Structure-based design of guanosine analogue inhibitors targeting GTP cyclohydrolase IB towards a new class of antibiotics

Article abstract of DOI:10.1016/j.bmcl.2019.126818

GTP cyclohydrolase (GCYH-I) is an enzyme in the folate biosynthesis pathway that has not been previously exploited as an antibiotic target, although several pathogens including N. gonorrhoeae use a form of the enzyme GCYH-IB that is structurally distinct from the human homologue GCYH-IA. A comparison of the crystal structures of GCYH-IA and -IB with the nM inhibitor 8-oxo-GTP bound shows that the active site of GCYH-IB is larger and differently shaped. Based on this structural information, we designed and synthesized a small set of 8-oxo-G derivatives with ether linkages at O⁶ and O⁸ expected to displace water molecules from the expanded active site of GCYH-IB. The most potent of these compounds, G3, is selective for GCYH-IB, supporting the premise that potent and selective inhibitors of GCYH-IB could constitute a new class of small molecule antibiotics.

Full text of DOI:10.1016/j.bmcl.2019.126818

Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxxx
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Structure-based design of guanosine analogue inhibitors targeting GTP
cyclohydrolase IB towards a new class of antibiotics
George N. Samaana, Naduni Paranagamaa, Ayesha Haquea, David A. Hechtb,c
,
⁎
⁎
Manal A. Swairjoa, , Byron W. Pursea,
a Department of Chemistry and Biochemistry and the Viral Information Institute, San Diego State University, San Diego, CA 92182, USA
b Department of Chemistry and Biochemistry, San Diego State University, San Diego, CA 92182, USA
c Department of Chemistry, Southwestern College, Chula Vista, CA 92120, USA
A R T I C L E I N F O
A B S T R A C T
Keywords:
Antibiotic
Resistance
Folate
GTP cyclohydrolase (GCYH-I) is an enzyme in the folate biosynthesis pathway that has not been previously
exploited as an antibiotic target, although several pathogens including N. gonorrhoeae use a form of the enzyme
GCYH-IB that is structurally distinct from the human homologue GCYH-IA. A comparison of the crystal struc-
tures of GCYH-IA and -IB with the nM inhibitor 8-oxo-GTP bound shows that the active site of GCYH-IB is larger
and differently shaped. Based on this structural information, we designed and synthesized a small set of 8-oxo-G
derivatives with ether linkages at O6 and O8 expected to displace water molecules from the expanded active site
of GCYH-IB. The most potent of these compounds, G3, is selective for GCYH-IB, supporting the premise that
potent and selective inhibitors of GCYH-IB could constitute a new class of small molecule antibiotics.
GTP
Nucleoside
Antibiotic resistance is a growing global health crisis, exacerbated
by many factors.1 The economics of antibiotic development are less
favorable than those of drug development against chronic diseases and
resistance to new antibiotics now typically emerges within two years of
clinical approval, a result of horizontal transfer, high bacterial mutation
rates, and the prevalence of antibiotic overuse and misuse.2 For ex-
ample, many cases of treatment failure with third generation cepha-
losporins, the final option of treatment for gonorrhea, have been re-
ported.3 Therefore, to fight against the high resistance to antibiotics,
novel classes of drugs with new mechanisms of action must be devel-
oped. Ongoing efforts by many academic and industrial labs continue to
advance new antibiotic compounds towards the clinic, but as many
recent reviews have highlighted, the vast majority of these compounds
are synthetically modified derivatives of proven antibiotic scaf-
folds.1,2,4,5
dihydropteroate synthetase, an enzyme downstream of DHFR in the
folate biosynthesis pathway. Inhibition of folate biosynthesis depletes
this essential cofactor in prokaryotes, disrupting the synthesis of pur-
ines, thymidylate, pantothenate, glycine, serine, and methionine, and
resulting in cell death.9 Because of the clinical success of antifolate
antibiotics and the emerging resistance to these drugs, we believe that a
new class of antibiotics targeting an unexploited enzyme of folate
biosynthesis, GTP cyclohydrolase IB, would present a prime opportu-
nity to target resistant pathogens that rely on this enzyme.
GTP cyclohydrolase I (GCYH-I) is the first enzyme of the prokaryotic
de novo biosynthetic pathway to folate (Fig. 1). It has been deemed an
unsuitable drug target because prokaryotic GCYH-I and human GCYH-I,
which catalyzes
a step in the essential biopterin biosynthesis
pathway,10 are highly homologous. In 2006, however, a prokaryote-
specific type I GTP cyclohydrolase subfamily was discovered in a
bioinformatic analysis that revealed the absence of the folE gene en-
coding the canonical GCYH-I in a number of clinically important pa-
thogens, all of which possess the remaining genes for the folate bio-
synthesis pathway.11 Further investigation showed that these microbes
express an alternative GCYH-I enzyme that exhibits virtually no se-
quence homology to the canonical enzyme, yet carries out the same
catalytic function.12 The new enzyme is prokaryote-specific and was
named GCYH-IB (and the corresponding gene folE2) while the canonical
The folate biosynthesis pathway is an established therapeutic target
for antibiotics.6 Drugs that block this pathway, specifically trimetho-
prim and sulfamethoxazole, are listed on the WHO’s 2017 Model List of
Essential Medicines. The trimethoprim-sulfamethoxazole combination
is widely used as a first-line treatment for methicillin-resistant strains of
S. aureus, but high levels of resistance to trimethoprim have emerged
from a F98Y point mutation in prokaryotic dihydrofolate reductase
(DHFR), its enzymatic target.7,8 Sulfamethoxazole inhibits
⁎ Corresponding authors.
E-mail addresses: mswairjo@sdsu.edu (M.A. Swairjo), bpurse@sdsu.edu (B.W. Purse).
https://doi.org/10.1016/j.bmcl.2019.126818
Received 15 October 2019; Received in revised form 7 November 2019; Accepted 8 November 2019
0960-894X/©2019ElsevierLtd.Allrightsreserved.
Pleasecitethisarticleas:GeorgeN.Samaan,etal.,Bioorganic&MedicinalChemistryLetters,https://doi.org/10.1016/j.bmcl.2019.126818

G.N. Samaan, et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxxx
Fig. 1. Inhibition of prokaryote-specific GCYH-IB, the first enzyme in the de
novo tetrahydrofolate (THF) biosynthesis pathway, is proposed for the creation
of a new class of antifolate antibiotics.
enzyme was renamed GCYH-IA (and its gene folE1). GCYH-IB provides
the sole enzymatic activity for the first step of folate biosynthesis in the
clinically important pathogen genera Staphylococcus and Neisseria,
making it an attractive and novel drug target.11 Indeed yciA, the S.
aureus gene encoding GCYH-IB, is an essential gene, and in bacteria that
possess both the IA and IB enzymes, GCYH-IB, which uses Zn2+ for
catalysis, is essential in a ΔfolE background or when Zn2+ is lim-
iting.12,13
Fig. 2. Surface representations of the active site cavities of (A) N. gonorrhoeae
GCYH-IB (PDB ID 5 K95),14 and (B) T. thermophilus GCYH-IA (PDB ID 1WUQ),15
both harboring bound 8-oxo-GTP and Zn2+, showing the additional space
available in Pockets 1 and 2 of GCYH-IB. 8-oxo-GTP is shown in stick re-
presentation. The metal ion and water molecules are shown as yellow and red
spheres, respectively. For additional representations of these cavities, see Figs.
S1 and S2 in the Supplementary Data. (For interpretation of the references to
colour in this figure legend, the reader is referred to the web version of this
article.)
Both GCYH-IA and -IB catalyze the conversion of GTP to 7,8-dihy-
droneopterin triphosphate (H2NTP; Fig. 1), a multistep reaction that
begins with addition of water to C8 of the guanine nucleobase, includes
the elimination of formaldehyde, and leads to the incorporation of C1′
and C2′ of ribose into the 7,8-dihydroneopterin heterocycle following
Amadori rearrangement.14–17 8-oxo-GTP binds strongly in the active
site of GCYH-I but is not a substrate for this reaction because it displaces
the nucleophilic water molecule from the active site and fills the va-
cated space with its C8-carbonyl group, which coordinates to the active
site Zn2+ 14–16,18,19
Against N. gonorrhoeae GCYH-IB, 8-oxo-GTP is the
.
most potent known inhibitor with Ki = 150 nM, but this compound has
a 28-fold lower Ki = 5.4 nM against T. thermophilus GCYH-IA.14,15
Crystallographic studies show that both GCYH-IA and -IB are members
of the tunneling-fold (T-fold) structural superfamily, but the active sites
are shaped significantly differently.12,14,15 A comparison of the crystal
structures of N. gonorrhoeae GCYH-IB and T. thermophilus GCYH-IA (the
only available crystal structures of enzymes from each GCYH-I sub-
family that contain bound 8-oxo-GTP) identifies three predominant
regions of difference that could be exploited to improve inhibitor se-
lectivity (Figs. 2, S1, and S2). The largest difference is in the region that
we name Pocket 1 (size ~ 40 Å3), a site that is occupied by two water
molecules when 8-oxo-GTP is bound to GCYH-IB.14 This pocket is ex-
pected to be the easiest to address synthetically, because it projects
directly outward from O6 of the 8-oxoguanine nucleobase. Pocket 2
(size ~ 100 Å3) is predominantly above the plane of the nucleobase (as
oriented in Fig. 2). Pocket 3 is the opening of the active site and ac-
commodates the ribose and phosphate groups. The glycosidic bond
angle is anti when 8-oxo-GTP is bound to GCYH-IA and syn in GCYH-IB,
resulting in a substantially different conformation of the inhibitor.
Based on these crystallographically observed active-site differences,
we propose that a new class of antifolate antibiotics can be developed
by modifying the structure of 8-oxo-GTP so as to enhance potency
against bacterial GCYH-IB and ablate binding to human GCYH-IA,
which exhibits 45% overall sequence identity to T. thermophilus GCYH-
IA (70% similarity) and identical active site residues and 3D structure
(r.m.s.d. 0.86 Å over 817 Cα atoms, see supplementary Fig. S1). We set
Fig. 3. Prototype inhibitor designs to probe the filling of Pockets 1 and 2 of
GCYH-IB. For G1, G3, and S-G3, R2 = H. For G2, R1 = H.
out to design a small set of test compounds with increasingly large
substituents oriented towards the larger active site pockets 1 and 2 of
GCYH-IB (Fig. 3). To build up the inhibitor structure in the direction of
Pocket 1, we envisioned modifying the enol tautomer at O6 of 8-ox-
oguanine as a set of ethers decorated with alcohols to replace the hy-
drogen bonding interactions of the two water molecules that fill Pocket
1 when 8-oxo-GTP is bound. This enol ether design converts N1 from a
protonated hydrogen bond donor to a hydrogen bond acceptor. Because
of the position of N1 near Glu216, we anticipate that this change from
2

G.N. Samaan, et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxxx
N. gonorrhoeae GCYH-IB (NgGCYH-IB) and a H. sapiens GCYH-IA
(HsGCYH-IA) construct lacking the 42 N-terminal amino acids and
previously reported to possess robust activity and improved solubility
(for details on enzyme overexpression and purification, see the
Supplementary Data).19 Enzyme activity was measured using two pre-
viously established assays: an absorbance-based assay quantifying
product H2NTP formation by its absorption at 330 nm; and a fluores-
cence-based assay that relies on post-reaction oxidation of the enzy-
matic product H2NTP to the fluorescent neopterin, and monitoring the
neopterin emission peak at 446 nm with excitation at 365 nm (details of
the assays are provided in the Supplementary Data).14 First, steady-
state kinetic analysis of HsGCYH-IA was performed using the fluores-
cence assay with GTP as substrate, and the analysis yielded a KM of
867
57 µM (details are in Supplementary Data). Inhibition assays
were then conducted by pre-incubating each enzyme with increasing
concentrations of inhibitor compound (0–1600 µM) for 30 min, and
measuring enzyme activity using the absorbance assay (for G1, G2 and
S-G3) or the fluorescence assay (for G3), and a substrate concentration
set at or near the enzyme’s KM for GTP: 867 µM for HsGCYH-IA (this
study), and 9 µM for NgGCYH-IB (experimental details are in the
Supplementary Data; see also Fig. S3).14 All enzymatic assays were
performed in triplicate.
IC50 values obtained from the inhibition data show that G1 has
modest selectivity for HsGCYH-IA, G2 is not selective, and both G3
compounds are modestly selective for NgGCYH-IB, with G3 showing the
greater selectivity (Table 1). G3 is three-fold more potent against
NgGCYH-IB, a 31-fold reversal of selectivity as compared with our
starting point, 8-oxo-GTP. Previous work shows that 8-oxo-GTP is 28-
fold more potent against GCYH-IA from T. thermophilus, which harbors
a nearly identical active site to HsGCYH-IA in sequence and 3D struc-
ture.14,15 While these IC50 measurements are indicative of only modest
potency, the 31-fold reversal of selectivity in favor of the bacterial
enzyme is significant and supports our structure-guided hypothesis on
inhibitor optimization. We note also that G1–3 lack the phosphate
groups of 8-oxo-GTP, which engage in ion pairing with arginine and
lysine residues of the enzyme, leading to the phosphorylated inhibitor’s
sub-μM affinity.14 Introducing one or more phosphate groups, or a
suitable phosphate surrogate, in a future round of inhibitor refinement
is expected to enhance the potency of G3 and related, future deriva-
tives.
Scheme 1. Synthesis of nucleoside analogue inhibitors. Reagents and condi-
tions: (a) sodium benzyloxide, DMSO, 65 °C, 16 h (65%) (to make 2); (b) so-
dium methoxide, DMSO, 50 °C, 16 h (39%) (to make G2); (c) TBDMSCl, imi-
dazole, DMF, 50 °C, 5 h (65%); (d) (i) PPh3, dioxane, 80 °C, 45 min, 2-
hydroxyethyl acetate (to make 4) or (S)-2,3-O-isopropylidene glycerol (to make
8) or (R)-2,3-O-isopropylidene glycerol (to make 11) (ii) DIAD, 60 °C, 2 h; (e)
1 M TBAF, THF, rt, 6 h (41% over 2 steps); (f) (i) 30% sodium methoxide in
methanol, rt, 48 h, (ii) 2% HCl in methanol and H2O (61% for G1); (g) 2% HCl
in methanol and H2O (61% for G3, 51% for S-G3).
donor to acceptor could be accommodated by a change in the proto-
nation state of the neighboring glutamate (see below).
In silico docking studies were performed in which we docked 8-oxo-
GTP and G3 into the GTP binding sites of the x-ray crystal structures of
NgGCYH-IB (PDB ID 5K95),14 and HsGCYH-IA (PDB ID 1FB1).19 For the
binding of G3, we performed docking both with and without the car-
boxylate of Glu216 protonated, and found a significantly better fit of G3
with the side chain protonated. This protonation change is predicted to
accommodate the change of G3′s N1 from hydrogen bond donor to
acceptor in the course of synthesizing its enol ether appendage. The
bound poses of 8-oxo-GTP docked into NgGCYH-IB as well as HsGCYH-
IA are consistent with the inhibitor’s crystallographically observed
mode of binding to NgGCYH-IB. Likewise, the conformation of the pose
of G3 docked into the active site of NgGCYH-IB is consistent with that of
8-oxo-GTP as seen bound in the crystal structure to NgGCYH-IB (Fig. 4).
To synthesize the first compound G1, 8-bromoguanosine was used
as the starting material (Scheme 1). Substitution of the bromine atom
with a benzyloxy group was performed using freshly prepared sodium
benzyloxide under conditions suitable for nucleophilic aromatic sub-
stitution (SNAr). The resulting benzyl ether 2 was then allowed to react
with tert-butyldimethylsilyl chloride (TBDMSCl) under basic conditions
to protect the ribose hydroxyl groups, which would otherwise interfere
with the Mitsunobu reaction. The protected nucleoside 3 was then
treated with triphenylphosphine, diisopropylazodicarboxylate (DIAD)
and 2-hydroxyethyl acetate under Mitsunobu conditions to give 4. Re-
moval of the TBDMS groups was performed using tetrabutylammonium
fluoride (TBAF) followed by removal of the acetyl and benzyloxy
groups using basic and acidic conditions, respectively, to give the final
product G1. G3 and its epimer S-G3 were synthesized using the same
strategy as G1, using different substrates in the Mitsunobu reaction.
Starting from the protected nucleoside 3, (S)- or (R)-2,3-O-iso-
propylidene glycerol was added to the reaction mixture along with PPh3
and DIAD to give the acetal-protected Mitsunobu products 8 and 11,
respectively. Treating 8 and 11 with TBAF and then aqueous acid
provided the target compounds G3 and S-G3, respectively. For the
synthesis of G2, 8-bromoguanosine was subjected to an SNAr reaction
using freshly prepared sodium methoxide to replace the bromine with a
methoxy group, giving the final product G2.
Table 1
Measured half maximal inhibitory concentrations (IC50) of designed inhibitors
against HsGCYH-IA and NgGCYH-IB. Values are presented with respect to their
standard errors.
Inhibitor
IC50 (µM)
HsGCYH-IA
NgGCYH-IB
G1
221
514
372
409
10
32
14
21
413
372
164
134
6
G2
25
11
11
With inhibitors G1–3 in hand, we measured their half-maximal in-
hibitory concentrations (IC50) in vitro against heterologously expressed
S-G3
G3
3

G.N. Samaan, et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxxx
Fig. 4. A Bound pose of G3 (carbon in gray) superimposed on 8-oxo-GTP (carbon in orange) as seen bound in the active site in the crystal structure of NgGCYH-IB
(5 K95). B Key protein-ligand interactions between the bound pose of G3 and NgGCYH-IB. (For interpretation of the references to colour in this figure legend, the
reader is referred to the web version of this article.)
However, because of steric clashes in Pocket 1 with Q182, H210 and
L165, we were not able to obtain a bound pose of G3 in HsGCYH-IA.
These molecular modeling studies support the preferential inhibition of
NgGCYH-IB by G3 on the grounds that its projections into Pocket 1 are
too large to be accommodated by HsGCYH-IA.
characterization data, enzyme expression and assay descriptions, and
methods for docking studies, is available online at https://doi.org/10.
1016/j.bmcl.2019.126818.
References
GCYH-IB is a prokaryotic enzyme in the folate biosynthesis pathway
that is essential for the proliferation of some pathogens, including
Staphylococcus and Neisseria. Based on the architecture of its active site
in comparison with the human orthologue GCYH-IA, we identified two
active site regions, Pockets 1 and 2, that are larger and geometrically
distinct in GCYH-IB. The use of a novel synthetic route allowed for the
preparation of a small set of four 8-oxo-G analogue inhibitors that build
into these active site pockets. Two of the analogues, G3 and S-G3, in-
vert the selectivity of the parent inhibitor 8-oxo-GTP by up to 31-fold in
the case of G3, now providing three-fold greater potency against
NgGCYH-IB as compared with the human enzyme. Docking studies
support the structure-based rationale for the inhibitor design and the
observed change in selectivity for the bacterial enzyme. The glycerol
ether of G3 projects into Pocket 1 where it is accommodated, but its
bulk reduces compatibility and potency against the human enzyme.
While these new inhibitors are not potent enough to be drug leads, their
structure-guided design and the observation of the predicted change in
selectivity to target the bacterial enzyme supports our premise that
GCYH-IB can be exploited as a target for a new class of antibiotics.
1. Friedman D, Alper J. Technological Challenges in Antibiotic Discovery and Development:
A Workshop Summary. Washington, DC: The National Academies Press; 2014https://
doi.org/10.17226/18616.
2. Fischbach MA, Walsh CT. Antibiotics for emerging pathogens. Science.
2009;325:1089–1093. https://doi.org/10.1126/science:1176667.
3. Gelband H, Miller-Petrie M, Pant S, et al. The State of the World’s Antibiotics.
Washington, D.C: CDDEP; 2015.
4. Blair JMA, Webber MA, Baylay AJ, Ogbolu DO, Piddock LJV. Molecular mechanisms
of antibiotic resistance. Nat Rev Microbiol. 2014;13:42. https://doi.org/10.1038/
nrmicro3380.
5. Hofer U. The cost of antimicrobial resistance. Nat Rev Microbiol. 2019;17:3. https://
doi.org/10.1038/s41579-018-0125-x.
6. Bourne CR. Utility of the biosynthetic folate pathway for targets in antimicrobial
discovery. Antibiotics. 2014;3:1–28. https://doi.org/10.3390/antibiotics3010001.
7. Dale GE, Broger C, D’Arcy A, et al. A single amino acid substitution in Staphylococcus
aureus dihydrofolate reductase determines trimethoprim resistance 1 1 Edited by T.
Richmond. J Mol Biol. 1997;266(1):23–30. doi:10.1006/jmbi.1996.0770.
8. Heaslet H, Harris M, Fahnoe K, et al. Structural comparison of chromosomal and
exogenous dihydrofolate reductase from Staphylococcus aureus in complex with the
potent inhibitor trimethoprim. Proteins Struct Funct Bioinforma. 2009;76:706–717.
https://doi.org/10.1002/prot.22383.
9. Scocchera E, Wright DL. The Antifolates. In: Fisher JF, Mobashery S, Miller MJ, eds.
Cham: Springer International Publishing; 2018:123-149. doi:10.1007/7355_2017_16.
10. Thöny B, Auerbach G, Blau N. Tetrahydrobiopterin biosynthesis, regeneration and
functions. Biochem J. 2000;347:1–16. https://doi.org/10.1042/bj3470001.
11. El Yacoubi B, Bonnett S, Anderson JN, Swairjo MA, Iwata-Reuyl D, De Crécy-Lagard
V. Discovery of a new prokaryotic type I GTP cyclohydrolase family. J Biol Chem.
2006;281:37586–37593. https://doi.org/10.1074/jbc.M607114200.
12. Sankaran B, Bonnett SA, Shah K, et al. Zinc-independent folate biosynthesis: Genetic,
biochemical, and structural investigations reveal new metal dependence for GTP
cyclohydrolase IB. J Bacteriol. 2009;191:6936–6949. https://doi.org/10.1128/JB.
00287-09.
Declaration of Competing Interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influ-
ence the work reported in this paper.
13. Chaudhuri RR, Allen AG, Owen PJ, et al. Comprehensive identification of essential
Staphylococcus aureus genes using Transposon-Mediated Differential Hybridisation
(TMDH). BMC Genomics. 2009;10:291. https://doi.org/10.1186/1471-2164-10-291.
14. Paranagama N, Bonnett SA, Alvarez J, et al. Mechanism and catalytic strategy of the
prokaryotic-specific GTP cyclohydrolase-IB. Biochem J. 2017;474:1017–1039.
https://doi.org/10.1042/BCJ20161025.
Acknowledgments
Financial support for this research was provided by San Diego State
University; the National Institutes of Health [GM110588 to M.A.S.];
and the California Metabolic Research Foundation.
15. Tanaka Y, Nakagawa N, Kuramitsu S, Yokoyama S, Masui R. Novel reaction me-
chanism of GTP cyclohydrolase I. High-resolution X-ray crystallography of thermus
thermophilus HB8 enzyme complexed with a transition state analogue, the 8-ox-
oguanine derivative. J Biochem. 2005;138:263–275. https://doi.org/10.1093/jb/
mvi120.
Appendix A. Supplementary data
16. Narp H, Huber R, Auerbach G, et al. Active site topology and reaction mechanism of
GTP cyclohydrolase I. Proc Natl Acad Sci U S A. 1995;92:12120–12125. https://doi.
org/10.1073/pnas.92.26.12120.
Supplementary data to this article, including supplementary figures,
general experimental information, synthetic procedures and compound
4

G.N. Samaan, et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxxx
17. Yaylayan VA, Huyghues-Despointes A, Feather MS. Chemistry of Amadori re-
arrangement products: analysis, synthesis, kinetics, reactions, and spectroscopic
properties. Crit Rev Food Sci Nutr. 1994;34:321–369. https://doi.org/10.1080/
10408399409527667.
Biochim Biophys Acta. 1969;192:468–478. https://doi.org/10.1016/0304-4165(69)
90396-1.
19. Auerbach G, Herrmann A, Bracher A, et al. Zinc plays a key role in human and
bacterial GTP cyclohydrolase I. Proc Natl Acad Sci. 2000;97:13567–13572. https://
doi.org/10.1073/pnas.240463497.
18. Wolf WA, Brown G. The biosynthesis of folic acid. X. Evidence for an Amadori re-
arrangement in the enzymatic formation of dihydroneopterin triphosphate from GTP.
5