Molecular mechanisms of ischemic preconditioning in the kidney

Ischemic/Hypoxic Preconditioning: Two Phases of Protection

Preconditioning is characterized by two phases of protection, an acute or early phase, which immediately follows the preconditioning stimulus and lasts no longer than several hours, and a late phase, which starts ∼24 h after the preconditioning stimulus and can last for several days (73, 107, 124). Traditionally, the acute phase has been viewed as being mediated by ion channels and signaling molecules produced during ischemia (61), while the late phase has been thought to depend on transcriptional responses that modulate glucose and mitochondrial metabolism, suppress ROS production and inflammation, and inhibit the expression of proapoptotic molecules. However, recent studies in a model of cardiac ischemia have challenged this concept and have provided experimental evidence that the acute phase of IPC is also transcription factor dependent (139). In mammals, transcriptional responses to hypoxia are primarily mediated by hypoxia-inducible factors (HIFs), oxygen-regulated transcription factors that allow cells to respond and adapt to low tissue Po₂.

Oxygen-Dependent Regulation of HIF

HIF-1 and HIF-2 belong to the PER/aryl-hydrocarbon-receptor nuclear translocator (ARNT)/single minded (PAS) family of basic helix-loop-helix transcription factors and consist of an oxygen-sensitive α-subunit and a constitutively expressed β-subunit, also known as ARNT (105, 147). Both transcription factors facilitate oxygen delivery and cellular adaptation to hypoxia by stimulating multiple hypoxia responses including anaerobic glucose metabolism, adenosine and nitric oxide (NO) metabolism, angiogenesis, erythropoiesis, and iron metabolism (59, 167). In the presence of sufficient oxygen, HIF-1α and HIF-2α are targeted for rapid proteasomal degradation by the von Hippel-Lindau (VHL) tumor suppressor, which functions as the substrate recognition component of an E3-ubiquitin ligase complex (109). VHL-mediated targeting of HIF-α requires oxygen- and iron-dependent hydroxylation of specific proline residues located within the oxygen-dependent degradation domain of HIF-α (Fig. 1). HIF prolyl hydroxylation is carried out by three Fe (II)- and 2-oxoglutarate (2OG)-dependent prolyl-4-hydroxylase domain (PHD) proteins, PHD1, 2 and 3 (also known as EGLN2/HPH3, EGLN1/HPH2, and EGLN3/HPH1, respectively) (65, 76), each displaying different expression profiles and specific subcellular localization patterns (for a recent review on this topic, see Ref. 143). PHD enzymes function as oxygen sensors and belong to a large family of 2OG-dependent dioxygenases (∼60 family members have been identified in mammals) that utilize molecular oxygen for catalysis and regulate multiple biological processes (76, 101, 115). Under normoxia, PHD enzymes are catalytically active and HIF-α is degraded, whereas under hypoxia hydroxylation is inhibited and HIF signaling is activated (Fig. 1). PHD catalytic activity is furthermore inhibited by NO and ROS (76), an important consideration for the study of oxygen-sensing mechanisms under conditions where intracellular concentrations of these signaling molecules has changed.

For the purpose of experimental and/or therapeutic IPC, HIF-α stabilization and activation of HIF-dependent signaling pathways can be achieved in several ways: 1) exposure to hypoxia, 2) genetic inactivation of the VHL-ubiquitylation machinery, 3) proteasomal inhibition, and 4) inhibition of HIF-PHDs by genetic or pharmacological means. Pharmacological inactivation of PHD enzymes can be accomplished by exposing cells and/or tissues to either cobalt chloride (CoCl₂), iron chelators, such as desferrioxamine, or to small-molecule compounds that reversibly inhibit PHD catalytic activity, several of which are now in clinical trials for the treatment of renal anemia (91). However, pharmacological inhibition of HIF-PHDs may also exert HIF-independent effects, as HIF-PHDs have been shown to hydroxylate targets other than HIF-α. Such non-HIF targets include IκB kinase-β (IKKβ), β2-adrenergic receptor (β2AR), sprouty homolog 2 (SPRY2), and pyruvate kinase muscle isoform 2 (PKM2) (170). To what degree HIF-activating PHD inhibitors, especially those that are in clinical trials, inhibit other non-HIF-related 2OG-dependent dioxygenases remains to be investigated.

Preischemic HIF Activation Protects From Renal Ischemia

In renal cells, the HIF system is activated under conditions of systemic or regional hypoxia, or in response to PHD inhibition. In acute hypoxia, HIF-1α has been detected in glomerular and renal tubular epithelial cells, whereas HIF-2α was detected in glomeruli, peritubular endothelial, and interstitial fibroblast-like cells, the source of renal erythropoietin (EPO) (91, 135). Compared with systemic hypoxia induced by either anemia or exposure to carbon monoxide (CO) (135), the pattern of HIF stabilization in the ischemic mouse kidney appears to be restricted to cells that are predominantly located at the corticomedullary junction (134). With respect to the reperfused rat kidney, significant HIF-1α accumulation was found in proximal tubular cells and felt to be due to increased HIF-1α translation via enhanced mammalian target of rapamycin (mTOR)/AKT activity (28). Consistent with these findings are immunohistochemical studies in human renal postengraftment biopsies, which were obtained between 25 and 40 min after completion of vascular anastomoses, and which revealed significant variability in the extent of HIF-α stabilization. These studies demonstrated a strong positive correlation between HIF-1α expression and allograft function, suggesting a cytoprotective role for HIF-1 in the setting of acute transplant ischemia (136).

Different experimental strategies have been used to test whether enhancement of HIF expression before ischemic kidney injury could afford cytoprotection. Matsumoto and colleagues (108), for example, found that cobalt, a “hypoxia-mimic” and inhibitor of PHD catalysis, conferred significant protection from renal ischemic injury in rats. Similarly, preconditional HIF activation in mice via either administration of CO or administration of PHD inhibitors l-mimosine, dimethyloxaloylglycine (DMOG), or FG-4487 protected the kidney from injury (9, 12, 64, 166). In keeping with this notion, HIF-α stabilization with xenon gas provided morphological and functional renoprotection in a mouse model of IRI. Interestingly, xenon did not activate HIF by inhibiting its degradation, but rather by increasing its translation via the mTOR pathway (104). Similarly, treatment of donor rats with FG-4497 improved tissue injury scores, early mortality rates, as well as long-term graft survival in an allogenic Fisher-Lewis rat kidney transplant model (11).

Strong support for a role of HIF signaling in renal IPC also comes from the study of genetic mouse models. Hill and colleagues (64) found exacerbation of ischemic renal injury in mice that were heterozygously deficient for HIF-1α or HIF-2α and furthermore demonstrated beneficial effects of preischemic HIF activation via the administration of l-mimosine. A cytoprotective role for HIF signaling in ischemic kidney injury is also supported by acute or constitutive deletion of Vhl in mice, which phenocopied the cytoprotective effects of hypoxic preconditioning (68, 142).

Cell Type-Dependent Effects of HIF Activation

A major challenge in the IPC field is the identification of the cell types that mediate HIF-induced renoprotection. Using a genetic approach, our laboratory has shown recently that endothelial cells are critical cellular mediators of ischemic/hypoxic preconditioning in the kidney. We demonstrated that endothelial HIF-2 plays a key role in renoprotection and suppresses IRI-associated inflammation through the modulation of endothelial vascular cell adhesion molecule 1 (VCAM1) expression and inflammatory cell adhesion (79). Specifically, we found that 1) endothelial inactivation of HIF-2α, but not of HIF-1α, was associated with increased expression of renal injury markers and inflammatory cell infiltration in kidneys injured by either ischemia reperfusion or by ureteral obstruction; and 2) the negative effects of HIF-2α deletion were reversed by pharmacological blockade of VCAM1 and its ligand very late antigen 4 (VLA4). In contrast, activation of HIF by either genetic means or via systemic PHD inhibition protected from ischemic kidney injury and inflammation (Fig. 2). In this context, endothelial HIF-2 activation may be useful for the prevention or amelioration of renal fibrogenesis and CKD progression, which are well-recognized, long-term sequelae of AKI. In support of our findings are genetic studies in mice subjected to acute lung injury. In this setting, endothelial HIF-2 was shown to enhance the integrity of adherens junctions and endothelial barrier function (51), supporting the concept that hypoxic signaling in endothelial cells induces a broad cytoprotective response that is directed toward maintaining tissue homeostasis during injury. Therefore, the endothelial PHD/HIF axis represents an attractive therapeutic target for the prevention and treatment of tissue injury associated with ischemia. The renoprotective effects of HIF activation, however, do not appear to be restricted to renal endothelial cells alone, as genetic activation of HIF signaling in other cell types, e.g., epithelial and myeloid cells, has been shown to be renoprotective (43, 87, 142). The role of HIF signaling in immune responses and renal inflammation is discussed in greater detail in a separate section.

While renal epithelial HIF-1 had previously been identified as a profibrotic transcription factor in experimental CKD (63, 84), transient global pre- or postischemic HIF stabilization did not promote fibrosis (77). Context, tissue, and cell-type dependence are expected and conceptually supported by several genetic studies in nonrenal injury models. For example, conditional inactivation of Hif1a in neurons resulted in protection from cerebral ischemic injury (62), whereas Baranova and colleagues (7) reported enhanced brain injury in neuron-specific Hif1a^−/− mice following transient focal cerebral ischemia. In the heart and kidney, germ line inactivation of one copy of Hif1a worsened both myocardial and renal ischemia (22, 64).

Timing and Duration of HIF Activation Determine Clinical Outcome

In contrast to the robust renoprotection observed with preischemic administration of PHD inhibitors, postischemic HIF-PHD inhibition failed to ameliorate renal injury (77, 164), suggesting that the timing of HIF activation is critical for renoprotection. In contrast, PHD inhibition with DMOG increased HIF-dependent gene expression and reduced injury when administered 30 or 60 min post-IRI in a model of cerebral ischemia (120). Although tissue-specific changes in HIF-mediated gene expression may explain these differences, the regulation of HIF signaling during the actual ischemic-reperfusion phase remains poorly understood and requires further investigation. Studies performed in our laboratory indicate that transcriptional HIF responses are suppressed in the immediate reperfusion phase, when PHD inhibitors are administered postinjury, which may be due to tissue acidification or the increased production of ROS (77).

In addition to timing, the duration of PHD inhibition/HIF activation appears to be critical for disease outcome and may produce adverse effects on tissue homeostasis. For example, cardiac function in mice with cardiomyocyte-specific Phd2 deletion was significantly improved 3 wk after myocardial IRI (66), whereas sustained Phd2 ablation or tissue-specific HIF-1α stabilization resulted in cardiomyopathy (113). In the kidney, chronic HIF activation in proximal tubules induced by Vhl gene loss caused tubular microcysts, which showed evidence of dedifferentiation and increased proliferation (133). In addition to renal cysts, inflammatory and fibrotic lesions were observed after inactivation of Vhl in collecting ducts and a subset of distal tubules (129). Furthermore, expression of a nondegradable form of HIF-2α in renal tubules led to cyst formation and fibrosis (140), while expression of a nondegradable form of HIF-1α resulted in lipid and glycogen deposition and as well as microcyst formation (47), supporting the notion that chronic activation of the renal epithelial HIF system promotes inflammation, fibrosis, cell proliferation, and dysplasia.

HIF Attenuates IRI Through the Coordinated Activation of Renoprotective Metabolic and Signaling Pathways

Although the molecular and cellular mechanisms underlying HIF-mediated cytoprotection in renal IRI are complex and only poorly understood, the concerted transcriptional activation of genes involved in cellular energy metabolism, antioxidant responses, antiapoptotic pathways, and other HIF-regulated biological processes is key to effective and robust cytoprotection (Fig. 3).

Differential Activation of Stress-Related Kinases

Park and colleagues (125) have evaluated the effect of IPC induced by 30 min of bilateral renal ischemia on the activation pattern of stress-related kinases. They found that the ischemia-related activation of JNK and p38 were markedly diminished by IPC, whereas an effect on postischemic activation of ERK1/2 was not observed. Similarly, the phosphorylation of MKK7, MKK4, and MKK3/6, upstream activators of JNK and p38, was markedly reduced. Given the involvement of JNK and p38 in the expression of adhesion molecules and cytokine production, the authors speculated that these alterations may lessen leucocyte-endothelial interactions and therefore explain the remarkable IPC-induced reduction in outer medullary cellular congestion. A link between enhanced cell survival, decreased JNK, and increased ERK1/2 activation was shown in an in vitro model of endoplasmic reticulum stress preconditioning (67). Joo and colleagues (73), using a different IPC protocol, found rapid phosphorylation of both ERK and AKT, but only the inhibition of PI3K-AKT pathway was able to blunt renoprotection afforded by IPC. Since AKT regulates several cellular survival pathways, it is likely that AKT acts as an important mediator of IPC-induced cytoprotection. This concept is supported by recent pharmacological studies, in which inhibition of AKT signaling with PI3K inhibitor wortmannin abrogated cytoprotection induced by single ischemia-reperfusion preconditioning (69).

microRNAs in IPC

microRNAs (miRs) are small noncoding RNA molecules that in their mature form regulate gene expression, primarily through binding to the 3′-untranslated region (UTR) of target mRNAs, leading either to their degradation or to translational inhibition. Maturation and binding to 3′-UTRs involve several enzyme complexes. Dicer cleaves miR precursors and generates a double-stranded RNA product, which then in association with argonaute endonuclease proteins becomes part of the RNA-induced silencing complex (RISC) that directly binds to target RNAs (13). A subset of miRs is hypoxia regulated and is differentially expressed in ischemic tissues, where they modulate cell survival and tissue repair, and either positively or negatively impact the clinical outcome of IRI. In the heart, for example, miR-499 inhibited apoptosis and improved injury in animal models of myocardial infarction (162), while inhibition of miR-92a promoted angiogenesis and functional recovery from limb and myocardial IRI (15). Interestingly, some miRNAs seem to play a dual role in the recovery from IRI, which appears to be cell type dependent. miR-24, for example, inhibits cardiomyocyte apoptosis through the repression of proapoptotic molecule Bcl-2 interacting mediator of cell death (BIM) (130), while it also promotes apoptosis of cardiac endothelial cells (44).

The relevance of miRs to the pathophysiology of renal IRI is underscored by the proximal renal tubule-specific deletion of dicer, which suppressed miR maturation and made kidneys resistant to IRI (165). Several miRs are differentially expressed following warm IRI (50). These include miR-20a, miR-21, miR-192, miR-194, miR-214, and others. Of those, miR-21 is well characterized and has been shown to be required for the late-phase of IPC in the kidney (172) and IPC in the heart (26). miR-21 is expressed in multiple cell types and is hypoxia inducible in a HIF-1-dependent manner (99); its expression increases after IRI (35, 50, 99). miR-21 reduces the expression of programmed cell death 4 (PDCD4), phosphatase and tensin homolog (PTEN), and other proteins, and has been shown to have antiapoptotic properties (35, 50, 98, 99). However, whether miR-21 represents a suitable therapeutic target for the prevention of ischemic injuries is unclear, as it regulates other signaling pathways that may promote inflammation and fibrosis (179). Several reports indicate the involvement of multiple miRs in IPC of other organs (94, 138); to what degree these miRs contribute to renal IPC is currently under investigation.

While their role as single therapeutic agents for the prevention or treatment of IRI is uncertain, it is clear that miRs regulate and fine-tune cellular hypoxia responses. There are numerous examples of miRs that are hypoxia regulated, either positively or negatively. A prototypical example is miR-210, which is HIF inducible and was initially identified in cancer cells (92). More recently, miR-210 was shown to function as a negative regulator of HIF-1α in T cells (161), illustrating that hypoxia- and also non-hypoxia-regulated miRs operate in complex feedback loops that can amplify or suppress HIF-controlled pathways (31) and thus have the potential to modulate IPC. For example, miR-199a is suppressed under hypoxic conditions and modulates HIF-1-induced preconditioning in cardiomyocytes by directly targeting HIF-1α and SIRT1-induced suppression of PHD2 (131).

The Role of Immune Responses in IPC

Immune cells are key players in tissue injury and repair, and recent studies have established their crucial role in renal IPC. Burne-Taney and colleagues (20) transferred immune cells from ischemic and sham-operated mice into T cell-deficient mice, which then underwent IRI. The investigators found that T cell-deficient mice, which received leukocytes from mice with prior exposure to ischemia, had reduced injury compared with mice which received leukocytes from sham-operated animals. Furthermore, experiments using iNOS-deficient leukocytes demonstrated that immune cell-mediated IPC was not dependent on iNOS. Among lymphocytes, regulatory T cells (Tregs) have been identified as mediators of IPC. Tregs are CD4⁺CD25^highFoxp3⁺ lymphocytes that are recruited into areas of inflammation and have immune-suppressive functions (111). Kinsey and colleagues (86) found that an ischemic insult caused a significant increase in CD4⁺CD25^highFoxP3⁺ Tregs in the kidney at day 7 post-IRI, which was furthermore associated with an absolute increase in the number of IL-10-expressing Tregs. Treatment of preconditioned mice with a Treg-depleting anti-CD25 antibody blunted the protective effects of IPC with regard to preserving renal function and morphology and reducing neutrophil infiltration. Conversely, adoptive transfer of Tregs to naive mice before IRI produced cytoprotection and mimicked the protective and anti-inflammatory effects of IPC in the kidney (86). Remarkably, in a follow-up study the same investigators showed that adenosine generation by CD73 and adenosine signaling through ADORA2A were essential in Treg-mediated renoprotection (85). Because CD73 and ADORA2A are both hypoxia regulated (Fig. 1), enhanced adenosine signaling may be central to Treg-mediated IPC. Additional support for a crucial role of immune responses in IPC comes from a study, where systemic CD11c⁺ macrophage/dendritic cell depletion with clodronate diminished renoprotection associated with IPC (27). However, in a separate study a role for infiltrating macrophages could not be established (70), suggesting that the role of macrophages in IPC may depend on the conditions of the experimental strategies used.

While the molecular mechanisms regulating immune cell function in the context of IPC are not well studied, it is likely that the PHD/HIF axis is critically involved, as it controls metabolic and functional adaptation of immune cells to hypoxic microenvironments (122). For example, inflammatory functions and lifespan of neutrophils under hypoxic conditions are promoted by HIF-1-dependent NF-κB activity, and also require PHD3 expression (159, 160), while HIF-2 regulates neutrophilic apoptosis impacting on the resolution of inflammation (155). The impact of HIF-1-dependent metabolic reprogramming on innate immunity (shift toward anaerobic glycolysis) is exemplified by a recent elegant study showing that myeloid HIF-1 is critical in β-glycan-derived trained immunity responses against bacterial sepsis (25). However, whether IPC has the potential to trigger trained immunity via HIF, and to what degree this impacts renoprotection, is not known. It is also unclear whether IPC in the kidney signals to resident immune cells only or whether renal IPC can also signal and condition immune cells in remote locations such as the bone marrow or spleen. Nevertheless, genetic deletion of Hif1a in myeloid cells prevented functional recovery after renal IRI (141), a finding, which supports the notion that myeloid cell-derived hypoxia responses may have significant positive impact on injury outcomes in the setting of IPC. Accordingly, a protective role for myeloid HIF-1 was shown in a model of renal fibrosis (87). The differential effects of HIF homologs on immune cell function add additional complexity to understanding IPC-induced renoprotection. For instance, in response to different cytokines, HIF-1 and HIF-2 act antagonistically in the polarization of macrophages via the induction of either iNOS or arginase 1 expression, respectively, regulating NO levels differentially (153). With regard to adaptive immune responses, HIF controls T cell activation, the function of Tregs, the balance between Tregs and T helper 17 cells, and enhances the effector responses of CD8⁺ T cells via modulating the expression of transcription factors, effector molecules, and costimulatory receptors (34, 122). Furthermore, both B cell development and hypoxia-induced cell cycle arrest require intact HIF-1 signaling (49, 88). Given the increasing evidence that hypoxic signaling plays a pivotal role in the regulation of immune cell function (122), additional studies are needed that investigate the contribution of innate and adaptive immune responses to IPC.

Conclusions

Recent advances in understanding the molecular and cellular basis of renal hypoxia responses have identified the PHD/HIF axis as a pivotal oxygen-sensing mechanism that mediates IPC through the coordinated activation of renoprotective signaling pathways. Preischemic activation of these pathways through pharmacological HIF stabilization mimics normal hypoxia responses and would represent a comprehensive physiological approach to maintaining renal tissue homeostasis, protecting from ischemia, and promoting renal tissue repair. Effective HIF-activating compounds are currently being investigated in clinical trials for renal anemia and would be readily available for pilot studies in patients at high risk for development of AKI. However, the physiological consequences of systemic HIF activation in humans are only incompletely understood, and more investigations are needed to establish treatment protocols that permit safe and effective targeting of the HIF oxygen-sensing pathway for therapy of ischemic kidney diseases.

GRANTS

P. P. Kapitsinou is supported by National Institutes of Health (NIH) Grant P20 GM104936 and the Carl Gottschalk Award of the American Society of Nephrology (ASN). V. H. Haase receives support from the Krick-Brooks Chair in Nephrology, from NIH Grants R01-DK080821, R01-DK081646, and R01-DK101791, and from a Department of Veterans Affairs Merit Award (1I01BX002348).

DISCLOSURES

V. H. Haase has received honoraria from AstraZeneca and Daiichi Sankyo, and serves on the Scientific Advisory Board of Akebia Therapeutics, a company that develops prolyl-4-hydroxylase inhibitors for the treatment of renal anemia.

AUTHOR CONTRIBUTIONS

Author contributions: P.P.K. and V.H.H. prepared figures; P.P.K. and V.H.H. drafted manuscript; P.P.K. and V.H.H. edited and revised manuscript; P.P.K. and V.H.H. approved final version of manuscript.