Available via license: CC BY 4.0
Content may be subject to copyright.
RESEARCH ARTICLE
Metabolomic Profiling in Individuals with a
Failing Kidney Allograft
Roberto Bassi
1,2☯
, Monika A. Niewczas
3☯
, Luigi Biancone
4
, Stefania Bussolino
4
,
Sai Merugumala
5
, Sara Tezza
1
, Francesca D’Addio
1,2
, Moufida Ben Nasr
1
,
Alessandro Valderrama-Vasquez
2
, Vera Usuelli
2
, Valentina De Zan
6
, Basset El Essawy
7
,
Massimo Venturini
8
, Antonio Secchi
2,6
, Francesco De Cobelli
6,8
, Alexander Lin
9
,
Anil Chandraker
10
, Paolo Fiorina
1,2
*
1Nephrology Division, Boston Children’s Hospital, Harvard Medical School, Boston, MA, United States
of America, 2Transplant Medicine, IRCCS Ospedale San Raffaele, Milan, Italy, 3Section on Genetics
and Epidemiology, Joslin Diabetes Center, Harvard Medical School, Boston, MA, United States of
America, 4San Giovanni Battista Hospital and University of Turin, Division of Nephrology, Dialysis, and
Transplantation, Turin, Italy, 5Biomedical Engineering, University of Texas, Austin, TX, United States of
America, 6Universita’ Vita-Salute San Raffaele, Milan, Italy, 7Medicine, Al-Azhar University, Cairo, Egypt,
8Radiology, San Raffaele Scientific Institute, Milan, Italy, 9Center for Clinical Spectroscopy, Department of
Radiology, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, United States of America,
10 Transplantation Research Center, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA,
United States of America
☯These authors contributed equally to this work.
*paolo.fiorina@childrens.harvard.edu
Abstract
Background
Alteration of certain metabolites may play a role in the pathophysiology of renal allograft
disease.
Methods
To explore metabolomic abnormalities in individuals with a failing kidney allograft, we ana-
lyzed by liquid chromatography-mass spectrometry (LC-MS/MS; for ex vivo profiling of
serum and urine) and two dimensional correlated spectroscopy (2D COSY; for in vivo study
of the kidney graft) 40 subjects with varying degrees of chronic allograft dysfunction strati-
fied by tertiles of glomerular filtration rate (GFR; T1, T2, T3). Ten healthy non-allograft indi-
viduals were chosen as controls.
Results
LC-MS/MS analysis revealed a dose-response association between GFR and serum con-
centration of tryptophan, glutamine, dimethylarginine isomers (asymmetric [A]DMA and
symmetric [S]DMA) and short-chain acylcarnitines (C4 and C12), (test for trend: T1-T3 =
p<0.05; p = 0.01; p<0.001; p = 0.01; p = 0.01; p<0.05, respectively). The same association
was found between GFR and urinary levels of histidine, DOPA, dopamine, carnosine,
SDMA and ADMA (test for trend: T1-T3 = p<0.05; p<0.01; p = 0.001; p<0.05; p = 0.001;
p<0.001; p<0.01, respectively). In vivo 2D COSY of the kidney allograft revealed significant
PLOS ONE | DOI:10.1371/journal.pone.0169077 January 4, 2017 1 / 14
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
OPEN ACCESS
Citation: Bassi R, Niewczas MA, Biancone L,
Bussolino S, Merugumala S, Tezza S, et al. (2017)
Metabolomic Profiling in Individuals with a Failing
Kidney Allograft. PLoS ONE 12(1): e0169077.
doi:10.1371/journal.pone.0169077
Editor: Valquiria Bueno, Universidade Federal de
Sao Paulo, BRAZIL
Received: April 18, 2016
Accepted: December 12, 2016
Published: January 4, 2017
Copyright: ©2017 Bassi et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: All relevant data are
within the paper and its Supporting Information
files.
Funding: This work was supported by a AST
Genentech/Novartis Clinical Science Fellowship
grant to RB and an Italian Society of Diabetes
(AMD-SID) Pasquale di Coste Award to RB. RB
was supported by a JDRF Post-Doctoral Research
Fellowship grant. MAN received a JDRF Career
Development Award (5-CDA-2015-89-A-B). PF was
supported by an American Heart Association (AHA)
Grant-In-Aid and the Italian Ministry of Health
reduction in the parenchymal content of choline, creatine, taurine and threonine (all: p<0.05)
in individuals with lower GFR levels.
Conclusions
We report an association between renal function and altered metabolomic profile in renal
transplant individuals with different degrees of kidney graft function.
Introduction
Kidney transplantation has become the most widespread organ engrafting procedure [1].
While advances in immunosuppressive protocols have reduced the incidence of kidney acute
rejection over the years [2], long-term outcome of the kidney allograft remains affected by the
persistence of chronic allograft dysfunction [3–6]. The success of a renal transplant strictly
depends on the ability of monitoring transplant recipients and responsively changing their
medications. Unfortunately, we are still relying on the measurement of serum creatinine levels
and proteinuria to assess kidney function, which are non-specific and insensitive markers
[7,8,9] and whose increase may underlie an already predominantly lost kidney function [8,9].
Also, metabolic tests and imaging techniques which are routinely employed to detect graft dys-
function, in some circumstances do not provide adequate specificity, sensitivity, or accuracy
[7,10]. Thus, follow-up biopsies, both inconvenient to the patient and associated with expen-
sive histopathological analysis, are required to reach a definitive diagnosis [11]. The appear-
ance of novel techniques that allow the detection of unprecedentedly discovered pathways or
unidentified metabolites, may lead to a whole new era of patient management, particularly the
use of novel "omics" may generate opportunities unexplored thus far, ideally bypassing the
shortcomings of the current routine diagnostic tools. Metabolomics has the potential to per-
form an unbiased, non-targeted and dynamic analysis of low molecular mass cellular products,
thus making it an ideal candidate for the discovery of new potential markers of renal graft
function in the transplant patient [12,13,14,15]. Multiple studies report the association
between certain immunosuppressive schemes and specific metabolic alterations in urine and
serum of transplant patients [16–18] while others propose a relationship between acute renal
allograft rejection and urine metabolic profile [19]. Metabolite alteration may also accompany
the progression of chronic kidney allograft dysfunction and this may be relevant for the out-
come both in terms of graft survival and health of the patient. Thus, aiming to explore the pro-
file of metabolomic abnormalities induced by the progressive reduction of kidney function
and their potential impact on kidney graft function, we took advantage of two complementary
approaches: liquid chromatography-mass spectrometry (LC-MS/MS) for targeted metabolo-
mic profiling of serum and urine [20] and two dimensional correlated spectroscopy (2D
COSY) [21,22] for the in vivo metabolomic profiling of the kidney allograft, in a population of
individuals with different degrees of graft dysfunction, defined by progressively lower levels of
glomerular filtration rate (GFR) and a pool of healthy non-allograft individuals controls. We
thus performed an analysis of the transplant individual at the serum, urine and kidney graft
level by taking advantage of the latest analytical techniques, in order to gain insights into the
metabolomic abnormalities evident in individuals with failing kidney allografts.
Materials and Methods
A complete description of methods is offered in the S1 Data.
Metabolomic Profiling and Failing Kidney Allograft
PLOS ONE | DOI:10.1371/journal.pone.0169077 January 4, 2017 2 / 14
(grant RF-2010-2303119). PF also received
support from the EFSD/Sanofi European Research
Programme. The funders had no role in study
design, data collection and analysis, decision to
publish, or preparation of the manuscript.
Competing Interests: The authors have declared
that no competing interests exist.
Abbreviations: 1D-NMR, one dimensional-nuclear
magnetic resonance; 2D-COSY, two dimensional
correlated spectroscopy; ADMA, asymmetric
dimethylarginine; ATP, adenosine tri-phosphate;
AU, arbitrary unit; C4, butyrylcarnitine; C12,
Dodecenoylcarnitine; Ctrl, control; FIA-MS/MS,
Flow Injection Analysis-mass spectrometry; GFR,
glomerular filtration rate; H, hydrogen; HLA, human
leukocyte antigen; IDO, indoleamine 2,3-
dioxygenase; K, potassium; LC-MS/MS, liquid
chromatography-mass spectrometry; Na, sodium;
SDMA, symmetric dimethylarginine; T, tertile; TGF-
β1, trasforming growth factor-β1.
Patient characteristics
Forty kidney transplant individuals, with at least 6 months of follow-up after transplantation,
were admitted for post-transplantation routine analysis. After clinical evaluation, individuals
were enrolled in the study and assigned to different groups according to degree of allograft
function impairment. Transplant patients were then stratified in tertiles according to GFR
distribution (T1, T2 and T3) as shown in Table 1. Exclusion criteria were defined as (i)
GFR <25 ml/min; (ii) serum creatinine >3.0 mg/dl; (iii) severe uncontrolled arterial hyper-
tension; and (iv) arterial renal stenosis (assessed with Color Doppler Ultrasonography).
Finally, the control group (Ctrl) consisted of ten healthy individuals with normal renal func-
tion. Data were obtained after individuals’ written consent. The study protocol was conducted
after Institutional Review Board approval. A blinded code was assigned to each participating
patient. Kidney transplant recipients did not differ with regard to donor age, HLA match,
panel of reactive antibodies, cold ischemia time, rejection rate, cytomegalovirus infection and
lymphoproliferative diseases across the various renal function strata.
Metabolomics protocol
To gain greater insight into the metabolome of the kidney transplant patient, we opted to use a
novel and unique composite approach to define the of ex vivo (serum and urine) and in vivo
metabolomic profile of kidney transplant individuals by using LC-MS/MS, FIA-MS/MS
(n = 40 patients) and 2D COSY [22] with subsequent 3D-image transformation [23], per-
formed on a subgroup (n = 15) of renal transplant individuals. For additional details on the
metabolomics protocol, please refer to S1 Data. The Human Metabolome DataBase (HMDB,
http://www.hmdb.ca/) was used to study the metabolic pathways at the base of the observed
molecular alterations and to hypothesize potential effects of these on graft function.
Statistical analysis
Serum markers were presented as median (25th, 75th percentiles). Urinary markers normal-
ized to creatinine were presented as median (25th, 75th percentiles). Serum and urinary
metabolites present in at least 80% of the study subjects were designated as common and sub-
jected to further analysis. Kidney transplant recipients were stratified according to the distri-
bution of renal function (T1, T2, T3), in which T3 represented subjects with impaired graft
Table 1. Demographic and metabolic characteristics of kidney transplant individuals. Results are expressed as median (25
th
, 75
th
percentile).
T1 (56–108 ml/min) T2 (46–55 ml/min) T3 (21–39 ml/min) p-value
Age 56.0 (44.5, 62.0) 62.0 (53.0, 65.0) 55.0 (48.0, 65.0) ns
Pre-transplant dialysis duration (months) 35.0 (15.5, 112.5) 53.0 (43.0, 90.0) 78.0 (15.7, 102.0) ns
Follow-up (months) 75.0 (48.5, 115.0) 77.0 (30.0, 118.0) 64.5 (15.7, 185.8) ns
Systolic blood pressure (mmHg) 130.0 (127.5, 150.0) 130.0 (130.0, 140.0) 140.0 (121.3, 152.5) ns
Diastolic blood pressure (mmHg) 80.0 (75.0, 90.0) 80.0 (80.0, 85.0) 75.5 (70.0, 80.0) ns
Cholesterol (mg/dl) 160.0 (147.5, 187.0) 180.0 (155.0, 213.0) 204.5 (176.8, 240.0) ns
Triglycerides (mg/dl) 106.0 (72.5, 159.5) 166.0 (83.0, 209.0) 140.0 (100.8, 198.8) ns
BUN (mg/dl) 56.5 (49.5, 76.5) 76.0 (67.0, 103.8) 104 (88.75, 150.8) 0.009
GFR (ml/min/1.73m
2
) 65.0 (60.0, 83.5) 50.0 (48.0, 55.0) 34.5 (24.2, 35.7) by design
S-Creatinine (mg/dl) 1.3 (1.2, 1.6) 1.5 (1.5, 1.8) 2.4 (2.0, 2.7) <0.0001
AER (g/day) 0.1 (0.1, 0.2) 0.3 (0.1, 0.4) 1.0 (0.2, 2.5) 0.009
Abbreviations. Male (M); female (F); blood urea nitrogen (BUN); glomerular filtration rate (GFR); albumin excretion rate (AER).
doi:10.1371/journal.pone.0169077.t001
Metabolomic Profiling and Failing Kidney Allograft
PLOS ONE | DOI:10.1371/journal.pone.0169077 January 4, 2017 3 / 14
function, T2 subjects with fairly conserved renal allograft function and T1 represented subjects
with well-preserved renal function, respectively. Multivariate analysis (volcano plot) of com-
mon metabolites represents a fold difference (x-axis) between mean values of the metabolites
within T3 and T1 strata respectively, whereas nominal significance is presented on the y-axis
(Figs 1and 2). Differences among the groups were evaluated in the general linear model based
on the metabolites transformed to their logarithms (base 10). Study groups were treated in a
categorical (T3 vs. T1, T1 vs. Ctrl) or an ordinal way (T1-T3) appropriately. Spearman non-
parametric correlation matrix was created among kidney transplant recipients to evaluate cor-
relations among the metabolites. Correlation coefficients are presented. All tests were two-
sided, and a p value of less than 0.05 was considered indicative of statistical significance. Data
analysis was performed using SAS version 9.3 (SAS Institute, Cary, NC).
Results
Individual characteristics
Forty kidney transplant individuals were enrolled in our cross-sectional study and stratified
according to tertiles of GFR distribution as follows: T1 = 56–108 ml/min; T2 = 46–55 ml/min;
and T3 = 21–39 ml/min. Individuals among groups did not show major differences in terms of
demographic characteristics, lipid profile or blood pressure measurements (Table 1), while
mean group comparison revealed significant differences in blood urea nitrogen, serum creati-
nine and albumin excretion rate among T1, T2 and T3 (Table 1).
Ex vivo LC-MS/MS and FIA-MS/MS in kidney transplant individuals with
different degrees of graft function
We took advantage of the AbsoluteIDQ
TM
p180 kit assay (BIOCRATES Life Sciences AG) to
determine serum and urinary concentration of 190 metabolites divided as follows: amines
(amino acids and biogenic amines), acylcarnitines, phosphatidylcolines, sphingomyelins, lyso-
phosphatidylcolines and hexose. The majority of the biochemical classes of metabolites were
commonly detected in serum except for acylcarnitines, for which the detectability was 37%.
On the contrary, there were two major biochemical classes of easily detectable metabolites in
urine: amino acid and biogenic amines (88%) and acylcarnitines (46%). Finally, all lipid
metabolites were below the limit of method detection in the urine samples (S1 Table).
Serum metabolomic profiling
Protein or amino acid metabolism alterations, dietary deficiencies, increased catabolic degra-
dation and inflammation are some of the causes behind metabolite abnormalities in serum
among kidney graft individuals [24]. In our cohort, glutamine was progressively higher in kid-
ney transplant individuals with impaired GFR (T3) as compared to patients with more pre-
served kidney function (T1) (T3 = 765 [749, 827] vs. T1 = 658 [640, 685] μM, p = 0.01; Fig 1A
and 1B). Conversely, serum tryptophan was reduced in patients with lower GFR (T3) as com-
pared to T1 patients (T3 = 71 [59, 76] vs. T1 = 93 [84, 94] μM, p<0.05; Fig 1A and 1B). Low
serum tryptophan concentrations have been linked to inflammation and regulation of the
immune response; in particular, indoleamine 2,3-dioxygenase (IDO)-mediated tryptophan
catabolism has been reported during allograft rejection [25].
Among biogenic amines, dimethylarginine (DMA) analogues showed significant differ-
ences among groups. Specifically, asymmetric (A)DMA was increased in patients with reduced
GFR (T3 = 2.47 [2.06, 2.57] vs. T1 = 1.56 [1.34, 1.68] μM, p<0.001; Fig 1A and 1B). Compari-
son of ADMA between T1 patients and control individuals revealed increases in serum
Metabolomic Profiling and Failing Kidney Allograft
PLOS ONE | DOI:10.1371/journal.pone.0169077 January 4, 2017 4 / 14
ADMA levels in kidney recipients but with preserved renal function (T1 vs. Ctrl = 0.88 [0.73,
0.99] μM, p<0.0001; Fig 1B). Similarly, symmetric DMA (S)DMA was increased in patients
with low GFR (T3 = 1.64 [1.30, 1.95] vs. T1 = 1.02 [0.98, 1.28] μM, p = 0.01; Fig 1A and 1B)
and SDMA serum concentration variations were proportional to kidney graft performance
(test for trend [T1-T3]: p = 0.01; Fig 1 B). Reference individuals with normal renal function
displayed lower SDMA levels as compared to T1 patients (T1 vs. Ctrl = 0.48 [0.35, 0.52] μM,
p = 0.0004; Fig 1B). Methylarginine isomers have been previously reported to be altered in
individuals with chronic renal failure [26], perpetrating kidney damage through inhibition of
nitric oxide synthase activity, induction of collagen and TGF-β1 synthesis and constituting
independent causes of mortality and cardiovascular risk [27].
A common finding during renal insufficiency is the elevation of acylcarnitine serum con-
tent, most likely due to defective kidney excretion [28]. In the sample of patients under study,
butyrylcarnitine (C4) and dodecanoylcarnitine (C12) were significantly higher in patients with
Fig 1. (A) Multivariate analysis (volcano plot) of common metabolites measured in the serum on the Biocrates platform and their association with glomerular
filtration rate (GFR; T1-T3) are reported as fold difference (x-axis), and nominal significance is presented on the y-axis. (B) Serum metabolites significantly
different among patients with varying renal function are shown in the kidney transplant recipient (T1, T2, T3) and the control (Ctrl) group. (C) Spearman
nonparametric correlation matrix among the metabolites in serum significantly associated with varying kidney transplant function. Correlation coefficients are
presented. Significant associations are marked with an asterisk (*).
doi:10.1371/journal.pone.0169077.g001
Metabolomic Profiling and Failing Kidney Allograft
PLOS ONE | DOI:10.1371/journal.pone.0169077 January 4, 2017 5 / 14
worse graft function (C4: T3 = 0.60 [0.35, 0.74] vs. T1 = 0.30 [0.26, 0.42] μM, p = 0.01; C12:
T3 = 0.12 [0.08, 0.15] vs. T1 = 0.10 [0.08, 0.11] μM, p<0.05; Fig 1A and 1B). Both C4 and C12
were significantly different between control individuals and kidney graft patients with con-
served renal function (C4: T1 vs. Ctrl = 0.15 [0.11, 0.23] μM, p = 0.0007; C12: T1 vs. Ctrl = 0.08
[0.07, 0.10] μM, p = 0.0015; Fig 1B). These alterations are also consistent with the impaired
fatty acid metabolism and subsequent acylcarnitine accumulation that occur during renal
failure.
Spearman correlation matrix of serum metabolites significantly associated with kidney
graft function revealed that only certain metabolites were correlated with each other. Total
DMA, SDMA and acylcarntine C4 were significantly correlated. Interestingly, there was also
an inverse association between acylcarnitine C12 and tryptophan. Glutamine did not correlate
with any other metabolite in the studied matrix (Fig 1C).
Urine metabolomic profiling
Urinary levels of amino acids and biogenic amines were overall reduced in individuals with
poor graft function, pointing to reduced biosynthesis, enhanced catabolism or poor filtration
of these classes of metabolites. Urinary histidine was reduced in T3 kidney graft patients as
compared to patients with more conserved graft function (T3 = 12.0 [10.0, 20.0] vs. T1 = 31.0
[22.0, 54.0] μM, p<0.05; Fig 2A and 2B). Histidine is an anti-inflammatory and anti-oxidant
factor, and its decrease has been associated with systemic inflammation and increased mortal-
ity in individuals with poor kidney function [29].
Among biogenic amines, the urinary concentration of carnosine was reduced in patients
with a failing graft (T3 = 0.11 [0.09, 0.17] vs. T1 = 0.46 [0.20, 0.56] μM, p<0.05; Fig 2A and
2B). Similarly, free urinary dopamine was decreased in T3 individuals compared to T1 (T3 =
0.08 [0.07, 0.10] vs. T1 = 0.14 [0.12, 0.22] μM, p<0.001) and Ctrl individuals (T1 vs. Ctrl = 0.18
[0.17, 0.28] μM, p<0.05; Fig 2A and 2B). Urinary DOPA (a metabolic precursor of dopamine),
followed the same pattern as dopamine (Fig 2A and 2B), and finally, total DMA and its two
analogues ADMA and SDMA were lower in patients in the T3 group (total DMA: T3 = 4.31
[4.19, 4.80] vs. T1 = 7.25 [5.26, 7.48] μM, p = 0.001; ADMA: T3 = 1.22 [1.02, 1.26] vs. T1 = 2.41
[1.72, 3.30] μM, p<0.001; SDMA: T3 = 3.19 [2.97, 3.32] vs. T1 = 4.34 [4.19, 4.98] μM, p<0.01;
Fig 2A and 2B). Notably, reduction in urinary ADMA, and in general disturbance of nitric
oxide metabolism, have been recently associated with renal graft failure and increased mortal-
ity in individuals following kidney transplantation [30].
Spearman correlation matrix of urinary metabolites significantly associated with kidney
graft function revealed that all the respective metabolites were significantly correlated with
each other (Fig 2C).
In vivo 2D COSY spectroscopy
A subgroup of fifteen individuals (n = 5 from each GFR tertile subgroup) underwent 2D
COSY examination for in vivo analysis of the transplanted kidney (Fig 3B1–3C1). Subse-
quently, additional 3D image post-processing was performed to better visualize and compare
the differences among resonance and crosspeaks composing the spectra obtained from T1 and
T3 groups (Fig 3A and 3B2–3C2), and 25 metabolites were identified. Mean comparison of the
crosspeak volumes revealed differences in the concentrations of the amino acids taurine,
which acts as an antioxidant agent and prevents lipid peroxidation of mesangial and tubular
epithelial cells [31], and threonine, whose role remains obscure in the context of renal function
(Fig 3A, 3C1 and 3C2). Both amino acids were found to be significantly reduced in the T3
group as compared to the T1 (Taurine: T3 = 0.07 [0.03, 0.10] vs. T1 = 0.12 [0.07, 0.16] arbitrary
Metabolomic Profiling and Failing Kidney Allograft
PLOS ONE | DOI:10.1371/journal.pone.0169077 January 4, 2017 6 / 14
units [AU], p = 0.04; Threonine: T3 = 0.11 [0.06, 0.19] vs. T1 = 0.25 [0.19, 0.27] AU, p = 0.006;
Fig 3A, 3B1, 3B2, 3C1 and 3C2). Among other metabolites identified by in vivo 2D COSY
examination, choline, an essential nutrient with a pivotal role in the synthesis of cell mem-
branes and neurotransmitters [32] (e.g. acetylcholine), was significantly reduced in T3 individ-
uals as compared to T1 (Choline 1: T3 = 0.12 [0.04, 0.17] vs. T1 = 0.21 [0.16, 0.33] AU,
p = 0.01; Choline 2: T3 = 0.15 [0.09, 0.45] vs. T1 = 0.47 [0.29, 0.65] AU; p = 0.04, Fig 3A, 3B1,
3B2, 3C1 and 3C2). Similarly, creatine, whose major function is to transport high energy
groups from their site of production (mitochondria) to the site of ATP consumption in the
cytoplasm [33], was depleted in kidneys from T3 allograft individuals compared to T1 patients
(T3 = 0.04 [0.03, 0.08] vs. T1 = 0.11 [0.06, 0.16] AU, p = 0.03, Fig 3A, 3B1, 3B2, 3C1 and 3C2).
Taken together, these data suggest that in kidney transplant individuals, low GFR may be asso-
ciated with reduced metabolism/high-energy levels and reduced cellularity of the renal graft.
Fig 2. (A) Multivariate analysis (volcano plot) of common metabolites measured in the urine on the Biocrates platform and their association with glomerular
filtration rate (GFR; T1-T3) are reported as fold difference (x-axis), and nominal significance is presented on the y-axis. (B) Urinary metabolites significantly
different among patients with varying renal function are presented in the kidney transplant recipient (T1, T2, T3) and the control (Ctrl) group. (C) Spearman
nonparametric correlation matrix among the metabolites in urine significantly associated with varying kidney transplant function. Correlation coefficients are
presented. Significant associations are marked with an asterisk (*).
doi:10.1371/journal.pone.0169077.g002
Metabolomic Profiling and Failing Kidney Allograft
PLOS ONE | DOI:10.1371/journal.pone.0169077 January 4, 2017 7 / 14
Discussion
In this work, we have characterized the metabolite profile of biofluids (i.e. serum, urine) and of
kidney allograft parenchyma in transplanted individuals with varying degrees of filtration
impairment, by using novel analytical techniques that allow unbiased quantification of the
molecular alterations associated with chronic allograft dysfunction [34], with the goal of defin-
ing the association between kidney allograft dysfunction and metabolomic fingerprint. Modifi-
cations in specific metabolites have been shown to be involved, either as a cause or symptom,
in kidney disease. For instance, circulating amines (i.e. amino acids and biogenic amines) are
promptly altered during the early phases of kidney impairment [34], and the more the graft
fails, the more the imbalance becomes clear. These alterations can usually be attributed to
increased protein degradation, inflammation [24,35] or protein malnutrition. Accordingly,
Fig 3. Two dimensional Correlated Spectroscopy (2D COSY) results of the kidney allograft. (A) Table of 2D COSYcrosspeak volumes shows
significantly lower threonine, taurine, creatine and choline content in T3 individuals with low glomerular filtration rate and severe allograft dysfunction when
compared to T1 individuals with more conserved graft function. “-”indicates a p value greater than 0.05. (B) Representative 2D COSY spectra show higher
content of lipid-derived metabolites and reduced levels of threonine, taurine, creatine and choline in T3 individualscarrying a failing allograft. B1 shows a
topological map of crosspeaks and B2 shows the 3D reconstruction. (C) Representative 2D COSY of T1 allograft patients with more conserved graft function
with two-dimensional (C1) and three-dimensional (C2) reconstruction of the 2D COSY data. Data are expressed as median (25
th
, 75
th
percentile).
Abbreviations. Arbitrary Units (AU).
doi:10.1371/journal.pone.0169077.g003
Metabolomic Profiling and Failing Kidney Allograft
PLOS ONE | DOI:10.1371/journal.pone.0169077 January 4, 2017 8 / 14
we showed that serum tryptophan alterations began to appear in T1 kidney allograft patients,
to worsen in poor allograft function patients (T3) with a dose response trend, and this decline
was not explained by urinary losses. Low serum tryptophan can be explained by an accelerated
breakdown rate by the immunomodulatory enzyme IDO, due to an excess of inflammation/
immune activation [25]. Evidence of a parallel reduction in urinary tryptophan in T3 individu-
als, who had poor allograft function, points to systemic exhaustion of this amino acid, rather
than localized waste. Conversely, high serum glutamine can be explained by the substantial
reduction in glutamine uptake that often takes place during chronic renal disease [36], and
this is further confirmed by the reduction in urinary glutamine in individuals with GFR
impairment. However, apart from the evidence that glutamine catabolism is one of the major
determinants of ammonemia in these patients, the kinetics of glutamine in renal dysfunction
are still largely unknown [37]. Progressive increase in the concentration of serum DMA deriv-
atives coupled with decrease in their urinary excretion was also evident in individuals with
more severe graft dysfunction. Low GFR can explain reduced excretion and serum accumula-
tion of ADMA and SDMA, also confirming their classification as toxic uremic retention sol-
utes [38]. DMA isomers appear to induce kidney damage through inhibition of nitric oxide
synthase, induction of the synthesis of collagen and TGF-β1 and sodium retention (22), sup-
porting the hypothesis that there is a relationship between ADMA and hypertension or glo-
merulosclerosis, two main determinants of kidney injury progression [39]. Finally, higher
serum concentration of short-chain acylcarnitines (C4 and C12) in T3 patients can be attrib-
uted to the loss of renal parenchyma typical of long-term renal failure that, by removing a
source of endogenous carnitine synthesis (thus reducing the handling and consumption of
acylcarnitines), impairs the ability of the kidney to excrete acylcarnitine into the urine [40].
Finally, decreased levels of the branched chain amino acids have been described in the pres-
ence of advanced chronic kidney disease in some reports [41]. In our study, serum levels of
leucine, isoleucine and valine did not differ between control groups and kidney transplant
recipients with varying renal function, most probably due to the overall good nutritional status
across groups of subjects under study.
Interesting results are also evident from urine mass spectrometry analysis. DOPA and
dopamine were reduced in T3 transplant individuals as compared to patients with more con-
served GFR. In the kidney, dopamine, when coupled to D
1
-like receptors in the proximal
tubule, causes inhibition of sodium reabsorption by blocking Na/H-exchanger and Na/
K-ATPase activity, thus regulating blood pressure. Notably, the absence of the same findings
in the serum points to a reduction in dopamine synthesis at the kidney level—evidence previ-
ously linked with onset of hypertension. Reduction in urinary ADMA has also been associated
with reduction in the lifespan of the kidney graft and overall mortality in kidney transplant
patients [30].
Changes in the in vivo NMR spectroscopy profile of certain metabolites often precede mor-
phological or symptomatic changes in the kidney, brain, breast, and other organs [42–46].
Although traditional 1D-NMR spectroscopy is sufficient to observe distinct functional groups
in small molecules, many overlapping resonances in complex molecules can render the inter-
pretation of peaks more difficult [47]. The use of 2D COSY circumvents this challenge by
introducing a second dimension to the spectrum derived from the graft [48], while additional
3D image transformation adds further spatial detail to the examination. In our study, the novel
application of in vivo allograft 2D COSY spectroscopy revealed a 50% reduction in peak inten-
sity from threonine, taurine, choline and creatine in individuals with advanced allograft dys-
function. Notably, taurine concentration in our patients was significantly altered only at the
kidney graft level. Recent studies suggest that during kidney injury, transcriptional repression
of the taurine transporter by p53 determines intracellular depletion of taurine, causing
Metabolomic Profiling and Failing Kidney Allograft
PLOS ONE | DOI:10.1371/journal.pone.0169077 January 4, 2017 9 / 14
necrotic cell death [49]. On the other hand, taurine supplementation protects mesangial and
tubular cells from high glucose or hypoxia in vitro, ameliorates nephrotic syndrome or diabetic
nephropathy in vivo in animal models [31], and provides better outcomes in patients trans-
planted with kidneys from donors submitted to taurine preconditioning [50]. T3 patients also
displayed intra-graft reduction in choline, a pivotal factor for the synthesis of cell membranes
and cell-signaling components, a condition that can translate to acute renal failure and hyper-
tension in animal models [51]. Finally, in vivo 2D COSY spectroscopy of the renal allograft
revealed a reduction in creatine content, whose major function is the transport of high energy
groups from mitochondria to cytoplasm, and is produced/stored in the kidney cortex; how-
ever, the implications of lack of creatine in kidney pathology are not clear yet. Notably,
although not statistically different, a general increase in intra-graft lipid content among T3
patients was evident.
The limitations of our study include minor overlap of metabolomic profile of the imaging
study with the targeted metabolomics of the biofluids, rendering it impossible for us to evalu-
ate whether serum or urinary metabolites reflected systemic or local changes within the trans-
planted kidney, therefore, metabolomic disturbances identified here, will need to be studied
further using tools of the functional studies. We also acknowledge a relatively small sample
size as well as the cross-sectional nature of our study design. Finally, the patients included in
our analysis were heterogenous in terms of both immunosuppressive schemes and other treat-
ments that they were submitted to: the decision to opt for heterogeneous groups was based on
the assumption of generalizability of our analysis. irrespective of underlying metabolomic
alterations induced by exogenous treatments, while focusing on common patterns of meta-
bolic abnormalities merely determined by the extent of kidney function. However, potential
value of the candidate metabolites in predicting worsening kidney graft function will need to
be evaluated in the subsequent follow-up studies.
We report the existence of a relationship between different levels of kidney graft impair-
ment and imbalance of specific metabolites possibly linked to the pathophysiology of renal
graft dysfunction. Low GFR was significantly associated with serum circulating factors linked
to negative immunomodulation, hypertension, micro-ischemic events, fibrosis and cytotoxic-
ity. Metabolic alterations at the parenchymal level of the transplanted kidney were also evident
with significant reduction in high-energy and structural components of the graft parenchyma,
and finally analysis of the urinary matrix highlighted the existence of a pro-hypertensive and
pro-inflammatory environment within the transplanted organ.
Supporting Information
S1 Data.
(DOCX)
S1 Table. Overview of the total number of metabolites analyzed per biofluid (i.e. serum
and urine) before and after threshold selection based on detection in at least 80% of the
study patients (commonly detected). Metabolite concentrations are expressed as μM.
(DOCX)
S2 Table. Numerical report of amino acids significantly different among groups in urine
based on their alteration in serum. Data expressed as median (25th, 75th percentiles). Metab-
olite concentrations are expressed as μM.
(DOCX)
S3 Table. No differences were detected among groups in branched chain amino acid (iso-
leucine, leucine, valine) concentration in serum and urine. Data expressed as median (25th,
Metabolomic Profiling and Failing Kidney Allograft
PLOS ONE | DOI:10.1371/journal.pone.0169077 January 4, 2017 10 / 14
75th percentiles). Metabolite concentrations are expressed as μM.
(DOCX)
S4 Table. 2D Correlated Spectroscopy Crosspeak Assignments.
(DOCX)
S1 Fig. Representative 2D COSY voxel location as shown on 3 plane T2-weighted Magnetic
Resonance Imaging.
(DOCX)
Acknowledgments
This work was supported by a AST Genentech/Novartis Clinical Science Fellowship grant to
RB and an Italian Society of Diabetes (AMD-SID) Pasquale di Coste Award to RB. RB was sup-
ported by a JDRF Post-Doctoral Research Fellowship grant. MAN received a JDRF Career
Development Award (5-CDA-2015-89-A-B). PF was supported by an American Heart Associ-
ation (AHA) Grant-In-Aid and the Italian Ministry of Health (grant RF-2010-2303119). PF
also received support from the EFSD/Sanofi European Research Programme. The funders had
no role in study design, data collection and analysis, decision to publish, or preparation of the
manuscript.
Author Contributions
Conceptualization: RB PF MAN AC AL SM.
Data curation: RB PF AL SM ST MBN FD.
Formal analysis: RB MAN SM AL AC.
Funding acquisition: PF AC AL.
Investigation: RB MAN LB SM ST FD MBN AVV VU VDZ BEE AS FDC AL.
Methodology: RB MAN PF.
Project administration: RB PF MAN SM AL AC.
Resources: PF RB LB SB.
Software: RB MAN.
Supervision: AC AL AC BL.
Validation: RB MAN LB SM ST FD MBN AVV VU VDZ BEE AS FDC AL.
Writing – original draft: RB MAN PF.
Writing – review & editing: RB MAN LB SM ST FD MBN AVV VU VDZ BEE MV AS FDC
AL AC PF.
References
1. Murray JE, Merrill JP, Harrison JH. Renal homotransplantation in identical twins. 1955. J Am Soc
Nephrol. 2001; 12(1):201–4. PMID: 11317972
2. Kasiske BL, Skeans MA, Leighton TR, Ghimire V, Leppke SN, Israni AK. OPTN/SRTR 2011 Annual
Data Report: International Data. Am J Transplant. 2013; 13 Suppl 1:199–225.
3. Nankivell BJ, Chapman JR. Chronic allograft nephropathy: current concepts and future directions.
Transplantation. 2006; 81(5):643–54. doi: 10.1097/01.tp.0000190423.82154.01 PMID: 16534463
Metabolomic Profiling and Failing Kidney Allograft
PLOS ONE | DOI:10.1371/journal.pone.0169077 January 4, 2017 11 / 14
4. http://www.srtr.org. 2010.
5. Amico P. Evolution of graft survival in kidney transplantation: an analysis of the OPTN/UNOS Renal
Transplant Registry. Clin Transpl. 2010:1–15. PMID: 21698830
6. Dharnidharka VR, Fiorina P, Harmon WE. Kidney transplantation in children. N Engl J Med. 2014; 371
(6):549–58. doi: 10.1056/NEJMra1314376 PMID: 25099579
7. Filler G, Sharma AP. How to monitor renal function in pediatric solid organ transplant recipients. Pediatr
Transplant.
8. Inker LA, Schmid CH, Tighiouart H, Eckfeldt JH, Feldman HI, Greene T, et al. Estimating glomerular fil-
tration rate from serum creatinine and cystatin C. N Engl J Med. 2012; 367(1):20–9. doi: 10.1056/
NEJMoa1114248 PMID: 22762315
9. Niewczas MA, Gohda T, Skupien J, Smiles AM, Walker WH, Rosetti F, et al. Circulating TNFreceptors
1 and 2 predict ESRD in type 2 diabetes. J Am Soc Nephrol. 2012; 23(3):507–15. doi: 10.1681/ASN.
2011060627 PMID: 22266663
10. Singh AK, Sahani DV. Imaging of the renal donor and transplant recipient. Radiol Clin North Am. 2008;
46(1):79–93, vi. doi: 10.1016/j.rcl.2008.01.009 PMID: 18328881
11. Stillman IE, Pavlakis M. Allograft biopsies: studying them for all they’re worth. J Am Soc Nephrol. 2009;
20(11):2282–4. doi: 10.1681/ASN.2009090930 PMID: 19797470
12. Foxall PJ, Mellotte GJ, Bending MR, Lindon JC, Nicholson JK. NMR spectroscopy as a novel approach
to the monitoring of renal transplant function. Kidney Int. 1993; 43(1):234–45. PMID: 8433564
13. Serkova N, Fuller TF, Klawitter J, Freise CE, Niemann CU. H-NMR-based metabolic signatures of mild
and severe ischemia/reperfusion injury in rat kidney transplants. Kidney Int. 2005; 67(3):1142–51. doi:
10.1111/j.1523-1755.2005.00181.x PMID: 15698456
14. Le Moyec L, Pruna A, Eugene M, Bedrossian J, Idatte JM, Huneau JF, et al. Proton nuclear magnetic
resonance spectroscopy of urine and plasma in renal transplantation follow-up. Nephron. 1993; 65
(3):433–9. PMID: 8289995
15. Rush D, Somorjai R, Deslauriers R, Shaw A, Jeffery J, Nickerson P. Subclinical rejection—a potential
surrogate marker for chronic rejection—may be diagnosed by protocol biopsy or urine spectroscopy.
Ann Transplant. 2000; 5(2):44–9. PMID: 11217206
16. Kim CD, Kim EY, Yoo H, Lee JW, Ryu DH, Noh DW, et al. Metabonomic analysis of serum metabolites
in kidney transplant recipients with cyclosporine A- or tacrolimus-based immunosuppression. Trans-
plantation. 2010; 90(7):748–56. doi: 10.1097/TP.0b013e3181edd69a PMID: 20842074
17. Dieme B, Halimi JM, Emond P, Buchler M, Nadal-Desbarat L, Blasco H, et al. Assessing the metabolic
effects of calcineurin inhibitors in renal transplant recipients by urine metabolic profiling. Transplanta-
tion. 2014; 98(2):195–201. doi: 10.1097/TP.0000000000000039 PMID: 24598938
18. Wang J, Zhou Y, Xu M, Rong R, Guo Y, Zhu T. Urinary metabolomics in monitoring acute tubular injury
of renal allografts: a preliminary report. Transplant Proc. 2011; 43(10):3738–42. doi: 10.1016/j.
transproceed.2011.08.109 PMID: 22172837
19. Wang JN, Zhou Y, Zhu TY, Wang X, Guo YL. Prediction of acute cellular renal allograft rejection by uri-
nary metabolomics using MALDI-FTMS. J Proteome Res. 2008; 7(8):3597–601. doi: 10.1021/
pr800092f PMID: 18620448
20. Niewczas MA, Sirich TL, Mathew AV, Skupien J, Mohney RP, Warram JH, et al. Uremic solutes and
risk of end-stage renal disease in type 2 diabetes: metabolomic study. Kidney Int. 2014; 85(5):1214–24.
doi: 10.1038/ki.2013.497 PMID: 24429397
21. Ramadan S, Andronesi OC, Stanwell P, Lin AP, Sorensen AG, Mountford CE. Use of in vivo two-dimen-
sional MR spectroscopy to compare the biochemistry of the human brain to that of glioblastoma. Radiol-
ogy. 2011; 259(2):540–9. doi: 10.1148/radiol.11101123 PMID: 21357517
22. Ramadan S, Ratai EM, Wald LL, Mountford CE. In vivo 1D and 2D correlation MR spectroscopy of the
soleus muscle at 7T. J Magn Reson. 2010; 204(1):91–8. doi: 10.1016/j.jmr.2010.02.008 PMID:
20206561
23. Lin AP, Ramadan S, Stern RA, Box HC, Nowinski CJ, Ross BD, et al. Changes in the neurochemistry of
athletes with repetitive brain trauma: preliminary results using localized correlated spectroscopy. Alzhei-
mer’s research & therapy. 2015; 7(1):13.
24. Suliman ME, Qureshi AR, Stenvinkel P, Pecoits-Filho R, Barany P, Heimburger O, et al. Inflammation
contributes to low plasma amino acid concentrations in patients with chronic kidney disease. Am J Clin
Nutr. 2005; 82(2):342–9. PMID: 16087977
25. Brandacher G, Cakar F, Winkler C, Schneeberger S, Obrist P, Bosmuller C, et al. Non-invasive monitor-
ing of kidney allograft rejection through IDO metabolism evaluation. Kidney Int. 2007; 71(1):60–7. doi:
10.1038/sj.ki.5002023 PMID: 17136028
Metabolomic Profiling and Failing Kidney Allograft
PLOS ONE | DOI:10.1371/journal.pone.0169077 January 4, 2017 12 / 14
26. Fleck C, Schweitzer F, Karge E, Busch M, Stein G. Serum concentrations of asymmetric (ADMA) and
symmetric (SDMA) dimethylarginine in patients with chronic kidney diseases. Clin Chim Acta. 2003;
336(1–2):1–12. PMID: 14500028
27. Leiper J, Nandi M, Torondel B, Murray-Rust J, Malaki M, O’Hara B, et al. Disruption of methylarginine
metabolism impairs vascular homeostasis. Nat Med. 2007; 13(2):198–203. doi: 10.1038/nm1543
PMID: 17273169
28. Wanner C, Schollmeyer P, Horl WH. Serum carnitine levels and carnitine esters of patients after kidney
transplantation: role of immunosuppression. Metabolism. 1988; 37(3):263–7. PMID: 3278191
29. Zhang ZH, Wei F, Vaziri ND, Cheng XL, Bai X, Lin RC, et al. Metabolomics insights into chronic kidney
disease and modulatory effect of rhubarb against tubulointerstitial fibrosis. Sci Rep. 2015; 5:14472. doi:
10.1038/srep14472 PMID: 26412413
30. Frenay AR, van den Berg E, de Borst MH, Beckmann B, Tsikas D, Feelisch M, et al. Plasma ADMA
associates with all-cause mortality in renal transplant recipients. Amino Acids. 2015;47(9):1941–9. doi:
10.1007/s00726-015-2023-0 PMID: 26077715
31. Trachtman H, Sturman JA. Taurine: A therapeutic agent in experimental kidney disease. Amino Acids.
1996; 11(1):1–13. Epub 1996/03/01. doi: 10.1007/BF00805717 PMID: 24178634
32. Montes de Oca M, Perazzo JC, Monserrat AJ, Arrizurieta de Muchnik EE. Acute renal failure induced
by choline deficiency: structural-functional correlations. Nephron. 1980; 26(1):41–8. PMID:7393377
33. Heimburger O, Stenvinkel P, Barany P. The enigma of decreased creatinine generation in acute kidney
injury. Nephrol Dial Transplant. 2012; 27(11):3973–4. doi: 10.1093/ndt/gfs459 PMID: 23144066
34. Ceballos I, Chauveau P, Guerin V, Bardet J, Parvy P, Kamoun P, et al. Early alterations of plasma free
amino acids in chronic renal failure. Clin Chim Acta. 1990; 188(2):101–8. PMID: 2379310
35. Tizianello A, De Ferrari G, Garibotto G, Gurreri G, Robaudo C. Renal metabolism of amino acids and
ammonia in subjects with normal renal function and in patients with chronic renal insufficiency. J Clin
Invest. 1980; 65(5):1162–73. doi: 10.1172/JCI109771 PMID: 7364943
36. Adibi SA. Renal assimilation of oligopeptides: physiological mechanisms and metabolic importance.
Am J Physiol. 1997; 272(5 Pt 1):E723–36.
37. Fadel FI, Elshamaa MF, Essam RG, Elghoroury EA, El-Saeed GS, El-Toukhy SE, et al. Some amino
acids levels: glutamine,glutamate, and homocysteine, in plasma of children with chronic kidney disease.
Int J Biomed Sci. 2014; 10(1):36–42. PMID: 24711748
38. Vanholder R, De Smet R, Glorieux G, Argiles A, Baurmeister U, Brunet P, et al. Review on uremic tox-
ins: classification, concentration, and interindividual variability. Kidney Int. 2003; 63(5):1934–43. doi:
10.1046/j.1523-1755.2003.00924.x PMID: 12675874
39. Duranton F, Lundin U, Gayrard N, Mischak H, Aparicio M, Mourad G, et al. Plasma and urinary amino
acid metabolomic profiling in patients with different levels of kidney function. Clin J Am Soc Nephrol.
2014; 9(1):37–45. doi: 10.2215/CJN.06000613 PMID: 24235289
40. Moder M, Kiessling A, Loster H, Bruggemann L. The pattern of urinary acylcarnitines determined by
electrospray mass spectrometry: a new tool in the diagnosis of diabetes mellitus. Anal Bioanal Chem.
2003; 375(2):200–10. doi: 10.1007/s00216-002-1654-7 PMID: 12560963
41. Garibotto G, Sofia A, Saffioti S, Bonanni A, Mannucci I, Verzola D. Amino acid and protein metabolism
in the human kidney and in patients with chronic kidney disease. Clin Nutr. 2010; 29(4):424–33. doi: 10.
1016/j.clnu.2010.02.005 PMID: 20207454
42. Perseghin G, Fiorina P, De Cobelli F, Scifo P, Esposito A, Canu T, et al. Cross-sectional assessment of
the effect of kidney and kidney-pancreas transplantation on resting left ventricular energy metabolism in
type 1 diabetic-uremic patients: a phosphorous-31 magnetic resonance spectroscopy study. J Am Coll
Cardiol. 2005; 46(6):1085–92. doi: 10.1016/j.jacc.2005.05.075 PMID: 16168295
43. Fiorina P, Bassi R, Gremizzi C, Vergani A, Caldara R, Mello A, et al. 31P-magnetic resonance spectros-
copy (31P-MRS) detects early changes in kidney high-energy phosphate metabolism during a 6-month
Valsartan treatment in diabetic and non-diabetic kidney-transplanted patients. Acta Diabetol. 2012; 49
Suppl 1:S133–9.
44. D’Addio F, Maffi P, Vezzulli P, Vergani A, Mello A, Bassi R, et al. Islet transplantation stabilizes hemo-
static abnormalities and cerebral metabolism in individuals with type 1 diabetes. Diabetes Care. 2014;
37(1):267–76. doi: 10.2337/dc13-1663 PMID: 24026546
45. Fiorina P, Vezzulli P, Bassi R, Gremizzi C, Falautano M, D’Addio F, et al. Near normalization of meta-
bolic and functional features of the central nervous system in type 1 diabetic patients with end-stage
renal disease after kidney-pancreas transplantation. Diabetes Care. 2012; 35(2):367–74. doi: 10.2337/
dc11-1697 PMID: 22190674
46. Fiorina P, Perseghin G, De Cobelli F, Gremizzi C, Petrelli A, Monti L, et al. Altered kidney graft high-
energy phosphate metabolism in kidney-transplanted end-stage renal disease type 1 diabetic patients:
Metabolomic Profiling and Failing Kidney Allograft
PLOS ONE | DOI:10.1371/journal.pone.0169077 January 4, 2017 13 / 14
a cross-sectional analysis of the effect of kidney alone and kidney-pancreas transplantation. Diabetes
Care. 2007; 30(3):597–603. doi: 10.2337/dc06-1324 PMID: 17327327
47. Hene RJ, van der Grond J, Boer WH, Mali WP, Koomans HA. Pre-transplantation assessment of renal
viability with 31P magnetic resonance spectroscopy. Kidney Int. 1994; 46(6):1694–9. PMID: 7700029
48. Delaglio F, Wu Z, Bax A. Measurement of homonuclear proton couplings from regular 2D COSY spec-
tra. J Magn Reson. 2001; 149(2):276–81. doi: 10.1006/jmre.2001.2297 PMID: 11318630
49. Ying Y, Kim J, Westphal SN, Long KE, Padanilam BJ. Targeted deletion of p53 in the proximal tubule
prevents ischemic renal injury. J Am Soc Nephrol. 2014; 25(12):2707–16. doi: 10.1681/ASN.
2013121270 PMID: 24854277
50. Guan X, Dei-Anane G, Liang R, Gross ML, Nickkholgh A, Kern M, et al. Donor preconditioning with tau-
rine protects kidney grafts from injury after experimental transplantation. J Surg Res. 2008; 146(1):127–
34. doi: 10.1016/j.jss.2007.06.014 PMID: 18061615
51. Keith MO, Tryphonas L. Choline deficiency and the reversibility of renal lesions in rats. J Nutr. 1978;
108(3):434–46. PMID: 627918
Metabolomic Profiling and Failing Kidney Allograft
PLOS ONE | DOI:10.1371/journal.pone.0169077 January 4, 2017 14 / 14