Mass spectrometric analysis of purine de novo biosynthesis intermediates

Article abstract of DOI:10.1371/journal.pone.0208947

Purines are essential molecules for all forms of life. In addition to constituting a backbone of DNA and RNA, purines play roles in many metabolic pathways, such as energy utilization, regulation of enzyme activity, and cell signaling. The supply of purines is provided by two pathways: the salvage pathway and de novo synthesis. Although purine de novo synthesis (PDNS) activity varies during the cell cycle, this pathway represents an important source of purines, especially for rapidly dividing cells. A method for the detailed study of PDNS is lacking for analytical reasons (sensitivity) and because of the commercial unavailability of the compounds. The aim was to fully describe the mass spectrometric fragmentation behavior of newly synthesized PDNS-related metabolites and develop an analytical method. Except for four initial ribotide PDNS intermediates that preferentially lost water or phosphate or cleaved the forming base of the purine ring, all the other metabolites studied cleaved the gly-cosidic bond in the first fragmentation stage. Fragmentation was possible in the third to sixth stages. A liquid chromatography-high-resolution mass spectrometric method was developed and applied in the analysis of CRISPR-Cas9 genome-edited HeLa cells deficient in the individual enzymatic steps of PDNS and the salvage pathway. The identities of the newly synthesized intermediates of PDNS were confirmed by comparing the fragmentation patterns of the synthesized metabolites with those produced by cells (formed under pathological conditions of known and theoretically possible defects of PDNS). The use of stable isotope incorporation allowed the confirmation of fragmentation mechanisms and provided data for future fluxomic experiments. This method may find uses in the diagnosis of PDNS disorders, the investigation of purinosome formation, cancer research, enzyme inhibition studies, and other applications.

Full text of DOI:10.1371/journal.pone.0208947

RESEARCH ARTICLE
Mass spectrometric analysis of purine de novo
biosynthesis intermediates
1,2
Lucie Madrova , Matyas Krijt , Veronika Baresova , Jan Vaclavık
3
3
1,2
´
David Friedecky
´
´ ˇ
ˇ
´
´
´
,
ID
1,2,4
1
3
3
ˇ
´
ˇ ´ ´ ´ ´
, Dana Dobesova , Olga Součkova , Vaclava Skopova ,
3
ID
1,2,4
´ ˇ
Tomas Adam
´
´
ID
, Marie Zikanova
*
´
1 Institute of Molecular and Translational Medicine, Faculty of Medicine and Dentistry, Palacky University and
University Hospital in Olomouc, Olomouc, Czech Republic, 2 Department of Clinical Biochemistry, University
Hospital in Olomouc, Olomouc, Czech Republic, 3 Research Unit for Rare Diseases, Department of
Pediatrics and Adolescent Medicine, First Faculty of Medicine, Charles University and General University
Hospital in Prague, Prague, Czech Republic, 4 Laboratory of Inherited Metabolic Disorders, Department of
Clinical Chemistry, University Hospital in Olomouc, Olomouc, Czech Republic
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
* marie.zikanova@lf1.cuni.cz
Abstract
Purines are essential molecules for all forms of life. In addition to constituting a backbone of
DNA and RNA, purines play roles in many metabolic pathways, such as energy utilization,
regulation of enzyme activity, and cell signaling. The supply of purines is provided by two
pathways: the salvage pathway and de novo synthesis. Although purine de novo synthesis
(PDNS) activity varies during the cell cycle, this pathway represents an important source of
purines, especially for rapidly dividing cells. A method for the detailed study of PDNS is lack-
ing for analytical reasons (sensitivity) and because of the commercial unavailability of the
compounds. The aim was to fully describe the mass spectrometric fragmentation behavior
of newly synthesized PDNS-related metabolites and develop an analytical method. Except
for four initial ribotide PDNS intermediates that preferentially lost water or phosphate or
cleaved the forming base of the purine ring, all the other metabolites studied cleaved the gly-
cosidic bond in the first fragmentation stage. Fragmentation was possible in the third to sixth
stages. A liquid chromatography-high-resolution mass spectrometric method was devel-
oped and applied in the analysis of CRISPR-Cas9 genome-edited HeLa cells deficient in the
individual enzymatic steps of PDNS and the salvage pathway. The identities of the newly
synthesized intermediates of PDNS were confirmed by comparing the fragmentation pat-
terns of the synthesized metabolites with those produced by cells (formed under pathologi-
cal conditions of known and theoretically possible defects of PDNS). The use of stable
isotope incorporation allowed the confirmation of fragmentation mechanisms and provided
data for future fluxomic experiments. This method may find uses in the diagnosis of PDNS
disorders, the investigation of purinosome formation, cancer research, enzyme inhibition
studies, and other applications.
OPEN ACCESS
´
´
ˇ
´
´
´
Citation: Madrova L, Krijt M, Baresova V, Vaclavık
´
ˇ ´
J, Friedecky D, Dobesova D, et al. (2018) Mass
spectrometric analysis of purine de novo
biosynthesis intermediates. PLoS ONE 13(12):
e0208947. https://doi.org/10.1371/journal.
pone.0208947
Editor: Timothy James Garrett, University of
Florida, UNITED STATES
Received: September 20, 2018
Accepted: November 26, 2018
Published: December 10, 2018
´
´
Copyright: © 2018 Madrova et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: All relevant data are
within the manuscript and its Supporting
Information files.
Funding: This work was supported by the Ministry
of Health of the Czech Republic [grant AZV 15-
28979A, http://www.azvcr.cz, recipient MZ], by The
Charles University Grant Agency [grant GAUK
818416, recipient MK and GAUK 1102217,
recipient OS, https://www.cuni.cz/UK-33.html] and
by The Czech Science Foundation [grant GACR 18-
12204S, recipient TA, https://gacr.cz/en/].
PLOS ONE | https://doi.org/10.1371/journal.pone.0208947 December 10, 2018
1 / 19

Mass spectrometric analysis of purine de novo biosynthesis
Institutional support was provided by Charles
University [programmes PRIMUS/17/MED/6,
recipient MZ, PROGRES Q26/LF1, UNCE 204064
and SVV 260367/2017 http://www.cuni.cz] and by
the Ministry of Education, Youth and Sports of CR
[LO1304 National Sustainability Programme I and
LQ1604 National Sustainability Programme II,
http://www.msmt.cz/]. The funders had no role in
study design, data collection and analysis, decision
to publish, or preparation of the manuscript.
Introduction
Purine nucleotides have vital functions in numerous pathways in both prokaryotes and
eukaryotes. As a part of many essential biomolecules, purine nucleotides participate in nucleic
acid synthesis, transcription, translation, cell signaling processes, and maintaining energetic
homeostasis and act as cofactors, neuromodulators, and cotransmitters. The supply of purine
nucleotides is provided by two pathways: the salvage pathway and de novo synthesis. Purine de
novo synthesis (PDNS) is a sequence of ten reactions catalyzed by six enzymes. Three of these
enzymes are multifunctional in PDNS (Fig 1), and bifunctional adenylosuccinate lyase (ADSL)
also participates in the purine nucleotide cycle, catalyzing the conversion of adenylosuccinic
acid (SAMP) to adenosine monophosphate (AMP).
Competing interests: The authors have declared
that no competing interests exist.
Initially, ribose-5-phosphate (R5P), which is synthesized in the pentose phosphate pathway,
is converted by ribose-phosphate pyrophosphokinase (PRPS, EC 2.7.6.1) to PRPP, the first
metabolite of PDNS. The formation of PRA is catalyzed by amidophosphoribosyltransferase
(PPAT, EC 2.4.2.14). The trifunctional enzyme GART (phosphoribosylglycinamide synthetase,
EC 6.3.4.13; phosphoribosylglycinamide formyltransferase, EC 2.1.2.2; and phosphoribosyla-
minoimidazole synthetase, EC 6.3.3.1) catalyzes steps 2, 3, and 5, respectively. Formation of
FGAMR from FGAR is catalyzed by phosphoribosylformylglycinamidine synthase (PFAS, EC
6.3.5.3). The bifunctional enzyme PAICS consists of phosphoribosylaminoimidazole carboxyl-
ase (EC 4.1.1.21, step 6) and phosphoribosylaminoimidazolesuccinocarboxamide synthase
(EC 6.3.2.6, step 7). The enzyme ADSL (EC 4.3.2.2) catalyzes the formation of AICAR from
SAICAR. The last two steps are catalyzed by the bifunctional ATIC enzyme phosphoribosyla-
minoimidazolecarboxamide formyltransferase (EC 2.1.2.3, step 9) and IMP cyclohydrolase
(EC 3.5.4.10, step 10). PDNS contributes to the cellular pool of purines at times when there is
an increased demand for purines, particularly during the G1 and S phases of the cell cycle, and
supplies enough purines for the synthesis of RNA and DNA [1]. The highest activity of PDNS
has been found in the skeletal muscles, even exceeding the activity in the liver. In contrast,
minimal activity has been found in the heart and brain (except developing brains). Erythro-
cytes, as an example of nondividing cells, lack an intact purine de novo synthetic pathway [2].
Recent studies focusing on protein-protein interactions have revealed that the individual
enzymes of PDNS group together sequentially to form multienzyme complexes called purino-
somes, which facilitate the synthesis of purines. It has been proposed that the formation of pur-
inosomes inside cells occurs in the G1 and S phases of the cell cycle [1]. This is well described
by immunofluorescence imaging of mammalian cells under purine-rich or purine-depleted
conditions [1, 3, 4] and by monitoring the overall flux through the pathway [4]. Direct tracing
of the intermediates of the pathway has thus far only been possible with the use of radioactive
labeling [5].
To date, two genetically determined defects of this metabolic pathway (out of ten that are
theoretically possible) have been identified: ADSL deficiency (OMIM 103050) and AICA-ribo-
siduria (ATIC deficiency, OMIM 608688). Almost eighty cases of ADSL deficiency [6] and
only one case of AICA-ribosiduria [7] have been reported worldwide to date. There are three
possible explanations for this limited number of cases: the first two are that the clinical condi-
tions associated with these particular enzyme deficiencies are either benign or incompatible
with fetal development. The third reason is that the available screening/diagnostic tools are rel-
atively laborious and expensive, which limits their widespread use. The methods used for
studying purine metabolism and diagnosing related metabolic diseases include proton nuclear
magnetic resonance, electrophoresis, and chromatography. Chromatography dominates the
field and has historically used UV absorbance, radiometric and tandem mass spectrometric
detection [4, 8–10]. PDNS ribosides excreted into growth media in four genetically modified
PLOS ONE | https://doi.org/10.1371/journal.pone.0208947 December 10, 2018
2 / 19

Mass spectrometric analysis of purine de novo biosynthesis
Fig 1. Purine de novo synthesis in humans. Abbreviations: phosphoribosylamine (PRA), glycinamideribonucleotide (GAR), N-formylglycinamide ribonucleotide
(FGAR), N-formylglycinamidine ribonucleotide (FGAMR), aminoimidazole ribonucleotide (AIR), carboxyaminoimidazole ribonucleotide (CAIR), N-
succinocarboxamide-5-aminoimidazole ribonucleotide (SAICAR), 5-aminoimidazole-4-carboxamide ribonucleotide (AICAR), 5-formamido-4-imidazolecarboxamide
ribonucleotide (FAICAR), inosine-5-monophosphate (IMP), glutamine (Gln), glutamate (Glu), pyrophosphate (PPi), adenosine-5’-triphosphate (ATP), adenosine 5’-
diphosphate (ADP), glycine (Gly), phosphate (Pi), N10-formyl tetrahydrofolate (N10-formyl THF), tetrahydrofolate (THF), hydrogen carbonic acid (HCO3-), aspartate
(Asp). For enzyme abbreviations, see the second paragraph of the Introduction.
https://doi.org/10.1371/journal.pone.0208947.g001
cell lines were detected in a recent study [11]. Currently, there is no published method for the
detection of all natural PDNS intermediates. The only published report utilizes radioactive
tracing [5]. Mass spectrometric fragmentation analysis of the compounds is of value because
of the biological importance of such compounds and the lack of experimental fragmentation
spectra. Mass spectral databases contain either no PDNS metabolites or in silico fragmentation
spectra of the metabolites, as these compounds are not commercially available (except AICAR,
AICAr and SAMP).
The aim of this work is to fully describe the mass spectrometric fragmentation behavior of
PDNS metabolites. Based on this knowledge, we analyzed individual PDNS-defective cell lines
to describe metabolic changes in the pathway with the objective of predicting metabolic
changes in analogous human conditions. The method that was developed for the detection of
PDNS intermediates could serve in pathway kinetic studies (enzyme activity measurements or
inhibition experiments, since a number of compounds are claimed to be PDNS inhibitors) and
in the diagnostics of inherited metabolic disorders. The method can also be applied in cancer
research, as some genes encoding PDNS enzymes have been reported to be up- or downregu-
lated in some tumors [12, 13]. In the first part, we synthesized 16 metabolites of PDNS and
their dephosphorylated analogues and performed in-depth multistage mass spectrometric
fragmentation analysis.
PLOS ONE | https://doi.org/10.1371/journal.pone.0208947 December 10, 2018
3 / 19

Mass spectrometric analysis of purine de novo biosynthesis
In the second part, we analyzed edited HeLa cells that were enzymatically deficient in the
individual steps of PDNS and treated with isotopically labeled glycine. The aim was to confirm
the identity of the newly synthesized intermediates of PDNS by comparing the fragmentation
patterns of the synthesized intermediates with those produced by cells (formed under patho-
logical conditions of known and theoretical possible defects of PDNS). Data obtained from the
comparison of high-resolution mass spectrometry (HRMS) fragmentation and mass shifts in
labeled and natural intermediates provided additional structural certainty in the form of
knowledge of the labeling position.
Finally, experiments with PDNS-deficient HeLa cell lines provided data for stable isotope
fluxomic analysis and allowed a deeper understanding to be gained of the biochemical changes
under those circumstances.
Materials and methods
Chemicals
Isotopically labeled glycine (U-13C2, 15N) was purchased from Cambridge Isotope Laboratories
(Andover, MA, USA). AICAR, 5-aminoimidazole-4-carboxamide riboside (AICAr) and adenylo-
succinic acid (SAMP) were purchased from Toronto Research Chemicals Inc. (North York, Can-
ada). SAICAR, succinylaminoimidazolecarboxamide riboside (SAICAr), succinyladenosine
(SAdo), AIR, 5-aminoimidazole riboside (AIr), CAIR, carboxyaminoimidazole riboside (CAIr),
and N10-formyl-tetrahydrofolate (N10-formyl-THF) were prepared as previously described [11, 14].
Calf intestinal alkaline phosphatase (CIP) and NEB3 buffer were purchased from New England Bio-
labs (NEB, Ipswich, MA, USA), and Dulbecco’s minimum essential medium (DMEM), F12 nutri-
ent mix, and fetal bovine serum (FBS) were obtained from Life Technologies, ThermoFisher
Scientific (MA, USA). Minimum Essential Medium (MEM) was obtained from BioSera (Nuaille,
France). All other chemicals were purchased from Sigma-Aldrich (St. Louis, USA).
Preparation and purification of substrates
GAR, FGAR, and FGAMR intermediates were synthesized biochemically using bacterial
recombinant enzymes. FAICAR was prepared by inorganic synthesis. We did not attempt to
synthesize PRA, as it is known to be unstable in vivo [15, 16].
The initial concentration of all compounds ranged from 57 μM in samples of FGAMR/r to
124 μM in samples of GAR/r (see S1 Table).
The bacterial recombinant enzymes phosphoribosylglycinamide synthetase (GARS) and phos-
phoribosylglycinamide formyltransferase (GARTF) were expressed and purified as fused protein
maltose binding protein MBP-GARS and MBP-GARTF using the pMALTM Protein Fusion and
Purification System (New England Biolabs Inc., USA), as described previously [17].
To produce recombinant bacterial phosphoribosylformylglycinamide synthetase fused with
a C-terminal polyhistidine tag (6H-PurL), the gene was introduced into the p6H vector,
expressed in Escherichia coli, and purified on a Co2+-immobilized metal affinity chromatogra-
phy column (GE Healthcare) according to standard procedure.
GAR/r preparation. The reaction mixture containing 5.7 mM ribose-5-phosphate, 0.7
mM ATP, 10 mM glycine, 10 mM ammonium hydroxide, 12.7 mM magnesium chloride, 20
mM phosphate buffer pH 7.4, and 0.4 μg/μL purified MBP-GARS was incubated at 37˚C for
four hours. The reaction was analyzed by high-performance liquid chromatography coupled
with mass spectrometry (HPLC-MS). The riboside form was prepared by dephosphorylation
with 1 U of CIP from NEB at 37˚C for four hours.
FGAR/r preparation. The reaction mixture containing 5.7 mM ribose-5-phosphate, 0.7
mM ATP, 10 mM glycine, 10 mM ammonium hydroxide, 12.7 mM magnesium chloride, 0.1
PLOS ONE | https://doi.org/10.1371/journal.pone.0208947 December 10, 2018
4 / 19

Mass spectrometric analysis of purine de novo biosynthesis
mM N10-formyl-THF, 20 mM phosphate buffer pH 7.4, 0.4 μg/μL MBP-GARS, and 0.4 μg/μL
MBP-GARTF was incubated at 37˚C for four hours. The subsequent procedure was the same
as in GAR/r preparation.
FGAMR/r preparation. A total of 200 μL of the reaction mixture from the synthesis of
FGAR was incubated with 2 mM glutamine, 2 mM ATP, and 0.25 μg/μL of purified 6H-PurL
at 37˚C for four hours. The subsequent procedure was the same as in GAR/r preparation.
FAICAR/r preparation. FAICAr was prepared according to Lukens et al. [18]. FAICAR
was prepared by adjusting the procedure used for the synthesis of FAICAr. In brief, we incu-
bated 10 mg of AICAR with 11 mg of NaOH, 136 μL of formic acid and 250 μL of acetic anhy-
dride for 1 hour at 37˚C. The product of the reaction was analyzed by HPLC-MS.
Cell cultivation and harvesting
We used the CRISPR-Cas9 genome-edited HeLa cells CR-GART, CR-PFAS, CR-PAICS,
CR-ADSL, and CR-ATIC prepared by Baresova et al. in 2016 [11]. CR-HGPRT cells (hypoxan-
thine-guanine phosphoribosyltransferase deficient) were prepared analogously [19]. HeLa cells
were cultured in a humidified atmosphere and incubated with 5% CO2 at 37˚C. All cells (knockout
and control) were maintained in DMEM/F12 nutrient mix medium supplemented with 10% FBS
(Gibco, Invitrogen) and 1% penicillin/streptomycin. The medium of the knockout cells was
enriched with 3x10-5 M of adenine. Twenty-four hours prior to the experiment, all the cell types
were cultivated in purine-depleted DMEM supplemented with dialyzed 10% FBS [11] and 1% pen-
icillin/streptomycin. Two hours prior to cell harvesting, the cells were washed with PBS and placed
into 5 mL of glycine-free MEM with 500 μM of isotopically labeled glycine (U-13C2, 15N) added.
Each deficient cell line was cultivated in hexaplicate in 75-cm2 flasks (approx. 5 million cells).
Cells were harvested by means of the modified quenching method described by Wojtowicz
et al. [20]. Initially, the medium was transferred into a 15-mL plastic tube for subsequent ana-
lytical preparation. Cellular metabolism was quenched by spraying 40 mL of 60% aqueous cold
methanol (v/v, -50˚C) by means of a plastic syringe with a needle. The culture flasks were kept
on ice and extracted with 1 mL of 80% methanol (v/v, -50˚C), and the cells were mechanically
detached using a cell scraper. The cell debris was drained out with a pipette. For an additional
extraction, another 2 mL of cold methanol was added. The methanol extracts were combined,
sonicated (30 s), and centrifuged (1800 g, 5 min, 4˚C), and the supernatants were freeze-dried.
Preparation of cell lysates
A total of 500 μL of cold 80% methanol was added to each lyophilizate and thoroughly mixed.
The samples were centrifuged at 15,000 g for 15 min at 4˚C, and the supernatants were taken
for analysis.
Preparation of cell media
The media were mixed using a vortex; then, 100 μL of each sample was taken and 300 μL of
80% methanol was added. The samples were left at -80˚C overnight. The extracts were centri-
fuged at 15,000 g for 15 min at 4˚C, and the supernatants were analyzed.
Fragmentation analysis of PDNS intermediates and their
dephosphorylated analogues (HPLC-HRMSn)
Chromatographic separation was achieved with hydrophilic interaction liquid chromatogra-
phy using an Ultimate 3000 RS (ThermoFisher Scientific, MA, USA). The aminopropyl col-
umn (Luna NH2 3 μm 100 Å, 100 x 2 mm, Phenomenex, Torrance, USA) was maintained at
PLOS ONE | https://doi.org/10.1371/journal.pone.0208947 December 10, 2018
5 / 19

Mass spectrometric analysis of purine de novo biosynthesis
35˚C. The mobile phase consisted of 20 mM ammonium acetate in water at pH 9.75 (mobile
phase A) and acetonitrile (mobile phase B). The gradient elution was performed as follows:
t = 0.0, 95% B; t = 7.0–13.0, 10% B; t = 14.0–17.0, 95% B. The flow rate was set to 0.3 mL/min,
and the injection volume was 2 μL.
Multistage fragmentation analysis was performed on an Orbitrap Elite (ThermoFisher Sci-
entific, MA, USA) operating in positive mode using electrospray ionization (capillary tempera-
ture 350˚C, source heater temperature 300˚C, sheath gas 10 arb. units, auxiliary gas 35 arb.
units, sweep gas 0 arb. units). The electrospray voltage was set at +3.0 kV. Fragmentation for
the most abundant fragments with intensities higher than 1E4 was performed using data-
dependent analysis (DDA) or TreeRobot (HighChem, SK); otherwise, the selection of frag-
ments was performed manually. Up to five of the most intense signals in MS2 were isolated
and further fragmented. Then, one to six of the most intense signals from each MS3 spectrum
were subjected to fragmentation to MS4. The subsequent MSn stages were also dependent on
the intensities of the emerging fragments, usually producing spectra of the one or two most
intense fragments from MS4/MS5. The fragmentation spectra were produced via collision-
induced dissociation (CID) using 30 units of normalized collision energy; the isolation width
was 2 Da, and the injection time was 200 ms. All the fragmentation spectra were measured
with a resolution of 120,000 full-width at half-maximum (FWHM) and with a mass error
below 3 ppm. The multistage fragmentation spectra of each compound were organized into
mass spectral trees. In every spectrum, the structures of the fragments belonging to the precur-
sor (target) compound/fragment were identified with the predictive fragmentation software
MassFrontier 7.0.5.09 SP3 (HighChem, SK).
Analysis of cell lysates and media
The chromatographic conditions were the same as in the HPLC-HRMSn analysis mentioned
above. Detection was performed on an Orbitrap Elite operating in positive ionization mode
with the same setting as above. The detection method was divided into four time segments.
Full scan analysis within the mass range m/z 70–1000 was performed in the first (0.0–3.0 min)
and fourth (12.0–17.0 min) segments. The selected ion monitoring (SIM) method was applied
in the second segment (3.0–7.0 min) for the analysis of ribosides (m/z 177–417) and in the
third segment (7.0–12.0 min) for the analysis of ribotides (m/z 257–497) to enhance the sensi-
tivity towards these metabolites (except for the measurement of SAdo and SAMP, which had
different m/z ranges: 379–389 and 459–469, respectively). The resolution was set to 60,000
FWHM. The mass error was below 3 ppm. All cell lines were measured in hexaplicate, and the
intensity values are presented as averages. The identities of the accumulated compounds in
both cell lysates and media were confirmed by MS2 fragmentation analysis. Fragmentation
spectra were produced via CID with the fragmentation energy set to 30 units of normalized
collision energy.
Results and discussion
Preparation and purification of PDNS intermediates and their
dephosphorylated analogues
The ribosides AIr, CAIr, SAICAr, AICAr, FAICAr, and SAdo and ribotides AIR, CAIR, and
SAICAR were synthesized according to previously published procedures [11, 14]. AICAR and
SAMP were commercially available.
New methods have been developed for the preparation of GAR, FGAR, and FGAMR. The
strategies were based on the preparation of the bacterial recombinant enzymes MBP-GARS,
PLOS ONE | https://doi.org/10.1371/journal.pone.0208947 December 10, 2018
6 / 19

Mass spectrometric analysis of purine de novo biosynthesis
MBP-GARTF, and 6H-PurL. The GAR metabolite was produced in two reactions that
occurred in one step from ribose-5-phosphate. The first reaction was the formation of PRA
from ribose-5-phosphate and ammonium hydroxide as a function of pH as described earlier
[11]. In the second reaction, PRA was converted by MBP-GARS to GAR with a production
yield of 10%. Then, GAR was dephosphorylated to GAr via CIP. FGAR was produced in the
same reaction as GAR with added N10-formyl THF and MBP-GARTF. The resulting FGAR
was dephosphorylated to FGAr by CIP. FGAMR was prepared from FGAR in the reaction cat-
alyzed by the 6H-PurL enzyme and dephosphorylated by CIP to FGAMr.
FAICAR/FAICAr was prepared inorganically from AICAR/AICAr. We applied a formylat-
ing environment consisting of formic acid, acetic anhydride, and NaOH to convert AICAR to
FAICAR within one hour at 37˚C.
MSn fragmentation analysis of PDNS intermediates and their
dephosphorylated analogues
In total, eight biologically stable PDNS intermediates (GAR, FGAR, FGAMR, AIR, CAIR,
AICAR, SAICAR, and FAICAR) and their dephosphorylated forms (GAr, FGAr, FGAMr, AIr,
CAIr, AICAr, SAICAr, and FAICAr) were subjected to HRMSn fragmentation analysis. More-
over, other PDNS-related purine metabolites, SAMP, SAdo, and IMP, were measured.
The compounds were sequentially fragmented up to MS6. The structure of the frag-
mented compound or fragment is shown in every spectrum, and up to six of the most
intense fragments are depicted in the structure (see one example in Fig 2; the rest are pro-
vided in S1 Fig). More than half of the compounds were fragmented up to MS4. The number
of fragmentation steps was determined by the concentration of the synthesized compound
in the reaction mixture (57–124 μM), molecular mass, ionization efficiency in positive
mode and ion suppression (relatively minor effect due to experimental set-up). In our view,
the richest information about the fragmentation behavior of PDNS metabolites lies within
the MS3 and MS4 spectra. Higher MS stages (MS5 and MS6) usually offered just one or two
additional fragments to the whole spectral tree (e.g., FAICAR, SAdo, SAICAr). The loss of
intensity observed in increasing MS stages caused higher noise levels, making the spectra
less reliable for interpretation. Nevertheless, even these spectra could help with the identifi-
cation of unknown molecules in mass spectral databases. One of the greatest advantages of
spectral trees is based on the comparison of substructural information of similar com-
pounds—precursor ion fingerprinting (PIF)—assuming that similar compounds share
identical substructures and the uniqueness of a compound is given by the way in which sub-
structures combine [21–23].
Generally, the instability of the majority of ribotides and ribosides was observed as a result
of ion-source fragmentation. MS2 fragments were visible in full MS scans of nearly all com-
pounds, most often not exceeding 5% of the intensity of a molecular ion. The intensity of some
MS2 fragments in full MS scans reached 17, 20, and 540% of the intensity of the molecular ions
for GAr, SAdo, and FGAr, respectively. GAr was the first compound to be identified with such
profound in-source fragmentation and was chosen for the optimization of the measurement
conditions. The initial parameters of the electrospray ionization (ESI) source were a 3 kV
spray voltage, 300˚C source heater temperature, 350˚C capillary temperature, and 60% S-Lens
RF level. ESI measurements under various conditions were performed (spray voltage 2.5 and 2
kV; heater and capillary temperatures lowered to 250 and 300˚C, respectively; and S-lens
parameters of 10, 20, and 40%) to reduce the ion-source fragmentation of fragile ions. No
change in these parameters resulted in significantly better stability (see S2–S4 Figs). For com-
pounds suffering from ion-source fragmentation and with low molecular ion intensity, the
PLOS ONE | https://doi.org/10.1371/journal.pone.0208947 December 10, 2018
7 / 19

Mass spectrometric analysis of purine de novo biosynthesis
Fig 2. Fragmentation spectral tree of AIR. Every m/z assigned with accurate mass in the spectra (in red) belongs to the fragmented structure. The other m/z represent
coeluting compounds or masses belonging to the fragmented structure that could not be identified using the given procedure.
https://doi.org/10.1371/journal.pone.0208947.g002
strategy for obtaining MSn spectra was modified. MS2 analysis of the ion-source fragments
(corresponding to MS3) belonging to the compound in the full MS spectrum was performed.
Ribotides showed considerable variability in their fragmentation patterns compared to
ribosides (Fig 3). Losses of water or the phosphate group from ribose were the dominant
events for the most intense fragments of the first part of the PDNS pathway (up to AIR). Other
fragments appeared with the build-up of the purine ring, starting with CAIR and proceeding
to consecutive PDNS intermediates. The most intense fragments of the ribotides of the second
part of the pathway originated from the breaking of the N-glycosidic bond and the loss of an
amino group from the base, similar to their dephosphorylated analogues.
All the ribosides shared common fragmentation behavior when the MS2 spectra were com-
pared (see S5 Fig). The most intense fragment ion always belonged to the part of the base
forming the purine ring, suggesting that the N-glycosidic bond had the highest susceptibility
to breaking. In some cases (FAICAr and SAdo), the formation of this fragment greatly pre-
dominated, and all the other fragments were found at below 1% of the intensity of the domi-
nant fragment ion but still above the lowest detectable signal. The majority of the second-
PLOS ONE | https://doi.org/10.1371/journal.pone.0208947 December 10, 2018
8 / 19

Mass spectrometric analysis of purine de novo biosynthesis
Fig 3. MS2 fragmentation spectra of PDNS ribotides. The structure of the fragmented compound is shown next to every spectrum, and up to six of the most intense
fragments are depicted in the structure.
https://doi.org/10.1371/journal.pone.0208947.g003
most-intense fragment ions were represented by a basic part that had lost either the amino or a
hydroxyl group. The loss of one or more molecules of water from the ribosidic part of nucleo-
sides was characteristic of the subsequent fragments.
PLOS ONE | https://doi.org/10.1371/journal.pone.0208947 December 10, 2018
9 / 19

Mass spectrometric analysis of purine de novo biosynthesis
To identify unknown structures, free online databases of fragmentation spectra can be used
in advance. In the case of PDNS metabolites, databases (Metlin, HMDB, mzCloud) [24] do not
contain experimental fragmentation spectra, just in silico predictions. Moreover, in silico frag-
mentation spectra are available for PDNS ribotides but not for ribosides. The MS2 spectra of
the PDNS ribotides acquired in our study were compared with in silico spectra provided by the
Metlin database (Fig 4), except for AICAR and IMP, which both have only experimental spec-
tra in the database.
Generally, the experimental spectra match the in silico predictions by roughly one-third.
Metlin in silico spectra generated by a machine learning approach are characterized by priori-
tizing low-molecular-weight ions (less than m/z 100) under all three different fragmentation
sets of conditions tested in positive mode (10, 20, and 40 eV). In contrast, our experimental
spectra offered many ions higher than m/z 100; the low-molecular-weight ions were sup-
pressed because of the mechanism of CID fragmentation. The diversity of the ions in the
experimental spectra can be attributed to combined sites of fragmentation leading to a particu-
lar ion. For example, in the MS2 experimental spectrum of GAR (Fig 4), the most significant
fragments originate from the loss of one or two molecules of water from ribose (m/z 251 and
269) or the loss of one or two molecules of water from ribose in combination with the loss of
the phosphate group (m/z 153 and 171). Of these four most significant ions, the in silico MS2
spectrum of GAR (MID 3411) offers just the ion of m/z 269, suggesting that only one fragmen-
tation site is preferred for in silico fragment prediction. This assumption could explain the dif-
ference observed between the in silico and experimental spectra of the PDNS metabolites.
Another reason for the spectral disagreement could arise from the existence of predicted ions
that would probably not be generated in a given ionization mode (in our case, positive). An
example is the ion of m/z 98.9842, annotated as a positively charged phosphate group (using
MassFrontier for annotation). This ion appears in all the in silico spectra of PDNS ribotides
but is unlikely to arise in positive ESI mode. In silico fragmentation approaches often use
experimental spectra of commercially available compounds in the prediction process of com-
mercially unavailable compounds (e.g., PDNS metabolites). AICAR is an example of a com-
mercially available PDNS metabolite whose experimental spectra are present in the Metlin
database. Generally, a greater number of equivalent fragment ions were found when compar-
ing the experimental and predicted spectra of metabolites that structurally resemble the
AICAR metabolite from the second part of PDNS. In contrast, the fragmentation spectra of
metabolites from the first part of PDNS (GAR, FGAR) differ significantly from the in silico
predictions. Despite the known limitations, in silico spectra can be helpful in the process of the
identification of unknown structures or confirmation of supposed structures.
The analysis of the HeLa cells enzymatically deficient in the individual steps of PDNS pro-
vided data for confirmation of the structures of the synthesized PDNS intermediates. The frag-
mentation patterns of the PDNS intermediates produced by cells were compared with those
obtained by fragmentation analysis of the synthesized compounds. The HeLa cell lines (defi-
cient and control) were cultivated in a glycine-free medium with the addition of isotopically
labeled glycine. Glycine enters PDNS in the second enzymatic reaction catalyzed by a trifunc-
tional GART enzyme, incorporating a glycine backbone into the forming purine structure.
The PDNS pathway was stimulated by cultivating cells in purine-depleted media 24 hours
before stable isotope labeling. Applying this procedure, we obtained naturally labeled PDNS
intermediates serving as another means of confirmation of identity. Identity was confirmed by
comparing the MS2 spectra of synthesized compounds to the MS2 spectra of these isotopically
labeled metabolites on the basis of mass shift and considering the position of glycine incorpo-
ration and fragmentation. An example of the results of the mass shift experiment is shown in
Fig 5. The majority of the PDNS metabolites detected in both cell lysates and growth media
PLOS ONE | https://doi.org/10.1371/journal.pone.0208947 December 10, 2018
10 / 19

Mass spectrometric analysis of purine de novo biosynthesis
PLOS ONE | https://doi.org/10.1371/journal.pone.0208947 December 10, 2018
11 / 19

Mass spectrometric analysis of purine de novo biosynthesis
Fig 4. MS2 spectral comparison. PDNS metabolites present in the Metlin database (either as in silico or experimental
fragmentation spectra) were compared with experimental spectra acquired in our study. The mass range of fragmentation spectra
from the database was adjusted to the mass range of the spectra from our study.
https://doi.org/10.1371/journal.pone.0208947.g004
were found in the labeled forms. PRA was not detected as a result of its known chemical insta-
bility [16]. We were not able to detect FAICAR, probably because of the specific kinetic prop-
erties of ATIC [25]. ATIC is a bifunctional enzyme catalyzing the final transformylation of
AICAR and cyclization of FAICAR to IMP. The spatial proximity of the two active sites, in
which the formylation reaction favors a backward direction while the terminal cyclohydrolase
is essentially unidirectional, makes FAICAR difficult to diffuse.
Metabolic changes in edited HeLa cell line cultures
The regulation of PDNS has been studied since the 1950s. Enzyme kinetics experiments, as
well as theoretical calculations, have proposed that only the first step of PDNS is limiting for
the pathway and that no change in other steps has an impact on the overall flux [26–28].
AICAr and SAICAr are the only PDNS intermediates detectable in the bodily fluids of healthy
humans by current analytical technologies [10, 17, 29, 30]. These observations imply that other
regulatory places might exist within the PDNS pathway.
The first knockout human cellular models of both known and potentially novel genetically
determined defects of PDNS, published in 2016 by Baresova et al. [11], were used in the study.
The aim was to model the situation in the bodily fluids of patients. All cells were transferred
into purine-depleted medium for 24 hours prior to the experiment to stimulate PDNS. The
HPLC-HRMS method was applied to analyze the accumulated PDNS metabolites in the cell
lysates and growth media. Overall, six CRISPR-Cas9 genome-edited defective HeLa cell lines
(CR-GART, CR-PFAS, CR-PAICS, CR-ADSL, CR-ATIC, and CR-HGPRT) and control HeLa
cells were analyzed. All the ribotides and ribosides detected in control and ADSL-defective
HeLa cells are listed in Fig 6 (see other defective cell lines in S6 Fig). The majority of the PDNS
metabolites detected in both cell lysates were found in the labeled forms. In all the deficient
cell lines, we were able to detect both intracellular and extracellular metabolites, except GART-
deficient cells, because of the instability of the theoretically expected accumulating PRA, which
has a half-life of 5 s under physiological conditions [15, 16]. The accumulation of intermediate
ribotides of a particular defective enzyme causes a shift in the equilibria of upstream enzymatic
reactions. Because of this effect, we also observed an accumulation of intermediates of PDNS
enzymes up to five steps prior to the defective step. In these cases, we usually did not observe
an accumulation of metabolites one by one in the opposite direction preceding the main
enzyme block. Not all such intermediates were detected because of the chemical instability of
the metabolite, differing enzyme kinetics across PDNS, and different ionization efficiencies of
PDNS intermediates. The fragmentation method used generally favors ribosides over the more
negatively charged ribotides. This might be the reason why ribotides are not detected in cells
in some cases but the dephosphorylated analogues of the ribotides are present. Some of the
dephosphorylated forms of PDNS intermediates were detected in both cell lysates and media.
This is a usual route for defective cells in both the de novo and salvage purine pathways to elim-
inate toxic accumulated intermediates via equilibrative nucleoside transporters. These sug-
gested mechanisms hold for purine salvage pathway defects [31] and are also supported by
biochemical observations in patients deficient in ADSL and ATIC, where only ribosidic forms
of substrates are detected in the body fluids of the patients [7, 32]. These mechanisms are of
the utmost clinical importance because extracellular nucleosides are used as diagnostic bio-
markers of these diseases [31]. Human equilibrative transporters possess highly different
PLOS ONE | https://doi.org/10.1371/journal.pone.0208947 December 10, 2018
12 / 19

Mass spectrometric analysis of purine de novo biosynthesis
PLOS ONE | https://doi.org/10.1371/journal.pone.0208947 December 10, 2018
13 / 19

Mass spectrometric analysis of purine de novo biosynthesis
Fig 5. Confirmation of the identity of FGAR. The MS2 spectrum of synthesized FGAR was compared to the MS2 spectra of FGAR (A) and isotopically
labeled FGAR-L (B), both of which were accumulated in CR-PFAS HeLa cells. In B, fragments with a mass shift of 3.0038 Da (because of the U-13C2, 15
labeling) are marked with an asterisk. The retention time of all compounds was 8.98 min.
N
https://doi.org/10.1371/journal.pone.0208947.g005
catalytic activities towards their substrates, and these substrates compete for the active site
[33]. Additionally, there is an unknown specificity of purine nucleoside phosphorylase towards
our metabolites. These facts make the interpretation of the levels of both natural and labeled
PDNS metabolites difficult, and Fig 6 represents just an illustrative view.
In the cytoplasm of control HeLa cells cultured in purine-free medium (Fig 6), AICAR and
SAICAR were detected, as were their dephosphorylated forms, which is in agreement with pre-
vious reports [29, 30]. Moreover, we were able to detect natural GAR/r and AIr in control cell
Fig 6. Ribotides and ribosides detected in cell lysates of ADSL-deficient and control HeLa cells. The deficient cell line is graphically represented by a PDNS pathway
with an enzyme block marked by a red rectangle. Accumulating metabolites are marked with red bold letters with an arrow. Cell lines were measured in hexaplicate, and
the value of the unlabeled peak intensity given above the baseline represents the average of a particular metabolite. Chromatographic peaks of metabolites that were
detected also in labeled form are shown as an asterisk. Note: The ionization efficiency of ribotides and ribosides is substantially different, so the responses of these
compounds are not directly comparable.
https://doi.org/10.1371/journal.pone.0208947.g006
PLOS ONE | https://doi.org/10.1371/journal.pone.0208947 December 10, 2018
14 / 19

Mass spectrometric analysis of purine de novo biosynthesis
lysates for the first time, suggesting higher sensitivity of the method. PDNS is considered to be
a pathway with an initial rate-limiting step regulating the overall flux, leading to an assumption
that its substrates are at low and mutually related concentrations. Thus, their detection proba-
bly depends on the ionization efficiency of the individual PDNS intermediates. The experi-
ments could be influenced by the differing viability of the individual defective cell lines used,
providing highly variable numbers of cells in each cultivation flask. There is a single study
reporting the partial detection of intermediates at trace levels by radiolabeling with autoradi-
ography [5]. The purine salvage metabolites SAMP and SAdo were also detected. Two ribo-
sides (AICAr and AIr) were excreted into the growth media of the control cells.
PFAS-deficient cells (CR-PFAS cells) accumulated FGAR and its riboside, which was also
detected in growth media. Most likely, accumulating FGAR causes a shift in equilibrium for
trifunctional GART, resulting in a detectable amount of GAR. FGAR and GAR were also
detected as isotopically labeled analogues.
The accumulation of AIR in CR-PAICS cells was previously described in Chinese hamster
ovary cells. These cells deficient in PAICS activity also accumulated a second compound that
the authors were not able to identify [8]. In our experiment, we detected three other ribotides
preceding the enzymatic block in addition to the expected accumulation of AIR. The abundant
ribotides were excreted into growth media in dephosphorylated forms. This observation sug-
gests a shift in the equilibria of enzymatic reactions up to the PFAS enzyme, as FGAR and
FGAr were the last detectable metabolites.
The deficiency of the bifunctional enzyme ADSL is characterized by the presence of the suc-
cinylpurines SAICAr and SAdo in bodily fluids [34]. In our experiment, CR-ADSL cells accu-
mulated SAICAR and SAICAr; however, SAMP and SAdo were detected but not accumulated
compared to the controls. This behavior could be explained by the experimental set-up, where
the purine-free medium forces cells to use PDNS extensively. The purine nucleotide cycle,
however, might not be used because of the low energy demands of the cultured cells, which
do not possess high proliferation. Two other PDNS intermediates were detected in the cell
lysates—AIR and GAR. These metabolites probably originate from the shifted equilibria of
enzymes preceding the ADSL blockage. AIr, FGAMr, and GAr were other ribosides detected
in cells. SAICAr, AIr, and FGAMr were found in growth media. The accumulation of AIR/AIr
in cells and growth media is in accordance with the findings from ADSL-deficient Chinese
hamster ovary cells (AdeI), where the authors detected the accumulation of AIR [8].
In the one patient suffering from ATIC deficiency reported so far, a massive accumulation
of AICAr, SAICAr, and SAdo in bodily fluids was detected [7]. CR-ATIC cells accumulated
AICAR and SAICAR. Corresponding ribosides were also found in growth media. SAMP and
its riboside SAdo were not detected.
Patients with partial or complete HGPRT deficiency (Kelley-Seegmiller syndrome and
Lesch-Nyhan syndrome, respectively) are known to have increased levels of AICAR in their
erythrocytes [35] and AICAr in their urine [36]. As mentioned previously, erythrocytes lack
an intact PDNS pathway but are capable of metabolizing exogenous ribosides to their corre-
sponding mono-, di-, and triphosphates and accumulating the metabolites [35]. Several studies
reported increased AICAR contents in the brains of HGPRT-deficient mice [37, 38]. We
detected AICAR in cell lysates and AICAr in both the cell lysates and media of CR-HGPRT
cells. The bifunctional enzyme ATIC, responsible for the last two steps of the pathway, is
strongly inhibited by xanthosine-5’-monophosphate, which is accumulated in patients with
Lesch-Nyhan syndrome because of the massive flux to uric acid production [39]. Another pos-
sible hypothesis for the accumulation of AICAR is enhanced histidine biosynthesis in HGPRT
patients; however, there are no experimental data to support this hypothesis [38, 40]. As a
result of shifted enzyme equilibria, labeled SAICAR was also detected.
PLOS ONE | https://doi.org/10.1371/journal.pone.0208947 December 10, 2018
15 / 19

Mass spectrometric analysis of purine de novo biosynthesis
Conclusion
In this study, we provide the first report of a comprehensive mass spectrometric fragmentation
analysis of synthesized PDNS intermediates and their dephosphorylated analogues. HRMS
fragmentation was possible in the third to sixth stages, and the spectra were compared with in
silico spectra present in free online databases. The data acquired allowed the development of
the method for the detection of these compounds in HeLa cells deficient in the individual
steps of PDNS. Our method is applicable in various areas of PDNS research, such as studying
cell cycle/purinosome formation. The detection of PDNS intermediates (ribotides) can be
valuable in enzyme kinetics studies since numerous compounds are claimed to be inhibitors of
this metabolic pathway. Medical applications include the development of diagnostic methods
for known/putative inherited metabolic disorders of PDNS and understanding their
pathobiochemistry.
Supporting information
S1 Table. List of initial concentrations of PDNS metabolites that were subjected to HRMSn
fragmentation analysis.
(TIF)
S1 Fig. Fragmentation spectral trees of GAR/r, FGAR/r, FGAMR/r, AIR/r, CAIR/r, SAI-
CAR/r, AICAR/r, FAICAR/r, IMP, SAdo and SAMP. All ribotides are marked with a capital
R in the name; ribosides are marked with a lowercase r. Every m/z assigned with accurate mass
in the spectra (in red) belongs to the fragmented structure. The other m/z represent coeluting
compounds or masses belonging to fragmented structures that could not be identified using
the given procedure.
(PDF)
S2 Fig. An example of in-source fragmentation. MS2 spectra of GAr measured using decreas-
ing spray voltages: 3 kV (A), 2.5 kV (B), and 2 kV (C). Intensities of the depicted fragments do
not significantly change with altered conditions.
(TIF)
S3 Fig. An example of in-source fragmentation. MS2 spectra of GAr measured using increas-
ing S-lens parameters: 10% (A), 20% (B), and 40% (C). Intensities of the depicted fragments do
not significantly change with altered conditions.
(TIF)
S4 Fig. An example of in-source fragmentation. MS2 spectra of GAr measured using differ-
ent temperatures of the heater (H) and capillary (CA): H 300˚C, CA 350˚C (A); H 300˚C, CA
300˚C (B); and H 250˚C, CA 250˚C (C). Intensities of the depicted fragments do not signifi-
cantly change with altered conditions.
(TIF)
S5 Fig. MS2 fragmentation spectra of PDNS ribosides. The structure of the fragmented com-
pound is shown next to every spectrum, and up to six of the most intense fragments are
depicted in the structure.
(TIF)
S6 Fig. Ribotides and ribosides detected in cell lysates of PFAS-, PAICS-, ATIC- and
HGPRT-deficient HeLa cells. Deficient cell lines are graphically represented by a PDNS path-
way with an enzyme block marked by a red rectangle. Accumulating metabolites are marked
with red bold letters with an arrow. Cell lines were measured in hexaplicate, and the value of
PLOS ONE | https://doi.org/10.1371/journal.pone.0208947 December 10, 2018
16 / 19

Mass spectrometric analysis of purine de novo biosynthesis
the unlabeled peak intensity given above the baseline represents the average of a particular
metabolite. Chromatographic peaks of metabolites that were also detected in labeled form are
shown with asterisks. Note: The ionization efficiency of ribotides and ribosides is substantially
different, so the responses of these compounds are not directly comparable.
(TIF)
Acknowledgments
The authors acknowledge Vladimir Cermak for providing the p6H vector.
Author Contributions
´
´
´ˇ
ˇ
´
´
´
Data curation: Lucie Madrova, Matyas Krijt, Veronika Baresova, Jan Vaclavık, Dana Dobe-
ˇ
ˇ
´
´
´
´
´ˇ
´
´
sova, Olga Součkova, Vaclava Skopova, Tomas Adam, Marie Zikanova.
´ˇ
Formal analysis: David Friedecky, Tomas Adam.
´
´
´
´ˇ
Investigation: Lucie Madrova, Matyas Krijt, Veronika Baresova, Jan Vaclavık, David Frie-
ˇ
´
´
´
ˇ
ˇ
´
´
´
´
´ˇ
´
´
´
decky, Dana Dobesova, Olga Součkova, Vaclava Skopova, Tomas Adam, Marie Zikanova.
´
Methodology: Marie Zikanova.
´
´
Software: David Friedecky.
´ˇ
Supervision: Tomas Adam, Marie Zikanova.
´
´
´
Writing – original draft: Lucie Madrova, Tomas Adam.
´
´ˇ
´
Writing – review & editing: Marie Zikanova.
´
References
1. An S, Kumar R, Sheets ED, Benkovic SJ. Reversible compartmentalization of de novo purine biosyn-
thetic complexes in living cells. Science. 2008; 320(5872):103–6. https://doi.org/10.1126/science.
1152241 PMID: 18388293
2. Stone TW, Simmonds HA. Purines: basic and clinical aspects. Dordrecht; London: Kluwer Academic
Publishers; 1991.
3. Baresova V, Skopova V, Sikora J, Patterson D, Sovova J, Zikanova M, et al. Mutations of ATIC and
ADSL affect purinosome assembly in cultured skin fibroblasts from patients with AICA-ribosiduria and
ADSL deficiency. Hum Mol Genet. 2012; 21(7):1534–43. https://doi.org/10.1093/hmg/ddr591 PMID:
22180458
4. Zhao H, Chiaro CR, Zhang L, Smith PB, Chan CY, Pedley AM, et al. Quantitative analysis of purine
nucleotides indicates that purinosomes increase de novo purine biosynthesis. J Biol Chem. 2015; 290
(11):6705–13. https://doi.org/10.1074/jbc.M114.628701 PMID: 25605736
5. Sant ME, Poiner A, Harsanyi MC, Lyons SD, Christopherson RI. Chromatographic analysis of purine
precursors in mouse L1210 leukemia. Anal Biochem. 1989; 182(1):121–8. PMID: 2604037
6. Jurecka A, Zikanova M, Kmoch S, Tylki-Szymanska A. Adenylosuccinate lyase deficiency. J Inherit
Metab Dis. 2015; 38(2):231–42. https://doi.org/10.1007/s10545-014-9755-y PMID: 25112391
7. Marie S, Heron B, Bitoun P, Timmerman T, Van Den Berghe G, Vincent MF. AICA-ribosiduria: a novel,
neurologically devastating inborn error of purine biosynthesis caused by mutation of ATIC. Am J Hum
Genet. 2004; 74(6):1276–81. https://doi.org/10.1086/421475 PMID: 15114530
8. Duval N, Luhrs K, Wilkinson TG 2nd, Baresova V, Skopova V, Kmoch S, et al. Genetic and metabolomic
analysis of AdeD and AdeI mutants of de novo purine biosynthesis: cellular models of de novo purine
biosynthesis deficiency disorders. Mol Genet Metab. 2013; 108(3):178–89. https://doi.org/10.1016/j.
ymgme.2013.01.002 PMID: 23394948
9. Hornik P, Vyskocilova P, Friedecky D, Adam T. Diagnosing AICA-ribosiduria by capillary electrophore-
sis. J Chromatogr B Analyt Technol Biomed Life Sci. 2006; 843(1):15–9. https://doi.org/10.1016/j.
jchromb.2006.05.020 PMID: 16798121
PLOS ONE | https://doi.org/10.1371/journal.pone.0208947 December 10, 2018
17 / 19

Mass spectrometric analysis of purine de novo biosynthesis
10. Zidkova L, Krijt J, Sladkova J, Hlobilkova A, Magner M, Zikanova M, et al. Oligodendroglia from ADSL-
deficient patient produce SAICAribotide and SAMP. Mol Genet Metab. 2010; 101(2–3):286–8. https://
doi.org/10.1016/j.ymgme.2010.06.014 PMID: 20674424
11. Baresova V, Krijt M, Skopova V, Souckova O, Kmoch S, Zikanova M. CRISPR-Cas9 induced mutations
along de novo purine synthesis in HeLa cells result in accumulation of individual enzyme substrates and
affect purinosome formation. Mol Genet Metab. 2016; 119(3):270–7. https://doi.org/10.1016/j.ymgme.
2016.08.004 PMID: 27590927
12. Chakravarthi BV, Goswami MT, Pathi SS, Dodson M, Chandrashekar DS, Agarwal S, et al. Expression
and Role of PAICS, a De Novo Purine Biosynthetic Gene in Prostate Cancer. Prostate. 2017; 77
(1):10–21. https://doi.org/10.1002/pros.23243 PMID: 27550065
13. Goswami MT, Chen G, Chakravarthi BV, Pathi SS, Anand SK, Carskadon SL, et al. Role and regulation
of coordinately expressed de novo purine biosynthetic enzymes PPAT and PAICS in lung cancer.
Oncotarget. 2015; 6(27):23445–61. https://doi.org/10.18632/oncotarget.4352 PMID: 26140362
14. Zikanova M, Krijt J, Hartmannova H, Kmoch S. Preparation of 5-amino-4-imidazole-N-succinocarboxa-
mide ribotide, 5-amino-4-imidazole-N-succinocarboxamide riboside and succinyladenosine, com-
pounds usable in diagnosis and research of adenylosuccinate lyase deficiency. J Inherit Metab Dis.
2005; 28(4):493–9. https://doi.org/10.1007/s10545-005-0493-z PMID: 15902552
15. Rudolph J, Stubbe J. Investigation of the mechanism of phosphoribosylamine transfer from glutamine
phosphoribosylpyrophosphate amidotransferase to glycinamide ribonucleotide synthetase. Biochemis-
try. 1995; 34(7):2241–50. PMID: 7532005
16. Schendel FJ, Cheng YS, Otvos JD, Wehrli S, Stubbe J. Characterization and chemical properties of
phosphoribosylamine, an unstable intermediate in the de novo purine biosynthetic pathway. Biochemis-
try. 1988; 27(7):2614–23. PMID: 2454658
17. Kmoch S, Hartmannova H, Stiburkova B, Krijt J, Zikanova M, Sebesta I. Human adenylosuccinate
lyase (ADSL), cloning and characterization of full-length cDNA and its isoform, gene structure and
molecular basis for ADSL deficiency in six patients. Hum Mol Genet. 2000; 9(10):1501–13. PMID:
10888601
18. Lukens L, Flaks J. Intermediates in Purine Nucleotide Synthesis. Method Enzymol. 1963; 6:671–702.
19. Baresova V, Skopova V, Souckova O, Krijt M, Kmoch S, Zikanova M. Study of purinosome assembly in
cell-based model systems with de novo purine synthesis and salvage pathway deficiencies. PLoS One.
2018; 13(7):e0201432. https://doi.org/10.1371/journal.pone.0201432 PMID: 30059557
ˇ
ˇ
´
ˇ
´
´
´
´
´
´
20. Wojtowicz P, Zrostlıkova J, Stastna V, Dostalova E, Zıdkova L, Bruheim P, et al. Comprehensive Two-
Dimensional Gas Chromatography Coupled to Time-of-Flight Mass Spectrometry in Human Metabolo-
mics. Salih B, editor: InTech; 2012.
21. Kasper PT, Rojas-Cherto M, Mistrik R, Reijmers T, Hankemeier T, Vreeken RJ. Fragmentation trees for
the structural characterisation of metabolites. Rapid Commun Mass Spectrom. 2012; 26(19):2275–86.
https://doi.org/10.1002/rcm.6340 PMID: 22956319
22. Ridder L, van der Hooft JJ, Verhoeven S, de Vos RC, van Schaik R, Vervoort J. Substructure-based
annotation of high-resolution multistage MS(n) spectral trees. Rapid Commun Mass Spectrom. 2012;
26(20):2461–71. https://doi.org/10.1002/rcm.6364 PMID: 22976213
23. Sheldon MT, Mistrik R, Croley TR. Determination of ion structures in structurally related compounds
using precursor ion fingerprinting. J Am Soc Mass Spectrom. 2009; 20(3):370–6. https://doi.org/10.
1016/j.jasms.2008.10.017 PMID: 19041260
24. METLIN database [database on the Internet] [cited 22 May 2018]. Available from: https://metlin.scripps.
edu.
25. Bulock KG, Beardsley GP, Anderson KS. The kinetic mechanism of the human bifunctional enzyme
ATIC (5-amino-4-imidazolecarboxamide ribonucleotide transformylase/inosine 5’-monophosphate
cyclohydrolase). A surprising lack of substrate channeling. J Biol Chem. 2002; 277(25):22168–74.
https://doi.org/10.1074/jbc.M111964200 PMID: 11948179
26. Becker MA, Kim M. Regulation of purine synthesis de novo in human fibroblasts by purine nucleotides
and phosphoribosylpyrophosphate. J Biol Chem. 1987; 262(30):14531–7. PMID: 2444588
27. Curto R, Voit EO, Sorribas A, Cascante M. Mathematical models of purine metabolism in man. Math
Biosci. 1998; 151(1):1–49. PMID: 9664759
28. Hartman SC, Buchanan JM. Biosynthesis of the purines. XXI. 5-Phosphoribosylpyrophosphate amido-
transferase. J Biol Chem. 1958; 233(2):451–5. PMID: 13563519
29. Thomas A, Vogel M, Piper T, Krug O, Beuck S, Schanzer W, et al. Quantification of AICAR-ribotide con-
centrations in red blood cells by means of LC-MS/MS. Anal Bioanal Chem. 2013; 405(30):9703–9.
https://doi.org/10.1007/s00216-013-7162-0 PMID: 23828211
PLOS ONE | https://doi.org/10.1371/journal.pone.0208947 December 10, 2018
18 / 19

Mass spectrometric analysis of purine de novo biosynthesis
30. Zikanova M, Krijt J, Skopova V, Krijt M, Baresova V, Kmoch S. Screening for adenylosuccinate lyase
deficiency using tandem mass spectrometry analysis of succinylpurines in neonatal dried blood spots.
Clin Biochem. 2015; 48(1–2):2–7. https://doi.org/10.1016/j.clinbiochem.2014.10.004 PMID: 25445730
31. Hommes FA. Techniques in diagnostic human biochemical genetics: a laboratory manual. New York,
NY: Wiley-Liss; 1991.
32. Zikanova M, Skopova V, Hnizda A, Krijt J, Kmoch S. Biochemical and structural analysis of 14 mutant
adsl enzyme complexes and correlation to phenotypic heterogeneity of adenylosuccinate lyase defi-
ciency. Hum Mutat. 2010; 31(4):445–55. https://doi.org/10.1002/humu.21212 PMID: 20127976
33. Huang W, Zeng X, Shi Y, Liu M. Functional characterization of human equilibrative nucleoside trans-
porter 1. Protein Cell. 2017; 8(4):284–95. https://doi.org/10.1007/s13238-016-0350-x PMID: 27995448
34. Jaeken J, Van den Berghe G. An infantile autistic syndrome characterised by the presence of succinyl-
purines in body fluids. Lancet. 1984; 2(8411):1058–61. PMID: 6150139
35. Sidi Y, Mitchell BS. Z-nucleotide accumulation in erythrocytes from Lesch-Nyhan patients. J Clin Invest.
1985; 76(6):2416–9. https://doi.org/10.1172/JCI112255 PMID: 4077987
36. Newcombe DS. The urinary excretion of aminoimidazolecarboxamide in the Lesch-Nyhan syndrome.
Pediatrics. 1970; 46(4):508–12. PMID: 5503686
37. Jinnah HA, Page T, Friedmann T. Brain purines in a genetic mouse model of Lesch-Nyhan disease. J
Neurochem. 1993; 60(6):2036–45. PMID: 8492116
38. Tschirner SK, Bahre H, Kaever A, Schneider EH, Seifert R, Kaever V. Non-targeted metabolomics by
high resolution mass spectrometry in HPRT knockout mice. Life Sci. 2016; 156:68–73. https://doi.org/
10.1016/j.lfs.2016.05.031 PMID: 27221022
39. Lopez JM. Is ZMP the toxic metabolite in Lesch-Nyhan disease? Med Hypotheses. 2008; 71(5):657–
63. https://doi.org/10.1016/j.mehy.2008.06.033 PMID: 18710792
40. Neychev VK, Mitev VI. The biochemical basis of the neurobehavioral abnormalities in the Lesch-Nyhan
syndrome: a hypothesis. Med Hypotheses. 2004; 63(1):131–4. https://doi.org/10.1016/j.mehy.2004.
01.019 PMID: 15193365
PLOS ONE | https://doi.org/10.1371/journal.pone.0208947 December 10, 2018
19 / 19