Results of an explorative clinical evaluation suggest immediate and persistent post-reperfusion metabolic paralysis drives kidney ischemia reperfusion injury

Please cite this article as: Lindeman JH, Wijermars LG, Kostidis S, Mayboroda OA, Harms AC, Hankemeier T, Bierau J, Sai Sankar Gupta KB, Giera M, Reinders ME, Zuiderwijk Bsc MC, Le Dévédec SE, Schaapherder AF, Bakker JA, Results of an explorative clinical evaluation suggest immediate and persistent post-reperfusion metabolic paralysis drives kidney ischemia reperfusion injury., Kidney International (2020), doi: https://doi.org/10.1016/j.kint.2020.07.026.


Results of an explorative clinical evaluation suggest immediate and persistent post-reperfusion metabolic paralysis drives kidney ischemia reperfusion injury.
Clinical Ischemia reperfusion injury (DGF) relates to an almost instantaneous and persistent post-reperfusion metabolic collapse Delayed graft function is the manifestation of ischemia reperfusion injury in the context of kidney transplantation. While hundreds of interventions successfully reduce ischemia reperfusion injury in experimental models, all clinical interventions have failed. This explorative clinical evaluation examined possible metabolic origins of clinical ischemia reperfusion injury 5 combining data from 18 pre-and post-reperfusion tissue biopsies with 36 sequential arteriovenous blood samplings over the graft in three study groups. These groups included living and deceased donor grafts with and without delayed graft function. Group allocation was based on clinical outcome. Magic angle NMR was used for tissue analysis and mass spectrometrybased platforms were used for plasma analysis. All kidneys were functional at one-year. 10 Integration of metabolomic data identified a discriminatory profile to recognize future delayed graft function. This profile was characterized by post-reperfusion ATP/GTP catabolism (significantly impaired phosphocreatine recovery and significant persistent (hypo)xanthine production) and significant ongoing tissue damage. Failing high-energy phosphate recovery occurred despite activated glycolysis, fatty-acid oxidation, glutaminolysis and autophagia, and 15 related to a defect at the level of the oxoglutarate dehydrogenase complex in the Krebs cycle. Clinical delayed graft function due to ischemia reperfusion injury associated with a postreperfusion metabolic collapse. Thus, efforts to quench delayed graft function due to ischemia reperfusion injury should focus on conserving metabolic competence, either by preserving the integrity of the Krebs cycle and/or by recruiting metabolic salvage pathways.

Introduction
Ischemia reperfusion injury (IRI) is the phenomenon of increased tissue damage following reperfusion of previously ischemic tissue. 1,2 IRI is a main contributor to organ damage following myocardial or brain infarction, 3 and graft damage following organ transplantation. 4 While a myriad of interventions quench IRI in preclinical models, clinical success remains missing. 3,4 5 Consequently, there appears a translational gap between preclinical models and clinical context.
Delayed Graft Function (DGF) is the manifestation of IRI in the setting of kidney transplantation. 5 DGF is defined as the need for dialysis in the first week(s) following transplantation. 6 While DGF is extremely rare in the context of living-donor graft procedures, it 10 affects up to 90% of deceased donor graft transplantations. 6 Previous work demonstrated an association between incident DGF and post-reperfusion normoxic glycolysis. 7 This observation implies that DGF relates to a defect in graft energy homeostasis as result of mitochondrial dysfunction in the reperfusion phase. 7 On this basis we hypothesized that clinical DGF involves, and may be driven by (a) metabolic defect(s). The objective of this study was to perform an in-15 depth analysis of metabolic responses to ischemia-reperfusion with-and without IRI (DGF).
This explorative metabolic evaluation is based on an integrated, time-resolved approach that involved sequential assessment of arterial-venous concentration (AV-) differences over reperfused grafts, and parallel profiling of graft (tissue) biopsies. Three different study groups were included: grafts from deceased donor grafts with and without later IRI, and living donor 20 grafts. Group allocation of deceased donor grafts (+DGF and -DGF, respectively) was done retrospectively on basis of their clinical outcome. Living donor grafts were included as a reference since these grafts associate with an instantaneous functional recovery following reperfusion. In order to cover all primary aspects metabolic homeostasis, it was decided to focus on the following gross metabolic clusters: nucleotide triphosphate metabolism, fatty acid 25 J o u r n a l P r e -p r o o f

Results
This study involves 53 patients. Paired tissue biopsies were obtained in 18 patients and sequential AV sampling was performed in 36 patients. One patient had both biopsies taken and underwent AV sampling. Clinical details for the different study groups are shown in 5 supplemental tables 1A (tissue biopsies) and 1B (AV-sampling). All DGF cases required multiple dialyses over a time course of at least 7 days, and all showed adequate functional recovery. None of the deceased donor without DGF required dialysis after transplantation. Oneyear graft survival was 100%.
We first explored putative differences in metabolic signatures for the three donor groups 10 (the living (reference) donor grafts, the -DGF deceased donor grafts, and +DGF (IRI) deceased donor grafts) by mapping the plasma metabolome (AV-differences) for the 30 min. postreperfusion time point (Fig 1a), and the tissue metabolome (tissue biopsies) for the 40 min. post-reperfusion time point (Fig 1b). These time points were chosen to avoid interference from washout of metabolites that have accumulated during the ischemic, cold-storage and/or of 15 constituents of the preservation fluid (e.g. histidine wash-out from living donor grafts (Fig. S1) reflects the selective use of H(istidine)TK preservation fluid in these grafts). 7 Results (Z-scores) for these time points are summarized in the heat maps shown in Fig. 1a (AV-differences) and Heat maps for the AV-differences indicate parallel metabolic signatures for the living donor and -DGF grafts, and a clearly distinctive signature for the +DGF grafts (Fig 1a). A similar, though less pronounced pattern was observed for the tissue metabolites (Fig. 1b). Exclusive Collectively, the data provide a gross metabolic signature for renal IRI.
For the sake of clarity, it was decided to present the data of the individual metabolites 5 along the lines of the 6 metabolic clusters. To avoid interference from the initial wash-out of metabolites that have accumulated during cold-storage within the first minutes of reperfusion, estimations for net post-perfusion release or uptake are based on integration of AV-differences for the 10 to 30 min. post-reperfusion time intervals (area between the curves).
The first cluster of metabolites ('nucleoside triphosphate catabolism') signals a persistent 10 post-reperfusion metabolic incompetence ('power shut down') in grafts with later DGF (+DGF). This conclusion is based on an impaired post-reperfusion recovery of the high-energy phosphate-buffer phosphocreatine in +DGF grafts (P<0.001), (Fig. 2a), and by persistent postreperfusion ATP/GTP catabolism. The latter is reflected in the continued release (AVdifferences) of hypoxanthine and xanthine ( Fig. 2b and 2c, P resp. <0.0001 and 0.02), the 15 terminal degradation products of ATP and GTP from these grafts. Data for the pre-reperfusion tissue biopsies showed graded degrees of inosine and hypoxanthine accumulation at the end ischemic storage period; with the lowest contents found in living, and the highest in deceased donor grafts ( Fig. 2d and 2e). Post-reperfusion (t=40 min) hypoxanthine and inosine tissue contents were similar and low in all three donor groups ( Fig. 2d and 2e). A final cluster of discriminatory metabolites relates to on-going cell damage. This cluster includes post-reperfusion release of uracil, an established marker of cell damage 12,13 (supplemental Fig. S3a, P <0.0001), and of amino acid derivates that associate with hydrolysis 5 of plasmalogens (viz. phospho-ethanolamine, ethanolamine, and phospho-serine (supplemental Above observations associate incident IRI with persistent post-reperfusion ATP catabolism and on-going cell damage in a context of mitochondrial failure and activation of glycolytic and lipolytic pathways (summarized in Fig. 6). Considering the vital role of ATP in 15 cellular homeostasis and survival, it was reasoned that recruitment of auxiliary ATPregenerative pathways (viz. independent of mitochondrial respiration) would be beneficial. In this context we considered inosine, a nucleoside that can generate ATP through nontraditional pathways. As shown in Fig. 7 neither preventive nor rescue inosine delivery (in concentrations up to 10 mMol/L) rescued ATP exhaustion following chemically induced metabolic paralysis.

Discussion
From this study, performed in the context of clinical kidney transplantation, the picture emerges of IRI (DGF) being the consequence of an almost instantaneous and persistent post-reperfusion failure of oxidative phosphorylation, and an activated normoxic glycolysis that is unable to sustain energy homeostasis. As a consequence, high energy phosphate pools are progressively 5 exhausted and cellular integrity cannot be preserved, resulting in perpetuation of tissue damage.
This clinical study is based on integration of metabolic data derived from tissue biopsies taken immediately prior to (pre) and 40 min after (post) reperfusion, and from sequential assessment of AV-differences over the reperfused graft. These AV-differences not only provide 10 an indication for the pace and duration of metabolic (mal)adaptions, but also allow for directing trends observed in the paired tissue biopsies and for appreciation of metabolite clearance into (elimination, e.g. lactate) or uptake from (e.g. medium chain fatty acids) the circulation. 15, 16 The resolution of the AV approach is clearly illustrated by the acylcarnitine data; not only showing selective uptake of medium chain fatty acids, but also suggesting that unsaturated C14 carnitine 15 species, tetradecenoyl-and tetradecadienyl carnitine, behave similar to medium chain fatty acids (Data. S1) and may not rely on specific fatty acid transporters. 17 In fact, in the process of data analysis, it was realized that sole reliance on tissue biopsies would have obscured most conclusions in this study since most metabolites formed are efficiently cleared into the circulation. Stable arterial blood concentrations show blood homeostasis is maintained, and 20 consequently metabolites released or absorbed are effectively disposed or replenished elsewhere. 15,16 Observed stable tissue contents, but clear AV differences challenge the validity of tissue-based metabolomic evaluations. Note that, for the context of deceased donor kidneys and the timeframe of the study, urinary clearance is not an interfering factor since all deceased donor grafts in this study were anuric for the 40 minutes measurement interval.

J o u r n a l P r e -p r o o f
Mapping of the data identifies a metabolic footprint that is fully discriminatory for IRI. To be more specific, the reperfusion phase of grafts with future DGF is uniformly and distinctively characterized by a severely impaired oxidative phosphorylation (histotoxic hypoxia), 18 and a compensatory normoxic glycolysis that is unable to sustain ATP regeneration. This latter conclusion is based on the incomplete recovery of the high-energy phosphate buffer 5 phosphocreatine, 19  grafts, and persistent release in +DGF grafts may reflect different degrees and rates of metabolic recovery in these grafts. However, while the earlier reports imply a role of a cytosolic )ethanolamine imply a more comprehensive activation of phospholipases that also involves type C-(phospho-ethanolamine) and D-(ethanolamine/choline) phospolipases. Along similar lines, depletion of tissue asparagine (Fig. 4J) and release of aspartate (Fig. 4G) from +DGF grafts may reflect impaired asparagine synthase activity as result of ATP depletion.
Post-reperfusion ATP catabolism in +DGF grafts occurred despite comprehensive 5 activation of catabolic pathways: glycolysis, β-oxidation of medium chain fatty acids (uniformly activated in all graft types); glutaminolysis (also transiently activated upon reperfusion in living donor and -DGF grafts); as well as activated autophagy. In fact, post-reperfusion release of isovaleryl-and butyrylcarnitine, deamination products of the branched-chain amino acids isoleucine and leucine, 11 were identified as discriminatory biomarkers for future DGF. Impaired oxoglutarate dehydrogenase activity may be caused by ischemia-related 20 damage to the complex, 30 but may also involve, or be exaggerated by, impaired postreperfusion availability of its co-factors acetyl-CoA, FAD + and NAD + . 31 For +DGF grafts such deficiencies could occur because of post-reperfusion acetyl-CoA washout, and a compromised cellular redox status (reductive stress with impaired NAD + availability). A notion supported by the low lactate/pyruvate ratio in +DGF grafts. 32 This metabolic approach taken, does not allow for evaluation of involvement of aspects of the respiratory chain. Yet, we earlier identified ischemia reperfusion-related defects in both respiratory complex I and II. 7,29 On the basis of these data in this study and previous mitochondrial work the picture emerges of clinical renal IRI being the consequences of (an) primary (eliciting) insult(s) to the mitochondrial Krebs cycle/redox shuttle that occurs prior to, 5 and/or within the first minutes of reperfusion. Failure to reinstate ATP levels results in a sustained and comprehensive activation of catabolic pathways, which actually perpetuates the energy crisis by progressively exhausting the cellular NAD + and FAD + pool (reductive stress). 33 It was reasoned that in this specific context with failing mitochondrial respiration, the purine inosine would be beneficial. Unlike adenosine, 34 inosine is stable in plasma, and has been 10 identified as alternate source of ATP in obligatory glycolytic cells (i.e. cells lacking mitochondria) such as erythrocytes 35 and in hypoxic renal cells, 36 and is exhausted following reperfusion.
Unfortunately, inosine supplementation did not rescue cellular ATP depletion following a forced metabolic shut-down, leaving little room for metabolic rescue strategies aimed at quenching IRI, and stressing the reliance on preventive strategies for limiting IRI.

15
Limitations: owing to the large number of comparisons the potential for significant findings due to random chance in the setting of multiple comparisons is high. Although the potential might be attenuated by the fact that the conclusions are supported by sound biological relationships, conclusions of the study might be interfered by multiple comparison problems.
A further limitation of this study is that it is fully based on clinical samples, as such 20 clamp-freezing required for direct assessment of ATP and redox status was not possible. Since the metabolome observed is clearly distinct from that reported in animal models, and as it reflects a system failure we were unable to perform more detailed evaluations in animal models or ex-vivo systems such as respirometry. Results in this study are for the kidney, as such conclusions for other organ may be different. The relative high donor age in this study is a outcomes for the Netherlands are at least equal to countries with younger donors such as in the USA. 37 As expected the majority of +DGF cases were DCD grafts. We noticed similar metabolic profiles for DBD and DCD grafts; yet the power of this explorative study is obviously too low to detect subtle differences between these two donor types. 5 In conclusion, this study shows that clinical renal IRI is preceded by an almost instantaneous metabolic collapse, and an accompanying high-energy phosphate crisis. It came to our attention that this deep and persistent metabolic deficit, and it's instantaneous character (and consequently a minimal window of therapeutic opportunity) will interfere with any pharmaceutical intervention that relies on ATP availability. This aspect may explain the poor  29 In this context, it is important to point out that all transplanted kidneys are exposed to ischemia reperfusion, and that only a subgroup of grafts develops IRI (DGF). Group allocation (+DGF or -DGF) in this study was performed retrospectively, as result this study discriminates between ischemia reperfusion and IRI. It cannot be excluded that the ischemia-reperfusion in experimental models 28,38-40 is insufficient to trigger IRI.

Patients and Methods
The Leiden University Medical Center medical ethics committee approved the study protocol.
Written informed consent was obtained from each patient. This single center study included 53 patients who underwent kidney transplantation: 37 underwent deceased donor graft procedures The study is based on an integration of metabolomic data obtained from sequential arterio venous (AV) blood sampling during first half hour of reperfusion, and from paired tissue  Table 1B, one patient had both biopsies and AV sampled)).
Targeted metabolomics analyses were performed using standard operating procedures using established mass spectrometry-based platforms or magic angle NMR (tissue biopsies). 43 Metabolites covered by the platforms are summarized in supplementary Table 1. 20 The potential of inosine to rescue the metabolic deficit during a metabolic collapse was

Supplementary information is available on Kidney International's website
Supplemental Tables S1A and B. Patient and transplantation characteristics of the procedures in which paired tissue biopsies were collected (S1A) and in which AV-sampling was performed 15 (S1B).   Curves for the arterial venous differences (red curve is arterial, blue curve is venous).

Supplemental
Tissue biopsies (bar graphs): white bars represent pre-reperfusion biopsies; grey bars represent  Curves for the arterial venous differences (red curve is arterial, blue curve is venous).  Curves for the arterial venous differences (red curve is arterial, blue curve is venous).  Metabolites in blue indicate net release from the graft. Metabolites in red indicate net uptake by the graft (AV-differences). Black: no AV-differences, Grey information not available. The six metabolic clusters are indicated: 1) nucleoside triphosphate catabolism; 2) β-oxidation (uptake of medium chain fatty acids (MCFA); 3) glycolysis/glutaminolysis (pyruvate/lactate release; glutamine uptake and alanine release); 4) autophagy (release of serine, methionine and tyrosine, and the oxidation products proprionyl-, butyryl-, and isovalerylcarnitine); 5) Krebs cycle defects ((iso)citrate uptake, but ketoglutarate release), and 6) cell damage: release of (phospho-5 )ethanolamine, betaine, and phospho-serine implying hydrolysis of plasmalogens. Fig. 7. Both preventive and rescue inosine treatment fail to recover ATP levels.
The PK-1 renal cell line was stably transfected with the PercevalHR fluorescent biosensor of ATP-to-ADP ratio. Chemically induced metabolic-paralysis was induced by adding