Ischaemia reperfusion injury: mechanisms of progression to chronic graft dysfunction

An increasing number of clinical epidemiology studies have reported an association between AKI and the development of CKD [1, 2]. Despite the well-established epidemiological link between AKI and CKD, evidence to support a causative relationship between the two clinical entities is still lacking. Transplant patients serve as an ideal cohort to investigate the link between acute ischaemic injury and development of late organ dysfunction. Ischaemic injury in kidney transplantation initially manifests as delayed graft function (DGF), which is associated with prolonged hospitalisation and the need of renal replacement therapy post-transplant. Nevertheless, recovery of organ function is usually achieved in patients with DGF, indicating a degree of resolution of acute ischaemic injury. However, several studies have reported that patients with DGF have increased risk of acute rejection and poorer long-term renal function [3]. These findings provide a link between acute ischaemic injury to the kidney and long-term deterioration in graft function, despite relatively “normal” function in the earlier stages. Several mechanisms have been proposed in the development of chronic dysfunction, as illustrated in Fig. 1. It is important to consider these mechanisms not as separate pathophysiologic entities, rather as components of an intricate network with many overlapping, co-existing pathways.

The anatomy of the nephron and renal microcirculation plays a crucial role in understanding the effect of ischaemia on the kidney. In physiological conditions, kidneys receive approximately 20% of cardiac output. This blood flow is primarily channelled to the cortex with blood flow to the medulla predominantly from the vasa recta, a continuation of efferent arterioles of the juxtamedullary glomeruli. In comparison to other parts of the body, the kidney, particularly the outer medulla, operates at lower oxygen tensions, both during normoxia and hypoxia [4]. This is due to post-glomerular arterio-venous shunting and high oxygen demands [5]. In response to hypoxia, kidney blood flow may alter dramatically, especially to the outer medullary region, reducing oxygen delivery and increasing hypoxic injury in this region.

The capacity of a kidney to withstand an ischaemic injury or to undergo repair after an ischaemic injury is highly dependent upon the available nephron mass, pre-existing glomerulosclerosis and arteriosclerosis [6, 7], features typically associated with increasing donor age. In experimental models of transplantation, older kidneys were more susceptible to ischaemic injury even after a brief ischaemia time [6]. A retrospective analysis of the Australian and New Zealand transplant registry by Wong et al. [8] also suggests a significant interaction between total ischaemic time, donor age and graft function, with higher DGF rates in recipients receiving kidneys from older donors.

The mechanism by which a further reduction in oxygen delivery to the kidney induces AKI has been well established. However, demonstrating the role of hypoxia in progression of kidney injury in the longer-term is more challenging. Several studies have shown an association between chronic tubulointerstitial hypoxia, oxidative stress and chronic inflammation, and that these factors are involved in the progression of CKD [9, 10].

Cellular adaptation to hypoxia is largely regulated by a heterodimeric transcription factor, hypoxia-inducible factor (HIF), which consists of two sub-units; α and β. Whereas the β sub-unit is relatively insensitive to alteration in oxygen levels, the level of the alpha sub-unit (HIF-α) is highly dependent on cellular oxygen tension [11, 12]. HIF-α has two major isoforms; HIF-1α, HIF-2α and one additional minor isoform HIF-3α [11, 13]. HIF-α is expressed at a basal level in cells but in normoxic conditions HIF-α is hydroxylated by HIF-prolyl hydroxylase (PHD) leading to ubiquitination and degradation. In order to perform their catalytic function, HIF-prolyl hydroxylases require O₂, Fe and 2-oxoglutarate (an intermediate in the Tri-carboxylic acid cycle) [14]. In normoxic conditions, degradation occurs very rapidly, resulting in very short half-life of HIF-α, making it almost undetectable in healthy cells. During hypoxia, HIF-propyl hydroxylase function is reduced, leading to accumulation of HIF-α in the cytoplasm. HIF-α forms a complex with the constitutively expressed HIF-β, leading to nuclear translocation and binding of the complex to hypoxia-response elements (HRE), initiating gene transcription (see Fig. 2).

Conde et al. found that HIF-1α is expressed during ischaemia, but disappears 24 h after hypoxia is reversed, then reappears in late reperfusion [15], suggesting recurrence of tissue hypoxia during the cell regenerative phase [16]. In human kidney allograft biopsies, upregulation of HIF-1α is detected immediately after engraftment, at 10–14 days post procedure, but not after 3 months [17]. These findings suggest that renal hypoxia occurs not only during the acute phase of ischaemic injury but also during the recovery phase, presumably due to the activity of oxygen-consuming regenerative processes.

Current evidence is conflicting regarding the role of HIF in progression of CKD. Experiments using cobalt chloride (CoCl₂) and dimethyloxalyglycine (DMGO) to stabilise HIF and increase expression of HIF target genes in an ablation/infarction mouse model showed up-regulation of vascular endothelial growth factor (VEGF), glucose transporter 1 (GLUT1) and cell proliferation, indicating a reno-protective effect of HIF [18]. Another study in mice assessed the relationship between HIF-1, ischaemic acute kidney injury (AKI) and the development of fibrosis by increasing HIF level using pre-ischaemic pharmacological inhibition of HIF-propyl hydroxylases. Increased HIF level was associated with reduced fibrosis and less alpha-smooth muscle actin (α-SMA) expression at 21-day post ischaemia reperfusion injury (IRI) [19]. This anti-fibrotic effect was not observed when HIF-propyl hydroxylase inhibition was given after IRI. In addition, Kobayashi et al. documented reduced fibrosis in mice subjected to unilateral ureteral obstruction (UUO) and global activation of HIF [20].

In contrast to these observations, there is a significant body of evidence to support a pro-fibrotic role of HIF. Wang et al. showed that silencing HIF-1α in a rat model of chronic renal ischaemia decreased collagen and α-SMA induction [21]. Moreover, HIF-1α knockout prevented transforming growth factor-β1 (TGF-β1) induced epithelial-to-mesenchymal transition (EMT) in mouse proximal tubular epithelial cells following unilateral ureteral obstruction (UUO) [22]. Our unpublished data shows that hypoxia or stabilisation of HIF-1α is associated with inhibition of SMAD7 and increased in SMAD3 activity, also suggesting a pro-fibrotic role of HIF. Post-ischaemic renal fibrosis can occur through several mechanisms, including direct transcriptional regulation of pro-fibrotic genes, epithelial-to-mesenchymal transition (EMT) and induction of epigenetic changes. All of these mechanisms can be driven by HIF activation, either through direct regulation or indirectly, involving crosstalk with multiple signalling pathways [23, 24], as discussed in the subsequent sections.

HIF binding to HREs in gene promotor regions allows direct transcriptional regulation of many genes essential in promoting renal fibrosis. As examples, in a mouse glomerulosclerosis model, Baumann et al. has shown HIF-1α binding to HREs in the promotor region of the collagen type-1 alpha 2 chain (COL1A2) gene [25]. Nuclear accumulation of HIF-1α induced by ischaemia was shown to promote extracellular matrix deposition by renal proximal tubular epithelial cells by direct regulation of plasminogen activator inhibitor-1 (PAI-1) [26].

Exposure of human renal fibroblasts to hypoxia in vitro led to significant increase in Col-1 and tissue inhibitor of metalloproteinase (TIMP)-1 production, accompanied with decreased expression of collagenase [27]. Although hypoxia alone induced TGF-β1, the introduction of an inhibitory anti-TGF-β1 antibody had no effect in preventing hypoxia-induced Col-1 and TIMP-1 mRNA expression, suggesting a direct effect of HIF-1α on gene transcription [27]. In addition, a study on human skin fibroblasts demonstrated the contribution of HIF-1α to the development of post-ischaemic fibrosis by directly regulating pro-Collagen prolyl (P4HA1 and P4HA2) and lysyl (PLOD2) genes essential for collagen deposition, extracellular matrix stiffening and collagen fibre alignment (see Fig. 3) [28].

TGF-β is one of the main mediators of renal fibrosis. A relationship between HIF and TGF- β was shown by the increasing level of TGF- β1 produced by tubular cells in response to hypoxia [29]. HIF has been shown to directly regulate the transcription of many genes involved in the TGF- β pathway. In a glomerulosclerosis model, HIF-1α was shown to form a complex with phophoSMAD3 at the COL1A2 promoter, inducing Col-1 synthesis [25]. Similar synergistic interaction of HIF and TGF-β1/SMAD3 signalling has also been reported in the regulation of VEGF, endothelin and erythropoietin expression [24]. Furthermore, accumulation of HIF in the endothelial cells of endothelial specific propyl hydroxylase-2 knockout mice upregulated TGF-β1 expression, resulting in significantly poorer renal function [30]. In a study using a renal tubular cell line in which transcription of HIF-α was absent, there was a decrease in basal and TGF-β1 stimulated Col-1 expression [31]. Altogether, these findings highlight the close interaction between HIF and TGF-β pathway.

Despite being implicated in promoting fibrosis, TGF-β has also been shown to have an immunosuppressive effect, which may be beneficial in a kidney transplant setting. Yoshimura et al. proposed several mechanisms by which TGF-β inhibits the immune response [32]. These mechanisms include suppression of helper T cell differentiation, suppression of T cell activation and proliferation, suppression in macrophages, dendritic cells (DCs) and natural killer (NC) cells, conversion of naive T cells to regulatory T cells, and inhibition of cytokine production (IL-2, IL-4). The effect of the interaction between HIF and TGF-β pathway on the recipients’ immune status, and how this may affect long-term allograft function is not known. However, profiling of biopsy tissues from chronic allograft nephropathy (CAN) patients showed significantly higher expression of both HIF-1α and TGF-β1 compared to the group without CAN [33].

Post-ischaemic induction of HIF also plays a pro-fibrotic role by indirect crosstalk with other pro-fibrotic pathways including Notch, NF-κB and PI3K/Akt pathways [23]. The interaction between HIF and these other pathways is complex (see Fig. 3). HIF can increase the transcription of proteins involved in these other pathways [30, 34]. HIF can also interact with other transcription factors, augmenting their transcriptional activity [35]. In addition, these other pathways can increase HIF gene expression, creating a positive feedback loop [35]. Recently, a transcription factor CCAAT/enhancer-binding protein δ (CEBPD) was discovered as a potential link between hypoxia and inflammation. CEBPD is known to be rapidly induced by inflammatory cytokines, such as IL-1β, in a NF-κB dependent manner. In both an acute and chronic murine model of renal hypoxia, CEBPD was induced in the nuclei of tubular epithelial cells and by direct promoter binding increased HIF transcription and activity [36].

Neutrophil adhesion to injured endothelial cells is a rapid and important component in the initiation of damage in the ischaemic kidney. Neutrophils have been shown to migrate into the transplanted organ within 6 h of reperfusion and are attracted by a set of chemokines, including CXCL8 (IL-8), CXCL10, RANTES, IL-17 and MCP-I [50‐53]. Damaged cells will be killed by neutrophils by direct phagocytosis or degranulation, releasing proteases, myeloperoxidase, nitrogen species, antimicrobial peptides and cytokines, which further contribute to the generation of ROS [54]. Recruitment of neutrophils also involves endothelial cell expression ICAM1, E and P selectin, which cross-talk with integrins and L-selectin on neutrophils [51, 55, 56]. Inhibiting the accumulation of neutrophils in the kidney may prevent acute kidney injury [53, 56‐58]. In contrast, other studies have failed to reproduce beneficial effects of neutrophil depletion and suggest neutrophil independent mechanism in the pathophysiology of acute tubular injury [59, 60]. Nevertheless, the majority of evidence supports a role for neutrophils in the development of post-ischaemic injury, by mechanisms including obstruction of renal microvasculature and release of free radicals and proteases [61].

Several studies have observed a decrease in IRI severity after macrophages depletion prior to injury [62, 63] indicting a role for macrophages in promoting tubular injury during the initial phase of IRI. However, suppressing macrophage function during the repair process has been shown to suppress tubular proliferation, thus impairing the normal recovery process. Therefore, the role of macrophages in renal response to IRI is complex. Pre-clinical studies have described the involvement of macrophages in the early inflammatory response, during cellular regeneration and tissue repair as well as during the development of fibrosis. These diverse roles are played by different sub-types of macrophages based on their activation and functional states. Classical activation of macrophages typically involves interferon gamma (IFNγ). Ischaemia-induced cellular injury also produces danger-associated molecular patterns (DAMPs), which will be recognised by pattern recognition receptors (PRRs) and contribute to classical macrophage activation. These classically activated M1 macrophages are pro-inflammatory and associated with tissue damage. However, they also play an important role in clearing apoptotic cells and debris, thereby initiating repair process [64]. Alternatively activated macrophages include M2a macrophages, which are responsible for wound healing and M2b macrophages, also known as immunoregulatory macrophages. M2a macrophages are activated through IL-4/IL-13 binding to IL-4 receptor, which leads to production of growth factors, collagen precursor synthesis and generation of extracellular matrix. M2b macrophages regulate inflammatory response by producing of immunosuppressive cytokines, IL-10 and TGF-β. Production of TGF-β limits inflammation, but at the same time may contribute to activation of pro-fibrotic pathways. When injury persists, chemokines, macrophage colony-stimulating factors (M-CSF) and IL-34 are secreted to sustain recruitment and retention of macrophages [65]. Blockade of the M-CSF receptor has a protective effect following experimental transplantation [66]. Retention of M2b macrophages in the injured tissue will produce macrophage-derived factors, which subsequently activate and support myofibroblasts, inducing extracellular matrix deposition and fibrosis. The signals responsible for retaining pro-fibrotic macrophages in the kidney remain unclear, but studies using unilateral ureteral obstructive (UUO) rodent model suggest a role for the chemokine receptors CCR1, CCR2, CX3CR1 [65].

The complement system has been well identified as an important, early mediator of the post-ischaemic inflammatory response [67, 68]. Activation of complement is an important factor in the progression of renal disease, and targeting complement has been an attractive therapeutic option due to its involvement in both innate and adaptive immune response to IRI [69, 70]. Ischaemic insult to the kidney has been shown to involve the anaphylotoxins, C3a and C5a, acting through their respective receptors (C3aR and C5aR). Stimulation of C3a/C5a receptors during IRI was shown to increase pro-inflammatory cytokine/chemokine production and tubular injury. The membrane attack complex C5b-9 has also been shown to contribute to the progression of renal damage [71]. Recently, C-type lectin collectin-11 (CL-11/Colec11) was discovered as an activator of mannan-binding lectin (MBL) pathway in the kidney in response to ischaemia. CL-11 acts by recognising L-fucose on kidney tubules following ischaemia [72]. The study showed that global or kidney-specific deficiency of CL-11 reduces post-ischaemic tubular injury and functional loss [72].

Substantial evidence is available to link natural killer (NK) and NKT cells, renal dendritic cells (DCs), T cells and B cells to early IRI, mainly linking their actions to direct targeting of injured tubular and endothelial cells, activation of neutrophils and macrophages, and secretion of pro-inflammatory cytokines, such as IFN-γ, TNF-α, IL-4 and IL-10. If NK and NKT cell function is reduced, the severity of renal injury following IRI is also reduced [73‐75]. DCs contribute to post-ischaemic kidney injury by the secretion of TNF [76]. Biopsies taken from DGF patients suggested an association between DGF and acute rejection due to an imbalance between myeloid DCs (involved in graft rejection) and plasmacytoid DCs (which may play role in graft tolerance) [77].

Although initially regarded as by-standers, current evidence suggests an active role for T cells in the pathogenesis of IRI. Depletion of CD4 and CD8 T cells in murine IRI has been shown to improve renal function and reduce neutrophil infiltration and tubular atrophy [78]. Furthermore, reconstitution of T cells in T cell-deficient mice restores injury to the level seen in normal mice [79]. Depletion of either αβ or γδ T-cells in mice was associated with reduction in renal injury [80]. Nevertheless, a subset of T cells is also recognised to play a role in preventing injury and promoting repair. Pre- and post-ischaemic adoptive transfer of regulatory T cells has been shown to protect the kidney from ischaemic injury, reduce TNF-α and IFN-γ production and accelerate repair [81, 82].

B cells are involved in the adaptive immune response to IRI. To date, studies have shown a predominantly harmful effect of post-ischaemic B cell activation [83]. Depleting B cell in mice was shown to improve renal function and reduce tubular injury after ischaemia [84]. A study by Jang et al. documented that B cell-deficient mice subjected to ischaemia showed more tubular proliferation, less tubular atrophy and higher expression of IL-10 and VEGF [85]. Adoptive transfer of B cells into these mice blocked this effect, suggesting that B cells may interfere with post-ischaemic repair processes. In contrast, another study found worse post-ischaemic renal injury in mice lacking all mature B cells [86], suggesting a more complex and divergent role for B cells in the progression of IRI.