 5 –7 311.

[49] K. W. Temburnikar, S. C. Zimmermann, N. T. Kim, C. R. Ross, C. Gelbmann,

C. E. Salomon, G. M. Wilson, J. Balzarini, K. L. Seley-Radtke, Bioorg. Med.

Chem. 2014 , 22 , 211 3 –2 122.

[50] C. Thibaudeau, P. Acharya, J. Chattopadhyaya, Stereoelectronic effects in

nucleosides and nucleotides and their structural implications, 2nd ed.„

Department of bioorganic chemistry, Uppsala University, Uppsala, 2005,

p. 166.

[59] J. He, F. Seela, Nucleic Acids Res. 2002 , 30 , 548 5 –5 496.

[51] H. M. Kalckar, Biochim. Biophys. Acta 1953 , 12 , 25 0 –2 64.

[52] A. E. Nixon, J. L. Hunter, G. Bonifacio, J. Eccleston, M. R. Webb, Anal. Bio-

chem. 1998 , 265 , 29 9 –3 07.

Received: April 5, 2015

Published online on July 31, 2015

Chem. Eur. J. 2015, 21, 13401 – 13419

www.chemeurj.org

13419

ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Article Doi

Recognition of Artificial Nucleobases by E. coli Purine Nucleoside Phosphorylase versus its Ser90Ala Mutant in the Synthesis of Base-Modified Nucleosides

DOI: 10.1002/chem.201501334

Source and publish data:

Authors:

Article abstract of DOI:10.1002/chem.201501334

Full text of DOI:10.1002/chem.201501334

Products guided by the article

R&D Labs maybe for 16220-07-8

Relevant to this article

Hot Product