Quantification of Metabolites by NMR Spectroscopy in the Presence of

Subscriber access provided by HACETTEPE UNIVERSITESI KUTUPHANESI

Article

Quantification of metabolites by NMR spectroscopy in the presence of protein Jens Wallmeier, Claudia Samol, Lisa Ellmann, Helena Ursula Zacharias, Franziska C. Vogl, Muriel Garcia, Katja Dettmer, Peter J. Oefner, and Wolfram Gronwald J. Proteome Res., Just Accepted Manuscript • DOI: 10.1021/acs.jproteome.7b00057 • Publication Date (Web): 15 Mar 2017 Downloaded from http://pubs.acs.org on March 18, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Journal of Proteome Research is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

Quantification of metabolites by NMR spectroscopy in the presence of protein

1

Jens Wallmeier, 1Claudia Samol, 1Lisa Ellmann, 1Helena U. Zacharias, 1Franziska

C. Vogl, 1Muriel Garcia, 1Katja Dettmer, 1Peter J. Oefner, 1Wolfram Gronwald* and 2

GCKD study investigators

1

Institute of Functional Genomics, University of Regensburg, Am Biopark 9, 93053

Regensburg, Germany

2

GCKD investigators are listed in the Acknowledgements

*Corresponding Author: Wolfram Gronwald, Institute of Functional Genomics, Am BioPark 9, 93053 Regensburg, Germany, E-mail: [email protected], Phone +49-(0)941-943-5015, Fax: +49-(0)941-943-5020

Abbreviated title: Metabolite quantification from CPMG spectra

1 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Abstract The high reliability of NMR spectroscopy makes it an ideal tool for large-scale metabolomic studies. However, the complexity of biofluids and, in particular, the presence of macromolecules poses a significant challenge. Ultrafiltration and protein precipitation are established means of deproteinization and recovery of free or total metabolite content, but neither is ever complete. In addition, aside from cost and labor, all deproteinization methods constitute an additional source of experimental variation. The Carr-Purcell-Meiboom-Gill (CPMG) echo-train acquisition of NMR spectra obviates the need for prior deproteinization by attenuating signals from macromolecules, but concentration values of metabolites measured in blood plasma will not necessarily reflect total or free metabolite content. Here, in contrast to approaches that propose the determination of individual T1 and T2 relaxation times for the computation of correction factors, we demonstrate their determination by spike-in experiments with known amounts of metabolites in pooled samples of the matrix of interest to facilitate the measurement of total metabolite content. Provided that the protein content does not vary too much among individual samples, accurate quantitation of metabolites is feasible. Moreover, samples with significantly deviating protein content may be readily recognized by inclusion of a standard that shows moderate protein binding. It is also shown that urinary proteins when present in high concentrations may effect detection of common urinary metabolites prone to strong protein binding such as tryptophan.

Keywords 1D NMR, quantification, plasma, CPMG, correction factors


Page 2 of 52

Page 3 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Introduction Nuclear magnetic resonance (NMR) spectroscopy is well suited for the quantification of small molecules in biological fluids.1,2 Advantages include that signal volumes scale linearly with concentrations over several orders of magnitude and that in most cases

very

limited

sample

pretreatment

is

required.

Furthermore,

NMR

measurements are highly robust and reproducible making NMR well suited for highthroughput analyses of large sample sets.3 However, chemically complex specimens such as plasma, serum and urine are prone to considerable signal overlap in onedimensional

1

H spectra. This may be overcome by the superior resolution of

multidimensional NMR spectra, albeit the time required for their acquisition becomes prohibitive for large cohort studies.4 A further alternative are 1D NMR spectra in combination with mathematical fitting of overlapping metabolite signals to signals modeled from pure compound spectra.5,6 However, this approach is limited to known compounds, for which reference spectra are available, and it will only work reasonably well in the absence of proteins, lipids and lipoproteins, which are typical components of blood plasma.7 These macromolecules may be removed by ultrafiltration or protein precipitation.8,9 However, ultrafiltration limits analysis to the unbound fraction of metabolites, though protein-bound metabolites may be recovered in part by extraction from the filter residues.9 In contrast, protein precipitation will release metabolites that bind with low affinity to proteins, thus enabling at least theoretically the determination of the total amounts of metabolites present.9,10 However, it has been reported that depending on the precipitating reagent employed protein removal is not always complete and signals arising from lipids and lipoproteins may be partially retained.9 Furthermore, all protein precipitation methods include a subsequent drying step that may lead to the loss of volatile metabolites.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Finally, both ultrafiltration and protein precipitation increase cost and labor of analysis, and they present an additional source of experimental variation. Application of diffusion and relaxation based spectral editing techniques allows the suppression of signals of selected classes of compounds such as macromolecules, thereby enhancing small molecular signals.7,11,12 Alternatively, by diffusion editing with strong gradients, spectra containing mostly macromolecular signals may be obtained, which are then subtracted from spectra containing both metabolite and macromolecular signals to yield spectra consisting predominantly of small molecular signals.11 For the determination of selected metabolites such as L-carnitine techniques based on single quantum coherence filtering may be used.13 For the simultaneous investigation of multiple metabolites, the Carr-Purcell-Meiboom-Gill (CPMG) pulse sequence, which employs a T2 based relaxation filter for the suppression of macromolecular signals, has gained wide-spread application.2,14 However, for metabolites showing interactions with macromolecules some amount of magnetization will be lost during the CPMG cycle and, therefore, their signal volumes will be decreased during signal acquisition, which in turn might lead to an underquantification of these molecules.7,12 These interactions can be described by the individual dissociation constants KD and the corresponding on- and off-rates. Depending on these values, a given metabolite will be found more or less in its free and protein-bound state, respectively, with constant exchange between these states. In case of fast exchange, an average of these two states will be observed that goes along with a shift in signal position compared to the purely free state in case the bound signal position is different. A potential increase in T2 relaxation of the bound form will reduce the amount of detectable magnetization during the CPMG cycle, leading to a subsequent reduction in observed signal volume. An increase in T2 relaxation of the bound form will also lead to an increased signal width of the 4 ACS Paragon Plus Environment

Page 4 of 52

Page 5 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


observed signal, which might hamper proper signal integration due to an increase in signal overlap. In case of slow exchange, only the unbound metabolite fraction will be detected, as greatly reduced signal volumes and increased signal widths will preclude detection of the protein-bound fraction. A good example is the standard reference substance 3-(trimethylsilyl)propionate (TSP), which shows considerable protein interaction. Compared to its free state, a shifted signal of reduced volume and increased width is observed indicating the presence of fast exchange. But even for metabolites showing only limited protein binding, such as creatinine,15,16 some influence of the plasma matrix on T2 relaxation is expected. Hence, in a complex biofluid such as human plasma observed T2 times will vary considerably for different metabolites7,12,17, ranging at 300 K from 63 ms for alanine to 280 ms for α-glucose.7 Furthermore, differences in T1 relaxation times may also effect signal intensities in case that the relaxation delay is shorter than 5 * T1. Addition of 0.8 mol/L NH4Cl to biospecimens has been reported as a means of releasing bound molecules from proteins prior to acquisition of Hahn-echo or CPMG spectra, but in case of the use of modern cryo-probes this will come at the expense of reduced detection sensitivity.12,17 In the present study, we explored the determination of calibration factors derived from spike-in experiments of different concentrations of metabolites to be quantitated as yet another means of determining the total amounts of metabolites present before applying the method to the quantitation of 22 selected metabolites in 267 plasma specimens from the German Chronic Kidney Disease (GCKD) study.18

Materials and Methods GCKD specimens



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The German Chronic Kidney Disease (GCKD) study was designed as a prospective observational cohort study that enrolled almost 5200 patients suffering from various stages of chronic renal impairment at nine study centers throughout Germany. Enrolled patients were aged between 18 and 74 years and exhibited either an estimated glomerular filtration rate (eGFR) of 30-60 mL/min per 1.73 m2 or an eGRF above 60 mL/min per 1.73 m2 and ‘overt’ albuminuria/ proteinuria.18 From all patients written informed consent was obtained.

Selection of reference compound In NMR, quantification of metabolite concentrations is often performed relative to a low-molecular weight reference. Ideally, the reference should not bind to protein to obtain a constant reference across samples. Here, formic acid has been selected.12 An alternative strategy considered is electronic referencing,19 but was discarded as it is difficult to obtain consistent results over hundreds of samples varying in their composition, with relative quantification being affected by differences in signal attenuation for different metabolites.

NMR sample preparation including different spike-in experiments For each of the 267 specimens of the GCKD cohort analyzed here, 400 µL of EDTAplasma were mixed with 200 µL of 0.1 mol/L phosphate buffer, pH 7.4, followed by the addition of 50 µL of 0.75% (w) 3-trimethylsilyl-2,2,3,3-tetradeuteropropionate (TSP) dissolved in deuterium oxide and 10 µL of 81.97 mmol/L formic acid (all from Sigma-Aldrich, Taufkirchen, Germany) dissolved in H2O. To analyze the recovery of creatinine in different biological matrices, two creatinine stock solutions were prepared containing creatinine at concentrations of 2.0 and 0.2 mmol/L, respectively. Two hundred µL of each stock solution were mixed with 200 µL 6 ACS Paragon Plus Environment

Page 6 of 52

Page 7 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


of either pure water, filtered or unfiltered EDTA plasma obtained from a volunteer with unimpaired kidney function, or unfiltered EDTA plasma randomly selected from the GCKD cohort. Filtered plasma was obtained by ultrafiltration employing a 10 kD cut-off filtering device.8,20 Buffer and reference substances were added as described above. For the simulation of matrix effects, three different stock solutions were prepared containing D-glucose, bovine serum albumin (BSA), and NaCl at concentrations of 20 mmol/L, 80g/L, and 308 mmol/L, respectively. These solutions were further used either undiluted, 1:1 or 1:2 diluted, resulting in a total of 9 preparations. From each of these preparations 350 µL were mixed with 50 µL of an 8 mmol/L creatinine stock solution giving a total of 9 samples. For reasons of comparison, an additional blank sample was prepared containing 400 µL of pure water. Buffer and reference substances were added as described above. For the determination of the unbound metabolite fraction in plasma, three unfiltered specimens were selected at random from the GCKD cohort and mixed with buffer, formic acid, and TSP. After measurement, the samples were ultrafiltered using Centrifree® (Merck, Darmstadt, Germany) ultrafiltration units with a nominal cut-off of 30,000 Da, which are recommended for the separation of free from bound microsolute in physiological fluids, before repeated NMR analysis. For the determination of calibration factors, 100 specimens randomly chosen from the GCKD cohort were pooled and appropriate amounts of metabolites were added. For amino acids, a commercially (Sigma-Aldrich, Taufkirchen, Germany) available mixture of 16 standard amino acids at a concentration of 2.5 mmol/L each was used. From this mix 10, 20 and 30 µL were added to 660 µL of pooled plasma each. For all other metabolites including glutamine and asparagine, which were not contained in



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

the standard amino acid mix, two spike-in levels were chosen with respect to the expected average endogenous concentrations. For this, 660 µL of pooled plasma were mixed with 10 µL of appropriately diluted stock solutions. By adding only small amounts of stock solutions to the pooled plasma matrix it was ensured that the macromolecular composition of the resulting sample was still an accurate representation of the pooled plasma matrix. Based on this data, correction factors were obtained. For NMR measurements of urine specimens, 400 µL of urine were mixed with 200 µL of 0.1 mol/L phosphate buffer at pH 7.4 and 50 µL of 0.75% (w) 3-trimethylsilyl2,2,3,3-tetradeuteropropionate (TSP) in deuterium oxide as internal standard (Sigma–Aldrich, Taufkirchen, Germany).

NMR measurements NMR experiments were carried out on a Bruker Avance III 600 MHz spectrometer employing a triple-resonance (1H,

13

C, 31P, 2H lock) cryogenic probe equipped with z-

gradients and an automatic sample changer. For each sample, the probe was automatically locked, tuned, matched and shimmed. All spectra were measured at 298 K. For the suppression of signals from macromolecules such as proteins and other substances with short T2 values, water suppressed 1D 1H spin-echo CarrPurcell-Meiboom-Gill (CPMG) spectra were acquired.21 The total echo time determined by the pulse length of the 180o pulse of the CPMG cycle plus the delay τe before and after the 180o pulse multiplied by the number of repetitions has to be adjusted depending on the intensity of the macromolecular signals. Beckonert et al. recommended for blood plasma a total echo time of at least 64 ms.22 In our case we adjusted the total echo time to 80 ms consisting of 128 repetitions with a τe time of 8 ACS Paragon Plus Environment

Page 8 of 52

Page 9 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


300 µs and a 180o pulse of approximately 27 µs. Note that the pulse-lengths for both 90o and 180o pulses were individually determined for each sample. For τe short values between 250 and 400 µs are recommended to avoid J-modulation due to spin-spin couplings.7 For each spectrum, a total of 128 scans was collected into 72k data points using a relaxation delay of 4 s and an acquisition time of 3.07 s. Spectra of urine specimens were acquired following established protocols.4 Employing TopSpin 3.1 (Bruker BioSpin GmbH, Rheinstetten, Germany), spectra were Fourier transformed and phase corrected, applying a line broadening of 0.3 Hz and zerofilling to 128 k points. Spectra were baseline corrected by applying a polynomial baseline correction.

Metabolite quantification from NMR spectra Metabolites of plasma specimens were quantified relative to the formic acid reference signal, while for urinary specimens the TSP reference signal was employed. To allow for the accurate determination of peak integrals even from partially overlapping signals spectral deconvolution is required. For this, the CHENOMX 8.1 (Chenomx Inc., Edmonton, Canada) software suite was used. First, automatically fitted integrals were obtained. These were then manually checked, as automatic fitting did not always yield optimally fitted integrals. Note, that prior to analysis of plasma specimens Chenomx was thoroughly evaluated by us and others including synthetic mixtures of pure compounds.5 As EDTA-plasma specimens were analyzed we verified by manual inspection that neither the signals of free EDTA nor of its metal complexes interfered with data analysis.

Quantification by liquid-chromatography-tandem mass spectrometry



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 52

Creatinine and creatine in plasma and tryptophan in urine were analyzed by reversed-phase

liquid

chromatography-electrospray

ionization-tandem

mass

spectrometry in positive ion multiple reaction monitoring mode (MRM) using a 1200 SL HPLC (Agilent, Böblingen, Germany) and a API 4000 QTrap mass spectrometer (ABSciex, Darmstadt, Germany). Solvents for LC-MS analysis were HPLC grade and purchased from VWR (Darmstadt, Germany). The water used was purified by a PURELAB Plus system (ELGA, LabWater, Celle, Germany). HPLC separation was performed on a Waters (Eschborn, Germany) Atlantis T3 column (2.1×150-mm i.d., 3 µm) with an acetonitrile gradient in 0.1% formic acid (v/v). For sample preparation, 10 µL of a 40-µM solution of creatinine-D3 (CDN Isotopes, Pointe-Claire, Canada) were added to 50 µL of plasma, which was also used for the quantification of creatine. After mixing, 200 µL methanol (VWR, Darmstadt, Germany) were added for protein precipitation and samples were kept at – 80°C overnight. Following centrifugation, supernatants were transferred to a 96-well plate (Agilent Technologies, Santa Clara, CA, USA) and vacuum evaporated (CombiDancer, Hettich AG, Bäch, Switzerland). Afterwards the residues were redissolved in 100 µL of water and the plates were sealed with a silicone mat (Agilent Technologies). For urinary tryptophan measurements, samples were diluted 1:50 with water prior to analysis and also spiked with a stable isotope labeled standard. Data acquisition and analysis was performed using Analyst v.1.6.2 from Applied Biosystems/MDS SCIEX. Quantification was performed using a calibration curve with creatinine-D3 as internal standard for both creatinine and creatine. Amino acids were quantified by LC-MS/MS after derivatization with propylchloroformate/propanol using 10 µL of plasma. Derivatization and subsequent analysis was performed as recently described. 23


Page 11 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Statistical data analysis The significance of differences in metabolite abundance between different groups of specimens was determined by the limma package within R (v. 3.3.0).24 First, an Ftest was performed, followed by pair wise comparisons using a two-sided unpaired ttest. Comparison of ultrafiltered and unfiltered spectra of the same specimens was performed by employing a two-sided paired t-test. Corresponding p-values for both Fand t-test were corrected for multiple testing by controlling the false discovery rate

25

at the 5% level.

Results Impact of biological matrix on determination of plasma creatinine by NMR First, we assessed the degree to which variation in matrix composition of human plasma specimens might impact the quantification of metabolites and, in particular, that of creatinine, which is not only widely used to assess kidney function but is also known to bind to human serum albumin.26 To that end, 267 plasma specimens randomly selected from the GCKD study were analyzed. Figure 1 shows a 1D 1H CPMG spectrum of a typical GCKD specimen. For reasons of clarity, only a few of the clearly visible metabolite signals are marked. It is apparent, that protein signals were almost completely suppressed. Nevertheless, there is considerable overlap of signals. Hence, for metabolite quantification relative to the formate reference signal, the Chenomx NMR data analysis software suite, which deals with overlapping resonances by spectral deconvolution, was employed.

Figure 2A shows a histogram of the levels of creatinine determined in the 267 GCKD plasma specimens by 1D 1H CPMG NMR. Since normal reference values range from 57-93 µmol/L for males and 50-80 µmol/L for females, respectively,27 it is obvious, 11 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 52

that the creatinine concentrations determined are too low for CKD patients, in whom plasma creatinine concentration increases with increasing impairment of glomerular filtration. Furthermore, parallel determination of plasma creatinine levels by means of an enzymatic creatinine assay performed by a commercial clinical chemistry laboratory

(Synlab

GmbH,

Augsburg,

Germany)

indicated

a

systematic

concentration-dependent underquantification of plasma creatinine by NMR with an average under quantification of 71.7 µmol/L (Figure 2B). Note that the enzymatic method determines the total amount of creatinine including both protein-bound and free creatinine. Thus, the data clearly shows that standard NMR will not measure the total amount of plasma creatinine.

Analysis of matrix effects To elucidate the impact of matrix effects on small molecular signal attenuation, we spiked two different concentrations of creatinine into both unfiltered and ultrafiltered (using a cut-off of 10 kD) plasma of a volunteer with normal kidney function, and into unfiltered plasma of a GCKD patient (Table 1). Employing the Chenomx software suite, creatinine was quantified relative to formic acid and its recovery was determined. Results showed an almost complete recovery of creatinine from water and filtered plasma, whereas recovery dropped to 69.7% and 46.1%, respectively, for the unfiltered plasma samples of the volunteer and the GCKD patient. This data corroborated the strong influence that different sample matrices might exert on the recovery of small molecules such as creatinine. Next, we investigated in more detail the effect of matrix composition on individual signal integrals. For this, typical plasma constituents such as glucose, salt and proteins were added at different concentrations to an aqueous solution containing 2.05 mmol/L formic acid and 1 mmol/L creatinine (Table 2). Blood plasma specimens 12 ACS Paragon Plus Environment

Page 13 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


generally contain considerable amounts of salt (0.9 weight%) leading to an increase in conductivity of the sample. The performance of cryo-NMR probes such as the one used in this investigation is sensitive to the conductivity of the sample. Inductive losses result from the dissipation of power due to the induction of currents in the sample, whereas dielectric losses result from the passage through the sample of the electrical lines of force arising from the distributed capacitance of the rf coil.28 Therefore, for plasma samples a considerable loss in sensitivity is expected in comparison to no or low salt samples such as cell extracts. The data shown in Table 2 clearly supports this expectation: with increasing salt concentration the signals of both, creatinine and formic acid, are subject to considerable signal attenuation. However, salt has a similar effect on both molecules as evidenced by the almost constant creatinine to formic acid integral ratios between 1.02 and 1.10. Therefore, only negligible effects on relative quantification will be observed but the signal-tonoise ratio is considerably decreased. Data also showed that the addition of varying concentrations of glucose had only a minor influence on signal intensities. In contrast, increasing protein amounts showed a strong differential effect on the peak-integrals of creatinine and formic acid: the signal of the CH2 group of creatinine at 4.06 ppm decreased from 52.6 * 106 a.u. in pure water to 30.2 * 106 a.u. after the addition of 80g/L BSA, whereas the formic acid signal stayed almost constant with 41.3 * 106 a.u. and 41.0 * 106 a.u. in pure water and 80g/L BSA, respectively. Correspondingly, line-shape analysis yielded for the formic acid signal a nearly constant width at half height of 0.95 Hz and 1.04 Hz in pure water and 80g/L BSA, respectively. In contrast, the signal-width of the CH2 group of creatinine at 4.06 ppm increased from 1.46 Hz in pure water to 2.41 Hz in the presence of 80g/L BSA, clearly indicating effects of the BSA matrix on the motional properties of creatinine. 13 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Variation of matrix effects Next, we investigated the impact of differences in matrix composition on metabolite recovery. For this, 1 mmol/L creatinine was added to a total of 36 different GCKD plasma specimens. For a realistic representation of the GCKD cohort, plasma specimens from patients with and without proteinuria (urinary levels of proteins above 3 g/L, respectively below 300 mg/L), with and without type 2 diabetes, and with a very low, a low and a relatively well preserved eGFR (300 mg/L), tryptophan could not be detected by NMR (n=3). In specimens of patients without proteinuria ( 300 mg/L) may significantly affect the NMR based detection of common urinary metabolites such as tryptophan, 3indoxyl sulfate and p-cresyl sulfate, which are known to bind strongly to protein.

Discussion The above results show, that even trace amounts of proteins in physiological fluids exert a profound effect on metabolite quantification. If not accounted for, they may lead to significant under-quantification of metabolites as demonstrated here exemplarily for creatinine in plasma and tryptophan in urine. This has also far ranging implications for the common practice of exploiting NMR metabolite fingerprints for the discovery of diagnostic and predictive biomarkers as differences in metabolite content between patients and controls or between different disease entities may rather indicate differences in content of protein than metabolites. Alternatively, differences in metabolite content may be missed because of differences in protein content. Regarding NMR analysis of metabolites in plasma, a number of strategies have been proposed to deal with the high abundance of proteins, which cause broad overlapping signals, obscure resonances of low-molecular-weight metabolites, and 22 ACS Paragon Plus Environment

Page 22 of 52

Page 23 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


impede quantification of these compounds due to T2-relaxation processes. Ultrafiltration, in theory, will produce two separate solutions: the ultrafiltrate, which contains free low-molecular weight compounds, and the retentate, which consists of macromolecules

and

low-molecular

weight

compounds

bound

to

those

macromolecules.37,38 However, aside from the fact that the free fraction of metabolites that bind strongly to protein, such as indoxyl-sulfate and p-cresyl sulfate,39 may be too low in concentration to be detected by NMR, factors such as nonspecific binding of metabolites to the plastic components or the ultrafiltration membrane of the filtration device, fouling of the ultrafiltrate with membrane preservatives or membrane components, or unwanted reactions such as glutamine cyclization to pyroglutamic acid,32 may lead to erroneous quantitative results. Moreover, the high cost of ultrafiltration may become prohibitive in large-cohort studies comprising thousands of specimens. Protein precipitation with various organic solvents may lead to varying recoveries of low-molecular weight compounds based on their polarity as well as to the unwanted retention of lipids and lipoproteins that may interfere with the detection of certain metabolites.37 Precipitation with perchloric acid also retains lipoproteins and, in addition, results in an overall loss of sensitivity, a poor signal-to-noise ratio, and shifts in peak position.37,38 Loss of sensitivity, which is even more pronounced for modern cryo-probes, also limits the use of salts such as of NH4Cl as a means of releasing bound molecules from proteins prior to acquisition of Hahn-echo or CPMG spectra.12,17 Consequently, spectral editing using a CPMG spin-echo train still offers the best approach to determining metabolite concentrations in plasma provided that the metabolite can be detected at all and appropriate steps are taken to account for protein binding. Spiking known amounts of metabolites into a pooled sample, as demonstrated here, represents a very practical way of accounting for the bound fraction of metabolites provided that 23 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

they do show fast exchange between their free and protein-bound state and that the protein content does not vary too much among the specimens under investigation as it is the case for the analyzed plasma specimens of the GCKD cohort. For these specimens an average serum albumin content of 38.5 ± 4.2 g/L was determined. Spiking a compound such as TSP that shows moderate protein binding and fast exchange may present at least a suitable means of detecting specimens that differ significantly in protein content. For such samples correction factors may then be determined separately. Ideally, in analogy to stable isotope-labeled internal standards frequently used in mass spectrometry for obtaining accurate quantification, one would wish for internal standards readily discriminated in NMR analysis but that do not differ in their physicochemical properties from the compounds to be quantitated. However, to date no such approach has been described for NMR spectroscopy. The protein content of urine in contrast to plasma is low, typically less than 200 mg/L in healthy individuals. Hence, NMR spectra can be readily acquired without the need for deproteinization or spectral editing. But as shown here, even in urine containing only trace amounts of protein substantial protein binding of metabolites may occur. Moreover, contrary to plasma, protein content in urine may easily vary over two orders of magnitude due to both renal and non-renal causes. Consequently, determination of factors based on spike-in experiments into a pool of specimens will not work, for instance, in studies involving patients suffering from no (≤200 mg/L), mild (≤1 g/L), moderate (≤3 g/L), and severe proteinuria (>3g/L), respectively. Under such circumstances, the only remedy may be the use of correction factors individually calibrated to each condition or in extreme cases to each specimen.

Conclusion 24 ACS Paragon Plus Environment

Page 24 of 52

Page 25 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Accurate quantification of metabolites in protein containing specimens represents a continuing challenge. Neither complete precipitation of macromolecules nor complete recovery of all metabolites contained in a biospecimen is feasible. The use of CMPG spin-echo trains obviates the need for deproteinization, but concentrations determined do not necessarily reflect the unbound fraction of metabolites. Determination of correction factors using pooled samples representative of the study population under investigation will yield more accurate data as long as individual samples do not vary too much in protein content. More generally, metabolites identified as diagnostic or predictive biomarkers in NMR metabolite fingerprinting need to be validated by independent methods such as tandem mass spectrometry in conjunction with stable isotope-labeled internal standards to ensure that differences in metabolite content observed between healthy and diseased or between different disease entities truly reflect differences in metabolite rather than protein content.

Acknowledgement This work was supported by the German Federal Ministry of Education and Research (BMBF grant no. 01 ER 0821). The authors thank Prof. Dr. Werner Kremer for critical reading. We thank all the GCKD study participants for their time and important contributions, all participating nephrologists' practices and outpatient clinics for their continued support, as well as the GCKD study personnel and investigators for their enormous commitment.

GCKD investigators are: University of Erlangen-Nürnberg, Germany: Kai-Uwe Eckardt, Stephanie Titze, HansUlrich Prokosch, Barbara Bärthlein, André Reis, Arif B. Ekici, Olaf Gefeller, Karl F. Hilgers, Silvia Hübner, Susanne Avendaño, Dinah Becker-Grosspitsch, Nina Hauck, 25 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Susanne A. Seuchter, Birgit Hausknecht, Marion Rittmeier, Anke Weigel, Andreas Beck, Thomas Ganslandt, Sabine Knispel, Thomas Dressel, Martina Malzer Technical University of Aachen, Germany: Jürgen Floege, Frank Eitner, Georg Schlieper, Katharina Findeisen, Elfriede Arweiler, Sabine Ernst, Mario Unger, Stefan Lipski Charité, Humboldt-University of Berlin, Germany: Elke Schaeffner, Seema BaidAgrawal, Kerstin Petzold, Ralf Schindler University of Freiburg, Germany: Anna Köttgen, Ulla Schultheiss, Simone Meder, Erna Mitsch, Ursula Reinhard, Gerd Walz Hannover Medical School, Germany: Hermann Haller, Johan Lorenzen, Jan T. Kielstein, Petra Otto University of Heidelberg, Germany: Claudia Sommerer, Claudia Föllinger, Martin Zeier University of Jena, Germany: Gunter Wolf, Martin Busch, Katharina Paul, Lisett Dittrich Ludwig-Maximilians University of München, Germany: Thomas Sitter, Robert Hilge, Claudia Blank University of Würzburg, Germany: Christoph Wanner, Vera Krane, Daniel Schmiedeke, Sebastian Toncar, Daniela Cavitt, Karina Schönowsky, Antje BörnerKlein Medical University of Innsbruck, Austria: Florian Kronenberg, Julia Raschenberger, Barbara Kollerits, Lukas Forer, Sebastian Schönherr, Hansi Weißensteiner University of Regensburg, Germany: Peter J. Oefner, Wolfram Gronwald, Helena Zacharias Department of Medical Biometry, Informatics and Epidemiology (IMBIE), University of Bonn: Matthias Schmid 26 ACS Paragon Plus Environment

Page 26 of 52

Page 27 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Competing interests The authors declare no competing financial interests.

Supporting information available Supplemental Figure S1. Bland-Altman plot of the individual differences in creatine concentration between corrected NMR and LC-MS/MS data versus their mean. Supplemental Figure S2. Bland-Altman plot of the individual differences in isoleucine concentration between corrected NMR and LC-MS/MS data versus their mean. Supplemental Figure S3. Bland-Altman plot of the individual differences in valine concentration between corrected NMR and LC-MS/MS data versus their mean. Supplemental Figure S4. Bland-Altman plot of the individual differences in threonine concentration between corrected NMR and LC-MS/MS data versus their mean. Supplemental Table S1. Comparison between MS and NMR data for selected amino acids.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

References

(1)

Holmes, E.; Foxall, P. J. D.; Spraul, M.; Farrant, R. D.; Nicholson, J. K.; Lindon, J. C. 750 MHz 1H NMR Spectroscopy Characterisation of the Complex Metabolic Pattern of Urine from Patients with Inborn Errors of Metabolism: 2-hydroxyglutaric Aciduria and Maple Syrup Urine Disease. J. Pharm. Biomed. Anal. 1997, 15, 1647–1659.

(2)

Nicholson, J. K.; Foxall, P. J. D.; Spraul, M.; Farrant, R. D.; Lindon, J. C. 750 MHz 1H und 1H-13C NMR Spectroscopy of Human Blood Plasma. Anal. Chem. 1995, 67, 793–811.

(3)

Dumas, M. E.; Maibaum, E. C.; Teague, C.; Ueshima, H.; Zhou, B.; Lindon, J. C.; Nicholson, J. K.; Stamler, J.; Elliott, P.; Chan, Q. et al. Assessment of Analytical Reproducibility of 1H NMR Spectroscopy Based Metabonomics for Large-Scale Epidemiological Research: the INTERMAP Study. Anal. Chem. 2006, 78, 2199–2208.

(4)

Gronwald, W.; Klein, M. S.; Kaspar, H.; Fagerer, S.; Nürnberger, N.; Dettmer, K.; Bertsch, T.; Oefner, P. J. Urinary Metabolite Quantification Employing 2D NMR Spectroscopy. Anal. Chem. 2008, 80, 9288–9297.

(5)

Weljie, A. M.; Newton, J.; Mercier, P.; Carlson, E.; Slupsky, C. M. Targeted Profiling: Quantitative Analysis of 1H NMR Metabolomics Data. Anal. Chem. 2006, 78, 4430–4442.


Page 28 of 52

Page 29 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(6)

Astle, W.; Iorio, M. de; Richardson, S.; Stephens, D.; Ebbels, T. A Bayesian Model of NMR Spectra for the Deconvolution and Quantification of Metabolites in Complex Biological Mixtures. J. Am. Stat. Soc. 2012, 107, 1259–1271.

(7)

Tang, H.; Wang, Y.; Nicholson, J. K.; Lindon, J. C. Use of relaxation-edited one-dimensional and two dimensional nuclear magnetic resonance spectroscopy to improve detection of small metabolites in blood plasma. Anal. Biochem. 2004, 325, 260–272.

(8)

Klein, M. S.; Almstetter, M.; Schlamberger, G.; Nürnberger, N.; Dettmer, K.; Oefner, P. J.; Meyer, H. H. D.; Wiedemann, S.; Gronwald, W. Nuclear Magnetic and Mass Spectrometry-based Milk Metabolomics in Dairy Cows During Early and Late Lactation. J. Dairy Sci. 2010, 93, 1539–1550.

(9)

Tiziani, S.; Emwas, A.-H.; Lodi, A.; Ludwig, C.; Bunce, C. M.; Viant, M. R.; Günther, U. L. Optimized metabolite extraction from blood serum for 1H nuclear magnetic resonance spectroscopy. Anal. Biochem. 2008, 377, 16–23.

(10) Sedgwick, G. W.; Fenton, T. W.; Thompson J.R. Effect of protein precipitating agents on the recovery of free amino acids. Can. J. Anim. Sci. 1991, 71, 953– 957.

(11) Liu, M.; Nicholson, J. K.; Lindon, J. C. High-resolution diffusion and relaxation edited one- and two-dimensional 1H NMR spectroscopy of biological fluids. Anal. Chem. 1996, 68, 3370–3376.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(12) Kriat, M.; Confort-Gouny, S.; Vion-Dury, J.; Sciaky, M.; Viout, P.; Cozzone, P. J. Quantitation of metabolites in human blood serum by proton magnetic resonance spectroscopy. A comparative study of the use of formate and TSP as concentration standards. NMR Biomed. 1992, 5, 179–184.

(13) Tsiafoulis, C. G.; Exarchou, V.; Tziova, P. P.; Bairaktari, E.; Gerothanassis, I. P.; Troganis, A. N. A new method for the determination of free L-carnitine in serum samples based on high field single quantum coherence filtering 1HNMR spectroscopy. Anal. Bioanal. Chem. 2011, 399, 2285–2294.

(14) Van, Q. N.; Chmurny, G. N.; Veenstra, T. D. The depletion of protein signals in metabonomics analysis with the WET-CPMG pulse sequence. Biochem. Bioph. Res. Co. 2003, 301, 952–959.

(15) Fagugli, R. M.; Smet, R. de; Buoncristiani, U.; Lameire, N.; Vanholder, R. Behavior of non-protein-bound and protein-bound uremic solutes during daily hemodialysis. Am. J. Kidney Dis. 2002, 40, 339–347.

(16) Vanholder, R.; Smet, R. de; Glorieux, G.; Argilés, A.; Baurmeister, U.; Brunet, P.; Clark, W.; Cohen, G.; De Deyn, Peter Paul; Deppisch, R. et al. Review on uremic toxins: classification, concentration, and interindividual variability. Kidney Int. 2003, 63, 1934–1943.

(17) Bharti, S.; Sinha, N.; Joshi, B.; Mandal, S.; Roy, R.; Khetrapal, C. Improved quantification from 1H-NMR spectra using reduced repetion times. Metabolomics 2008, 4, 367–376. 30 ACS Paragon Plus Environment

Page 30 of 52

Page 31 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(18) Titze, S.; Schmid, M.; Köttgen, A.; Busch, M.; Floege, J.; Wanner, C.; Kronenberg, F.; Eckardt, K.-U. Disease burden and risk profile in referred patients with moderate chronic kidney disease: composition of the German Chronic Kidney Disease (GCKD) cohort. Nephrol. Dial. Transpl. 2015, 30, 441–451.

(19) Mehr, K.; John, B.; Russell, D.; Avizonis, D. Electronic referencing techniques for quantitative NMR: pitfalls and how to avoid them using amplitude-corrected referencing through signal injection. Anal. Chem. 2008, 80, 8320–8323.

(20) Zacharias, H. U.; Hochrein, J.; Klein, M. S.; Samol, C.; Oefner, P. J.; Gronwald, W. Current Experimental, Bioinformatic and Statistical Methods used in NMR Based Metabolomics. Curr. Metabol. 2013, 1, 253–268.

(21) Meiboom, S.; Gill, D. Modified Spin Echo Method for Measuring Nuclear Relaxation Times. Rev. Sci. Instr. 1958, 29, 688–691.

(22) Beckonert, O.; Keun, H. C.; Ebbels, T. M. D.; Bundy, J.; Holmes, E.; Lindon, J. C.; Nicholson, J. K. Metabolic Profiling, Metabolomic and Metabonomic Procedures for NMR Spectroscopy of Urine, Plasma, Serum and Tissue Extracts. Nat. Protocols 2007, 2, 2692–2702.

(23) van der Goot, Annemieke T; Zhu, W.; Vázquez-Manrique, R. P.; Seinstra, R. I.; Dettmer, K.; Michels, H.; Farina, F.; Krijnen, J.; Melki, R.; Buijsman, R. C. et al. Delaying aging and the aging-associated decline in protein homeostasis by 31 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

inhibition of tryptophan degradation. Proc. Natl. Acad. Sci. U.S.A. 2012, 109, 14912–14917.

(24) Smyth, G. K. Linear Models and Empirical Bayes Methods for Assessing Differential Expression in Microarray Experiments. Statist. Appl. Gen. Mol. Biol. 2004, 3, Art. 1.

(25) Benjamini, Y.; Hochberg, Y. Controlling the False Discovery Rate: a Practical and Powerful Approach to Multiple Testing. J. Roy. Stat. Soc. B 1995, 57, 289–300.

(26) Varshney, A.; Rehan, M.; Subbarao, N.; Rabbani, G.; Khan, R. H. Elimination of endogenous toxin, creatinine from blood plasma depends on albumin conformation: site specific uremic toxicity & impaired drug binding. PLOS ONE 2011, 6, e17230.

(27) Geigy Scientific Tables; Lentner, C., Ed. 8th Rev edition; West Cadwell, N.J. : Medical education Div., Ciba-Geigy Corp: Basel, Switzerland, 1992.

(28) Flynn, P.; Mattiello, D.; Hill, H.; Wand, A. J. Optimal Use of Cryogenic Probe Technology in NMR Studies of Proteins. J. Am. Chem. Soc. 2000, 122, 4823– 4824.

(29) Kaysen, G. A. Plasma composition in the nephrotic syndrome. Am. J. Nephrol. 1993, 13, 347–359.


Page 32 of 52

Page 33 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(30) Joven, J.; Cliville, X.; Camps, J.; Espinel, E.; Simo, J.; Vilella, E.; Oliver, A. Plasma protein abnormalities in nephrotic syndrome: effect on plasma colloid osmotic pressure and viscosity. Clin. Chem. 1997, 43, 1223–1231.

(31) Anderson, N. L.; Anderson, N. G. The human plasma proteome: history, character, and diagnostic prospects. Mol. Cell. Prot. 2002, 1, 845–867.

(32) Nagana Gowda, G. A.; Gowda, Y. N.; Raftery, D. Massive glutamine cyclization to pyroglutamic acid in human serum discovered using NMR spectroscopy. Anal. Chem. 2015, 87, 3800–3805.

(33) Martineau, E.; Tea, I.; Akoka, S.; Giraudeau, P. Absolute quantification of metabolites in breast cancer cell extracts by quantitative 2D 1H INADEQUATE NMR. NMR Biomed. 2012, 25, 985–992.

(34) Fritzsche, I.; Bührdel, P.; Melcher, R.; Böhme, H. J. Stability of ketone bodies in serum in dependence on storage time and storage temperature. Clin. Lab. 2001, 47, 399–403.

(35) Hochrein, J.; Zacharias, H. U.; Taruttis, F.; Samol, C.; Engelmann, J. C.; Spang, R.; Oefner, P. J.; Gronwald, W. Data Normalization of 1H NMR Metabolite Fingerprinting Data Sets in the Presence of Unbalanced Metabolite Regulation. J. Proteome Res. 2015, 14, 3217–3228.

(36) Fuller, R. W.; Roush, B. W. Binding of tryptophan to plasma proteins in several species. Comp. Biochem. Phys. B 1973, 46, 273–276. 33 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(37) Daykin, C. A.; Foxall, P. J. D.; Connor, S. C.; Lindon, J. C.; Nicholson, J. K. The Comparison of Plasma Deproteinization Methods for the Detection of Low-Molecular-Weight-Metabolites by 1H Nuclear Magnetic Resonance Spectroscopy. Anal. Biochem. 2002, 304, 220–230.

(38) Mutsaers, H. A. M.; Engelke, U. F. H.; Wilmer, M. J. G.; Wetzels, J. F. M.; Wevers, R. A.; van den Heuvel, Lambertus P; Hoenderop, J. G.; Masereeuw, R. Optimized metabolomic approach to identify uremic solutes in plasma of stage 3-4 chronic kidney disease patients. PLOS ONE 2013, 8, e71199.

(39)

Bohringer, F.; Jankowski, V.; Gajjala, P. R.; Zidek, W.; Jankowski, J. Release of uremic retention solutes from protein binding by hypertonic predilution hemodiafiltration. ASAIO J. 2015, 61, 55–60.


Page 34 of 52

Page 35 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Tables Table 1 Recovery of creatinine in different matricesa Water

Filtered

Unfiltered

Unfiltered

plasmab,c

plasmab,c

GCKD plasmab,d

100 µmol/L

92.0 µmol/L

98.2 µmol/L

68.5 µmol/L

46.1 µmol/L

1000 µmol/L

954.0 µmol/L

925.5 µmol/L

709.3 µmol/L

461.4 µmol/L

a

Concentrations were determined using Chenomx.

b

1

For the spike-in experiments performed employing a 1D H CPMG pulse sequence in filtered plasma,

unfiltered plasma and GCKD plasma the reported values were corrected by the corresponding independently determined blank values. c

EDTA plasma of a volunteer not suffering from kidney problems.

d

EDTA plasma from a patient of the GCKD study.

Table 2 Raw integrals of creatinine and formic acid in artificially created matricesa Salt (NaCl) c,d

100%b

c,d

d

c,d

d

Formate

ratio

Creatinine

Formate

ratio

Creatinine

Formate

ratio

34.0*106

33.3*106

1.02

30.2*106

41.0*106

0.74

50.6*106

44.8*106

1.13

6

1.08

36.2*10

6

0.85

51.0*10

6

1.13

6

1.10

41.3*10

6

0.94

49.7*10

6

1.16

6

1.27

b

39.3*10

25%

b

42.8*10

b

d

Glucose

Creatinine

50%

0%

Protein (BSA)

6

36.4*10

6

38.9*10

6

41.3*10

52.6*10

6

42.5*10

6

44.1*10


6

45.1*10

6

42.9*10


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 36 of 52

a

Raw peak integrals were determined with AMIX 3.9.13 (Bruker, BioSpin GmbH, Rheinstetten,

Germany). Note that integrals were not corrected for number of contributing protons and for differences in T1 relaxation. Integral values are given in arbitrary units as determined by AMIX. b

Glucose (100%: 20 mmol/L), NaCl (100%: 308 mmol/L), BSA (100%: 80 g/L). Note that either

glucose, salt or protein was added. No combinations of the above three substances were used. The last row contains data where no glucose, salt or protein was added. c

For creatinine the singlet signal of the CH2 group at 4.06 ppm was analyzed.

d

1 mmol/L creatinine and 2.05 mmol/L formic acid were added.

Table 3 Differential matrix effects. Spike-in of 1 mmol/L creatinine in 36 selected GCKD specimens mean recoverya

n

Proteinuria >3 g/L

56 ± 4%

17

No proteinuria

Quantification of Metabolites by NMR Spectroscopy in the Presence of

Recommend Documents