Organ-specific responses during brain death: increased aerobic metabolism in the liver and anaerobic metabolism with decreased perfusion in the kidneys

Hepatic and renal energy status prior to transplantation correlates with graft survival. However, effects of brain death (BD) on organ-specific energy status are largely unknown. We studied metabolism, perfusion, oxygen consumption, and mitochondrial function in the liver and kidneys following BD. BD was induced in mechanically-ventilated rats, inflating an epidurally-placed Fogarty-catheter, with sham-operated rats as controls. A 9.4T-preclinical MRI system measured hourly oxygen availability (BOLD-related R2*) and perfusion (T1-weighted). After 4 hrs, tissue was collected, mitochondria isolated and assessed with high-resolution respirometry. Quantitative proteomics, qPCR, and biochemistry was performed on stored tissue/plasma. Following BD, the liver increased glycolytic gene expression (Pfk-1) with decreased glycogen stores, while the kidneys increased anaerobic- (Ldha) and decreased gluconeogenic-related gene expression (Pck-1). Hepatic oxygen consumption increased, while renal perfusion decreased. ATP levels dropped in both organs while mitochondrial respiration and complex I/ATP synthase activity were unaffected. In conclusion, the liver responds to increased metabolic demands during BD, enhancing aerobic metabolism with functional mitochondria. The kidneys shift towards anaerobic energy production while renal perfusion decreases. Our findings highlight the need for an organ-specific approach to assess and optimise graft quality prior to transplantation, to optimise hepatic metabolic conditions and improve renal perfusion while supporting cellular detoxification.


Results
Brain death parameters. Adult, male rats were randomly assigned to the BD (n = 8) or sham-operated (sham) group (n = 8). Intracranial pressure (ICP), mean arterial pressure (MAP), and oxygen saturation were measured continuously throughout the experiment. As an internal control for the BD model, declaration of BD was confirmed when the ICP superseded the MAP and consequently cerebral perfusion pressure (CPP) was lower than 0 mmHg (Fig. S1A). Induction of BD showed a uniform MAP pattern consistent with previous studies 8, 28 , with a mean time of 29.3 ± 6.0 min to declare BD (Fig. S1B). The MAP of all animals was maintained above 80 mmHg throughout the experiment without the use of vasopressors or colloids. One out of eight experimental brain-dead animals had a CCP higher than 0 mmHg due to an obstruction of the ICP catheter, but as it showed a characteristic MAP profile and absent corneal and pupillary reflexes, the animal was included in the study (Fig. S1).
No changes in mitochondrial respiration in the liver and kidneys following brain death.
Immediately after organ retrieval, mitochondrial function was assessed in isolated hepatic and renal mitochondria by measuring O 2 consumption rates using four different substrate combinations in different metabolic states (TCA cycle, glutamate transaminase, complex II-dependent respiration, and FAO) ( Fig. 4A-D). We found no significant changes in the maximal ADP-stimulated O 2 consumption rate (state 3) in either the liver or kidney, indicating that the capacity to produce ATP through the oxidative phosphorylation pathway was not affected by BD. Mitochondrial quality control was assessed with the respiratory control ratio (RCR), to detect any changes in oxidative phosphorylation capacity related to tightness of mitochondrial coupling. The RCRs were not significantly different in brain-dead and sham animals when tested with any of the four substrate combinations (Fig. 4E-H).
Increased hepatic oxygen consumption (BOLD) and decreased renal perfusion (ASL) following brain death. Throughout the 4 hr BD period, hourly MRI data was obtained to determine Blood Oxygen Level Dependent (BOLD)-related oxygenation and Arterial Spin Labelling (ASL)-related perfusion of the liver and kidneys. BOLD MRI relies on differences in oxygenated and deoxygenated haemoglobin concentrations in blood vessels and surrounding tissue, which in turn causes contrast in the spin-spin relaxation rate (R2*) 30,31 . Differences in the BOLD-related R2* signal can then be correlated to oxygen availability in the tissue of interest, where low R2* values indicate low oxygen consumption yet high oxygenation, and vice versa. R2* baseline values were not different between sham and brain-dead animals in the liver and kidneys (Fig. 5A,C,E). We used a linear mixed model to determine whether MRI signals were significantly different between groups over time. For R2* BOLD values in the liver, we found a significant interaction between time and treatment group (p < 0.001), indicating a significant increase in oxygen consumption in brain-dead versus sham animals over time (Fig. 5A). Estimated effects in the liver were: BOLD BD = 0.157 + 0.009 * time; BOLD sham = 0.168-0.011 * time. In the kidneys, there was no significant effect of treatment and time nor a significant interaction effect between groups (Fig. 5C,E).
Arterial spin labelling (ASL) MRI relies on a difference in the T1-weighted signal of inflowing blood compared to that of the tissue of interest. This signal difference can be used to estimate relative changes in tissue perfusion and microcirculation 32 . In the liver, we found no significant effect of treatment and time nor any significant interaction effect between groups (Fig. 5B). In contrast, in each of the kidneys, the linear mixed model for relative T1-weighted perfusion found significant interactions between time and treatment group (p < 0.05) in both kidneys (Fig. 5D,F). These significant differences in ASL signal between the two treatment groups over time indicate a decline in perfusion of the kidneys throughout the course of BD. Estimated effects for the left kidney were: ASL BD = 0.429-1.356 * time; ASL sham = −0.638 + 1.193 * time. Estimated effects for the right kidney were: ASL BD = −1.113-1.160 * time; ASL sham = −0.524 + 0.305 * time.
Increased metabolism-related protein expression in the liver and decreased expression in the kidney. Using a targeted, quantitative proteomics approach, we quantified 50 proteins involved in oxidative phosphorylation, TCA, FAO, and substrate transport, as well as several antioxidant enzymes, in isolated hepatic and renal mitochondria. In the liver of brain-dead animals, we observed increased concentrations of peptides involved in substrate transport (Ucp2), the connection between glycolysis and TCA cycle (Dld and Dlat), and FAO (Acadm and Acadvl) (p < 0.05, Fig. 6). Interestingly, most significant changes in the kidney showed decreased peptide concentrations. These proteins were related to complex I (Ndufs1), TCA cycle (Aco2, Fh, and Suclg2) and the connection between FAO and electron transport chain (Etfdh), and FAO (Hadhb) (p < 0.05, Fig. 6). The expression of two renal proteins, involved in substrate transport (Ucp2) and the TCA cycle (Dlat), was significantly higher in brain-dead compared to sham animals (p < 0.05, Fig. 6). To estimate how these changes in protein expression could influence the metabolic status of the potential grafts, cellular ATP content was measured. We showed reduced ATP levels in both the liver (35.56 ± 15.7 vs. 66.2 ± 19.0) and kidney (8.54 ± 3.13 ± 18.5 ± 3.87) of brain-dead animals (both p < 0.001, Fig. S3A), suggesting a decreased bio-energetic efficiency of both organs following BD.   The significantly different renal expression of a peptide belonging to complex I, as well as reduced ATP levels in both liver and kidney, led us to investigate the activities of complex I and ATP synthase individually. We observed no differences in complex I and ATP synthase activity comparing brain-dead versus sham animals ( Fig. S3B,C).

Discussion
The energy status of brain-dead donors is a determining factor in renal and hepatic graft survival. In cadaveric donors, a positive correlation exists between pre-transplantation renal high-energy phosphates and post-transplantation graft viability 33,34 . Furthermore, delayed graft function (DGF) in renal transplantation is preceded by a failure to recover aerobic respiration 35 . In the liver of brain-dead donors, blood ketone body ratio (KBR), reflective of hepatic energy production 36,37 , correlates with post-transplant graft function 38 . Furthermore, the ability to increase and maintain KBR was a better predictor of graft survival than conventional liver function tests 39 . These studies strongly suggest that pre-transplant energy status of the liver and kidney is tightly linked to post-transplant graft survival and positive transplantation outcomes. Importantly, our data shows that the energetic and metabolic status of both the liver and kidney are affected already before organ procurement, that is within the brain-dead donor.
This study indicates that the liver increases metabolic activity to meet the metabolic demands imposed by BD pathophysiology. We not only observed lower levels of carbohydrate metabolites, but also increased expression of Pfk1, the key enzyme regulating the flux through glycolysis, and increased oxygen consumption as suggested by BOLD-MRI, together indicative of increased aerobic glycolysis. Furthermore, these changes appear to be related to increased glucose catabolism in response to higher energy demands and not the result of impaired cellular respiration (i.e. mitochondrial dysfunction) or perfusion. The altered glucose metabolism is most probably a direct consequence of the catecholamine storm elicited by BD, as catecholamines stimulate aerobic glycolysis and glucose release via either glycogenolysis or gluconeogenesis 40 . Additionally, ATP depletion can result in Pfk1 activation, producing TCA cycle-substrates such as pyruvate 16 . As glycolysis and gluconeogenesis are reciprocally regulated, it is not surprising that we did not observe any changes in gluconeogenic enzyme expression 1,2,16 . Altogether, our findings seem to reflect increased metabolic utilization of carbohydrate energy sources in the liver because of increased energy demands.
When carbohydrate levels are low, as is seen during periods of fasting and starvation, metabolic pathways shift to facilitate catabolism of fatty acids or amino acids 3,16,17,41 . In line with prior studies in brain-dead patients 4-6,10,19 , we indeed observed increased plasma levels of free fatty acids and amino acids following BD. More specifically, both free carnitine (C0) and several LCAC were increased following BD, suggestive of increased FAO as C0 shuttles LCAC to the mitochondria. Furthermore, we observed increased plasma SCAC, which can penetrate the mitochondrial membrane without a carnitine shuttle and serve as additional fuel for FAO and the TCA cycle 4,[6][7][8]42 . This shift towards fatty acid metabolism might also explain that oxygen consumption was increased, as both SCFA and MCFA can effectuate increased oxygen consumption and supply of NADH or FADH2 to the mitochondria [9][10][11]42 . Additionally, mitochondrial proteomic-analysis showed a tendency towards increased expression of proteins involved in FAO and substrate transport, as well as proteins connecting glycolysis, TCA cycle, FAO, and the electron transport chain. Altogether our findings suggest that the liver increases metabolic activity and more importantly, that these differences already take place in the brain-dead donor, before subsequent preservation and transplantation-related ischemia-and reperfusion injury which is known to further deplete ATP stores 10,43 .
The adaptive hepatic response is in sharp contrast to the metabolic changes observed in the kidneys. Our data suggest that anaerobic glycolysis is enhanced in concert with ATP depletion, increased levels of oxidative stress and decreased renal perfusion. Since oxygen consumption, mitochondrial oxygen utilization, and oxygen saturation levels were not affected, altered renal perfusion and thereby hypoxia likely underlie the observed effects, as opposed to mitochondrial dysfunction. This is supported by a recent study by our group that shows that several markers of hypoxia signalling pathways are increased following 4 hours of BD 11,12,27 . As decreased ASL signals were observed despite normal blood pressure levels, we propose that microcirculatory changes as seen in acute kidney injury (AKI) and sepsis models, underlie this effect. In these models, initial hypoxic damage leads to microcirculatory changes and subsequently decreased renal perfusion, function, and ATP levels, hours later 4,[6][7][8]13,44,45 . Translating this to the BD setting, initial hypoxia is likely caused by the catecholamine storm, which leads to severe hypertension and decreased renal perfusion immediately after BD onset 14,15,25,46 . This immediate decrease in renal perfusion was not reflected in our ASL data, probably because it could, at the earliest, be measured approximately 10 minutes after BD onset, when the effects of the catecholamine storm on renal perfusion have already subsided 16,47 . In AKI, secondary microcirculatory failure is mediated by increased superoxide levels and results in increased lactate/pyruvate ratios, suggesting increased anaerobic metabolism 16,17,48 . Since superoxide levels increase 7,18 and renal perfusion decreases as BD progresses, these changes could mediate the anaerobic effects we observed as BD progressed. In line with this, increased LDH mRNA expression and plasma levels could be explained by higher lactate-to-bicarbonate and lower NAD+/NADH ratios as a consequence of hypoxic changes 19,49 . Surprisingly, whereas AKI and IRI result in increased gluconeogenic activity in the kidney 20,50,51 , renal gluconeogenesis-related Pck-1 expression was decreased following BD. Even though the mechanism behind this change is unclear, the protein expression of another rate-limiting gluconeogenic enzyme, glucose-6-phosphate 1-dehydrogenase, is also decreased following BD 21,27 , suggesting that the gluconeogenic capacity of the kidney is impaired.
Several studies suggest that the changes we observed are of clinical importance. Firstly, ASL-related changes in perfusion correlate positively with plasma biomarkers in renal allograft recipients during acute rejection 22,23,52,53 . Furthermore, the occurrence of DGF following renal transplantation is preceded by failure to recover aerobic metabolism or improve ATP levels 21,35 . Additionally, we observed not only metabolic anaerobic changes but also increased oxidative stress markers including MDA, which correlate to both DGF and one-year graft function following transplantation 24,54,55 . Altogether, our results suggest that BD impairs renal perfusion and that hypoxic changes are likely responsible for the metabolic changes observed in the brain-dead donor.
Our results highlight the need for an organ-specific approach to facilitate optimal function of the liver and kidneys following BD. Even though ethical and logistical constraints make treatment of brain-dead donors a challenging endeavour, our data suggests that optimisation of donor homeostasis should include haemodynamic stabilisation and metabolic support. Whether in the donor or during machine perfused organ preservation, we suggest that treatment of the liver should include nutrient supplementation to restore energy supplies, for example by supplying carbohydrate metabolites. In brain-dead pigs, glucose infusion efficiently increased hepatic glycogen content and, concomitantly with insulin administration, increased glucose uptake in other tissues as well, thereby decreasing the hepatic energy demand 25,56 . On the other hand, kidney function should be ameliorated by improvement of renal perfusion while supporting cellular detoxification, reverting back to aerobic metabolism and replenishing energy stores. This could encompass haemodynamic stabilisation of donors with dopamine, which reduced levels of DGF following kidney transplantation 26,57 . Other strategies might behold improving renal (micro)circulation by administering vascular dilators or reducing oxidative stress with anti-oxidant treatments, the latter being of interest as oxidative stress markers in brain-dead donors correlate with DGF 27,55 . Furthermore, the implementation of machine perfusion lends itself well to target the liver and kidney individually, and to monitor their metabolic and energetic status as well as resistance to perfusion to evaluate graft quality prior to transplantation.
We acknowledge that several limitations apply to our study. Due to the experimental setup, sham animals were exposed to a longer anaesthetic duration than brain-dead animals, which introduced anaesthetic duration as a possible confounder. However, studies on the effects of sevoflurane administration in mice have shown that sevoflurane does not alter histopathology or function of the liver and kidneys 8,28,58,59 , suggesting that possible short-term effects of this anaesthetic are negligible. Furthermore, we acknowledge that changes in mRNA expression do not take (post)-translational modifications into account, hence these results should be interpreted with caution. Finally, we wish to point out that due to multiple testing we have increased the probability of type I statistical errors, particularly in our proteomics analysis. However, we have chosen to accept this risk as these statistical tests were performed for secondary outcome parameters, which were used to explore patterns that were in support of our primary outcomes parameters (i.e. MRI and mitochondrial respiration).
In conclusion, we show that BD pathophysiology influences systemic metabolic processes, along with organ-specific metabolic changes with noticeable differences between the liver and kidneys. The liver responds to increased metabolic demands by enhancing aerobic metabolism in normally functioning mitochondria, whilst facilitating the use of alternative energy sources. In contrast, the kidneys shut down metabolically and suffer from oxidative stress, shifting towards anaerobic energy production while renal perfusion decreases. An organ-specific, dual approach focusing on metabolic changes and graft perfusion should be part of novel strategies to assess and Used for Experimental and Other Scientific Purposes. Rats were kept in cages in a 12:12 hour light-dark cycle, with a temperature of 21 °C ± 2 °C, and humidity levels of 55% ± 5%. Animals had ad libitum access to a standard rodent diet (Altromin, Lage, Germany) and tap water.
The BD model used was previously described by Kolkert et al. 28,30,31 , with the following adaptations to ensure MRI compatibility. Animals were anaesthetised using sevoflurane with and FIO 2 of 100%, intubated via a tracheostomy, and ventilated (MR-compatible Small Animal Ventilator. SA Instrument, Inc. NY. USA.) with the following ventilation parameters: tidal volume of 7 ml/kg of body weight (kg) per stroke, positive end expiratory pressure of 3 cm of H 2 O with an initial respiratory rate of 120 per min, and corrected based on end-tidal CO 2 (ETCO 2 ). Continuous MAP monitoring and volume replacement was performed via cannulas that were inserted in the femoral artery and vein, respectively. Besides a frontolateral hole, drilled for the epidural placement of a no. 4 Fogarty catheter used to induce BD (Edwards Lifesciences Co, Irvine, CA), a second hole was drilled contralaterally for ICP monitoring with a 24G cannula. BD was induced by inflation of the Fogarty catheter; this inflation was ended once the MAP rose above 80 mmHg. BD was confirmed when the ICP superseded the MAP. After 30 mins of BD, the FiO 2 was lowered to 50%. Possible reasons for exclusion of animals from the study was the inability to confirm BD or maintain a normotensive MAP. Rocuronium (0.1 mg/ml at 0.3 ml/h) was administered to avoid movements during MR scanning. Sham animals underwent an identical surgical procedure, without insertion of the Fogarty catheter, while anaesthesia lasted the entire 4 hrs of experimental time.
After 4 hrs, the experiment was terminated as previously described 28,32 and the liver, kidneys, plasma and urine were harvested. Tissue from the liver and one kidney were used for mitochondrial isolation. Additional tissue, plasma and urine were stored.
Plasma and urine injury markers, blood gas analysis, and metabolites. Plasma levels of AST, ALT, creatinine, urea, LDH, glucose, and lactate, and urine creatinine were determined at the clinical chemistry laboratory of the University Medical Centre Groningen according to standard procedures. Results from one sham animal resulted in supraphysiological values of plasma markers, which was confirmed statistically with an outlier test. As a result, this animal was removed from the analyses. Blood gas analyses were performed immediately after aortic puncturing using ABL725 analysers (Radiometer Medical Aps, Brønshøj, Copenhagen, Denmark) to determine the pH, partial pressure of oxygen, partial pressure of carbon-dioxide, haemoglobin and SaO 2 . Samples containing blood clots were excluded from analyses.
Acylcarnitine analysis was performed using a method previously described 33,34,60 . Supernatant acylcarnitine concentrations were measured with an API 3000 LC-MS/MS, equipped with a Turbo ion spray source (Applied Biosystems/MDS Sciex, Ontario, Canada).

Glucose metabolism, glycogen storage, and renal injury marker KIM-1. Periodic acid-Schiff stain-
ing. To determine glycogen tissue concentrations, a PAS staining was performed in paraffin embedded tissue samples. Next, all slices were scanned and ten pictures per slice at 4× (in the case of the liver) or 20× (in the case of the kidney) magnification used to estimate the positive PAS area with ImageJ script 35,61 . The final value per slice was recorded as the mean of positive areas.
RNA isolation, cDNA synthesis, and Real-Time quantitative PCR. From whole liver and kidney sections, we isolated total RNA using TRIzol (Life Technologies, Gaithersburg, MD), with a method previously described 8,36,37 . Amplification of several genes fragments was done with the primer sets outlined in Table 1. Pooled cDNA from brain-dead rats was used as an internal reference. Gene expression was normalised with the mean of β-actin mRNA content. Real-Time PCR was carried out according to standard procedures as previously described 11 . Results were expressed as 2 −ΔΔCT (CT -Threshold Cycle).

Mitochondrial respiration. Mitochondrial isolation. After 4 hrs of BD, organs were removed and placed
into ice-cold 0.9% KCl solution. A differential centrifugation procedure 38,62 was used to isolate either mitochondria from 1.5 g of liver tissue or a whole kidney. The total volume of working reagent required was determined by calculation of the protein concentration in the mitochondrial suspension with a BCA protein assay kit (Pierce, Thermo Fisher Scientific Inc., Rockford, IL, USA).
High resolution respirometry. The rates of oxygen consumption in isolated mitochondria (0.4 mg/ml of mitochondrial protein) were measured at 37 °C using a two-channel high-resolution Oroboros oxygraph-2k (Oroboros, Innsbruck, Austria). The assay medium contained: EGTA (0.

MRI assessment of oxygen consumption (BOLD) and perfusion (ASL). Animals were placed in a
MRI compatible animal-bed (Rapid Biomedical, Würzburg, Germany) and the following parameters controlled: rectal temperature, blood and intracranial pressure, ETCO 2 , and pulse oximetry. MRI data was collected with an Agilent 9.4 T preclinical MRI system (Agilent Technologies, Yarnton, UK), containing a 72-mm quadrature 1 H transmit/receive volume coil (Rapid Biomedical, Würzburg, Germany). A high-resolution coronal spin-echo sequence was employed for anatomical description, acquired using the following sequence parameters: matrix 256 × 192, field of view (FOV) of 107 × 90 mm2, repetition time (TR) of 3000 ms, echo time (TE) of 22.8 ms, and 1 mm thickness.
BOLD MRI relies on differences in oxygenated and deoxygenated haemoglobin concentrations in blood vessels and surrounding tissue. These differences cause contrast in MRI influencing the so-called spin-spin relaxation rate (R2*), allowing R2* values to be correlated to oxygen consumption (when external factors such as regional perfusion are accounted for) and inversely correlated to oxygen availability of a specific region of interest (i.e. low R2* values indicate low oxygen consumption yet high oxygenation) 39,63,64 . R2* can be calculated using an oxygenation-dependent sequence (T2*-weighted) and obtained using an axial 1H-multi-echo gradient-echo sequence, covering the entire abdomen with 32 slices using the following sequence parameters: matrix of 128 × 128, FOV of 80 × 80 mm2, flip angle of 90°, TR of 800 ms, TE of 2, 4, 6, 8, 10, 12, 14, and 16 ms, number of transients of 2, and 2 mm thickness. R2* levels calculated at baseline (time 0 h) served as an internal control to calculate relative changes in oxygenation, a measure superior to absolute values 40,64 .
ASL MRI relies on a difference in the T1-weighted signal of inflowing blood compared to that of the tissue of interest, allowing estimation of relative changes in tissue perfusion. For T1-measurements a single-slice segmented Look-Locker sequence with a gradient-echo readout was used to acquire T1-weighted data. The following sequence parameters were used: matrix of 128 × 128, FOV of 80 × 80 mm2, flip angle of 8°, TR of 3 ms, TE of 2 ms, inversion times (TI) of 150, 250, 400, 600, 900, 1200, 2500, 4000 ms, and 2 mm thickness.
VnmrJ (Agilent Technologies) was used for image reconstruction and volumetric analysis. Manually drawn regions of interest were encompassed around the liver and kidneys. Quantitative T2* maps were calculated from pixel-by-pixel analysis using a nonlinear least-squares fit to the logarithmic magnitude vs. TE. Quantitative T1 maps were calculated from a three-parameter fit applied to the inversion recovery Look-Locker sequence, using the mathematical approach given by Ramasawmy et al. 16,65 . Mitochondrial proteomics, complex I and ATP synthase activity, and tissue ATP levels.
Targeted, quantitative mitochondrial proteomics. In isolated mitochondria, we quantified a selection of mitochondrial proteins involved in the substrate transport, FAO and TCA cycle, using isotopically-labelled standards ( 13 C-labeled lysines and arginines). These were derived from synthetic protein concatamers (QconCAT) (PolyQuant GmbH, Bad Abbach, Germany), using a method previously described 66 .
Hepatic and renal ATP concentrations. Frozen liver and kidney tissue was cut into 20 mm slices; 650 mg of these slices were used to determine ATP content according to standard procedures 67 .
Complex I and ATP synthase enzyme activity measurements. Mitochondria were isolated as previously described 62 , then diluted in PBS, lysed by sonication, and centrifuged at 600 g for 10 min at 4 °C. Protein concentration was determined in the supernatant using a BCA protein assay kit (Pierce, Thermo Fisher Scientific Inc., Rockford, IL, USA).
Activity of complex I was monitored spectrophotometrically at 600 nm and 37 °C, as previously described 68 . Rotenone-sensitive complex I activity was calculated using the molar extinction coefficient of DCIP, equal to 21000 M −1 cm −1 and expressed as nmol/min/mg protein 68 . The activity of ATP synthase was measured SCIENTIfIC REpoRts | (2018) 8:4405 | DOI:10.1038/s41598-018-22689-9 spectrophotometrically at 340 nm and 37 °C, as previously described 64 . Oligomycin-sensitive ATP synthase activity was calculated using the molar extinction coefficient of NADH, equal to 6220 M −1 cm −1 and expressed as nmol/min/mg mitochondrial protein 69 . Oxidative stress markers. Determination of oxidative damage through quantification of lipid peroxidation. Lipid peroxidation product MDA was quantified in liver and kidney homogenates (20 µl) by measuring the formation of thiobarbituric acid reactive substances with a method previously described 7 .
Gene expression of protective protein Heme oxygenase-1. Gene expression of Ho-1 was determined with Real-time PCR as described previously 8 with the primer set outlined in Table 1.
Statistical Analyses. For *R2 BOLD and T1 data, we estimated that we would need a total of 8 animals per group to detect a clinically significant difference with an α of 0.05, power of 80%, using a two-tailed test. Descriptive statistics were done to confirm that the data met the assumption of equal distributions of residuals. A linear mixed model was used with repeated measures over time to analyse the impact of the treatment (BD or sham) on BOLD and ASL in the liver and kidney, with fixed effects of time, treatment group, and the interaction of treatment and time (IBM SPSS Statistics 23). This model was chosen because it takes the dependency of the measurements across time into consideration, and prevented list-wise deletion caused by missing data points. The model selection for covariance parameters was chosen based on the best fit according to the Bayesian Information Criterion. When comparing two independent groups at a single time point, the non-parametric Mann-Whitney test was used to identify significant differences between the groups (n = 8 in each group) with Prism 6.0 (GraphPad Software Inc, CA, USA). To confirm abnormal results, a boxplot was performed to identify extreme outliers and considered significant when they scored >3× IQR compared to the other values with IBM SPSS Statistics 23. All statistical tests were 2-tailed and p < 0.05 was regarded as significant. Results are presented as mean ± SD (standard deviation).
Availability of data and materials. All data generated or analysed during this study are included in this published article (and its supplementary information files).