Mechanistic Analysis of Nonoxygenated Hypothermic Machine Perfusion's Protection on Warm Ischemic Kidney Uncovers Greater eNOS Phosphorylation and Vasodilation

Experimental design

In order to assess MP's endothelial protection, we used a preclinical model of DCD kidney autotransplantation in pigs, including 1 h of warm ischemia (WI-60), to evaluate the consequences of MP on the NO signaling pathway at the end of preservation and on renal circulation in vivo after reperfusion (Figure 1). At the end of preservation, we evaluated the levels of KLF2, eNOS and phospho-eNOS proteins and/or mRNAs in the renal cortex. Moreover, we assessed the vasodilation capacity of renal arteries isolated at the end of preservation with an ex situ organ bath. In vivo, we measured the early cortical microcirculation by laser Doppler in kidney grafts after autotransplantation, as a marker of endothelial function because cortical microcirculation has been reported to be NO-dependent 19.

Experimental groups

To be clinically relevant, we chose to evaluate the machine effect by comparing CS with the University of Wisconsin (UW) solution to MP with Belzer's Machine Perfusion Solution (MPS). UW being one of the two main solutions used for CS in the clinic and MPS being the only solution currently allowed for MP. However, these solutions have distinct compositions and do not offer the same level of preservation quality when compared in a similar setting 20; thus comparing UW-CS to MPS-MP provides data on both the effect of the machine and of the solution, which was not our goal. Therefore, it was necessary to include two supplementary groups of preserved kidneys (UW-MP and MPS-CS) to be able to isolate the effect of MP alone.

We constituted the following experimental groups using warm ischemic kidneys: UW-CS: CS with Viaspan® (Bristol-Myers Squibb, Fontenay, France) (UW) (n = 5); UW-MP: MP with UW (n = 5); MPS-CS: CS with KPS-1® solution (MPS; Kidney Preservation Solution-1, Organ Recovery Systems, Chicago, IL) (n = 5); MPS-MP: MP with MPS (n = 4) because one kidney was excluded after the machine had stopped for several hours during the night, following a “check tubing” error; control (CTL) group: kidneys from gender-, age- and weight-matched normal control animals (n = 5), WI-60: kidneys sampled at the end of warm ischemia (n = 5).

Kidney recovery and preservation

Male Large-White pigs (30–35 kg) were prepared as detailed in the Supplementary methods. Briefly, warm ischemia was induced by renal pedicle clamping for 60 min prior to recovery. Kidneys were flushed with the cold preservation solution and cold preserved for 24 h by CS or by nonoxygenated hypothermic MP using the LifePort® machine (Organ Recovery Systems). During MP, the resistance index were calculated by the machine as the ratio between the measured pressure and the flow rate (mmHg.min/mL) and downloaded at the end of the experiment.

At the end of the preservation period, kidneys were either used for the ex situ experiments or were transplanted in the same animal for the in vivo experiment.

Western blotting procedure

A standard Western blotting protocol was used to evaluate the cortical levels of flow-dependent proteins (KLF2, eNOS, thrombomodulin). In certain pathological conditions, inducible NOS (iNOS) is expressed in endothelial cells 21 and may represent an alternative source of NO leading to vasodilation 22. Therefore, we also quantified iNOS protein content. In addition, we evaluated the activation state of eNOS by quantifying the phosphorylation levels of the proteins on two activating residues: (phosphoserine-1179-eNOS, phosphoserine-635-eNOS), and one inhibiting residue (phosphothreonine-495-eNOS) as well as the activation state of the kinases involved in eNOS phosphorylation in response to flow (Akt, PKA and AMPK). Levels of the proteins of interest were expressed in percentage of the mean levels of the CTL group.

Reverse transcriptase quantitative polymerase chain reaction

Reverse transcriptase quantitative polymerase chain reaction (RT-qPCR) was used to evaluate the mRNA levels of KLF2, eNOS, thrombomodulin and iNOS. The mRNAs of the ribosomal phosphoprotein large P0 subunit (RPLPO) and the ribosomal protein L19 were used as reference genes. mRNA levels were expressed as fold change in comparison to the CTL group.

Ex situ vascular function study

The ex situ organ-bath technique is commonly used to determine vasomotor function of isolated blood vessels 23-25. In the present work, first and second divisions of the renal artery were dissected and cut into 3-mm-long rings. For each kidney, eight artery rings were individually mounted between two hooks and immersed in eight separate thermoregulated 5-mL organ baths containing oxygenated (95% O₂ and 5% CO₂) Krebs solution at 37°C. Contraction force was recorded using an isometric force transducer.

After a final fourth wash with Krebs solution, a sustained contractile response was evoked by adding 1 micromolar (µM) of norepinephrine to the bath. The contraction force was recorded and defined as the maximal contraction force. Then, relaxation was elicited by addition, to the norepinephrine-containing bath, of cumulatively increasing concentrations of acetylcholine (range: 0.01–100 μM). Acetylcholine-relaxing effect was expressed as a percentage of the maximal contraction force. Acetylcholine half-maximal effective concentration was defined as the drug concentration reducing the contraction force to 50% of its initial value. Acetylcholine-relaxing effects were determined in artery rings in the absence (four baths) or presence (four baths) of 100 µM NG-nitro-L-arginine methyl ester (L-NAME; a NOS inhibitor) in the bath.

In addition, we repeated the vasomotor activity experiments with kidneys flushed and preserved by MP with MPS supplemented with 100 µM of L-NAME (MPS-MP + L-NAME) to evaluate whether L-NAME during MP was able to block vasodilation after reoxygenation. Three rings were rinsed and mounted according to the standard protocol (four rinses group), three rings were cut in Krebs solution and directly mounted in warm oxygenated Krebs solution without further rinsing prior to norepinephrine and acetylcholine additions (one rinse group). L-NAME was added to the bath for the last two rings (data not shown).

Kidney autotransplantation: in vivo cortical evaluation of cortical microcirculation

The preserved kidneys were transplanted in the same animal they were recovered from. An independent experimenter, blinded to the experimental groups, performed the laser doppler measurements in situ in anesthetized animals, 15 min after unclamping of the grafted kidney, using the Periflux System 5000 Laser Doppler (Perimed, KB, Jarfalla, Sweden). Blood cell flow, velocity and concentrations of moving blood cells were recorded for 10 min. Mean arterial blood pressure was recorded at the same time by placing a catheter in the left carotid artery.

Statistical methods

Results are shown as mean ± standard error of the means (SEM). Statistical analyses were performed with GraphPad Prism® 5 (GraphPad Software, Inc., La Jolla, CA) and R packages (R Core Team, URL: http://www.R-project.org/). The Western blot, RT-qPCR and laser Doppler experiments were first analyzed by one-way analysis of variance (ANOVA) with a Dunnett post hoc test to compare the preserved groups to a reference group (either CTL or WI-60). Then, the values from the four preserved groups were analyzed by two-way ANOVA to isolate the machine effect, the solution effect and the potential interaction between the two factors (machine, solution). Comparison of the resistance index during MP was performed using a Mann–Whitney test. Statistical significance was accepted for p < 0.05.

MP did not alter KLF2 and eNOS content in renal cortex at the end of preservation

We evaluated the cortical levels of KLF2 and its downstream effectors (eNOS and thrombomodulin) by Western blot at the end of preservation. KLF2 levels were not significantly altered in the four groups compared to CTL (p = 0.6), whereas eNOS levels were increased approximately twofold (significant for MPS-MP only, p = 0.024) and thrombomodulin levels were decreased by approximately twofold (p = 0.001) compared to CTL (Figure 2A and B). Two-way ANOVA performed on KLF2 quantifications in the UW-CS, UW-MP, MPS-CS and MPS-MP groups confirmed the absence of machine effect (indicated by almost horizontal lines between the CS and MP columns in the graphical representation, p = 0.7), the absence of a solution effect (indicated by overlapping of lines, p = 0.2) and the absence of an interaction between the preservation mode and the solution used (line alignment almost parallel, p = 0.4) (Figure 2C). Two-way ANOVAs of eNOS and thrombomodulin levels provided similar results (Figure 2C). Thus, MP did not affect the protein levels of KLF2, thrombomodulin or eNOS.

Because the effects of flow on KLF2, eNOS and thrombomodulin have previously been demonstrated at the mRNA level 26, 27, we performed RT-qPCR analysis. Here again, their mRNA levels were not significantly different from the CTL group (Figure S1A) and there was no significant machine or solution effect (Figure S1B). We also measured the expression levels of iNOS, another source of NO, and confirmed it was not affected by MP (Figure S2)

MP increased cortical eNOS phosphorylations at the end of preservation

Decreases in bloodflow reduce NO production by decreasing eNOS phosphorylation on the activating serine residues 28. As there was no machine effect on eNOS expression, we postulated that MP modulated eNOS phosphorylation.

eNOS phosphorylation on serine-1179 in the UW- and MPS-CS groups were not different from the CTL group but appeared increased 1.6-fold in the UW-MP (p = 0.213) and 1.8-fold in the MPS-MP group (p = 0.105) compared to CTL (Figure 3A and B). In contrast, phosphorylation of eNOS on the serine-635 residue was significantly decreased compared to CTL in all groups of preserved kidneys (Figure 3A and B).

Two-way ANOVA showed a significant machine effect on eNOS phosphorylation on serine-1179 and serine-635 (p < 0.05), without a solution effect (Figure 3C). For serine-635, there was a significant interaction between the preservation mode and the solution used (p = 0.042), leading to a greater MP-induced increase in phosphoserine-635 content with MPS (3.5-fold between MPS-CS and MPS-MP) compared to UW (1.5-fold between UW-CS and UW-MP) (Figure 3C). These machine effects were specific of the two activating serine residues as there was no machine or solution effect on the phosphorylation of eNOS on the inhibiting threonine-497 (equivalent to threonine-495 in humans 10) (Figure S3).

MP increased cortical eNOS phosphorylation possibly through AMPKα activation

Akt, PKA and AMPK phosphorylate eNOS on the activating serine residues in response to flow 29-32. We measured the activation status of these kinases at the end of preservation and determined that neither Akt nor PKA were affected by MP (Figures S4 and S5). However, total AMPKα levels were significantly increased in the MPS-MP group (2.2-fold) compared to CTL (p = 0.006), whereas phospho-AMPKα levels were significantly decreased compared to CTL in all the preserved groups (p < 0.05) except the MPS-MP group (Figure 4A and B). There were significant machine effects and solution effects both on total- and phospho-AMPKα levels (Figure 4C).

Increased cortical AMPKα and eNOS phosphorylations are specific of MP, not of warm ischemia

To determine if MP truly increased cortical eNOS and AMPKα phosphorylation/activations rather than maintain a high phosphorylation state induced by warm ischemia, we reproduced the Western blot experiments replacing the CTL group by a group of kidneys sampled after WI-60.

Phosphoserine-1179-eNOS levels were similar between WI-60 and the CS groups (UW-CS, p = 0.9; MPS-CS, p = 0.7), but were significantly (p < 0.05) increased in the MP groups (UW-MP: 1.6-fold; MPS-MP: 2.1-fold) (Figure 5A and B). In addition, phosphorylated AMPKα was detectable in the preserved groups but not in the WI-60 group (Figure 5C and D). Thus, we demonstrated that MP specifically increased activation of eNOS and AMPKα. We then endeavored to demonstrate the physiological effects of this mechanism.

MP improved renal artery NO-dependent vasodilation

Vasomotor experiments were performed ex situ at the end of preservation after rewarming and oxygenation, partly mimicking reperfusion. Addition of increasing doses of acetylcholine to precontracted arteries revealed differences in the maximum vasodilation response between the experimental groups, for concentrations superior to 10^−4.5 M (30 µM) (UW-CS: 81.6 ± 9.8%; UW-MP:18.7 ± 9.7%; MPS-CS: 57.0 ± 8.0%; MPS-MP 26.2 ± 9.3% of the initial contraction value) (Figure 6A). Acetylcholine half-maximal effective concentrations could be calculated only in the MP groups (UW-MP: 738 ± 1.56 nM and MPS-MP: 205 ± 5 nM) as the contraction force was not reduced below 50% of the initial contraction value in the CS groups. Vasodilation in the CS groups remained reduced for the entire duration of the experiment (90 min).

Two-way ANOVA of the contraction values after administration of 10^−4.5 M of acetylcholine in all groups revealed a significant machine effect improving the vasodilation response compared to CS (p = 0.0001) with no solution effect, but with a significant interaction between the solution and the preservation mode (p = 0.047) (Figure 6B) reflected by a greater gain in dilation after MP with UW than with MPS.

To confirm that MP benefited grafts through NO signaling, we added of L-NAME in the bath, which completely abolished the acetylcholine-induced relaxation in all groups (Figure 6B). MP protection is thus an NO-dependent mechanism, likely relying on AMPKα-dependent regulation of eNOS. We next measured the relevance of this mechanism in vivo.

MP improved early cortical microcirculation at reperfusion

All the transplanted animals had similar time for vascular anastomoses (30 ± 5 min), with no visible signs of stenosis after unclamping, and mean arterial pressures similar to CTL animals (p = 0.66) (Table 1). Thus, any difference in cortical microcirculation between the experimental groups could not be linked to differences in systemic perfusion pressure.

All laser Doppler parameters were significantly decreased in the transplanted kidneys compared to CTL (Figure 7A). Within the transplanted groups, there was a significant machine effect on all three laser Doppler parameters (p = 0.0001) (Figure 7A and B). Indeed, compared to CS, MP with UW increased blood cell flow 3.4-fold, velocity 1.5-fold and concentration of moving blood cells 2.3-fold, and MP with MPS increased blood cell flow 1.8-fold, velocity 1.3-fold and concentration of moving blood cells 1.4-fold. There was also a significant solution effect in laser Doppler parameters (p = 0.0001) (Figure 7B). Finally, there was a significant interaction between the preservation mode and the solution on blood cell flow (p = 0.04) and on the concentration of moving blood cells (p = 0.0003) (Figure 7).

Thus, MP benefits translate into a better revascularization of the graft at reperfusion, with a superiority of MP-MPS over MP-UW.

Link between the machine effects on NO signaling at the end of preservation and on increased microcirculation in vivo: NOS inhibition during MP

Next, we aimed at demonstrating that the machine effect on cortical microcirculation in vivo was directly linked to the MP effect on eNOS phosphorylation/activation by inhibiting NOS activity at reperfusion. For this, we added a NOS inhibitor (L-NAME) in the preservation solution and not at reperfusion to avoid the systemic effects of L-NAME 33, 34. Before implementing this strategy in vivo, we verified that addition of L-NAME to the preservation solution was able to block the vasodilation measured, after rinsing, with the organ-bath technique. We performed this preliminary experiment in the MPS-MP group only, the most clinically relevant.

Using the standard protocol (four rinses), NOS inhibition during MP (MPS-MP + L-NAME) partially blocked the acetylcholine-induced vasodilation, increasing 6.6-fold the acetylcholine half-maximal effective dose compared to the MPS-MP group (p = 0.1) (Figure 8A). Rinsing the artery rings only once after MP-MPS with L-NAME blocked the vasodilation response to acetylcholine to a similar extent as the one observed in the MPS-CS group (Figure 8B).

The inhibition of vasodilation after one wash assures that 100 µM of L-NAME in the machine is sufficient to inhibit eNOS activity during perfusion. But, supplementation of MPS with L-NAME did not affect the decrease in resistance index calculated by the machine (Figure 8C).

Because NOS inhibition was rapidly lost after rinsing, we did not implement this blocking strategy in vivo as flushing of the inhibitor would be even faster by blood. A more robust NO-inhibiting strategy would be required to perform this experiment.