Abstract
Background
Preservation of deceased donor kidneys by hypothermic machine perfusion results in superior transplant outcomes as compared with static cold storage and provides the opportunity to measure biomarkers of cellular injury in perfusate samples. Identification of biomarkers predicting early graft dysfunction so far has met with limited success.
Methods
Two-dimensional difference gel electrophoresis and mass spectrometry were used to explore the proteome of perfusate samples from machine-perfused human donor kidneys (N = 18) and to discover novel biomarkers of ischaemic acute kidney injury.
Results
Thirty-two protein spots were successfully identified, representing 19 unique proteins that were derived from renal tissue and from residual plasma in the renal microcirculation. Two unidentified protein spots were significantly up-regulated, whereas one protein spot - identified as haptoglobin - was significantly down-regulated in the perfusate of ischaemically injured kidneys from donors after cardiac death as compared with kidneys from brain-dead donors who had not suffered warm ischaemic injury. Furthermore, two protein spots were up-regulated in kidneys that never functioned after transplantation, whereas one spot was up-regulated - identified as α1-antitrypsin - in kidneys with delayed graft function.
Conclusions
We provide the first description of the renal perfusate proteome and present preliminary evidence of differentially expressed biomarkers in human donor kidneys with different levels of acute ischaemic injury. Their diagnostic value for the selection of marginal kidneys in clinical transplantation should be determined in future studies.
Introduction
Preservation of deceased donor kidneys by hypothermic machine perfusion results in superior transplant outcomes as compared with static cold storage. 1 Furthermore, hypothermic machine preservation provides the opportunity to study the renovascular resistance to perfusate flow and to measure biomarkers of cellular injury in samples of the perfusion solution. High vascular resistance and increased levels of ‘viability markers’ such as lactate dehydrogenase (LDH), glutathione S-transferase (GST), fatty acid-binding proteins (FABPs) and redox-active iron are associated with early graft dysfunction and are used by transplant centres to decide whether or not to accept marginal donor kidneys for transplantation.2–7 However, it has recently been demonstrated that these parameters have insufficient predictive value to be clinically useful in identifying adverse transplant outcomes.8–10
Given the ongoing interest in the diagnostic potential of perfusate biomarkers, we explored the proteome of perfusate samples from human donor kidneys using two-dimensional difference gel electrophoresis (2D-DIGE) and tandem mass spectrometry (MS/MS) to discover potential biomarkers for ischaemic acute kidney injury. This unbiased approach may provide novel insight into the application of hypothermic machine perfusion as a diagnostic tool to identify marginal kidneys with a low risk of early graft failure that can safely be used for transplantation.
Material and methods
Study design
This observational cohort study included 18 donor kidneys preserved by hypothermic machine perfusion at our transplant centre from May 2007 until April 2009. All kidneys were recovered from different donors between 18 and 65 y of age and were transplanted within the Eurotransplant area. Among the 21 donor kidneys fulfilling these criteria, six kidneys from donors after brain death (DBD), six kidneys from donors after cardiac death (DCD) with controlled withdrawal of supportive treatment in the intensive care unit and six kidneys from uncontrolled DCD donors after failed cardiopulmonary resuscitation were selected. The perfusate proteome of these kidney types was compared with the aim to discover novel biomarkers of ischaemic acute kidney injury. This sample size is considered sufficient for statistical analysis of proteomics findings for biomarker discovery. 11
Graft function in the early postoperative period was defined as immediate function (no dialysis treatment after transplantation), delayed graft function (temporary dialysis treatment initiated in the first week after transplantation) or primary non-function (continuous dialysis treatment after transplantation). In a secondary analysis, the perfusate pro-teome of kidneys with different transplant outcomes was compared.
Hypothermic machine perfusion
DBD kidneys were recovered according to standard multi-organ procurement techniques. 12 DCD kidneys were recovered after rapid laparotomy and direct aortic cannulation for controlled donors and after in situ perfusion with double-balloon triple-lumen catheters (Coloplast A/S, Humlebaeck, Denmark) for uncontrolled donors. 13 After recovery, donor kidneys were transported to our institution for hypothermic perfusion on a LifePort machine with 1 L UW-MPS preservation solution (Organ Recovery Systems, Des Plaines, IL, USA). Maximum perfusion pressure was 30 mmHg for DBD and controlled DCD kidneys and 40 mmHg for uncontrolled DCD kidneys. The LifePort machine contains a 20 μm filter to clear large particles such as white blood cells from its circulation, leaving all proteins in the perfusate.
At one hour of perfusion, a sample of the preservation solution was taken from the machine and centrifuged at 900g at 48C for 10 min. Within 30 min from sampling, 100 μL and 1 mL aliquots of the perfusate were stored at -8°C until further analysis. Renovascular resistance was recorded and the perfusate concentrations of LDH and GST were measured. LDH was determined by standard colorimetric assay (Roche Diagnostics, Mannheim, Germany) and total GST activity was measured as described previously. 14 All perfusion characteristics were adjusted for kidney weight.
2D-DIGE
CyDye minimal labelling was performed according to the manufacturer's instructions (GE Healthcare, Uppsala, Sweden). Five microgram perfusate protein from three experimental groups was randomly labelled with either 400 pmol Cy3 or Cy5. A Cy2-labelled internal standard was generated by combining equal amounts of each perfusate sample in the proteomic analysis. Labelling was performed on ice in the dark for 30 min at pH 8.5. The labelling reaction was quenched by incubation with 10 mmol/L lysine (Sigma-Aldrich, St Louis, MO, USA) on ice in the dark for 10 min. Subsequently, three labelled samples, two analytical and an internal standard, were pooled and an equal volume of 2 × lysis buffer (7mol/L urea, 2 mol/L thiourea, 4% CHAPS, 0.04% bromophenol blue, 2% dithiothreitol (DTT), 2% (v/v) immobilised pH gradient (IPG) buffer pH 3-10) was added. Immobiline DryStrip gels (18 cm; pH 3-10 nonlinear; GE Healthcare) were passively pre-rehydrated for six hours with DeStreak Rehydration Solution (GE Healthcare) supplemented with 0.5% (v/v) IPG buffer pH 3-10. Consequently, perfusate samples were applied on these strips via cup-loading and isoelectric focusing was carried out at a constant temperature of 20°C as follows: 150 V for three hours, 300 V for three hours, gradient to 1000 V in six hours, gradient to 8000 V in one hour and 8000 V for two hours. Proteins on the focused strips were equilibrated for 15 min with gentle shaking at room temperature in equilibration solution (6 mol/L urea, 2% sodium dodecyl sulphate [SDS], 50 mmol/L Tris pH 8.8,0.02% bromophenol blue, 30% glycerol) supplemented with 1% DTT, followed by 2.5% iodoacetamide (IAA) in fresh equilibration solution for 15-min incubation with gentle shaking at room temperature. Second dimension SDS-polyacrylamide gel electrophoresis was carried out on polyacrylamide gels (12.5% T, 3% C) using the Ettan DALTsix (GE Healthcare). The IPG strips were loaded on these gels and fixed with 0.5% agarose containing a trace of bromophenol blue. Gels were run at 208°C, 0.5 W/gel for one hour and 15 W/gel until the bromophenol blue frontier reached the bottom of the gel. After completing the second dimension, gels were scanned on the Ettan Dalt Imager (GE Healthcare) using CyDye-specific excitation/ emission wavelengths. Gels were analysed using DeCyder 7.0 software (GE Healthcare) according to the manufacturer's recommendations. Briefly, differential in-gel analysis and biological variation analysis modules were used to calculate normalized spot volume/protein abundance.
In-gel digestion
Preparative gels, loaded with 16 μg unlabelled protein, were run using previous conditions. These gels were fixed for 30 min in 10% methanol and 7% acetic acid followed by an overnight incubation in Sypro Ruby (Bio-Rad, Hercules, CA, USA). Subsequently, gels were incubated for 45 min in 10% methanol and 7% acetic acid and then thoroughly rinsed with MilliQ water (Millipore, Bedford, MA, USA). Protein spots of interest were picked from the Sypro-stained gel using the automated Ettan Spot Picker (GE-Healthcare) into 96-well plates and in-gel digestion was carried out on the MassPREP digestion robot (Waters, Manchester, UK). The 2 mm diameter gel plugs were destained twice in 100 mmol/L ammonium bicarbonate (NH4HCO3), 50% (v/v) acetonitrile (ACN) for 10 min. Gel plugs were then dehydrated in 100% ACN for five minutes, supernatant was removed and gel plugs were allowed to air-dry for 10 min. Cysteines were reduced with 10 mmol/L DTT in 100 mmol/L NH4HCO3 for 30 min followed by alkylation with 55 mmol/L IAA in 100 mmol/L NH4HCO3 for 20 min. Spots were washed with 100 mmol/L NH4HCO3 and subsequently dehydrated with 100% ACN. Trypsin suspended in 50 mmol/L NH4HCO3 was added to the gel plugs (12 ng/μL) and allowed to digest at 40°C for five hours. Peptides were extracted twice with 1% (v/v) formic acid, 2% (v/v) ACN.
Protein identification by MS/MS
C18 ZipTip pipette tips (Millipore, Bedford, MA, USA) were first washed three times with 10 μL 0.1% trifluoroacetic acid (TFA) in 100% ACN and thereafter three times with 10 μL 0.1% TFA in 50% ACN. Following these wash steps, peptide digests were loaded through 10 up-down pipette draws with the C18 ZipTips which had been equilibrated with 0.1% TFA. Peptides were eluted through three up-down pipette draws in 0.1% TFA in 50% ACN. Peptide solutions were mixed at a 1:1 ratio with 5 mg/mL α-cyano-4-hydroxycinnamic acid matrix in 50% ACN, 0.1% TFA and spotted in duplicate on a stainless steel matrix-assisted laser desorption ionization (MALDI) sample plate (Applied Biosystems, Foster City, CA, USA). The spots were allowed to air-dry for homogeneous crystallization. MALDI-time of flight (TOF) mass spectra were acquired in positive ion reflectron mode on a 4800 MALDI-TOF/TOF analyzer (Applied Biosystems). Tandem MS fragmentation spectra were acquired for each sample averaging 200 laser shots per fragmentation spectrum on each of the eight most abundant ions present in each sample. The generated peak list was searched with the MASCOT search engine (http://www.matrixscience.com) against the Swiss-Prot protein database for protein identification with taxonomy set at Homo sapiens; trypsin and keratin peaks were excluded. One miss-cleavage was tolerated; carbamidomethylation was set as a fixed modification and oxidation of methionine as an optional modification. The protein charge was set at 1+. Mass tolerance for precursor ion was set to 150 ppm and MS/MS tolerance was set at 0.2 Da. No restrictions were made on the protein mass. Protein identification was considered reliable when the Mascot score was significant (protein and total ion score within 95% CI).
Ethics
Patient data were collected, stored and used in agreement with the code of conduct ‘Use of data in health research’ put forward by the Federation of Dutch Medical Scientific Societies (http://www.federa.org).
Statistics
Continuous variables were expressed as means with standard errors and categorical variables as percentages. Differences between groups were compared with Kruskal–Wallis tests for continuous variables and with Fisher exact tests for categorical variables. Spot abundances were compared with one-way analysis of variance after logarithmic transformation and standardization. Results with P < 0.05 were considered statistically significant.
Results
Transplant characteristics
Donor, preservation and recipient characteristics of the machine-perfused kidneys are presented in Table 1. As expected, uncontrolled DCD donors died significantly more often from cardiovascular causes than controlled DCD and DBD donors and warm ischaemia times of DCD kidneys were significantly longer than those of DBD kidneys. Furthermore, DBD kidneys were associated with significantly longer cold ischaemia times, since our policy was to preserve DBD kidneys by hypothermic machine perfusion only for recipients residing in the Dutch Antilles with expected prolonged cold ischaemia. Patients who received uncontrolled DCD kidneys were significantly older than recipients of controlled DCD and DBD kidneys, which may be explained by the tendency to transplant grafts of perceived lower quality into older recipients with a worse prognosis on dialysis therapy. Other baseline characteristics were similar between the study groups.
Transplant characteristics of machine-perfused donor kidneys*
DCD, donors after cardiac death; DBD, donors after brain death; CPR, cardiopulmonary resuscitation; LDH, lactate dehydrogenase; GST, glutathione S-transferase
Data are presented as mean (standard error) or as percentages
Ventilator switch-off time was defined as the time from withdrawal of supportive treatment until circulatory arrest
Warm ischaemia time was defined as the time from circulatory arrest until initiation of hypothermic organ perfusion and therefore does not include ventilator switch-off time and cardiopulmonary resuscitation time
Glomerular filtration rate was estimated using the abbreviated Modification of Diet in Renal Disease formula 30
During machine perfusion, renovascular resistance of the three kidney types was comparable, whereas LDH and GST concentrations in the perfusate were significantly higher for uncontrolled DCD kidneys than for controlled DCD and DBD kidneys. Immediate graft function was observed in 50% of controlled DCD and DBD kidneys but not in uncontrolled DCD kidneys, whereas 50% of controlled DCD and DBD kidneys and 67% of uncontrolled DCD kidneys suffered from delayed graft function. Primary non-function was observed in two grafts from uncontrolled DCD donors. The relatively high incidence of delayed graft function in DBD kidneys can be explained by the long cold ischaemic times needed to transfer their recipients across the globe. At three months after transplantation, estimated glomerular filtration rate of functioning grafts was not significantly different between the study groups.
Profiling of the renal perfusate
To characterize the renal perfusate proteome, MS/MS was used to identify proteins present on 2D-DIGE gels. No protein spots were detected in the preservation solution sampled before the start of renal perfusion (data not shown). In contrast, 68 protein spots were detected in the renal perfusate. Thirty-two of these spots (47%), representing 19 unique proteins, were successfully identified (Figure 1, Table 2). Of these proteins, only six (32%) cyto-plasmic proteins are known to be present in renal tissue whereas the remaining 13 (68%) proteins are probably derived from plasma that remained in the renal micro-circulation after the initial flush-out.
Two-dimensional-gel of perfusate samples from machine-perfused human donor kidneys. Thirty-two protein spots were identified by tandem mass spectrometry representing 19 unique proteins. Proteins were visualized by Sypro Ruby poststaining. VDBP, vitamin D-binding protein
Proteins identified in renal perfusate by two-dimensional difference gel electrophoresis and tandem mass spectrometry
Amino acid coverage of matched peptides
Theoretical molecular weight (kDa) and isoelectric point (pI) from the ExPASy database
Blood-borne proteins
Cytoplasmic proteins
Perfusate biomarker discovery with 2D-DIGE
To find novel biomarkers of ischaemic injury, the differential renal perfusate proteome between three types of donor kidneys was investigated using 2D-DIGE. To reduce the chances of random associations due to multiple comparisons, protein spots were only considered relevant when both the differential abundance had at least a 1.5-fold change (N = 11 for comparisons between donor types and N = 6 for comparisons between transplant outcomes) and the P value was less than 0.05.
Two protein spots were up-regulated in the perfusate of controlled and uncontrolled DCD kidneys as compared with DBD kidneys, whereas one protein spot was down-regulated in the perfusate of uncontrolled DCD kidneys as compared with DBD kidneys (Table 3). Confounding by recipient age or gender was excluded because these variables were not associated with donor type or spot abundance (data not shown). These three spots were automatically picked from a preparative gel and subjected to trypsin in-gel digestion for subsequent identification by MS/MS. The down-regulated spot in the perfusate of DCD kidneys was identified as haptoglobin, whereas the two up-regulated spots could not be identified.
Differentially expressed proteins in perfusion solution of donor kidneys*
DCD, donors after cardiac death; DBD, donors after brain death; ANOVA, analysis of variance
Data are presented as mean standardized spot abundance (standard error)
ANOVA P values are calculated using log standardized abundances
When the perfusate proteome was analysed according to transplant outcome, two protein spots were up-regulated in kidneys with primary non-function, whereas one spot was up-regulated in kidneys with delayed graft function (Table 3). The up-regulated spot in kidneys with delayed graft function was identified as α1-antitrypsin. Furthermore, the perfusates of the two donor kidneys with primary non-function contained greater quantities of spots 2 and 3, which could not be identified, than in functioning kidneys.
Discussion
Hypothermic machine preservation of donor kidneys provides the opportunity to study the renovascular resistance to perfusate flow and to measure biomarkers of cellular injury in samples of the perfusion solution. In the current study, we provide the first description of the renal perfusate proteome and present preliminary evidence of six novel bio-markers of ischaemic acute renal injury in human donor kidneys. Using 2D-DIGE and MS/MS, 19 unique proteins were identified that were derived both from renal tissue and from residual plasma in the renal microcirculation. Two unidentified protein spots were up-regulated, whereas a third protein spot representing haptoglobin was down-regulated in the perfusate of ischaemically injured DCD kidneys as compared with DBD kidneys that had not suffered warm ischaemic injury. Furthermore, two protein spots were up-regulated in kidneys that never functioned after transplantation, whereas one spot was up-regulated in kidneys with delayed graft function. Alpha-1 antitrypsin, up-regulated in kidneys with delayed graft function, has antiapoptotic and anti-inflammatory effects, which can contribute to the delayed type of protection. 15 These biomarkers may provide valuable information for the assessment of ischaemic injury in human donor kidneys before transplantation.
Until now, hypothesis-driven selection of parameters for evaluation of ischaemic injury to donor kidneys has failed to produce clinically useful predictors of graft viability. Characteristics of hypothermic machine perfusion such as renovascular resistance and perfusate concentrations of enzymes released from damaged tubular epithelial cells (LDH, GST and FABP) did not predict early graft failure with sufficient accuracy to justify a decision to discard scarce donor kidneys.8–10 Since the hypothesis-driven identification of biomarkers for acute ischaemic injury so far has met with limited success, we decided to take an alternative approach by performing unbiased data-driven analysis of the perfusate proteome. Rather than building on previous experimental findings, this approach may generate novel hypotheses about the pathophysiology and diagnosis of ischaemic acute kidney injury. 16 In renal transplant recipients, analyses of the urinary proteome have previously been successful in identifying biomarkers for acute rejection, chronic allograft nephropathy and BK virus nephro-pathy.17–23 Furthermore, acute kidney injury after cardio-pulmonary bypass or administration of iodinated contrast agents was also reflected by changes in the urinary pro-teome.24–26
A major strength of the current proteome analysis is the use of preservation solution from isolated perfused human donor kidneys. Since the preservation solution did not contain proteins before perfusion, all protein spots in the gels must have been derived from the donor kidneys, increasing the chances of discovering tissue-specific biomarkers. Interestingly, the majority of identified proteins in the renal perfusate were not synthesized by the kidney but were blood-borne proteins which remained in the renal microcirculation in spite of manual flush-out of the vasculature before machine perfusion. Proteome analysis has been criticized because of its low reproducibility between laboratories. Indeed, some techniques are sensitive to seemingly minor differences in the analytical protocol. 27 We therefore collected, processed and stored the perfusate samples in a highly standardized fashion. Moreover, the reproducibility of gel electrophoresis was improved by normalizing the spot intensities to an internal standard that was run on each gel. Furthermore, separating proteins by 2D-DIGE generally does not permit resolution of proteins with high (>150 kDa) or low (<10 kDa) molecular mass, nor of very basic or hydrophobic proteins, limiting the proteomic coverage of biological samples. However, in contrast to other proteomic techniques, 2D-DIGE can be used to detect post-translational protein modifications.
After an unbiased assessment of the perfusate proteome of machine-perfused donor kidneys, haptoglobin was identified as a novel biomarker of ischaemic acute kidney injury. Lower concentrations of haptoglobin in the perfusion solution were associated with more extensive ischaemic injury to the donor kidneys. Haptoglobin is an acute-phase plasma protein that scavenges free haemoglobin with high affinity, thereby preventing oxidative stress mediated by the redox-active iron in its haeme group. 28 We have previously reported that higher concentrations of redox-active iron in the renal preservation solution are associated with increased risk of primary non-function in clinical kidney transplantation. 5 Since lower levels of haptoglobin will lead to higher levels of free haemoglobin – containing redox-active iron in its haeme group – our current findings have biological plausibility and fit into an established theoretical framework. Although we were able to identify almost 50% of the spots picked from the Sypro Ruby-stained gel, the identity of the other differentially expressed protein spots unfortunately could not be identified by MALDI-TOF/TOF, or by additional attempts using electrospray ionization tandem mass spectrometry.
Validation of proteomics-based findings is an important next step in biomarker discovery research. Preliminary attempts to validate haptoglobin with enzyme-linked immu-nosorbent assay were inconclusive. However, proteomic results can sometimes not be confirmed with antibody-based techniques due to epitope reactivity of antibodies and post-translational modifications of proteins. Mass spectrometry-based assays provide an excellent alternative approach for quantitative biomarker validation and circumvent some limitations of antibody-based methods. Selected reaction monitoring can be used to specifically select and quantify protein biomarkers in serum and tissue. 29
In conclusion, after unbiased exploration of the protein content of perfusate samples from machine-perfused kidneys, we provided the first description of the renal perfusate proteome and presented preliminary evidence of a series of novel biomarkers of ischaemic acute renal injury in human donor kidneys. Their diagnostic value for the selection of ischaemically injured kidneys in clinical transplantation should be determined in future studies.