THE PURINE AND PYRIMIDINE METABOLISM IN LACTATING DAIRY COWS CHARLOTTE STENTOFT NIELSEN Ph.D. THESIS ∙ SCIENCE AND TECHNOLOGY ∙ 2014 Aarhus University Faculty of Science and Technology Department of Animal Science Blichers Allé 20 P.O. Box 50 DK-8830 Tjele
179
Embed
THE PURINE AND PYRIMIDINE METABOLISM IN LACTATING DAIRY …pure.au.dk/portal/files/84593569/PhD_Thesis_Charlotte_Stentoft_Niel… · THE PURINE AND PYRIMIDINE METABOLISM IN LACTATING
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
THE PURINE AND PYRIMIDINE METABOLISM IN
LACTATING DAIRY COWS
CHARLOTTE STENTOFT NIELSEN
Ph.D. THESIS ∙ SCIENCE AND TECHNOLOGY ∙ 2014
Aarhus University
Faculty of Science and Technology
Department of Animal Science
Blichers Allé 20
P.O. Box 50
DK-8830 Tjele
I
Supervisors and Ph.D. assessment committee
Supervisors
Head of Research Unit, Ph.D., Mogens Vestergaard
Aarhus University, Faculty of Science and Technology, Department of Animal Science, Denmark
Senior Scientist, Ph.D., Søren Krogh Jensen
Aarhus University, Faculty of Science and Technology, Department of Animal Science, Denmark
Assistant professor, Ph.D., Mogens Larsen
Aarhus University, Faculty of Science and Technology, Department of Animal Science, Denmark
Senior Scientist, Ph.D., Torben Larsen
Aarhus University, Faculty of Science and Technology, Department of Animal Science, Denmark
√; the parameter is required, X; the parameter is not required, LLOQ; lower limit of detection, ULOQ; upper limit
of quantification, RSD; relative standard deviation (modified from Nováková, 2013). 1To be analysed in replicates.
2Precision is further subdivided into within-day (repeatability) and across-day (in-
termediate) precision and reproducibility. 3Very detailed stability studies are required, for full details consult the
EMA guideline (Guideline EMA, 2011).
Concerning validation of the developed method in this study, once the pre-treatment, LC-ESI-
MS/MS, and calibration procedures had been established, the performance characteristics of the
method were established by assessment of selectivity, linearity (calibration curve), stability, preci-
sion, accuracy (recovery), and absolute matrix effects, followed by tests of the application range.
No single guideline was used for the validation studies, but efforts were made to cover all relevant
parts of the specific method while still keeping in line with conventional approaches. Method vali-
dation is a vast area in bioanalytical science and to present all details will be beyond the scope of
50
this thesis. However, the main principles and relevant applications for the validation of the purine
and pyrimidine LC-ESI-MS/MS method will be given in the following subsections, for full details
consult paper I (Stentoft et al., 2014).
With regard to selectivity, this is defined as; the ability of a bioanalytical method to measure and
differentiate the component(s) of interest and internal standard(s) in the presence of other compo-
nents which may be expected to be present in the sample (Guideline EMA, 2011). The selectivity
should be proved using at least six individual sources of the appropriate blank matrix, which are
individually analysed and evaluated for interference (Table 2). In this study, a blank sample matrix
was not available and other components at the same Rt could not be excluded. Instead, the absence
of component/SIL cross-talk was confirmed by comparing chromatographic responses for standards
and SIL alone and in a mixture. The calibration curve describes the response of the instrument with
regard to the concentration of component over the calibration range (Guideline EMA, 2011). Ac-
cording to the EMA guideline, the calibration standards should; first of all, be matrix-matched; sec-
ondly, there should be one calibration curve for each studied component, and for each analytical
run; thirdly, it should cover the calibration range, defined by the LLOQ and ULOQ (Table 2);
fourthly, a minimum of six calibrators should be used in addition to the blank sample (processed
plasma without component or SIL) and a zero sample (processed plasma with SIL); and finally, a
relationship which can simple and adequately describe the response of the instrument with regard to
the concentration of component should be applied. Calibration curve precision and accuracy is vital
for achieving high quality data. The calibration and quantification methodology applied has already
been described in a previous section. All calibration curves used in this study were matrix matched
and covered relevant concentration ranges. Logarithmic and linear calibration models were tested
and the linearity of the log calibration curves studied with a lack of fit hypothesis test. The quantifi-
cation ranges was determined by homogeneity of variance and the stability between run days ac-
cessed. Limits of detection and quantification were not determined, instead, the homogeneity of
variance of the calibration curves was considered. According to the EMA guideline, stability in
method validation is; the chemical stability of a component in a given matrix under specific condi-
tions for given time intervals (Guideline EMA, 2011). An evaluation of stability should be carried
out to ensure that every step taken during sample preparation and sample analysis as well as the
storage conditions, do not affect the concentration of the component. Stability studies should be
carried out so as to investigate conditions and time periods that equal those applied to actual study
samples. In this study, for continuous evaluation of long-term storage stability, a freshly thawed
quality control was analyzed and evaluated in all analytical runs. The stability within runs (6-24 h)
was evaluated by assessing a quality control at the beginning and at the end of each sequence and
51
by analysing a set of spiked standard samples at five different times (different vials) during a 30 h
sequence. To determine the stability of the calibration curves, the across-day variation was assessed
over five consecutive days. Freeze-thaw cycle stability was not explored. If working with very
small concentration differences, as in this study, precision is one of the most critical validation pa-
rameters. It is defined as; the ratio of standard deviation/mean (%) (Guideline EMA, 2011). The
precision of an analytical procedure expresses the closeness of agreement between a series of meas-
urements obtained under the prescribed conditions expressed as the CV%. In this study, precision of
the method was determined by analyzing replicate sets of spiked standard plasma samples on five
separate days. The accuracy, more commonly known as the recovery, of an analytical procedure
expresses the closeness of the determined value to the value which is accepted either as a conven-
tional true or an accepted reference value, defined as; (determined value/true value) × 100%
(Guideline EMA, 2011). Accuracy should be assessed on samples spiked with known amounts of
the component, independently from the calibrators, using separately prepared stock solutions. The
samples are analysed against the calibration curve, and the obtained concentrations compared with
the nominal value. Accuracy should be evaluated within-day and across-day as for precision. The
absolute accuracies of the developed method were calculated using the same set of spiked standard
plasma as for the precision evaluation. The LC-ESI-MS/MS analysis developed in this study was
established for use with blood plasma samples from multicatheterized cows. Since jugular vein
plasma was used for method development, to determine the application range of the method, rela-
tive matrix effects were evaluated in alternative types of plasma as well as water, urine, and milk
samples.
Based on the validation and the examination of relative matrix effects, it was determined that the
LC-ESI-MS/MS method was suitable for quantification of the 20 targeted purine and pyrimidine
metabolites in bovine blood plasma from the multicatheterized cow model.
52
5. Brief summary of papers and manuscripts included in the thesis
Paper I
Simultaneous quantification of purine and pyrimidine bases, nucleosides and their degrada-
tion products in bovine blood plasma by high performance liquid chromatography tandem
mass spectrometry. Stentoft C., M. Vestergaard, P. Løvendahl, N.B. Kristensen, J.M. Moorby and
S.K. Jensen. 2014. J. Cromatogr. A. 1356:197-210.
Hypothesis and objectives
The hypotheses were i) that LC-ESI-MS/MS can accurately be used to quantitatively determine a
range of purine and pyrimidine metabolites in cow blood plasma when incorporated with matrix-
matched calibration standards and SIL, and ii) purine and pyrimidine metabolites can be isolated
and concentrated from blood plasma by applying an appropriate pre-treatment protocol. The objec-
tive was to develop and validate a LC-ESI-MS/MS procedure and pre-treatment protocol for quanti-
fication of a range of purine and pyrimidine metabolites in cow blood plasma.
Materials and methods
A LC-ESI-MS/MS method for simultaneous quantification of 20 purine pyrimidines metabolites in
blood plasma from dairy cows were developed and validated. The technique was combined with
individual matrix-matched calibration standards and SIL and preceded by a novel pre-treatment
procedure.
Data presented
Method development including pre-treatment and LC-ESI-MS/MS procedure
The log-calibration model and quantification ranges
Method validation
Potential application
Conclusions
The method was developed and validated as intended. It was confirmed that using a log-calibration
model resulted in a satisfying linear regression. The method covered concentration ranges for each
metabolite according to that in actual samples. The CV% of the chosen quantification ranges were
below 25%. The method had good repeatability (CV% ≤ 25%) and intermediate precision (CV% ≤
25%) and excellent recoveries (91-107%). All metabolites demonstrated good long-term stability
and stability within-runs (CV% ≤ 10%). Different degrees of absolute matrix effects were observed.
The potential application of the method was demonstrated by evaluating its range of use in different
types of blood plasma from multicatheterized cows.
53
Paper II
Absorption and intermediary metabolism of purines and pyrimidines in lactating dairy cows.
Stentoft C., B.A. Røjen, S.K. Jensen, N.B. Kristensen, M. Vestergaard and M. Larsen. Accepted
November 11th
2014 by Br. J. Nutr.
Hypothesis and objectives
The hypotheses were i) that the purine and the pyrimidine metabolites, in the form of nucleosides,
bases, and degradation product, are absorbed from the small intestine of the dairy cow and undergo
degradation across the intestinal wall and the hepatic tissue and ii) that the purine and pyrimidine
nitrogen to a large extent ultimately are lost following degradation and excretion via the kidneys.
The objective was to describe the metabolism of purine and pyrimidines by studying postprandial
patterns of net PDV and hepatic metabolism and to evaluate the fate of nitrogen in this context.
Materials and methods
Eight ruminally cannulated Holstein cows in second lactation were permanently catheterised in the
artery and gastrosplenic, mesenteric, hepatic portal, and hepatic vein and randomly allocated to a
triplicate incomplete 3 x 3 Latin square design with 14 d periods. Cows were fed a basal total mixed
ration (TMR) supplying 80% of requirements for metabolisable protein. Four cows assigned to a
treatment of 8.5 g of feed urea/kg (ventral ruminal infusion, 15% CP) of dry matter intake (DMI)
were evaluated. Concentrations of purine and pyrimidine metabolites were determined in plasma
using LC-ESI-MS/MS, splanchnic fluxes calculated, and postprandial pattern evaluated.
Data presented
Plasma concentrations and concentration differences between veins of metabolites
Net portal, net hepatic and, and net splanchnic fluxes of metabolites
Purine and pyrimidine nitrogen metabolism
Conclusions
All of the 20 purine and pyrimidine metabolites were absorbed from the PDV; the purines mainly as
degradation products and only minimally as nucleosides and bases and, the pyrimidines mainly as
nucleosides and bases and, only minimally as degradation products. Most of the bases were degrad-
ed during absorption, in the blood or in the hepatic tissue. Eventually, an effective blood and hepat-
ic metabolism further degraded all of the purine metabolites into degradation products for excretion
into the kidneys. The pyrimidine nucleosides was to a much larger extend absorbed intact and an
outlet into other parts of the nitrogen metabolism was detected. The postprandial pattern was not
found to have an effect on neither the net PDV nor the net hepatic metabolism.
54
Manuscript III
Protein level influences the splanchnic metabolism of purine and pyrimidine metabolites in
lactating dairy cows. Stentoft C., C. Barratt, L.A. Crompton, S.K. Jensen, M. Vestergaard, M.
Larsen and C.K. Reynolds. To be submitted to J. Dairy Sci.
Hypothesis and objectives
The hypothesis was i) that the net PDV and net hepatic fluxes of the purine and pyrimidine nucleo-
sides, bases, and degradation products would reflect different degrees of microbial biosynthesis
with different dietary protein levels (12.5, 15.0, and 17.5% CP) and proportions of forage sources
(grass vs. corn silage) in the ration. The objectives were to study the net PDV, net hepatic and total
splanchnic metabolism of the purine and pyrimidine metabolites and evaluating how this was af-
fected by dietary protein level and forage source and, to evaluate the fate of the purine and pyrimi-
dine nitrogen by estimating nucleic acid nitrogen fluxes in the splanchnic tissues.
Materials and methods
Six ruminally cannulated Holstein Friesian cows in mid-late lactation were permanently catheter-
ised in the artery and mesenteric, hepatic portal, and hepatic vein and randomly allocated to a 2 × 3
factorial study design with 21 d periods. Cows were fed a TMR consisting of 50:50 mixture of for-
age:concentrate. There were six treatment periods with diets containing one forage type (DM was
either 25:75 or 75:25 grass silage:corn silage) and one protein level (12.5%, 15.0%, 17.5% CP) for
each period. Concentrations of purine and pyrimidine metabolites were determined in plasma using
LC-ESI-MS/MS, splanchnic fluxes calculated, and protein and roughage effects evaluated.
Data presented
Arterial concentrations
Net portal, net hepatic and, net splanchnic fluxes of metabolites
Epigastric concentration differences
Conclusions
Protein effects were detectable for metabolites with considerable levels of net fluxes and good pre-
cision in the method. The effect of protein level was most easily detectable at the level of release
from the PDV and became harder to trace when passing the hepatic tissue. None of the splanchnic
fluxes were influenced by forage source. Due to a very effective intermediary degradation depend-
ent on the level of protein, considerable amounts of purine nitrogen was found to be lost to the dairy
cow. The effect of protein level seemed to be less relevant in the case of the pyrimidine nitrogen,
since the pyrimidine metabolites has an anabolic outlet into other parts of the nitrogen metabolism.
55
6. Paper I
Simultaneous quantification of purine and pyrimidine bases, nucleosides and their degrada-
tion products in bovine blood plasma by high performance liquid chromatography tandem
mass spectrometry.
Stentoft C., M. Vestergaard, P. Løvendahl, N.B. Kristensen, J.M. Moorby and S.K. Jensen. 2014. J.
Cromatogr. A. 1356, 197-210.
Snh
CJa
b
c
d
a
ARRAA
KNRPPPL
1
s
m((
h0
Journal of Chromatography A, 1356 (2014) 197–210
Contents lists available at ScienceDirect
Journal of Chromatography A
j o ur na l ho me page: www.elsev ier .com/ locate /chroma
imultaneous quantification of purine and pyrimidine bases,ucleosides and their degradation products in bovine blood plasma byigh performance liquid chromatography tandem mass spectrometry
harlotte Stentofta,∗, Mogens Vestergaarda, Peter Løvendahlb, Niels Bastian Kristensenc,on M. Moorbyd, Søren Krogh Jensena
Department of Animal Science, Aarhus University, Blichers Allé 20, DK 8830 Tjele, DenmarkDepartment of Molecular Biology and Genetics, Aarhus University, Blichers Allé 20, DK 8830 Tjele, DenmarkKnowledge Centre for Agriculture, Cattle, Agro Food Park 15, DK 8200 Aarhus N, DenmarkInstitute of Biological, Environmental and Rural Sciences (IBERS), Aberystwyth University, Gogerddan, Aberystwyth, Ceredigion, SY23 3EE Wales, UK
r t i c l e i n f o
rticle history:eceived 13 July 2013eceived in revised form 9 May 2014ccepted 11 June 2014vailable online 27 June 2014
eywords:itrogenuminanturineyrimidinelasmaC–MS/MS
a b s t r a c t
Improved nitrogen utilization in cattle is important in order to secure a sustainable cattle production.As purines and pyrimidines (PP) constitute an appreciable part of rumen nitrogen, an improved under-standing of the absorption and intermediary metabolism of PP is essential. The present work describesthe development and validation of a sensitive and specific method for simultaneous determination of 20purines (adenine, guanine, guanosine, inosine, 2′-deoxyguanosine, 2′-deoxyinosine, xanthine, hypoxan-thine), pyrimidines (cytosine, thymine, uracil, cytidine, uridine, thymidine, 2′-deoxyuridine), and theirdegradation products (uric acid, allantoin, �-alanine, �-ureidopropionic acid, �-aminoisobutyric acid) inblood plasma of dairy cows. The high performance liquid chromatography-based technique coupled toelectrospray ionization tandem mass spectrometry (LC–MS/MS) was combined with individual matrix-matched calibration standards and stable isotopically labelled reference compounds. The quantitativeanalysis was preceded by a novel pre-treatment procedure consisting of ethanol precipitation, filtra-tion, evaporation and reconstitution. Parameters for separation and detection during the LC–MS/MSanalysis were investigated. It was confirmed that using a log-calibration model rather than a linear cal-ibration model resulted in lower CV% and a lack of fit test demonstrated a satisfying linear regression.The method covers concentration ranges for each metabolite according to that in actual samples, e.g.guanine: 0.10–5.0 �mol/L, and allantoin: 120–500 �mol/L. The CV% for the chosen quantification rangeswere below 25%. The method has good repeatability (CV% ≤ 25%) and intermediate precision (CV% ≤ 25%)and excellent recoveries (91–107%). All metabolites demonstrated good long-term stability and good
stability within-runs (CV% ≤ 10%). Different degrees of absolute matrix effects were observed in plasma,urine and milk. The determination of relative matrix effects revealed that the method was suitable foralmost all examined PP metabolites in plasma drawn from an artery and the portal hepatic, hepatic andgastrosplenic veins and, with a few exceptions, also for other species such as chicken, pig, mink, humanand rat.
. Introduction
The global efficiency of nitrogen in animal production is onlylightly over 10%, with the result that 102 Tg (1012 gram) nitrogen is
excreted annually (1998 figures) by domesticated animals globally[1]. The nitrogen efficiency in dairy cows is generally low [2], andnot only the environment, but also the productive efficiency, wouldbenefit from an optimization of diet and metabolism to improvenitrogen efficiency and utilization [1,3,4]. Most research hithertohas focused on refining protein and amino acid utilization, but thishas only led to minor improvements in efficiency [4–6]. A better
understanding of the quantitative absorption and intermediarymetabolism of other nitrogenous products such as the purines andpyrimidines (PP), the building blocks of nucleic acids and mainconstituents of DNA/RNA, could uncover new ways of improving
airy cow nitrogen use-efficiency and propose new feeding strate-ies [7,8]. So far, the possible significance of microbial PP in theutritional physiology of ruminants has not been investigated,egardless of the fact that they correspond to more than 20% of theotal microbial nitrogen supply [7–9]. Little is known about theuantitative aspects of PP metabolism. What is known, however, ishat the purines go through an effective multistep degradation toric acid and allantoin, and the pyrimidines are similarly degradedo �-alanine, before excretion [8,10].
Quantitative analysis of PP in dairy cattle research has almostolely focused on purines in urine, as excretion of purine deriva-ives can be used as an indirect measure of rumen microbialynthesis [11–14]. Most published methods have thus been devel-ped for purine metabolites in urine. Only recently, Boudra et al.ublished a method able to quantify the pyrimidine degrada-ion products (DP) �-alanine and �-aminoisobutyric acid as well14].
Different analytical separation methods have been used foretermining PP in biological matrices of which the majority haspplied high performance liquid chromatography (HPLC) [15–17]r capillary electrophoresis chromatography [17–20]. When higheparation selectivity and sensitivity were essential, electrokineticechniques [16] or ultra high performance liquid chromatography21] have been used. Concerning detection, spectrometric, elec-rochemical or mass spectrophotometric detection methods haveeen used, with ultra violet detection coupled to HPLC being theost common one [15–17]. HPLC coupled with tandem spectro-etric detection (LC–MS/MS) is currently considered the method
f choice for quantitative analysis of compounds in biological matri-es [22] and LC–MS/MS has been shown to be capable of quantifyingP and their derivatives accurately in urine.
For this study, we wanted to develop and validate an LC–MS/MSethod for quantification of a range of PP and their derivatives
n cow blood plasma. Into this procedure, we wanted to incor-orate matrix-matched calibration standards as well as stable
sotopically labelled reference compounds (SIL). As no appro-riate pre-treatment procedure was identified in the literature,e also wanted to develop a good, stable, simple, component-
pecific, and repeatable pre-treatment protocol for the plasmaamples.
Several sets of plasma samples from experiments thatttempted to manipulate urea-recycling and increase nitrogentilization using multicatheterized Danish Holstein cows weremployed in the development of this method [23] because theseere representative of the types of samples that this method is
ikely to be used for in the future.
. Materials and methods
.1. Chemicals, reagents and materials
Water quality was at all times secured by treatment on aillipore Synergy® UV water treatment system from Millipore
.S. (Molsheim, France). Methanol (MeOH) from Poch S.A. (Gli-ice, Poland) and ethanol (EtOH 99.9% vol.) from Kemetyl A/S
Køge, Denmark) were of HPLC grade. Formic acid (98–100%)HCOOH), acetic acid (100%) (CH3COOH), and ammonium solution25%) (NH4OH) from Merck (Darmstadt, Germany) were of ana-ytical reagent grade. Sodium hydroxide (NaOH), also from Merck,
as prepared in a 0.01 M aqueous solution. Tricholoroacetic acid≥99.0%) from Sigma-Aldrich (Brøndby, Denmark) was prepared in
12% (v/v) aqueous solution (TCA) daily. Contamination betweenamples was minimized by the use of disposable materials (vials,ottles, etc.) where practicable, or through the use of lab equipmenthat was cleaned without the use of detergents.
. A 1356 (2014) 197–210
2.2. Standards
The following compound standards (bases (BS), nucleo-sides (NS), DP) were obtained from Sigma-Aldrich (Brøndby,Denmark): adenine, guanine, cytosine, thymine, uracil, adeno-sine, guanosine, cytidine, uridine, inosine, 2′-deoxyadenosine,2′-deoxyguanosine, 2′-deoxycytidine, thymidine, 2′-deoxyuridine,2′-deoxyinosine, xanthine, hypoxanthine, uric acid, allantoin,�-alanine, �-ureidopropionic acid and �-aminoisobutyric acid.�-ureidoisobutyric acid, one important intermediate pyrimidinederivate metabolite, was not commercially available and could notbe included. No traces of either adenosine or 2′-deoxyadenosinewere identified during method development in plasma or urinesamples. 2′-deoxycytidine was present in trace amounts but evenafter extensive optimization the sensitivity remained too low forquantification. These three components were therefore not pur-sued further. The chemical structures of the targeted metabolitesare shown in Table 1.
Stable isotopically labelled reference compounds used asinternal standards were purchased from Cambridge Isotope Lab-oratories (Andover, USA). These were: adenine (8-13C), guanine (8-13C;7,9-15N2), thymine (15N2), uracil (U-13C4;U-15N2), guanosine(U-13C10;U-15N5), inosine (U-15N4), cytidine (U-13C9;U-15N3),uridine (U-13C9;U-15N2), 2′-deoxyguanosine (U-15N5), thymidine(U-15N2), xanthine (1,3-15N2), hypoxanthine (15N4), uric acid (1,3-15N2), and �-alanine (U-13C3;15N). Cytosine (2,4-13C2;15N3) waspurchased from Sigma-Aldrich (Brøndby, Denmark). All were 13Cand/or 15N labelled with purities of at least 95% (95–99%). Unfor-tunately, exact SIL were not available for all metabolites studied;a suitable SIL was consequently selected on its similarity to thecorresponding metabolite in terms of structure, retention time,fragmentation pattern and group. Individual stock solutions of allcompound standards and SIL were prepared and kept at −80 ◦C.Bases and purine DP were diluted in water and NS and pyrimidineDP were diluted in 0.01 M NaOH solution. Two stock concentrationsof 500 and 5000 �mol/L were made for each compound standard.The exception was for uric acid and allantoin, where the stock con-centration was 500/2000 and 500/40,000 �mol/L, respectively. ForSIL only the low concentration stock was prepared. All stocks werefiltered through 0.45 �m PALL GHP Membrane syringe filters pur-chased from VWR (Herlev, Denmark) and kept at −20 ◦C in darkvials. Appropriate dilutions of these solutions were made in waterto produce standard mixtures and SIL mixtures for external cali-bration and quantification.
2.3. Samples
A number of 5 mL aliquots of heparinized plasma to be usedfor external calibration and quality control were prepared from2 L of venous blood [23] drawn from a Danish Holstein dairy cowfed a traditional total mixed ration. Experimental plasma sampleswere obtained from a feeding experiment [24] with multicatheter-ized dairy cows [25,26]. This set of samples was drawn from fourblood vessels simultaneously, representing blood from an arteryand the portal hepatic, hepatic and gastrosplenic veins. Additionaltest plasma samples were obtained on site for relative matrixeffect evaluations. These samples were from five other species(chicken, pig, mink, human, and rat) for between species compar-isons, four multicatheterized cows (jugular vein) for intraspeciescomparisons, and bovine urine and milk samples for matrix effectevaluations.
2.4. Pre-treatment
Before pre-treatment, plasma samples for quantification of uricacid and uracil were diluted twenty-fold (5%, v/v) and four-fold
C. Stentoft et al. / J. Chromatogr. A 1356 (2014) 197–210 199
Table 1Names, types, empirical formulae and suggestions for fragmentations of the compounds analyzed by the LC–MS/MS method.
Purines Pyrimidines
Name Type Empirical formula Name Type Empirical formula
AdenineFrag. 1
Base
N
N
NH
NNH2
Cytosine Base N
NH
NH2
O
GuanineFrag. 2
Base
N
NH
NH
N
NH2
O
Thymine Base
NH
NH
O
O
CH3
GuanosineFrag. 3
NSN
OOH
OH
N
NHN
NH2
O
OH
Uracil Base
NH
NH
O
O
Inosine NS NOOH
OH
N
NHN
O
OH
Cytidine NS NOOH
OH
N
NH2
O
OH
2′-deoxyguanosineFrag. 4
NS NOOH
OH
N
NHN
NH2
O
Uridine NS NOOH
OH
NH
O
O
OH
2′-deoxyinosine NS NOOH
OH
N
NHN
O
ThymidineFrag. 7 NS NOOH
OH
NH
O
O
CH3
200 C. Stentoft et al. / J. Chromatogr. A 1356 (2014) 197–210
Table 1 (Continued)
Purines Pyrimidines
Name Type Empirical formula Name Type Empirical formula
Xanthine Base/DP
NH
NH
N
NH
O
O
2′-deoxyuridine NS NOOH
OH
NH
O
O
HypoxanthineFrag. 5
Base/DPNH
N
N
NH
O
�-alanine Frag. 8 DPNH2
O
OH
Uric acid DP
NH
NH
NH
NH
O
OO
�-ureidopropionic acid DP
NH
NH
NH
NH
O
OO
AllantoinFrag. 6
DP
NH
NH2
NH
NH
OO
O�-aminoisobutyric acid DP NH2
O
OH
N f sugg
(tctgiabi4Oanair(mpvs
2
swpp
S, nucleoside; DP, degradation product. Illustrated with lines are the eight types o
25%, v/v) in water, respectively. This was, in the case of uric acid,o avoid a non-linear calibration curve with the very high uric acidoncentrations in all samples, and, in the case of uracil, to be ableo distinguish the small uracil signal from the pronounced back-round noise. Pre-treatment: plasma samples were defrosted andmmediately put on ice. The sample (300 �L) was then added to
SIL mixture and a water/standard mixture (550 �L total vol.)efore being precipitated with 1.8 mL ice-cold ethanol (10 min, on
ce, −20 ◦C). This was followed by centrifugation (15 min, 5500 × g,◦C). The supernatant was ultrafiltered on a Pall Nanosep 10K,mega membrane spin filter purchased from VWR. A 500 �Lliquot of filtered supernatant was dried down under a flow ofitrogen on a SuperthermTM fitted with a Mini Oven for AI blocksnd evaporator with valves from Mikrolab A/S (Aarhus, Denmark)n conical autosampler vials from VWR until dryness (app. 75 min.,oom temp.). The pellet was re-suspended in 100 �L cold solventA) (30 min, 4 ◦C) and transferred to a clean dark LC-vial. Matrix-
atched external calibrators were treated similarly to standardlasma. Milk samples were cleared with ice-cold TCA 12% (end 50%,/v) before pre-treatment. Urine samples were handled as plasmaamples throughout.
.5. LC–MS/MS analysis
Chromatographic separation was performed on an Agilent 1100
eries HPLC system (Agilent Technologies, Hørsholm, Denmark)ith a SynergiTM Hydro-RP LC Column (250 mm × 2 mm, 4 �m)rotected by a conventional guard column of the same materialurchased from Phenomenex (Værløse, Denmark). Samples were
ested metabolite fragmentations.
analyzed in five separate runs, three in negative electrospray (ESI)mode and two in positive ESI mode. The five groups of metabolitesand their chromatographic profiles are shown in Table 2. Separa-tion was performed using a gradient solvent system. For each run,HPLC solvents were freshly prepared and cleared on a 0.45 �m Pallhydrophilic polypropylene membrane filter purchased from VWR.Both solvents (A) and (B) were prepared from a 0.05 mol/L aceticacid buffer containing 10% or 50% methanol, respectively. The aceticacid buffer was prepared by adjusting 0.05 mol/L acetic acid to pH4.0 with ammonium solution and readjusting to pH 2.8 with formicacid. The following elution gradient was used: initial percentage ofsolvent B was 5%, this was raised to 100% in 8 min and kept there for6 min, then lowered to 5% in 30 s, after which it was kept constantfor 3.5 min to re-equilibrate the column prior to the next injection.The flow rate was 200 �L/min and the injection volume was 5 �L.The column temperature was maintained at 30 ◦C while the autosampler temperature was set to 4 ◦C to stabilize the samples dur-ing time-consuming analyses. The total run time was 18 min persample.
A Waters (Hedehusene, Denmark) micromass triple quadropolemass spectrometer was used for electrospray mass spectrometricanalyses using massLynx 4.0 (Waters) software for data collectionand processing. Capillary voltage was set to 3.2 kV, source temper-ature to 120 ◦C, and desolvation temperature to 400 ◦C. The coneand desolvation gas flows (nitrogen and argon) were set at 29
and 628 L/h, respectively. Fragment ion spectra were recorded inboth polarities and promising selective fragment ions were testedand optimized along with the cone voltage in multiple-reactionmonitoring (MRM) mode. The values of the tune parameters were
C. Stentoft et al. / J. Chromatogr. A 1356 (2014) 197–210 201
Table 2The 20 metabolites were divided into five groups and run according to ESI−/+ mode and structure.
P, degradation product; NS, nucleoside. Plasma samples and standard plasma forv/v), respectively, in water. A group 5 chromatographic profile (uric acid) is not illuame shape, same RT).
ptimized by separately infusing a solution (500 �mol/L) of eachetabolite in its mobile phase at a flow rate of 10 �L/min. The MRM
ransitions and the applied cone voltages and collision energies areummarized in Table 3. Common transitions were originated fromhe loss of HCN, NH3, ribose, deoxyribose, HNCO, HNCONH2 and2O fragments for the various PP metabolites (Table 1). The most
ntense transition reaction was used for quantification (Table 3).ata were collected in centroid mode with a constant dwell timef 0.05 s and an interscan delay of 0.02 s.
.6. Calibration and quantification
Quantification was performed by matrix-matched external cal-bration applying standard plasma spiked with a two-fold serialilution of mixed standard solutions to obtain seven differentoncentration levels of each compound. The only exception wasith uracil where a two-third-fold serial dilution was applied.
tandard plasma (not spiked) was used for subtraction and qualityontrol but was not included in the regression analysis. In gen-ral, all samples and calibrators were analyzed in duplicate and atandard curve and quality control samples were analyzed at theeginning and at the end of each sequence. The response was calcu-
ated as the chromatographic peak area for all compounds. Whenpplying standard plasma, which contained unknown quantitiesf the metabolites under investigation, the measured metabo-ite response was initially normalized and the response fromhe standard plasma was subtracted. The mean of the measured
IL responses/SIL area for each sample was used as the normal-zation factor. During method development the focus of work
as on quantifying as low concentrations of metabolite as possi-le.
tification and external calibration of uracil and uric acid were diluted 25% and 5%d in the table since uric acid (1,3-15N2) can be observed with group 1 (same peak,
Matrix-matched calibration curves, within the relevant concen-tration ranges given in Table 4, were generated for each metaboliteat four (allantoin) or seven concentration levels on five consecutivedays for determining and evaluating the calibration model. As notedpreviously, uric acid and uracil were quantified from diluted sam-ples. The coefficient of variation (CV%) for each concentrate levelwas then calculated for a logarithmic and a linear calibration modelto test the use of log–log transformation. The linearity of the logcalibration curves were studied with a lack of fit hypothesis test.Subsequently, the homogeneity of variance was estimated for eachconcentration by plotting the CV% against log(concentration) andthe quantification range set to the lowest and highest quantifiedconcentration giving a CV% below 25%.
2.7. Validation procedure
The method was validated according to reports from the “Ana-lytical methods validation: bioavailability, bioequivalence andpharmacokinetic studies” conferences held in Washington in 1990[27] and 2000 [28], as described by Peters et al. [29]. It was vali-dated with respect to assessment of selectivity, stability, precision,recovery, and matrix effect.
2.7.1. SelectivityMetabolite and SIL cross-talk was evaluated by analyzing the
standard compounds alone and together with their correspond-ing SIL (no blank matrix was available). Three groups were studied
and their signals compared; a compound standard group (10%, v/v,50 �mol/L), a SIL group (10%, v/v, 50 �mol/L), and a combined group(5%, v/v, 25 �mol/L). Analyses of BS/DP and NS were carried outseparately.
202 C. Stentoft et al. / J. Chromatogr. A 1356 (2014) 197–210
Table 3Transition reactions monitored by LC–MS/MS, cone voltages and collision energy for the metabolite/stable isotopically-labelled reference compound (SIL) analyzed, andsuggested corresponding fragments lost.
Metabolite/SIL Mw (g/mol) Retentiontime (min)
Precursor ion (m/z) Conevoltage (V)
Product ion(m/z)
Collisionenergy (eV)
Neutralloss (NL)
Fragmentation 1–8
PurinesAdenine/
Adenine (8-13C)135.13136.12
3.81 134135
– 3536
107108
1617
27 –HCN 1
Guanine/Guanine(8-13C,7,9-15N2)
151.13154.11
3.86 150153
– 2830
133136
1313
17 –NH3 2
Guanosine/Guanosine(U-13C10;U-15N5)
283.24298.13
6.18 282297
– 3333
150160
1920
132137
–Deoxyribose 3
Inosine/Inosine (U-15N4)
268.23272.20
5.81 267271
– 2626
135139
2020
132 –Deoxyribose 3
2′-deoxyguanosine/2′-deoxyguanosine(U-15N5)
267.24272.17
7.31 266271
– 2628
150155
1920
116 –Ribose 4
2′-deoxyinosine/2′-deoxyguanosine(U-15N5)a
252.23–
6.74–
251–
– 27–
135–
20–
116 –Ribose 4
Xanthine/Xanthine (1,3-15N2)
152.11154.10
5.18 151153
– 2931
108109
1616
4344
–HNCO 5
Hypoxanthine/Hypoxanthine (15N4)
136.11140.09
4.56 135141
+ 3434
92113
1619
4327
–HNCO –HCN 51
Uric acid/Uric acid (1,3-15N2)
168.11170.10
4.28 167169
– 2629
124125
1614
4344
–HNCO 5
Allantoin/Uric acid (1,3-15N2)a
158.12–
3.05–
157–
– 16–
97–
16–
60 –HNCONH2 6
PyrimidinesCytosine/
Cytosine(2,4-13C2;15N3)
111.95116.08
2.91 112117
+ 2930
9599
2019
1718
–NH3 2
Thymine/Thymine (15N2)
126.11128.10
6.21 127129
+ 2727
110111
716
1718
–NH3 2
Uracil/Uracil(U-13C4;U-15N2)
112.09118.04
3.97 113119
+ 2627
96101
716
1718
–NH3 2
Cytidine/Cytidine(U-13C9;U-15N3)
243.22255.13
3.19 242254
– 2321
109116
1415
133138
–Deoxyribose 3
Uridine/Uridine(U-13C9;U-15N2)
244.20255.12
4.50 243254
– 2328
110116
1516
133138
–Deoxyribose 3
Thymidine/Thymidine (U-15N2)
242.23244.22
8.52 241243
– 2526
151153
1211
90 –Rearrangement 7
2′-deoxyuridine/2′-deoxyguanosine(U-15N5)a
228.20–
5.34–
227–
– 22–
184–
12–
43 –HNCO 5
�-alanine/�-alanine(U-13C3;15N)
89.0993.07
2.91 9094
+ 1314
7276
107
18 –H2O 8
�-ureidopropionicacid/ˇ-alanine(U-13C3;15N)a
132.12–
3.77–
133–
+ 11–
115–
10–
18 –H2O 8
�-aminoisobutyricacid/ˇ-alanine
13 15 a
103.12–
2.98–
104–
+ 13–
86–
10–
18 –H2O 8
S retenta ragme
2
qbqesfuio
(U- C3; N)
IL, stable isotopically-labelled reference compound. All metabolites had a specific
This SIL was selected as the most suitable according to structure, retention time, f
.7.2. StabilityFor continuous evaluation of long-term storage stability, a fresh
uality control sample was analyzed in all analytical runs. The sta-ility within runs (6–24 h) was evaluated in two ways. First, auality control sample was analyzed at the beginning and at thend of each sequence (data not shown). Secondly, a set of spikedtandard plasma samples were analyzed at five different times (dif-
erent vials) during a 30-h sequence. Analysis of variance (ANOVA)sing linear mixed models procedures was used to test the stabil-
ty over time, both with a trend element and with random changesver and above the linear trend (regression line) [30,31]. Applying
ion time and generated single peak shapes.ntation pattern and metabolite group.
ANOVA, the across-day variation of the PP calibration curves (inter-cepts and slopes as interactions with test day) was assessed overfive consecutive days and expressed by their P-values. The stabil-ity during repeated freeze-thaw cycles was not explored since allplasma samples in the present study were only thawed once.
2.7.3. Precision and recovery
Precision of the method, in terms of within-day variation
(repeatability) and across-day variation (intermediate precision),was determined by analyzing replicate sets of spiked standardplasma samples on five separate days expressed as their CV%. The
C. Stentoft et al. / J. Chromatogr. A 1356 (2014) 197–210 203
Table 4Concentration level, calibration range, lack-of fit, quantification range and precision of the metabolite calibration curves.
Metabolite Type Rangea Linearity Precision (test-day)d
NS, nucleoside; DP, degradation product. Only four curves were available for uric acid and �-ureidopropionic acid. In the case of allantoin, the three lower concentrationlevels were excluded to better fit the concentration range of actual samples. For uridine, one observation in one curve was considered an outlier following visual inspectionand was rejected.a External calibration was performed with seven concentrations of metabolite on five separate days (n = 5, days), except for allantoin where only four concentration levelswere available. The ranges where chosen according to concentration ranges in actual samples.b Lack of fit hypothesis test to validate the linearity of the calibration curves expressed by their P-values (n = 5, curves). P < 0.05 was considered significant.c The quantification range was set to the lowest and highest quantified concentration giving an acceptable CV% < 25% (see Fig. 2).d ractiose
ast
2
rsatsswbccmbtTm
2
msppuf
The intermediate precision of the calibration curves (intercepts and slopes as inteignificant, P < 0.1 a tendency.Value is above the highest calibrator concentration.
bsolute recoveries were calculated using the same set of spikedtandard plasma, at one level, by comparing the obtained concen-rations with the initial spiked level.
.7.4. Matrix effectEarly tests with spiked water, urine and plasma samples
evealed large variations in matrix effect-induced signal suppres-ion and enhancement between the metabolites included in thenalysis. Following optimization of the pre-treatment procedure,hese matrix effects were evaluated as the difference betweenamples of water and standard plasma, urine or milk samplespiked with constant amounts of SIL before pre-treatment. Thus,e took advantage of the fact that the incorporated SIL should
ehave as their matching metabolite in the ESI source [27]. Theonventional strategy of spiking a blank matrix sample with aompound standard was again not possible as completely blankatrices were not available for these metabolites. The applied SIL-
ased method was a modified version of the conventional methodo evaluate matrix effect described by Matuszewski et al. [32].he observed matrix effect was rendered insignificant by utilizingatrix-matched external calibration.
.8. Application
To determine the application range of the method, the relativeatrix effect was evaluated by comparing the response from PP SIL
piked in standard jugular vein plasma with the response in test
lasma samples. Four different sets of samples were assessed. First,lasma from the jugular vein of four multicatheterized cows wassed to investigate within-species variation. Next, plasma drawnrom the portal vein, the hepatic vein, the gastrosplenic vein, and
ns with test day) expressed by their P-values (n = 5, days). P < 0.05 was considered
an artery from a multicatheterized dairy cow to represent differ-ent possible sampling sites were examined. Third, plasma samplesfrom different species (chicken, pig, mink, human, rat) were usedfor between-species evaluation. Finally, water, urine and milk sam-ples were used to compare different matrices. The relative recoverydetermined which of the tested matrices were suitable for themethod. For the same reasons as described previously, SIL replacedcompound standards. Water, urine and milk samples were evalu-ated in the same manner as plasma samples.
3. Results and discussion
3.1. Method development
The aim of this study was to develop a quantitative LC–MS/MSanalysis and a sample pre-treatment procedure for the simulta-neous analysis of several metabolites of the PP metabolism in bloodplasma of dairy cows. The chemical properties of the metaboliteswere polar due to high contents of –OH, =O and –N groups. Based ontheir polarity, they were roughly divided into three groups: The verypolar group, containing �-alanine, �-aminoisobutyric acid and �-ureidopropionic acid, were all small molecules with similar linearpolar structures, as well as the also highly polar allantoin, cytosineand cytidine. The polar group included the majority of the BS, suchas adenine, guanine and uracil, as well as the intermediate DP withmore base-like structures, such as uric acid, xanthine and hypoxan-thine. Finally, the semi-polar group comprised the majority of the
NS with large but semi-polar sugar side groups, such as most ofthe ribonucleosides (2× –OH) and deoxyribonucleosides (1× –OH).Owing to their very non-polar methyl side groups, thymine andthymidine were also placed in the semi polar group. The very polar
2 atogr
mot
3
Lassetbw
mnatTlirtbawkpmten
mdttopo
3
leqat
ctbomaa(wt4itlat
04 C. Stentoft et al. / J. Chrom
etabolites were poorly retained on the C18 column with the aque-us solvents and eluted first as expected, offering a longer retentionime of the less polar components.
.1.1. Pre-treatment development and evaluationAn effective clean-up procedure is crucial when performing
C–MS/MS analysis as this diminishes cross-talk [33,34] as wells matrix effects [35] and at the same time enhances both theelectivity and the sensitivity of the analysis [29]. A novel multi-tep approach, consisting of protein precipitation, ultrafiltration,vaporation under nitrogen flow, and subsequent resolution, ableo purify and to concentrate all of the studied metabolites fromovine plasma simultaneously, in a simple and efficient manner,as developed and optimized.
Initially, different solvents (acetone, acetonitrile, ethanol,ethanol, sulfo-salicylic acid) were tested for precipitation (data
ot shown). Ethanol precipitation resulted in the highest recoveriesnd least noise when comparing chromatographic responses andhis less harmful solvent was therefore chosen for the procedure.he ultrafiltration step was added as this step caused markedlyower levels of background noise. As a consequence of the approx-mately eight-fold dilution during pre-treatment, evaporation andeconstitution steps were included. Overall this resulted in a 1.4imes concentration effect. To try to reduce degradation and insta-ility of the samples caused by reactive oxygen species or enzymectivities during pre-treatment, all centrifugations and incubationsere performed at 4 ◦C and samples, stocks, and solvents, etc., were
ept at −4 ◦C or on ice. Only during evaporation were the sam-les maintained at room temperature. Other types of pre-treatmentethods such as simple dilution (impractical), solid-phase extrac-
ion (different chemical properties) [36,37] and accelerated solventxtraction [38] were also investigated (data not shown) but wereot found useful.
The effectiveness of the pre-treatment and the stability of theetabolites during the multiple steps were evaluated during vali-
ation of the method, described in Section 3.3, and demonstratedhe ability of this pre-treatment to purify and concentrate all of theargeted PP simultaneously in an easy and efficient manner with-ut significant losses. To our knowledge, no other publications haveresented a similar and effective pre-treatment procedure, as mostther approaches include dilution of the samples.
.1.2. LC–MS/MS procedureBased on the chemical properties of the targeted metabo-
ites, experiences from similar studies [14,39], and availablequipment, a reversed-phase C18 column known to be able touantify the majority of the studied metabolites from urine waspplied with an acetic acid buffer/methanol HPLC solvent sys-em.
To achieve adequate separation and elution order, a series ofonditions were modified and implemented. The composition ofhe acetic acid buffer and the methanol extraction solvent wasased on the work of Hartmann et al. [39], and no other typesf solvent were tested. Having tested several acetic acid buffer toethanol ratios (95%, 90%, 85%, and 80%, v/v), assessing peak sep-
ration and shapes, it was concluded that the best separation wasccomplished with a 90% (v/v) solvent (A) and 50% (v/v) solventB). The chosen injection volume, 5 �L, and flow rate, 200 �L/min,as found by assessing the same parameters, testing first injec-
ions of 5, 10, 20 �L and then flow rates of 100, 200, 300 and00 �L/min. Concerning the elution gradient, we strived to make
t as short as possible, while still achieving as good a peak separa-
ion as possible. Different elution profiles were tested, with more oress steep gradients. The final profile, described in Section 2.5, gave
total run time of 18 min. By adding a small amount of methanolo the otherwise aqueous solvent (A), and, by keeping the baseline
. A 1356 (2014) 197–210
at 5% solvent (B), the solvent mixing became more smooth andtransitions between runs became more stable. A major improve-ment in precision between runs was achieved by maintainingthe column temperature at 30 ◦C instead of 25 ◦C. An improve-ment in the sample stability during the time-consuming analyseswas achieved by cooling the auto-sampler to 4 ◦C. In the end,useful combinations of retention times and peak shapes of eachmetabolite were achieved with the parameters described, and themethod was therefore adapted and brought on to further valida-tion.
3.2. The log-calibration model and quantification range
Calibration curves were prepared by linear regression oflog(area) against log(concentration) (log-calibration) and by linearregression in linear units on both axes (linear calibration) to verifythe use of the log-calibration model. Initially, the normality of resid-uals around the calibration lines were inspected visually (Q–Q plot)and found to be approximately normal. The CV% for each concen-tration level for both the log-calibration and the linear calibrationis illustrated in Fig. 1. A large group of the PP (panel I) considerablyimproved their CV% profiles using the log-calibration, especially inthe low ranges. However, a smaller group of PP (panel II) did notbenefit from the log transformation; and the transformation did notweaken as their CV% profiles either. Exceptions were with allantoin,�-ureidopropionic acid, cytosine and �-alanine, their CV% at thehigh end of their profiles were better without the log-log trans-formation. Given that quantification at low concentrations wasconsidered to be most important, these findings validated the useof log–log transformation in the analysis of all the applied PPs. Per-forming a lack of fit test, the linearity of the PP calibration curveswere evaluated and expressed by their P-values (Table 4). Noneof the PP curves resulted in a significant lack of fit except uridine,which had a very low sensitivity in the analysis, demonstrating asatisfying log–log regression.
The homogeneity of variance for the different concentration lev-els is illustrated in Fig. 2 and the quantification ranges (CV < 25%)in Table 4. Focusing on the lower concentration range, most ofthe PP demonstrated a typical precision profile where the CV%decreased with higher concentration levels. All purines had accept-able variation levels around the lowest concentration levels exceptallantoin, which should not be quantified at concentrations below∼100 �mol/L. The pyrimidine BS and cytidine and uridine hadlarger CV%’s with acceptable lower concentration levels from 0.66to 5.15 �mol/L. Thymidine and 2′-deoxyuridine demonstrated avery large variation with CV%’s above 25% over the entire con-centration range. In the case of the pyrimidine DP, they werereasonably stable over their concentration ranges, not counting �-alanine which only had a CV% < 25% at its highest calibrator. Theupper part of the quantification range was in all cases the highestquantified calibrator.
3.3. Method validation
Once the pre-treatment, LC–MS/MS procedure, and calibrationmodel had been set, the performance characteristics of the methodwere established by validation with spiked standard plasma. Interms of quantification purposes, selectivity, stability, precision,recovery, and matrix effects were evaluated.
The most intensive fragment ion from each precursor ion wasselected as the transition ion for detection and quantification. Pos-
itive identification was based on the correlation of retention timewith the standards and the selected precursor/product transition.Less intensive second transitions were used for confirmation. Allmetabolites generated single peak shapes.
C. Stentoft et al. / J. Chromatogr. A 1356 (2014) 197–210 205
1
10
100
1000
10001001010.10.01
CV
%
Standard concentration (μmol/L)
Linear calibration I
Ade Gu a Guo In o dGuo
dIno Xan Hyp Uac All
Urd β-ure β-am i
1
10
100
1000
10001001010.10.01
CV
%
Standard concentration (μmol/L)
Log-calibration I
Ade Gu a Gu o In o dGuo
dIno Xan Hy p Ua c Al l
Urd β-ure β-am i
1
10
100
1000
10001001010.10.01
CV
%
Standard concentration (μmol/L)
Linear calibration II
Cyt Thy Ur a Cyd
Thd dUrd β-ala
1
10
100
1000
10001001010.10.01
CV
%
Standard concentration (μmol/L)
Log-calibration II
Cyt Th y Ur a Cy d
Thd dUrd β-ala
Fig. 1. The coefficient of variation (CV%) for each concentration level using linear regression of area against concentration (linear calibration) and using linear regression oflog(area) against log(concentration) (log-calibration). Panel I present the 13 purines and pyrimidines that considerably improved their CV% profiles using the log-calibration.P log trG an, xau Ura, u
3
fptbptrssdra
3
a
anel II, present the seven purines and pyrimidines that did not benefit from theuo, guanosine; Ino, inosine; dGuo, 2′-deoxyguanosine; dIno, 2′-deoxyinosine; Xreidopropionic acid; �-ami, �-aminoisobutyric acid; Cyt, cytosine; Thy, thymine;
.3.1. SelectivityA blank sample for selectivity evaluation was not available
or these naturally occurring plasma metabolites. Hence, theresence of chromatographic peaks from standard plasma athe same retention times as the targeted metabolites could note excluded; such endogenous peaks would be expected to beresent. Instead, the absence of standard compound/SIL cross-alk contributions was verified by comparing chromatographicesponses for standards and SIL alone and in a mixture (data nothown). It was important to assess cross-talk contributions, asome of the applied SIL (Table 3) had less than three mass unitifferences (3–8) to the natural metabolite, which is normallyecommended as the lowest mass unit difference for LC–MS/MSnalysis [33,34].
.3.2. StabilityGood stability was achieved by optimizing the pre-treatment
nd LC–MS/MS parameters as described in Section 3.1. Long-term
ansformation. Abbreviations for the 20 metabolites: Ade, adenine; Gua, guanine;nthine; Hyp, hypoxanthine; Uac, uric acid; All, allantoin; Urd, uridine; �-ure, �-racil; Cyd, Cytidine; Thd, thymidine; dUrd, 2′-deoxyuridine; �-ala, �-alanine.
storage stability was tested by comparing chromatographic pro-files of quality control standard plasma on a daily basis. Within-runstability was evaluated by analyzing a control sample at the begin-ning and end of each sequence. Long sequence run times havebeen of concern and the within-run stability was consequentlyalso evaluated by performing ANOVA for measurements made attimes 0, 7, 15, 22 and 29 h, during a 30-h sequence with trip-licate determinations at each time-point, using either a slopemodel: yij = intercept + b × time hour + εij, or a combined model:yij = intercept + timei + b × time hour + εij, where yij is the area mea-sured in the sample at time i, replicate j, and b is the slope of the areachange per hour, and εij is the random error term. Significance of thetime effects were tested using an F-test with type 1 sum of squares.Residual mean square error was calculated as the square of the
residual variance estimate and expressed as CV%. The metaboliteresponses were normalized as usual but the SIL responses were notsince they could not be used to normalize themselves. The resultsare given in Table 5.
206 C. Stentoft et al. / J. Chromatogr. A 1356 (2014) 197–210
0
5
10
15
20
25
30
35
1010.10.01
CV
%
Standard concentration (μmol/L)
Purine bases
Ade Gua Xan Hyp
0
5
10
15
20
25
30
35
40
1010.10.01C
V%
Standard concentration (μmol/L )
Purine nucleosides
Guo In o dGuo dIno
0
5
10
15
20
25
30
35
40
45
50
1000100101
CV
%
Standard concentration (μmol/L)
Purine degradation products
Uac All
0
10
20
30
40
50
60
70
80
1010.10.01
CV
%
Standard concentration (μmol/L)
Pyrimidine bases
Cyt Th y Ur a
0
20
40
60
80
100
120
1010.10.01
CV
%
Standard concentration (μmol/L )
Pyrimidine nucleosides
Cyd Ur d Th d dUrd
0
10
20
30
40
50
60
1001010.10.01C
V%
Standard concentration (μmol/L)
Pyrimidine degradation products
β-ala β-ur e β-am i
Fig. 2. The homogeneity of variance for the different concentration levels of the purine and pyrimidine calibration curves divided into bases, nucleosides and degradationproducts (CV%). Abbreviations for the 20 metabolites: Ade, adenine; Gua, guanine; Guo, guanosine; Ino, inosine; dGuo, 2′-deoxyguanosine; dIno, 2′-deoxyinosine; Xan,xanthine; Hyp, hypoxanthine; Uac, uric acid; All, allantoin; Urd, uridine; �-ure, �-ureidopropionic acid; �-ami, �-aminoisobutyric acid; Cyt, cytosine; Thy, thymine; Ura,uracil; Cyd, Cytidine; Thd, thymidine; dUrd, 2′-deoxyuridine; �-ala, �-alanine.
Table 5Stability of each metabolite/stable isotopically labelled reference compound during a 30-h sequence.
SIL, stable isotopically labelled reference compound. An appropriate concentration level was chosen for each metabolite/SIL according to their sensitivity in the analysis.The stability (significance of time) of each metabolite/SIL was expressed by their CV% using either a slope- or a combined model. The data handling was conducted withmetabolite responses in area units. If the CV% ≤10% the stability was considered acceptable over time.a SIL used for more than one metabolite.
C. Stentoft et al. / J. Chromatogr. A 1356 (2014) 197–210 207
Table 6The recovery and within- and across-day variation of each metabolite investigated.
Only four curves were available for uric acid and �-ureidopropionic acid. In the case of allantoin, the three lower concentration levels were excluded to better fit theconcentration range of actual samples. For uridine, one observation in one curve was considered an outlier following visual inspection and was rejected. An appropriateconcentration level was chosen for each metabolite according to the metabolites sensitivity in the analysis.a Recovered quantified concentration.b The recovery (%) was calculated as: (mean recovery concentration/mean spiked concentration) × 100 (n = 8, samples). Recovery (%) was an average of recoveries obtainedoc
d
tcdC(fttc(2u(s
r(aiwara
3
pssduCo
ver 5 days (m = 5, days).The within-day variation (n = 8, samples) expressed as CV%.The across-day variation (m = 5, days) expressed as CV%.
In general, the combined model resulted in lower CV%’s thanhe slope model, as the irregular time effect was also taken intoonsideration in the combined model. All but a few metabolitesemonstrated very stable profiles over the 30-h time span withV% ≤ 10%. Exceptions were thymidine (136%), 2′-deoxyuridine46%) and �-alanine (13%), where especially the former two wereound to be unstable. This was probably due to low sensitivities inhe analysis. The SILs were found to be equally or more stable thanheir corresponding metabolites probably due to their higher spikeoncentrations. As expected, thymidine (U-15N2) and �-alanineU-13C3;15N) had the same instability issues as their partners. No′-deoxyuridine SIL was applied in this analysis. Surprisingly, theracil and cytidine SIL had CV%’s above 10%. In the case of uracil13%), excessive degradation was avoided by always placing uracilamples in the beginning of a sequence.
To assess the stability of the calibration curves betweenun-days, ANOVA was conducted determining the across-dayintermediate precision) precision (Table 4). Most PP demonstrated
significant (P < 0.05) difference between test days on either curventercept or slope, or at least a tendency (P < 0.1). Exceptions were
ith allantoin, cytosine, uridine, thymidine and 2′-deoxyuridine,ll of which revealed reasonably stable curves over test days. Theseesults demonstrated the need for renewing calibration curves on
daily basis.
.3.3. Precision and recoveryTo ensure correct quantification and to evaluate analytical
recision, within-day and across-day variation was determined bytudying replicate sets of spiked standard plasma samples (n = 8,amples) on five separate days (m = 5, days). Here, precision was
efined as the degree to which repeated measurements undernchanged conditions showed the same result, expressed as theV%. Absolute recoveries were identified by using the same setf spiked standard plasma samples, comparing the recovered
quantified concentrations with the initial spiked concentrations.Since linearity ranges were short and close to zero, a single, insteadof the traditional three, recovery concentration levels was chosen.Precision and recovery outcomes are given in Table 6. The obtainedresults showed very good extraction efficiency and precision. Therecoveries were between 91% and 107%, except for uric acid with alower recovery of 78%. Also, the low sensitivity and accompanyinginstability of cytidine, thymidine and 2′-deoxyuridine was againhighlighted with recoveries of 162%, 121%, and 149%, respectively.In general, the within- and across-day variations mirrored therecovery results. The exceptions were with allantoin and cytosine,both of which had good recoveries, 107% and 103%, but exhibitedlarge CV%’s, within-day variation 34% and 21%, and across dayvariation 49% and 24%, respectively.
3.3.4. Absolute matrix effectIt is useful to distinguish between two types of matrix effects:
absolute matrix effect, which is the difference in response betweenan undiluted solution and a post-extraction spiked sample, and rel-ative matrix effect (Section 3.4), which is the difference betweenvarious lots of post-extraction spiked samples [32]. Matrix effectsare very common problems when applying LC–MS/MS analysison biological samples [22,35,40]. The term describes the effectmolecules originating from the sample matrix can have on the ion-ization process in the mass spectrometer when co-eluting with thecompound of interest. It theoretically occurs in either the solutionor the gaseous phase and the main cause is a change in dropletsolution properties caused by the presence of non-volatile andless volatile solutes that change the efficiency of droplet forma-tion and evaporation, which in turn affects the amount of charged
ions in the gas phase that ultimately reach the detector [35]. Asthe effect occurs in the ESI source before detection, it is hard tocompensate for by mass spectrometry alone [41,42]. In this analy-sis, the matrix effect was quantified by comparing the response of
208 C. Stentoft et al. / J. Chromatogr. A 1356 (2014) 197–210
-2500-2000-1500-1000
-5000
5001000
MilkUrinePlasmaRes
pons
e re
lativ
e to
wat
er (a
rea)
Purine bases and nucelosides
Adenine (8-13C) Guanine (8-13C,7,9-15N2)
Guanosine (U-13C10;U-15N5) Inosine (U-15N4)
2’-deoxyguanosine (U-15N5) Xanthine (1,3-15N2)
-40000
-20000
0
20000
40000
60000
80000
MilkUrinePlasma
Res
pons
e re
lativ
e to
wat
er (a
rea)
Purine degradation products
Uric acid (1,3-15N2) Hypoxanthine (15N4)
-4000
-3000
-2000
-1000
0
1000
2000
espo
nse
rela
tive
to w
ater
(are
a)
Pyrimidine bases and nucleosides
Thymine (15N2) Uracil (U-13C4;U-15N2)
Cytidine (U-13C9;U-15N3) Uridine (U-13C9;U-15N2)
Thymidine (U-15N2)
-80000
-60000
-40000
-20000
0
20000
40000
60000
80000
Res
pons
e re
lativ
e to
wat
er (a
rea)
Cytosine and β-alanine
β-alanine (U-13C3;15N ) Cytosine (2,4-13C2;15N3)
ilk ex
SoF
sbmmane[idrmse
ntamueSnpit
MilkUrinePlasm aR
Fig. 3. Matrix effects in plasma, urine and m
IL in spiked matrix samples before extraction with the responsebtained in water. Matrix effects for all SILs are illustrated inig. 3.
Recognizing that the nature of matrix effects is varying and theensitivity between metabolites are very different the sizes of thears are relative indicators of the degree of suppression or enhance-ent. Signal enhancement was observed in plasma for almost alletabolites, and only a few, such as inosine, cytidine, �-alanine
nd cytosine, had their signals suppressed. These metabolites didot share any obvious similarities in polarity or structure; how-ver, matrix effects are known to be very compound-dependent22]. In contrast to the signal enhancement generally encounteredn plasma, in urine all metabolite signals were suppressed. Thisemonstrates the different matrix effects a given component expe-ience when present in different matrices in LC–MS/MS analysis. Inilk, only the purines had a common pattern, i.e., signal suppres-
ion, and the remaining metabolites were neither suppressed nornhanced.
Matrix effects can vary between measurements, hence, it isot possible to test for matrix effects only once and consider ito be constant [43]. Matrix effects were largely eliminated in thenalysis first of all by making the external calibrators matrix-atched, hence, quantifying calibrators and sample metabolites
nder the same conditions, secondly, by implementing a veryffective pre-treatment [33,44], and thirdly, by implementingIL [22,42]. Matrix-matching is necessary when specific SILs are
ot available for all metabolites [42]. These initiatives com-ensated quite well for the signal suppression or enhancement
n the plasma samples, thereby achieving accurate quantifica-ion.
This LC–MS/MS analysis was established for quantification of 20target metabolites of the PP metabolism in blood plasma samplesfrom multicatheterized cows. Since jugular vein plasma (represent-ing systemic circulating blood) was used for method developmentand because quantification relied on matrix-matched calibration(jugular vein plasma), the relative matrix effect was evaluated inalternative types of plasma. The relative matrix effect was evalu-ated by comparing the response from SIL spiked in standard jugularvein plasma with the response in tested plasma samples. A rel-ative recovery between 85% and 115% was considered good andbetween 75% and 125% acceptable, hence, tested samples exertedthe same matrix effect on the metabolite as the cow jugular veinplasma sample. The generosity of 75–125% was due to the smallsample size (n = 2 samples) inevitably resulting in less precision.The PP responses given as recovery (%) are depicted in Table 7.
First of all, it was confirmed that within-species variationwas not an issue with any of the metabolites examined, exceptfor uridine. Secondly, the results demonstrated that all theexamined metabolites, evaluated in all four plasma types fromfeeding experiments with multicatheterized cows with this par-ticular type of cow model, could appropriately be quantified withthe developed LC–MS/MS method. Only xanthine (67%), uridine(135%/148%/127%) and thymidine (132%) displayed recoveries out-side the acceptable range of 75-125% and especially thymidine will
be hard to quantify with this method due to other issues anyway.Surprisingly, the between-species range was very broad and mostmetabolites could be evaluated in plasma from other species testedwith a few exceptions. Further confirmed was also the results from
C. Stentoft et al. / J. Chromatogr. A 1356 (2014) 197–210 209
Table 7Comparison of the response from the metabolites (stable isotopically labelled reference compounds) spiked in standard jugular vein plasma with the response obtained intested plasma samples from four other cows, four other blood vessels, five other animal species and three other matrices, to evaluate relative matrix effect and the applicationrange of the method.
SIL Four cows Four vessels Five species Three matrices
nd between 75% and 125% was considered acceptable. Shaded areas show recover
ection 3.3.4, concluding that matrix effects varied significantlyetween different types of matrices such as water, plasma, urinend milk. Hence, it is necessary to design, optimize and validate apecific LC–MS/MS method for each applied matrix.
. Conclusions
This work presents the development and validation of a newethod for simultaneous and accurate quantification of 20 targetedetabolites of PP metabolism with different structures and physio-
hemical properties in blood plasma from dairy cows. Exceptionsere with cytidine, thymidine and 2′-deoxyuridine, where theethod’s sensitivity for these three PP metabolites was so low that
hey caused imprecise quantification over the examined concentra-ion ranges. The metabolites were purified and concentrated using
novel multi-step pre-treatment procedure consisting of proteinrecipitation, ultrafiltration, evaporation under nitrogen flow, andubsequent reconstitution. This procedure ensured efficient recov-ries for most investigated metabolites and efficient removal ofnterfering matrix components. The method is selective, sensi-ive, stable, and precise. The potential application of the methodas demonstrated by evaluating its range of use in different types
f blood plasma from multicatheterized cows, here, only uridine,howed undesirable matrix effects. The method is adaptable andan be further developed for the quantitative detection of the sameetabolites in other matrices such as urine or milk.
cknowledgements
We gratefully acknowledge Lis Sidelmann, Birgit Hørdum Løthnd the barn staff at Department of Animal Science, Aarhus Uni-ersity, Foulum, Denmark for skillful technical assistance. Stevenock, Application manager EMEA at ABSCIEX, is recognized for hisssistance in assessing MS/MS fragmentation patterns. We alsohank senior scientists Torben Larsen and Peter Lund for supplyinglasma samples for analytical application experiments. C. Stentoft
olds a PhD scholarship co-financed by the Faculty of Science andechnology, Aarhus University and a research project supportedy the Danish Milk Levy Fond, c/o Food and Agriculture, Aarhus, Denmark. Funding for the cow animal experiments from which
[
nic vein; A, artery; C, chicken; P, pig; M, mink; H, human; R, rat; W, water; U, urine;00 (n = 2, samples). A relative recovery between 85% and 115% was considered goodt fulfilling these criteria.
some of the plasma samples were obtained was partly provided bythe Commission of the European Communities (Brussels, Belgium;Rednex project FP7, KBBE-2007-1).
References
[1] H. Steinfeld, P. Gerber, T. Wassenaar, V. Castel, M. Rosales, C. de Haan, Live-stock’s long shadow: Environmental issues and options, 2006, www.fao.org,Accessed Oct.1, 2012.
[2] R.A. Kohn, M.M. Dinneen, E. Russek-Cohen, Using blood urea nitrogen to predictnitrogen excretion and efficiency of nitrogen utilization in cattle, sheep, goats,horses, pigs, and rats, J. Anim. Sci. 83 (2005) 879–889.
[3] C.K. Reynolds, N.B. Kristensen, Nitrogen recycling through the gut and the nitro-gen economy of ruminants: an asynchronous symbiosis, J. Anim. Sci. 86 (2008)293–305.
[4] S. Calsamiglia, A. Ferret, C.K. Reynolds, N.B. Kristensen, A.M. van Vuuren, Strate-gies for optimizing nitrogen use by ruminants, Animal 4 (2010) 1184–1196.
[5] S. Tamminga, Nutrition management of dairy cows as a contribution to pollu-tion control, J. Dairy Sci. 75 (1992) 345–357.
[6] J.L. Firkins, Maximizing microbial protein synthesis in the rumen, J. Nutr. 126(1996) 1347–1354.
[7] T. Fujihara, M.N. Shem, Metabolism of microbial nitrogen in ruminants withspecial reference to nucleic acids, Anim. Sci. J. 82 (2011) 198–208.
[9] R.H. Smith, A.B. McAllan, Some factors influencing the chemical compositionof mixed rumen bacteria, Br. J. Nutr. 31 (1974) 27–34.
10] X.B. Chen, M.J. Gomes, Estimation of Microbial Protein Supply to Sheep andCattle Based On Urinary Excretion of Purine Derivatives—An Overview Of TheTechnical Details, Occasional Publication of International Feed Resources Unit,Rowett Research Institute, Bucksburn, Aberdeen AB2 9SB, UK, 1992, pp. 1–21.
11] J.M. Moorby, R.J. Dewhurst, R.T. Evans, J.L. Danelon, Effects of dairy cow dietforage proportion on duodenal nutrient supply and urinary purine derivativeexcretion, J. Dairy Sci. 89 (2006) 3552–3562.
12] B.M. Tas, A. Susenbeth, Urinary purine derivates excretion as an indicator ofin vivo microbial N flow in cattle: a review, Livest. Sci. 111 (2007) 181–192.
13] M. Gonzalez-Ronquillo, J. Balcells, A. Belenguer, C. Castrillo, M. Mota, A com-parison of purine derivatives excretion with conventional methods as indicesof microbial yield in dairy cows, J. Dairy Sci. 87 (2004) 2211–2221.
14] H. Boudra, M. Doreau, P. Noziere, E. Pujos-Guillot, D.P. Morgavi, Simulta-neous analysis of the main markers of nitrogen status in dairy cow’s urineusing hydrophilic interaction chromatography and tandem mass spectrometrydetection, J. Chromatogr. A 1256 (2012) 169–176.
15] L. Liu, J. Ouyang, W.R.G. Baeyens, Separation of purine and pyrimidine bases byion chromatography with direct conductivity detection, J. Chromatogr. A 1193
(2008) 104–108.
16] M. Haunschmidt, W. Buchberger, C.W. Klampfl, Investigations on the migrationbehaviour of purines and pyrimidines in capillary electromigration techniqueswith UV detection and mass spectrometric detection, J. Chromatogr. A 1213(2008) 88–92.
[solid-phase dispersion for liquid chromatography tandem mass spectrome-
10 C. Stentoft et al. / J. Chrom
17] H. Lin, D.K. Xu, H.Y. Chen, Simultaneous determination of purine bases, ribonu-cleosides and ribonucleotides by capillary electrophoresis electrochemistrywith a copper electrode, J. Chromatogr. A 760 (1997) 227–233.
18] Y.X. Gong, S.P. Li, P. Li, J.J. Liu, Y.T. Wang, Simultaneous determination of sixmain nucleosides and bases in natural and cultured Cordyceps by capillaryelectrophoresis, J. Chromatogr. A 1055 (2004) 215–221.
19] N.-P. Hua, T. Naganuma, Application of CE for determination of DNA base com-position, Electrophoresis 28 (2007) 366–372.
20] H. Kazoka, Analysis of purines and pyrimidines by mixed partition-adsorptionnormal-phase high-performance liquid chromatography, J. Chromatogr. A 942(2002) 1–10.
21] M. Clariana, M. Gratacós-Cubarsí, M. Hortós, J.A. García-Regueiro, M. Castel-lari, Analysis of seven purines and pyrimidines in pork meat products by ultrahigh performance liquid chromatography–tandem mass spectrometry, J. Chro-matogr. A 1217 (2010) 4294–4299.
22] R.N. Xu, L. Fan, M.J. Rieser, T.A. El-Shourbagy, Recent advances in high-throughput quantitative bioanalysis by LC-MS/MS, J. Pharm. Biomed. Anal. 44(2007) 342–355.
23] B.A. Røjen, P.K. Theil, N.B. Kristensen, Effects of nitrogen supply on inter-organfluxes of urea-N and renal urea-N kinetics in lactating Holstein cows, J. DairySci. 94 (2011) 2532–2544.
24] B.A. Røjen, N.B. Kristensen, Effect of normal and high NaCl intake on PDV urea-Nflux and renal urea-N kinetics in lactating cows, in: EAAP—European Federa-tion of Animal Science, 63rd Annual Meeting, Bratislava 2012, WageningenAcademic Publishers, Bratislava, Slovakia, 2012, p. 116.
25] N.B. Kristensen, A. Storm, B.M.L. Raun, B.A. Røjen, D.L. Harmon, Metabolism ofsilage alcohols in lactating dairy cows, J. Dairy Sci. 90 (2007) 1364–1377.
26] G.B. Huntington, C.K. Reynolds, B.H. Stroud, Techniques for measuring blood-flow in splanchnic tissues of cattle, J. Dairy Sci. 72 (1989) 1583–1595.
27] V.P. Shah, K.K. Midha, S. Dighe, I.J. McGilveray, J.P. Skelly, A. Yacobi, T. Layloff,C.T. Viswanathan, C.E. Cook, R.D. McDowall, K.A. Pittman, S. Spector, K.S. Albert,S. Bolton, M. Dobrinska, W. Doub, M. Eichelbaum, J.W.A. Findlay, K. Gallicano,W. Garland, D.J. Hardy, J.D. Hulse, H.T. Karnes, R. Lange, W.D. Mason, G. McKay,E. Ormsby, J. Overpeck, H.D. Plattenberg, G. Shiu, D. Sitar, F. Sorgel, J.T. Ste-wart, L. Yuh, Analytical methods validation–bioavailability, bioequivalence andpharmacikinetic studies, Pharm. Res. 9 (1992) 588–592.
28] V.P. Shah, K.K. Midha, J.W. Findlay, H.M. Hill, J.D. Hulse, I.J. McGilveray, G.McKay, K.J. Miller, R.N. Patnaik, M.L. Powell, A. Tonelli, C.T. Viswanathan, A.Yacobi, Bioanalytical method validation – a revisit with a decade of progress,
Pharm. Res. 17 (2000) 1551–1557.
29] F.T. Peters, O.H. Drummer, F. Musshoff, Validation of new methods, ForensicSci. Int. 165 (2007) 216–224.
30] R.C. Littell, G.A. Milliken, W.W. Stroup, R.D. Russell, SAS® System for MixedModels, SAS Institute Inc., Cary, NC, 1996, 633 pp. ISBN: 1-55544-779-1.
[
. A 1356 (2014) 197–210
31] C.E. McCulloch, S.R. Searle, Generalized, Linear, and Mixed Models, Wiley, NewYork, 2004, 358 pp. ISBN: 978-0-471-65404-9.
32] B.K. Matuszewski, M.L. Constanzer, C.M. Chavez-Eng, Strategies for theassessment of matrix effect in quantitative bioanalytical methods based onHPLC-MS/MS, Anal. Chem. 75 (2003) 3019–3030.
33] A. Tan, I.A. Levesque, I.M. Levesque, F. Viel, N. Boudreau, A. Levesque, Analyteand internal standard cross signal contributions and their impact on quantita-tion in LC-MS based bioanalysis, J. Chromatogr. B 879 (2011) 1954–1960.
34] E. Stokvis, H. Rosing, J.H. Beijnen, Stable isotopically labeled internal standardsin quantitative bioanalysis using liquid chromatography/mass spectrometry:necessity or not? Rapid Commun. Mass Spectrom. 19 (2005) 401–407.
35] A. Van Eeckhaut, K. Lanckmans, S. Sarre, I. Smolders, Y. Michotte, Validation ofbioanalytical LC-MS/MS assays: evaluation of matrix effects, J. Chromatogr. B877 (2009) 2198–2207.
36] R. Bakhtiar, T.K. Majumdar, Tracking problems and possible solutions in thequantitative determination of small molecule drugs and metabolites in bio-logical fluids using liquid chromatography-mass spectrometry, J. Pharmacol.Toxicol. Methods 55 (2007) 262–278.
37] E. Chambers, D.M. Wagrowski-Diehl, Z. Lu, J.R. Mazzeo, Systematic and com-prehensive strategy for reducing matrix effects in LC/MS/MS analyses, J.Chromatogr. B 852 (2007) 22–34.
38] B.E. Richter, B.A. Jones, J.L. Ezzell, N.L. Porter, N. Avdalovic, C. Pohl, Acceleratedsolvent extraction: a technique for sample preparation, Anal. Chem. 68 (1996)1033–1039.
39] S. Hartmann, J.G. Okun, C. Schmidt, C.D. Langhans, S.F. Garbade, P. Burgard,D. Haas, J.O. Sass, W.L. Nyhan, G.F. Hoffmann, Comprehensive detection ofdisorders of purine and pyrimidine metabolism by HPLC with electrosprayionization tandem mass spectrometry, Clin. Chem. 52 (2006) 1127–1137.
40] P.J. Taylor, Matrix effects: the Achilles heel of quantitative high-performanceliquid chromatography-electrospray-tandem mass spectrometry, Clin.Biochem. 38 (2005) 328–334.
41] A. Kruve, A. Kunnapas, K. Herodes, I. Leito, Matrix effects in pesticide multi-residue analysis by liquid chromatography-mass spectrometry, J. Chromatogr.A 1187 (2008) 58–66.
42] A.K. Hewavitharana, Matrix matching in liquid chromatography-mass spec-trometry with stable isotope labelled internal standards – is it necessary? J.Chromatogr. A 1218 (2011) 359–361.
43] D. Mutavdzic Pavlovic, T. Pinusic, M. Perisa, S. Babic, Optimization of matrix
try analysis of 12 pharmaceuticals in sediments, J. Chromatogr. A 1258 (2012)1–15.
44] L.L. Jessome, D.A. Volmer, Ion suppression: a major concern in mass spectrom-etry, LC GC N. Am. 24 (2006) 498–510.
43. Giesecke D, Ehrentreich L, Stangassinger M et al. (1994) Mammary and renal excretion of 97
purine metabolites in relation to energy intake and milk yield in dairy cows. J Dairy Sci 77, 2376-98
2381. 99
44. Valadares RF, Broderick GA, Valadares Filho SC et al. (1999) Effect of replacing alfalfa silage 100
with high moisture corn on ruminal protein synthesis estimated from excretion of total purine 101
derivatives. J Dairy Sci 82, 2686-2696. 102
45. IUPAC, Abbreviations and Symbols for Nucleic Acids, Polynucleotides and their Constituents; 103
available at http://www.chem.qmul.ac.uk/iupac/misc/naabb.html (accessed May 2014). 104
46. KEGG: Kyoto Encyclopedia of Genes and Genomes, purine metabolism; available at 105
http://www.genome.jp/kegg/pathway/map/map00240.html (accessed May 2014). 106
47. KEGG: Kyoto Encyclopedia of Genes and Genomes, pyrimidine metabolism; available at 107
http://www.genome.jp/kegg/pathway/map/map00230.html (accessed May 2014)108
92
Table 1. Abbreviation, type and calibration range of investigated purine and pyrimidine metabolites
Abbreviations1
Type Range (min)2 Range (max)
2
µmol/L
Pyrines
Guanosine Guo NS 0∙0 5∙0
Inosine Ino NS 0∙0 5∙0
2’-deoxyguanosine dGuo NS 0∙0 5∙0
2’-deoxyinosine dIno NS 0∙0 5∙0
Adenine Ade BS 0∙0 5∙0
Guanine Gua BS 0∙0 5∙0
Hypoxanthine Hyp BS/DP 0∙0 5∙0
Xanthine Xan BS/DP 0∙0 5∙0
Uric acid Uac DP 0∙0 200
Allantoin Alo DP 125 500
Pyrimidines
Cytidine Cyd NS 2∙5 5∙0
Uridine Urd NS 1∙9 7∙5
Thymidine dThd NS 2∙5 5∙0
2’-deoxyuridine dUrd NS 0∙16 5∙0
Cytosine Cyt BS 1∙9 7∙5
Uracil Ura BS 0∙0 5∙0
Thymine
Thy BS 0∙0 5∙0
β-alanine β-ala DP 3∙1 13
β-ureidopropionic acid β-ure DP 0∙0 75
β-aminoisobutyric acid β-ami DP 0∙0 5∙0
BS, base; NS, nucleoside; DP, degradation product; min, minimum concentration; max, maximum concentration. 1Abbreviations from IUPAC, abbreviations and symbols for nucleic acids, polynucleotides and their constituents
(45).
2External calibration was performed with five concentrations
and bottom points were excluded to fit the concentration
range in actual samples.
93
Table 2. Concentrations (µmol/L) of purine and pyrimidine metabolites in plasma samples from lactating dairy cows
min, minimum concentration; max, maximum concentration. 1Mean ± SD.
94
Table 3. Concentration differences (µmol/L) between each of four blood veins and an artery of purine and pyrimidine metabolites in lactating dairy cows
ΔPA, concentration difference between hepatic portal vein and artery; ΔHA, concentration difference between hepatic vein and artery; ΔPH, concentration difference between
hepatic portal vein and hepatic vein; ΔGA, concentration difference between gastrosplenic vein and artery. 1Mean ± SEM (n = 4). Difference from zero declared when *P ≤ 0∙05, tendency when †P ≤ 0∙1 (t-test).
2P-values for main effect of time relative to feeding. Significance declared when *P ≤ 0∙05, tendency when †P ≤ 0∙1 (F-test).
95
Table 4. Blood plasma flows (L/h) and net fluxes (µmol/h or mmol/h) of purine and pyrimidine metabolites in lactating dairy cows
PDV, portal-drained viscera; TSP, total splanchnic tissue; HA, hepatic artery; HEP, hepatic tissue. 1Hourly samples (time) were collected during an 8-h period, 0∙5 h before feeding, and at 0∙5, 1∙5, 2∙5, 3∙5, 4∙5, 5∙5 and 6∙5 h after feeding, on d 14 of the experimental period.
2Overall mean ± SEM (n = 4, across the four cows).
3SEM (n = 4, across the four cows within each sampling time).
4P-values for linear (Lin), (Quad) and cubic (Cubic) time effects. Significance declared when *P ≤ 0∙05, tendency when †P ≤ 0∙1 (F-test).
97
Table 5. Hepatic fractional removal as percentage of net PDV release and percentage of total influx of purines and pyrim-
idine metabolites
Percentage of net PDV release (NP%) Percentage of total influx (TI%)
NP%, percentage of net PDV release; TI%, percentage of total influx. 1Overall mean ± SEM (n = 4, across the four cows). Only the overall mean and not individual time estimates are given
since almost no effects of time were detected. 4P-values for linear (Lin), (Quad) and cubic (Cubic) time effects. Significance declared when *P ≤ 0∙05, tendency when
†P ≤ 0∙1 (F-test).
98
Table 6. Renal purine variables in lactating dairy cows
Item Mean1 SEM
1
Renal plasma flow, L/h 346 36
Diuresis, L/h 0∙89 0∙072
Arterial concentration
Xan, μmol/L 0∙011 0∙0070
Hyp, μmol/L 0∙043 0∙013
Uac, mmol/L 73 33
Alo, mmol/L 122 30
Urine concentration
Xan, μmol/L 0∙0
Hyp, μmol/L 0∙0
Uac, mmol/L 1∙0 0∙11
Alo, mmol/L 11 1∙5
Renal influx, mmol/h
Uac 24 4∙8
Alo 41 3∙0
Net urine flux, mmol/h
Uac 0∙89 0∙11
Alo 10 1∙2
Urine/renal ratio
Uac 0∙47 0∙10
Alo 0∙25 0∙039
Urine/splanchnic ratio
Uac 0.13 0.036
Alo2
Renal clearance, L/h
Uac 15 4∙4
Alo 89 21
1Mean ± SEM (n = 4).
2The net splanchnic flux of Alo was negative hence, a urine/splanchnic ratio could not be determined.
99
Figure 1
NucleosideNucleotide Base Intermediate Degradation product
2'-deoxyinosine
C10H12N4O4
2'-deoxyadenosine
C10H13N5O3
Adenosine
C10H13N5O4
XanthosineC10H12N4O6
IMPC10H13N4O8P
GMPC10H14N5O8P
XMPC10H13N4O9P
GuanosineC10H13N5O5
InosineC10H12N4O5
dAMP
C10H14N5O6P
AMP
C10H14N5O7P
dGMPC10H14N5O7P
2'-deoxyguanosineC10H13N5O4
Adenine
C5H5N5
HypoxanthineC5H4N4O
Guanine
C5H5N5O
XanthineC5H4N4O2
Uric acidC5H4N4O3
AllantoinC4H6N4O3
1
1
1
1
3
2
4
1
1 6
6
6
6
6
6
7
7
6
11
12
13
9
10
10
8
8
5
11 14
NH3
NH3
NH3 NH3
NH3NH3
Fig. 1. Degradation pathways of the purine metabolism. Illustration modified from KEGG: Kyoto Encyclopedia of Genes and Genomes, Purine metabolism(46)
NucleosideNucleotide Base Intermediate Degradation product
Cytidine
C9H13N3O5
Uridine
C9H12N2O6
2'-deoxyuridine
C9H12N2O5
dCMPC9H14N3O7P
2'-deoxycytidineC9H13N3O4
UMP
C9H13N2O9P
dUMP
C9H13N2O8P
Cytosine
C4H5N3O
UracilC4H4N2O2
DihydrouracilC4H6N2O2
β -ureidopropionic acid
C4H8N2O3
β-alanineC3H7NO2
1
2
1
3
1
7
6
4
4
5
1312
11
NH3
CMP
C9H14N3O8P
9
10
8
NH3
NH3 NH3
NH3
Thymidine
C10H14N2O5
dTMP
C10H15N2O8PThymine
C5H6N2O2
DihydrothymineC5H8N2O2
β -ureidoisobutyric acid
C5H10N2O3
β-aminoisobutyricacid
C4H9NO2
2 8 1312
11
10
NH3
Fig. 2. Degradation pathways of the pyrimidine metabolism. Illustration modified from KEGG: Kyoto Encyclopedia of Genes and Genomes, pyrimidine metabolism(47)
Fig. 3. The purine N and pyrimidine N intestinal absorption and intermediary metabolism in the portal-drained viscera, hepatic and total splanchnic tissue in lactating dairy cows.
Purine-N, purine nitrogen; pyrimidine-N, pyrimidine nitrogen; NS-N, purine or pyrimidine nucleoside nitrogen; BS-N, purine or pyrimidine base nitrogen; Uac-N, uric acid nitro-
gen; Alo-N, allantoin nitrogen; β-ala-N, β-alanine nitrogen; β-ami-N, β-aminoisobutyric acid nitrogen; N-outlet, nitrogen outlet into the β-alanine metabolism(22)
and the valine,
leucine, and isoleucine metabolism and the citric acid cycle(23)
; NH3, ammonia release during degradation available for urea-recycling(24)
. The purine N and pyrimidine N were
estimated from the microbial crude protein in the small intestine and the notion that when degraded dietary nitrogen is reused by the microbial population, 75-85% (80%) N goes
to microbial protein and 15-25% (20%) N to microbial nucleic acids(7-8,10)
. Values are means ± SEM (n = 4).
Microbial purine-N in nucleic acids
40 g/d N
NS-N: 3∙9 ± 0∙92 g/d
BS-N: 0∙0029 ± 0∙015 g/d
Uac-N: 9∙3 ± 3∙0 g/d Uac-N: 1∙2 ± 0∙15 g/d
Alo-N: 13 ± 1∙6 g/d
Inte
stin
al m
uco
sa
Kid
ney
s
Microbial pyrimidine-N in nucleic acids
20 g/d N
NS-N: 3∙9 ± 0∙33 g/d
β-ala-N: 0∙28 ± 0∙13 g/d
β-ami-N: 0∙016 ± 0∙0012 g/d
Inte
stin
al m
uco
sa
NH3
β-ami-N: -0∙031 ± 0∙0097 g/d
β-ala-N: -0∙069 ± 0∙15 g/d
NS-N: -5∙1 ± 0∙18 g/d
Kid
ney
s
N-outletNH3NH3
NH3 NH3Sum 27 g/d N
Sum: 4∙7 g/d N
Portal-drained viscera Total splanchnic tissueIntestine Urine
Alo-N: 14 ± 5∙8 g/d
Total splanchnic tissueIntestine Hepatic tissue
NH3
Alo-N: - (-22 ± 3∙9 g/d)
Uac-N: 0∙71 ± 2∙0 g/d
BS-N: -0∙056 ± 0∙015 g/d
NS-N: -4∙0 ± 1∙0 g/d
Hepatic tissue
NS-N: 0∙018 ± 0∙034 g/d
BS-N: -0∙022 ± 0∙011 g/d
Uac-N: 10 ± 2∙1 g/d
Alo-N: - (-8 ± 5∙51 g/d)
NH3
NS-N: -1∙1 ± 0∙17 g/d
β-ala-N: 0∙19 ± 0∙13 g/d
β-ami-N: -0∙013 ± 0∙074 g/d
NH3
Portal-drained viscera
Pu
rin
eP
yri
mid
ine
102
8. Manuscript III
Protein level influences the splanchnic metabolism of purine and pyrimidine metabolites in
lactating dairy cows.
Stentoft C., C. Barratt, L.A. Crompton, S.K. Jensen, M. Vestergaard, M. Larsen and C.K. Reynolds.
To be submitted to J. Dairy Sci.
103
PURINE AND PYRIMIDINE METABOLISM IN DAIRY COWS
Protein Level influences the Splanchnic Metabolism of Purine and Pyrimidine metabolites in
Lactating Dairy Cows.
C. Stentoft,*1 C. Barratt, † L. A. Crompton, † S. K. Jensen,* M. Vestergaard,* M. Larsen,* C.
K. Reynolds, †
* Department of Animal Science, Aarhus University, Foulum, DK-8830 Tjele, Denmark
† School of Agriculture, Policy and Development, University of Reading, Early Gate, Reading RG6
1Cows were feed a TMR containing grass:corn silage (25:75 or 75:25) with 12.5%, 15.0%, or 17.5% CP of DM.
2mean ± SEM (pooled) (n = 6).
3P-values for protein (Pro) describe the effect of feeding different CP levels. P-values for forage (For) describe the effect of feeding mainly corn or grass silage. P-values for pro-
tein×forage (Pro × For) describes any interaction between CP level and either corn or grass silage. P-values for linear (Lin) and quadratic (Quad) effects describe the effect of
dietary CP. Significance declared when P ≤ 0.1 (F-test). 4All metabolites except hypoxanthine, cytosine, uracil, and thymine had one or more ΔPA, ΔHA, ΔPH or ΔMA values that differed from zero (P ≤ 0.10).
126
Table 3. Venous-arterial concentration differences (µmol/L) between each of four blood veins and an artery of purine and pyrimidine metabolites in lactating dairy cows1
1Cows were feed a TMR containing grass:corn silage (25:75 or 75:25) with 12.5%, 15.0%, or 17.5% CP of DM.
2ΔPA, concentration difference between hepatic portal vein and artery; ΔHA, concentration difference between hepatic vein and artery; ΔPH, concentration difference between
hepatic portal vein and hepatic vein; ΔEA, concentration difference between epigastric vein and artery. 3Mean ± SEM (n = 6).
4P-values for difference from zero. Significance declared when P ≤ 0.10 (t-test).
5All metabolites except hypoxanthine, cytosine, uracil, and thymine had one or more ΔPA, ΔHA, ΔPH or ΔEA values that differed from zero (P ≤ 0.10).
127
Table 4. Net splanchnic fluxes (µmol/h, unless otherwise noted) of purine and pyrimidine metabolites in lactating dairy cows1
1Cows were feed a TMR containing grass:corn silage (25:75 or 75:25) with 12.5%, 15.0%, or 17.5% CP of DM.
2PDV, portal-drained viscera; HEP, hepatic tissue; TSP, total splanchnic tissue.
3mean ± SEM (pooled) (n = 6).
4P-values for protein (Pro) describe the effect of feeding different CP levels. P-values for forage (For) describe the effect of feeding mainly corn or grass silage. P-values for pro-
tein×forage (Pro × For) describes any interaction between CP level and either corn or grass silage. P-values for linear (Lin) and quadratic (Quad) effects describe the effect of
dietary CP. Significance declared when P ≤ 0.10 (F-test).
129
Table 5. Venous-arterial concentration differences (µmol/L) between the epigastric vein and artery (ΔEA) of purine and pyrimidine metabolites in lactating dairy cows1
1Cows were feed a TMR containing grass:corn silage (25:75 or 75:25) with 12.5%, 15.0%, or 17.5% CP of DM.
2Mean ± SEM (pooled) (n = 6).
3P-values for protein (Pro) describe the effect of feeding different CP levels. P-values for forage (For) describe the effect of feeding mainly corn or grass silage. P-values for pro-
tein × forage (Pro × For) describes any interaction between CP level and either corn or grass silage. P-values for linear (Lin) and quadratic (Quad) effects describe the effect of
dietary CP. Significance declared when P ≤ 0.10 (F-test).
130
Table 6. The purine and pyrimidine nitrogen (g/d) intestinal absorption and intermediary metabolism in lactating dairy cows1
1Cows were feed a TMR containing grass:corn silage (25:75 or 75:25) with 12.5%, 15.0%, or 17.5% CP of DM.
2Purine N, purine nitrogen; pyrimidine N, pyrimidine nitrogen; Nucleic acid N, nucleic acid nitrogen.
3PDV, portal-drained viscera; HEP, hepatic tissue; TSP, total splanchnic tissue.
4Mean ± SEM (pooled) (n = 6).
5P-values for protein (Pro) describe the effect of feeding different CP levels. P-values for forage (For) describe the effect of feeding mainly corn or grass silage. P-values for pro-
tein × forage (Pro × For) describes any interaction between CP level and either corn or grass silage. P-values for linear (Lin) and quadratic (Quad) effects describe the effect of
dietary CP. Significance declared when P ≤ 0.10 (F-test).
131
Figure 1
NucleosideNucleotide Base Intermediate Degradation product
2'-deoxyinosine
C10H12N4O4
2'-deoxyadenosine
C10H13N5O3
Adenosine
C10H13N5O4
XanthosineC10H12N4O6
IMPC10H13N4O8P
GMPC10H14N5O8P
XMPC10H13N4O9P
GuanosineC10H13N5O5
InosineC10H12N4O5
dAMP
C10H14N5O6P
AMP
C10H14N5O7P
dGMPC10H14N5O7P
2'-deoxyguanosineC10H13N5O4
Adenine
C5H5N5
HypoxanthineC5H4N4O
Guanine
C5H5N5O
XanthineC5H4N4O2
Uric acidC5H4N4O3
AllantoinC4H6N4O3
1
1
1
1
3
2
4
1
1 6
6
6
6
6
6
7
7
6
11
12
13
9
10
10
8
8
5
11 14
NH3
NH3
NH3 NH3
NH3NH3
Fig. 1. Degradation pathways of the purine metabolism. Illustration modified from Kyoto Encyclopedia of Genes and Genomes; Purine metabolism (Kanehisa et al., 2014). Me-
NucleosideNucleotide Base Intermediate Degradation product
Cytidine
C9H13N3O5
Uridine
C9H12N2O6
2'-deoxyuridine
C9H12N2O5
dCMPC9H14N3O7P
2'-deoxycytidineC9H13N3O4
UMP
C9H13N2O9P
dUMP
C9H13N2O8P
Cytosine
C4H5N3O
UracilC4H4N2O2
DihydrouracilC4H6N2O2
β -ureidopropionic acid
C4H8N2O3
β-alanineC3H7NO2
1
2
1
3
1
7
6
4
4
5
1312
11
NH3
CMP
C9H14N3O8P
9
10
8
NH3
NH3 NH3
NH3
Thymidine
C10H14N2O5
dTMP
C10H15N2O8PThymine
C5H6N2O2
DihydrothymineC5H8N2O2
β -ureidoisobutyric acid
C5H10N2O3
β-aminoisobutyricacid
C4H9NO2
2 8 1312
11
10
NH3
Fig. 2. Degradation pathways of the pyrimidine metabolism. Illustration modified from Kyoto Encyclopedia of Genes and Genomes; Pyrimidine metabolism (Kanehisa et al.,
centration (%); PII, paper II; MIII, manuscript III (constructed from paper I, paper II and manuscript III). 1The quantification range was set to the lowest and highest quantified concentration giving an acceptable CV <
25% (Paper I). 2Value is above the highest calibrator concentration.
3Applied range (μmol/L) was determined by
external calibration with five concentrations and points were excluded to fit the concentration range in actual sam-
ples (Paper II and manuscript III). 4Mean arterial concentrations (µmol/L) of purine and pyrimidine metabolites in
plasma samples (Paper II and manuscript III). 5Within-day variation expressed as CV% (Paper I).
6ΔPA/A (%);
the hepatic portal venous-arterial difference (%) (Paper II and manuscript III). 7Not determined because the Δ[PA]
was essentially zero.
For the method to be precise enough to be used for determining fluxes when analysing sample sets
from the multicatheterized cow model, the within-day variation (%) preferable should be below that
of the venous-arterial difference (%). The hepatic portal venous-arterial concentration differences
were for most metabolites higher or similar to the within-day variation (%) (Table 5). Only for the
degradation products such as uric acid, allantoin and β-alanine, that had natural high levels of en-
dogenous metabolites, the precision of the method was unfortunately not higher than the within-day
variation (%). However, the within-day variation (%) was determined at relatively low concentra-
tion levels, resulting in larger CV% than what would be expected at the considerable higher concen-
tration levels used in the experimental sample sets (Paper I, paper II and manuscript III). Also, ow-
ing to the discovery of leaking in the HPLC system followed by a repair performed between the
analyse of the samples from the two experiments, as well as further small refinements of the analy-
sis procedures during the study, the across-day variation (CV%) and probably within-day variation
(CV%) of the method had improved when performing analyses for manuscript III (Table 6). Espe-
cially the variation of uric acid and allantoin benefitted from the maintenance repair of the instru-
ment. The within-day and across-day variation (CV%) of this method was in most cases in line with
or better than previously reported values (Hartmann et al., 2006).
Table 6. Within-day and across-day variation (CV%) established during method development, re-evaluated in
manuscript III, and reported by Hartmann et al. in human plasma (Hartmann et al., 2006).
Paper I Hartmann et al. Paper I Manuscript III Hartmann et al.
Purines
Adenine 2 8 5 5 7
Guanine 2 4 6
Guanosine 4 16 12 4 9
Inosine 2 8 9 6 11
2’-deoxyguanosine 4 19 7 6 17
2’-deoxyinosine 2 8 8 4 9 Xanthine 3 9 9 7 11
Hypoxanthine 1 10 6 8 16
Uric acid 16 18 55 5 6
141
Allantoin 34 49 23
Pyrimidines
Cytosine 21 24 9
Thymine 4 11 15 7 17
Uracil 5 10 4 - 13
Cytidine 18 24 6
Uridine 7 14 12 11 10
Thymidine 23 8 21 9 8
2’-deoxyuridine 33 14 37 52 9
β-alanine 12 8 5 20 7
β-ureidopropionic acid 14 13 22
β-aminoisobutyric acid 6 8 7 11 10
CV%, coefficient of variation (%) (constructed from paper I, manuscript III, and Hartmann et al., 2006). 1The within-day variation (CV%) determined in paper I (conc. level = 4-7 μmol/L, except allantoin 40 μmol/L, n
= 8, samples) and Hartmann et al. (2006) (conc. level = 35-50 μmol/L, except uric acid 200 μmol/L, n = 10, sam-
ples). 2The across-day variation (CV%) determined in paper I (conc. level = 4-7 μmol/L, except allantoin 40
μmol/L, n = 8, samples, m = 5, days), manuscript III (Two conc. levels: Low level = 0.5-5 μmol/L, except uric
acid, allantoin and β-alanine; 10, 60, 15 μmol/L and high level = 2-60 μmol/L, except uric acid, allantoin and β-
alanine; 180, 1000, 200 μmol/L, n = 4, samples, m = 6 days), and Hartmann et al. (2006) (conc. level = 35-50
μmol/L, except uric acid 200 μmol/L, n = 10, samples, m = 7 days).
The fact that the analysis variation (within and across-day) improved from experiment I (Paper II)
to experiment II (Manuscript III) was especially noteworthy with regard to the allantoin fluxes. In
paper II, allantoin could not be quantified as precisely as hoped for and the negative net hepatic flux
of allantoin did not agree with theory that allantoin passes the hepatic tissue without being metabo-
lised. On the other hand in manuscript III, allantoin was shown to pass the hepatic tissue unharmed
and the theory of allantoin functioning as a terminal product for excretion was thus confirmed.
9.1.5 Pre-treatment
An effective pre-treatment was vital in this study as complex biological matrices such as plasma can
easily clot the HPLC column resulting in a loss of efficiency, and ESI is sensitive to matrix effects
caused by salts, sugars, and proteins (Hopfgartner and Bourgogne, 2003; Nováková, 2013; Peters et
al., 2007; Praksah et al., 2007; Van Eeckhaut et al., 2009). Also, a proper clean-up enhances the
selectivity and the sensitivity of the analysis. A novel multi-step approach, consisting of PPT, ultra-
filtration, evaporation under nitrogen flow, and subsequent resolution, focused on isolation, clean-
up, and pre-concentration, was developed and optimized. In addition, the HPLC system was
equipped with a guard column to try to avoid blockage from contaminants escaping pre-treatment
and/or originating from the HPLC system itself (Ardrey, 2003). The pre-treatment procedure was
able to purify and to concentrate all of the purine and pyrimidine metabolites from bovine plasma
simultaneously, in a simple and efficient manner. Initially, different solvents (acetone, acetonitrile,
ethanol, methanol, sulfo-salicylic acid) were tested for PPT (Nováková, 2013; Polson et al., 2003).
Ethanol PPT was chosen for the procedure as it resulted in the highest recoveries and least noise,
and because it was the least harmful of the tested solvents. The ultrafiltration step was added to re-
move additional pollutants. Evaporation and reconstitution steps were included to obtain a concen-
142
tration effect. To try to reduce degradation and instability of the samples caused by reactive oxygen
species or enzyme activities during pre-treatment, all centrifugations and incubations were per-
formed at 4°C, and samples, stocks, and solvents etc. were kept at -4°C or on ice. Other types of
pre-treatment methods such as simple dilution (impractical) (Antignac et al., 2005), SPE (Bakhtiar
and Majumdar, 2007; Chambers et al., 2007; Poole, 2003), and accelerated solvent extraction (Rich-
ter et al., 1996), a form of LLE, were also investigated but were not found useful. Different types of
SPE from Waters were tested; HLB (polar components), C8, C18, WCX (basic conditions), and
MCX (acidic conditions), but none was found capable of a satisfactory purification of all the purine
and pyrimidine metabolites in one step. There probably exists other more sensitive and complicated
ways to quantify smaller groups of or even single purine and pyrimidine metabolites but, when ana-
lyzing this many components with such different chemical properties simultaneously, in such a
complex matrix, the procedure chosen herein seems like a better choice. The pre-treatment proce-
dure is able to purify and concentrate all of the targeted purine and pyrimidine metabolites simulta-
neously, in an easy and efficient manner without significant losses.
9.2 Absorption and intermediary metabolism of purine and pyrimidine metabolites
Taking advantage of the inbuilt ability of the multicatheterized cow model to describe the net PDV
and net hepatic fluxes of selected metabolites, the absorption pattern and intermediary metabolism
of the purine and pyrimidine nucelosides, bases, and degradation products were studied using two
feeding experiments (Experiment I and II). Besides describing the basics of these mechanisms and
how the purine and pyrimidine metabolism was influenced by postprandial pattern (Paper II), the
effects of protein level and forage source in the ration (Manuscript III) was assessed. In addition,
the fate of the purine and pyrimidine nitrogen in the dairy cows were evaluated (Paper II and
manuscript III). The purine and pyrimidine metabolites were found to be absorbed and metabolised,
and because they were affected differently, they will be discussed as two distinct groups.
The quantitative absorption and intermediary metabolism of purine and pyrimidine metabolites is an
almost unwritten chapter of the nitrogen metabolism in dairy cows and ruminants. Hence,
information and relevant litterature on the subject are at present very limited. The digestion of
monogastrics is very different from that of ruminants and to exchange knowledge between the two
animal types have therefore been difficult (McDonald et al., 2011). However, the results of the two
experiments (Experiment I and II) have provided a fairly clear picture of the mechanisms involved
and of the importance of the purine and pyrimidine metabolites.
9.2.1 The purine metabolism
All 10 purine metabolites were identified in all four types of experimental plasma samples from the
multicatheterized cows (Table 2 in paper II, data not shown in manuscript III) and arterial levels,
143
venous-arterial concentration differences, net PDV, net hepatic, and total splanchnic fluxes as well
as excretion parameters were determined.
Arterial levels of purine metabolites
It both experiments, arterial concentrations of purine degradation products were higher than the
concentrations of nucleosides and nucleoside concentrations higher than concentrations of bases
(Table 2 in paper II and manuscript III). Only in the case of the purine nucleosides, absorption from
the small intestine was indicated by higher concentrations in the hepatic portal vein compared to the
artery, and the hepatic, gastrosplenic, and epigastric veins. The arterial concentration levels of the
purine nucleosides and the bases detected in the two experiments were very similar. However, more
in line with other studies, lower concentration of uric acid and higher concentrations of allantoin
were identified in experiment II as compared to experiment I (Balcells et al., 1992b). It is believed
that the differences in concentration levels observed between experiments were caused by degrada-
tion during handling and/or storage in experiment II, where the samples went through three
freeze/thaw cycles prior to analysis. Storage degradation could have been avoided but the decision
to use the unique set of blood samples from experiment II for this study was taken after blood was
analysed for other purposes (Barratt et al., 2013). In experiment I, measures were taken to avoid
storage degradation i.e. samples were only thawed upon analysis. A difference in the activity of
degradation enzymes in the small intestine and/or intestinal mucosa between the two experiments
could also be the cause for the difference. In contradiction to this theory was that uricase [1.7.3.3],
the enzyme that catalyses the degradation of uric acid to allantoin, has only been detected in trace
amounts in bovine blood (Chen et al., 1990a). On the other hand, the degradation can also happen
spontaneously or be aided by hydroxyisourate hydrolase [3.5.2.17] (Kenehisa et al., 2014: KEGG
purine metabolism). Not surprisingly, as uric acid is the intermediate precursor of allantoin, only the
ratio between uric acid and allantoin changed between experiments and not the total amount (uric
acid + allantoin) (Berg et al., 2002; Carver and Walker, 1995; McDonald et al., 2011). This led to
the conclusion that estimation of purine degradation product concentrations in bovine blood plasma
should be based on the sum of uric acid and allantoin. As anticipated, no notable effects of post-
prandial pattern, protein level, or forage source was detected in the arterial levels of the purine me-
tabolites. The small effects detected in experiment II were assumed to be the result of influences of
diets on nutrient flow and other metabolic processes.
When calculating net fluxes, venous-arterial concentration differences are multiplied by the respec-
tive blood flows. However, the venous-arterial concentration differences across the PDV and hepat-
ic tissues were in both experiments, especially for the purine bases, very small (Table 3 in paper II
and manuscript III) (Kristensen et al., 2007; Reynolds et al., 1988; Seal and Reynolds, 1993).
144
Therefore, the a priori criteria for calculating net fluxes were that at least one of the five venous-
arterial concentration differences estimated between the 1) hepatic portal vein and artery, 2) hepatic
vein and artery, 3) hepatic portal vein and hepatic vein, 4) gastrosplenic vein and artery, and 5) epi-
gastric vein and artery, of the purine metabolites were different from zero (P ≤ 0.10). All purine
metabolites except adenine and xanthine in experiment I and hypoxanthine in experiment II met this
criterion. When not different from zero, individual venous-arterial concentration differences were
considered when interpreting net fluxes. In general, all of the concentrations and venous-arterial
concentration differences of the purine bases; adenine, guanine, hypoxanthine, and xanthine were
very small.
Release of purine metabolites from the portal-drained viscera
Large amounts of fully degraded purine metabolites in the form of uric acid and allantoin and very
low levels of purine bases and nucleosides were found to be released from the PDV in both experi-
ments (Table 4 in paper II and manuscript III). These findings suggested a very effective degrada-
tion of purine metabolites in the small intestine or in the intestinal mucosa, and most likely a com-
bination of these, in dairy cows. The almost non-existing release of bases in general corresponded
with previous findings by McAllan, demonstrating how purine and pyrimidine bases were removed
by 50-100% when infused into the intestine of steers (McAllan, 1980). The extensive purine degra-
dation was also in line with previous observations demonstrating a very effective degradation of
nucleic acids to nucleosides and bases in the small intestine, facilitated by the excreted pancreatic
polynucleotidases, nucleosidases, and phosphatases (Barnard et al., 1969; Berg et al., 2002; Carver
and Walker, 1995; McAllan, 1980; McDonald et al., 2011; Nakayama et al., 1981). If the purine
nucleosides and/or bases are not absorbed directly, a further degradation to uric acid and/or allanto-
in probably takes place (Fig. 1 in paper II and manuscript III). Also, an even further degradation to
ammonia and urea is possible. The purine nucleosides, bases, and degradation products are known
to be absorbed from the intestinal lumen and subjected to another level of degradation in the intesti-
nal mucosa (McAllan, 1980, Verbic et al., 1990). One of the degradation enzymes known to be
highly active in most tissues, and especially in the small intestinal mucosa, blood, and hepatic tissue
in cattle, is xanthine oxidase [1.17.3.2] (Al-Khalidi and Chaglassian, 1965; Balcells et al., 1992b;
Chen and Gomes, 1992; McDonald et al., 2011; Roussos, 1963; Verbic et al., 1990). Xanthine oxi-
dase in collaboration with additional degradation enzymes in the intestinal mucosa, produces uric
acid and removes purine nucleosides and bases. Some of the released uric acid and allantoin may
also originate from turnover of mucosal enterocytes and other parts of the PDV such as the rumen,
hind gut, pancreas, spleen, and fat. The exact location of degradation of uric acid to allantoin is un-
determined. Since uricase [1.7.3.3] is present in only trace amounts in bovine blood, it most likely
145
takes place in the small intestine, intestinal mucosa, and hepatic tissue (Chen et al., 1990a). The
effect of storage on the levels of uric acid and allantoin, but not of their sum, in the samples from
experiment II, points toward alternative mechanisms of uric acid degradation in the blood (Kenehisa
et al., 2014: KEGG purine metabolism). To try to differentiate between purine metabolites absorbed
from the small intestine and released from the forestomachs and other tissues drained by the gastro-
splenic vein, net gastrosplenic releases of metabolites were estimated (Paper II, data not shown).
Presuming the gastrosplenic plasma flow was around 20% of the hepatic portal plasma flow, a dis-
tinction between the release of purine metabolites from the forestomachs and the intestines was
made with the use of the gastrosplenic-arterial concentration difference (Remond et al., 1993; Storm
et al., 2011). Under these presumptions, allantoin was the only purine metabolite with a net gastro-
splenic release contributing considerably to the net PDV release (40% of PDV release). This could
indicate that allantoin was absorbed from the rumen which has been proposed for urea (Abdoun et
al., 2006; Kristensen et al., 2010; Reynolds and Kristensen, 2008; Røjen et al., 2011). Allantoin
would probably in that case, based on its relatively large chemical structure, be actively transported.
The gastrosplenic allantoin contribution could also, at least partly, originate from purine turnover in
the very large tissues of the forestomachs. Still, further investigations are needed to clarify the gas-
trosplenic contribution of allantoin. This result might also be a consequence of the problematic ac-
curate determination of allantoin in experiment I.
Only in the case of allantoin, a positive effect of postprandial pattern was detected in the PDV re-
lease (Table 4 in paper II). Most likely because of their complex digestion and absorption itinerary,
postprandial absorption profiles were not detected for the remaining purine metabolites (Fujihara
and Shem, 2011; McAllan, 1982; McDonald et al., 2011). In addition, any real effects were most
likely easiest to detect for metabolites with considerable fluxes, such as that of allantoin, and at
PDV release, as endogenous contributions of the hepatic metabolism was added to the hepatic flux-
es.
Positive effects of dietary protein level was detected for metabolites with large levels of net fluxes
and good precision in the method and mainly at release from the PDV (Table 4 in paper I and man-
uscript III). Hence, the absorption profile of 2’-deoxyguanosine, adenine, and xanthine were found
to be linearly and positively influenced by an increase in the dietary protein level (12.5, 15.0, and
17.5% CP) (Clark et al., 1992; Ipharraguerre and Clark, 2005; Nocek and Russell, 1988; Reynolds
et al., 2001). Most likely is due to their small net PDV releases, no influences of protein level in the
diet were detected for the remaining purine metabolites. This was also the case for uric acid, even
though considerable levels of uric acid were absorbed. In case of allantoin, the protein level effect
was quadratic and not linear (Ipharraguerre and Clark, 2005). The decline between the 15.0 and the
146
17.5% protein levels were most likely caused by a decrease in the microbial flow due to the fact that
the 17.5% level was achieved by feeding a greater amount of rumen protected protein. So, in the
case that the high protein level had been achieved by feeding protein sources with a lower degree of
rumen protection, the result could have been different. Reduced degradation of nucleic acids in the
small intestine or other negative effects on the absorption mechanisms on the high protein level
could also partly explain why the effect was quadratic.
Removal of purine metabolites in the hepatic tissue
As anticipated, a further and complete removal of the purine bases and nucleosides in the hepatic
tissue was observed in both experiments (Table 4 in paper II and manuscript III). Uric acid was also
almost completely removed, with only small amounts being released from the splanchnic tissues. In
paper II, it was indicated based on the negative net hepatic flux of allantoin, that allantoin was de-
graded in the hepatic tissue. This was unfortunately probably a result of the method being unable to
quantify allantoin as precisely as hoped for. Following the repair of the instrument performed be-
tween the two experiments, the close to zero net hepatic flux of allantoin in manuscript III very
nicely showed what was expected; i.e., that allantoin simple passes through the hepatic tissue. These
results were also reflected in the NP% of the purine nucleosides and bases (approx. 100%) as well
as of uric acid and allantoin (approx. 100%) (Table 5 in paper II). From these results, it becomes
evident that the considerable amounts of uric acid and allantoin excreted by dairy cows (Chen and
Ørskov, 2004; Tas and Susenbeth, 2007; Verbic et al., 1990) first of all, originates from the very
effective degradation in the small intestine and intestinal mucosa and secondly, from the final and
almost complete degradation across the hepatic tissue, and from endogenous losses (Chen and
Gomes, 1992; McAllan, 1980; Verbic et al., 1990). Due to the relatively small net hepatic fluxes
and endogenous contributions, effects of postprandial pattern were not detected in the hepatic me-
tabolism of the purine metabolites (Table 4 in paper II).
The small levels of hepatic removal and endogenous contributions of metabolites in the liver, was
also the reason why effects of dietary protein levels otherwise measurable at the level of PDV re-
lease became harder to detect across the hepatic tissue. This was especially the case for the net he-
patic removal of the purine nucleosides and bases. Still, a protein level effect was detected in the net
hepatic removal of 2’-deoxyguanosine. In contrast to the missing effect on the PDV release of uric
acid, a linear effect was observed for the net hepatic removal, demonstrating an almost complete
degradation of uric acid in the hepatic tissue. As anticipated, no effect was observed for allantoin,
since allantoin has been shown to pass through the hepatic tissue. Diet composition in experiment II
was adjusted to minimize differences in the total concentrations of starch, water soluble carbohy-
drates, or neutral detergent fiber across the 2 × 3 treatments. This meant that effects of subtle
147
changes in carbohydrate concentrations, forage source (grass vs corn silage), and rate of degrada-
tion on the rumen outflow of purine metabolites could possibly be hard to detect in this study (Clark
et al., 1992; Nocek and Russell, 1988; Reynolds et al., 2001). Hence, no effects of forage source
were detected in any part of the purine PDV or hepatic metabolisms (Table 4 in manuscript III).
This was in agreement with the findings by Baratt et al., who did not identify effects of forage
source in any measured nitrogen parameters either (Barratt et al., 2013).
Excretion of purine metabolites in urine and milk
Urinary excretion has been found to be the primary route of disposal of purine degradation products
(Balcells et al., 1991; Chen et al., 1990a; Vagnoni et al., 1997; Verbic et al., 1990) and the level of
excretion in especially urine but also in milk can be used as an indirect measure of rumen microbial
biosynthesis (Chen and Ørskov, 2004; Giesecke et al., 1994; Gonda and Lindberg, 1997; Gonzalez-
Ronquillo, 2004; Tas and Susenbeth, 2007; Verbic et al., 1990). In cattle, 82-93% of the urinary
excreted purine degradation products are allantoin, the remainder is uric acid but other products,
such as xanthine and hypoxanthine, have also been identified in bovine urine in small concentra-
tions (Chen et al., 1990a; Yanez-Ruiz et al., 2004). Renal variables were determined in experiment I
and they showed in full agreement with previous studies that large amounts of allantoin and uric
acid, with typical clearance rates, and not hypoxanthine and xanthine, were excreted in the urine
(Table 6 in paper II) (Bristow et al., 1992; Giesecke et al., 1994; Gonzalez-Ronquillo et al., 2004;
Martín-Orúe et al., 2000; Valadares et al., 1999; Verbic et al., 1990).
Some of the purine degradation products may also be cleared by secretion into milk (Giesecke et al.,
1994; Gonda and Lindberg, 1997; Tiemeyer et al., 1984). Studies have shown that concentrations of
uric acid and allantoin in milk correlate with their plasma and urine concentrations as well as feed
composition. In this study, this route of disposal was assessed by venous-arterial concentration dif-
ferences between the epigastric vein and artery. Keeping in mind that plasma flows are needed to
calculate actual fluxes, these results could give an indication about the flux of these metabolites
across the mammary gland. In contrast to previous reports, uptake of uric acid and allantoin into the
mammary gland were not detected (Table 5 in manuscript III). This suggested that the rate of trans-
fer from arterial blood to the mammary tissues and milk may be too small to be measured based on
venous-arterial concentration differences. However, 2’-deoxyinosine and guanine was shown to be
taken up by the mammary gland and guanosine and inosine released into the arterial blood. Inosine
was the purine metabolite with the highest venous-arterial concentration difference in the study and
by estimating a mammary plasma flow and a net mammary flux for this metabolite, it became clear
that a release from the mammary gland to the liver of this nucleoside probably exists in dairy cows
(Larsen et al., 2014). This was in agreement with reports showing that with advancing lactation, the
148
rate of cell proliferation in the mammary gland is exceeded by the rate of cell apoptosis and hence,
in all probability, release of degraded nucleic acids from the mammary gland into milk and arterial
blood (Capuco et al., 2001; Sørensen et al., 2006).
9.2.2 The pyrimidine metabolism
The 10 pyrimidine metabolites were identified in all four types of experimental plasma samples
from the multicatheterized cows (Table 2 in paper II and manuscript III) and arterial levels, venous-
arterial concentration differences, net PDV, net hepatic, and total splanchnic fluxes as well as some
excretion parameters were examined.
Arterial levels of pyrimidine metabolites
It both cow experiments, it was determined that the arterial concentrations of the pyrimidine nucle-
osides were generally higher than the concentrations of the purine nucleosides, whereas the concen-
trations of the pyrimidine bases were in the same range as the purine bases (Table 2 in paper II and
manuscript III). The pyrimidine nucleoside concentrations were higher than for the pyrimidine ba-
ses. The concentrations of the pyrimidine degradation products were more variable but generally
lower than that of the purine degradation products. Also in the case of the pyrimidine nucleosides,
PDV release was clearly indicated by relatively high concentrations levels in the hepatic portal vein.
The different handling/storage employed during the two experiments did not, as seen for uric acid
and allantoin, induce different arterial levels of the pyrimidine metabolites. The greater ability of
the pyrimidine metabolites to withstand degradation fits with the observation that the pyrimidine
metabolites to a much larger extend were released from the PDV as nucleosides. The differences in
concentration levels clearly indicated that the mechanisms of the purine and pyrimidine absorption
and intermediary metabolism differed (Loffler et al., 2005). The arterial concentrations were, with a
few exceptions, not affected by postprandial pattern, protein level, or forage source.
The calculations of pyrimidine net fluxes were performed as for the purine metabolites and using
the same criteria and considerations. The concentration levels and venous-arterial concentration
differences of the pyrimidine bases were just as small as those of the purine bases (Table 3 in paper
II manuscript III) (Kristensen et al., 2007; Reynolds et al., 1988; Seal and Reynolds, 1993). Thus,
net fluxes were calculated for all pyrimidine metabolites, except β-ureidopropionic acid (only exp.
I), cytosine, thymine, and uracil.
Release of pyrimidine metabolites from the portal-drained viscera
The pattern of net PDV release of the pyrimidine metabolites was found to be quite different from
that of the purine metabolites. The pyrimidine metabolites were to a much larger extend released
from the PDV intact as nucleosides and much smaller amounts of pyrimidine degradation products
than purine degradation products were observed (Table 4 in paper II and manuscript III). From