One concept does not fit all: the immune system in different forms of acute kidney injury

Abstract

Renal and immune systems maintain body homoeostasis during physiological fluctuations and following tissue injury. The immune system plays a central role during acute kidney injury (AKI), adapting evolutional systems programmed for host defence and minimizing unnecessary collateral damage. Indeed, depending upon the disease context, the impact of the immune system upon the manifestations and consequences of AKI can be quite different. Here we provide an overview of the known and unknown involvement of the immune system within the wide range of different forms of AKI, to oppose oversimplification and to endorse deeper insights into the pathogenesis of the different diseases causing kidney injury. This approach may help to overcome some of the current hurdles in translational AKI research and the development of specific treatments for the different diseases, all presenting with an acute increase in serum creatinine or decline in urinary output. One concept does not fit all.

INTRODUCTION

The current definition of acute kidney injury (AKI) is useful to predict mortality and the need for dialysis in patients in intensive care units [1]. It also endorses epidemiological studies and increases awareness of kidney disease in other medical disciplines, healthcare providers and the general population [2, 3]. However, the term ‘AKI’ is now also extensively used among clinical and experimental kidney researchers. Kidney meetings hold sessions and journals publish reviews about the ‘pathophysiology of AKI’ and research teams search for new cures for ‘AKI’ [4]. It should be noted that AKI is a clinical syndrome, not a disease. Indeed, the first Kidney Disease: Improving Global Outcomes (KDIGO) recommendation clearly states: ‘2.1.3. The cause of AKI should be determined whenever possible’ [1], because AKI is the functional consequence of numerous different diseases, each of them with its own pathophysiology and potential therapeutic consequences. As such, AKI should be considered as other clinical syndromes such as anaemia, syncope or thrombotic microangiopathy (TMA).

A systematic review of experimental studies on AKI reveals a broad diversity of animal models of AKI, with preferences towards post-ischaemic tubular necrosis (33.0%), sepsis-related AKI (18.0%) and cisplatin- (13.9%) and gentamycin-induced tubular necrosis (5.7%) (own unpublished results). Although quite frequent in clinical settings, models of rhabdomyolysis, haemorrhagic and traumatic shock or calcineurin inhibitor-related AKI are infrequently employed (Table 1). Hence knowledge gaps remain and an update of AKI pathogenesis may enable disease-specific therapies to be identified [5, 6].

Here we discuss the diverse roles of the immune system in diverse diseases clinically all presenting as AKI. This attempt follows the idea that ‘one concept does not fit all’ when talking about the ‘pathophysiology of AKI’. Indeed, the contribution of the immune system is quite diverse and strongly depends on the upstream trigger of AKI. Thus, targeting the immune system is most promising in forms of AKI with aberrant immunity as a central and upstream pathogenic mechanism, while it may be less valuable when inflammation is a secondary or downstream phenomenon.

KEEPING HOMOEOSTASIS, THE IMMUNE SYSTEM OF THE HEALTHY KIDNEY

Like other sterile solid organs, the kidney harbors resident mononuclear phagocytes, namely macrophages and conventional dendritic cells [7]. These immune sentinels exist as developmentally and functionally distinct subsets and form a peritubular network in the interstitial compartment of kidney cortex and medulla (Figure 1) [8]. Surface markers only partially dissect the different lineages, as surface marker expression varies during development and between different activation states [7]. Lineage tracing studies using transgenic reporter mice revealed that during early development yolk sac–derived MHC^neg/F4/80^hi macrophages populate many solid organs and constitute the majority of the resident immune cells thereafter [9, 10]. In the brain, these yolk sac–derived MHC^neg/F4/80^hi cells can self-renew from local progenitors into adults, with minimal contribution of bone marrow progenitors, making them a true tissue-resident population. In the kidney, however, these yolk sac–derived macrophages get replaced at an early age by MHCII⁺/F4/80^hi cells derived from haematopoietic stem cells. Fate mapping further suggests that in adulthood, MHCII⁺/F4/80^hi cells originate from Clec9a⁺ common dendritic cell precursors of the bone marrow, indicating that resident renal MHCII⁺/F4/80^hi cells arise via distinct developmental waves [11]. The diverse kidney dendritic cells also include the main classical dendritic cell types 1 and 2, as well as a CD64-expressing CD11b^hi dendritic cell subtype [12]. If or how this peritubular network of resident myeloid cells contributes to normal kidney physiology, i.e. maintaining homoeostasis, is still unknown. One may speculate that they sense metabolic, toxic or ischaemic stress of the proximal tubule and secrete homoeostatic signals as long as no cell necrosis occurs to minimize inflammatory reactions under physiological conditions. This may apply whenever stimuli transiently challenge the proximal tubule, e.g. after a large meal or during pregnancy. Even persistent exposures to glucose in diabetes or protein in minimal change disease may involve resident immune cells to limit a metabolic stress-related inflammatory response [13]. This may explain why such states of the increased tubular workload are not generally associated with secondary tubulointerstitial disease. This balance, however, may eventually turn into overt inflammation, e.g. once passing the threshold for tubular cell necrosis. Cell necrosis triggers the activation of numerous pattern recognition receptors and drives the auto-amplification loop of necroinflammation, where sterile inflammation and regulated necrosis induce each other [14].

PRE-RENAL AKI

Unlike hepatorenal syndrome, transient renal hypoperfusion due to hypovolaemia or haemorrhagic shock are frequent causes of AKI and activation of the renin–angiotensin system further impairs renal perfusion. Restoration of renal blood flow before ischaemic tubular necrosis occurred can rapidly recover renal function without obvious structural changes to the kidney [15]. The contribution, if any, of the immune system, for example, on the resident peritubular dendritic cells, to this prevalent but understudied form of AKI is unknown.

INTRARENAL AKI

Necroinflammation drives AKI in ischaemic and toxic acute tubular necrosis

At some point, hypoperfusion of the kidney causes ischaemic necrosis of those parts of the kidney where oxygen demand and oxygen delivery are most critical, i.e. the proximal tubule’s S3 segments in the outer stripe of the outer medulla. Necrotic tubular cells release numerous danger-associated molecular patterns that activate pattern recognition receptors of the innate immune system, such as Toll-like receptors (TLRs) and inflammasomes not only in adjacent parenchymal cells but also in resident immune cells [14]. Upon activation, especially the latter secrete large amounts of IL-1β and other pro-inflammatory cytokines and chemokines that initiate local inflammation, i.e. vascular permeability, interstitial oedema and the recruitment of neutrophils and other bone marrow–derived immune cells [3]. This inflammatory response, in turn, accelerates parenchymal necrosis, either via outside-in cytokine signals or by facilitating intrinsic pathways of regulated necrosis [16, 17]. The release of extracellular histones takes a central role in this process, as histones elicit direct cytotoxicity on renal parenchymal cells and also activate TLR2/-4 and the NLRP3 inflammasome in immune cells [18–20]. The self-perpetuating amplification loop of tubular necrosis and inflammation, i.e. necroinflammation, can lead to substantial renal cell loss, as becomes clinically obvious by cellular and granular casts in the urine (Figure 2). Casts of necrotic cell debris themselves contribute to AKI by obstructing tubules and by maintaining necroinflammation [21]. Interestingly, the apoptosis inhibitor of macrophage protein on such intraluminal cell debris interacts with kidney injury molecule-1 on remnant tubular epithelial cells for the phaygocytic clearance of such cell debris [21]. While neutrophils are the predominant bone marrow–derived immune cell type infiltrating the kidney in the early phase of necroinflammation, other leucocytes such as other granulocytes and pro-inflammatory CC-Chemokine Receptor -2 (CCR2⁺) macrophages are indicators of the late injury phase that can last for days or weeks in rodent ischaemia–reperfusion injury of the kidney [4]. Abrogating inflammatory signals in this process such as pattern recognition receptors, cytokines or immune cells attenuate AKI in animal models to the same extent as does interfering with cell necrosis itself [4, 14, 17, 20]. It should be noted that endogenous negative regulators of innate immunity limit necroinflammation at all levels of the signalling cascade [22–26]. Translation of this finding to human AKI remains difficult, as necroinflammation is an early event and the window of opportunity for therapeutic intervention might be clinically irrelevant [5, 6, 27].

Necroinflammation develops in a similar manner also in toxic and crystalline forms of tubular necrosis, although, depending on the type of toxin, the S1/2 segments of the proximal tubule, the thick ascending limb of the loop of Henle or the distal tubule may be the primary site of injury [28].

Both types of AKI elicit different dynamics of necroinflammation depending on the severity/dose as well as the length of exposure to ischaemia or toxin, respectively.

Immunothrombosis drives AKI in TMA and kidney infarction

Thrombotic arterial occlusions involve innate immune mechanisms, i.e. immunothrombosis [29], causing regional ischaemia and infarction. For example, in TMA, a heterogeneous group of diseases share endothelial cell injury in kidney arteries, arterioles and capillaries as an upstream mechanism of ischaemic kidney injury. Activation of the soluble and membrane-bound complement factors is central, especially in genetic forms of atypical haemolytic-uraemic syndrome (aHUS) and specifically involves the alternative pathway of C3 convertase activation (Figure 2) [30]. Although, many different immune cell subsets, cytokines, chemokines and other pro-inflammatory mediators are involved in post-ischaemic AKI [27], the dominant upstream pathogenic mechanism in aHUS is uncontrolled complement activation in the vascular compartment. Hence eculizumab, a monoclonal antibody that inhibits the cleavage of complement factor C5 into C5a, is particularly effective in aHUS-related AKI [31]. Also, Shiga toxin-producing Escherichia coli–related HUS or A disintegrin and metalloprotease with thrombospondin-1-like domains-13 deficiency-related AKI involve activation of the alternative complement pathway during microvascular injury and may in some circumstances benefit from complement inhibition [30].

Cholesterol crystal embolism is a complication of advanced atherosclerosis and is triggered by spontaneous or procedure-related ruptures of atheromatous plaques in the aorta [32]. Crystal embolism into intrarenal arteries and arterioles causes endothelial cell injury and microvascular thrombosis followed by territorial ischaemic infarction of the kidney and AKI [32, 33]. Such kidney infarcts are associated with strong perilesional inflammation, but only targeting the crystal clot, e.g. with anticoagulants, fibrinolytics or recombinant DNAse I, can prevent kidney infarction and AKI [34]. Complement activation may also contribute to crystal embolism, but this has not yet been proven experimentally.

Cytokine storm, neutrophil extracellular traps, histones and adenosine triphosphate (ATP) drive AKI in sepsis

The systemic inflammatory response in sepsis, referred to as ‘cytokine storm’, induces diffuse vasodilation and hence renal hypoperfusion, i.e. septic shock [35]. Hence the pathogenesis of sepsis-related AKI involves pre-renal AKI to a significant degree. Nevertheless, intrarenal mechanisms and injury also contribute [36]. Bacterial endotoxin and other bacterial components activating TLRs on endothelial cells trigger endothelial dysfunction, i.e. a pro-thrombotic state, as well as vascular leakage into the interstitial space causing diffuse interstitial oedema often requiring mechanical ventilation. Peritubular interstitial oedema is a contributing factor to AKI and promotes the exposure of resident dendritic cells to bacterial products, which induces intrarenal cytokine signalling, leucocyte recruitment and tubular injury (Figure 2). In addition, histones, either released from neutrophil extracellular traps (NETs) in the bloodstream or from dying renal parenchymal cells, have direct cytotoxic and pro-inflammatory effects and drive necroinflammation inside the kidney [18, 19]. Recombinant alkaline phosphatase injections can improve mortality after AKI in sepsis [37], which is thought to act by neutralizing the NLRP3 inflammasome agonist extracellular ATP [38]. The upstream triggers, the infection and hypovolaemia are the main therapeutic targets in this setting. Attempts to clear the circulation of pro-inflammatory cytokines have shown promising results in animal models and small patient series, but not yet in randomized clinical trials [39].

Host defence, autoimmunity or alloimmunity drive AKI in the various forms of acute interstitial nephritis (AIN)

When infections affect the kidney directly, e.g. in bacterial, viral or fungal nephritis, numerous mechanisms of host defence contribute to collateral kidney injury. Resident immune cells sense pathogens and release chemokines to attract professional phagocytes into the kidney. Intact neutrophils and other granulocytes release reactive oxygen species and, upon forming NETs, cytotoxic histones into the extracellular space. In viral nephritis, viral nucleic acids activate nucleic acid–specific TLRs in resident dendritic cells and parenchymal cells, which release Type I interferons to set off an antiviral immune response [40]. In infections as well as in alloimmunity, foreign antigens are processed by resident dendritic cells that migrate to regional lymph nodes to activate antigen-specific T helper cells via major histocompatibility complex (MHC) II–T-cell receptor interaction, while cytotoxic T cells recognize target cells by their surface expression of the foreign antigens via MHC-I. As such, infections and alloimmunity in kidney allograft recipients involve all aspects of innate and adaptive immunity inside and outside the kidney. Therefore AKI in this setting includes acute cellular allograft rejection.

Similar mechanisms apply for AIN triggered by certain drugs that act as haptens, are handled such as foreign antigens and induce delayed-type hypersensitivity reactions. In contrast, the AIN associated with checkpoint inhibitors, a new class of agents overcoming the capacity of certain cancers to suppress anti-cancer immunity, develops from bypassing immune tolerance to kidney tissue. As a result, interstitial nephritis caused by alloimmunity and autoimmunity is characterized by strong and diffuse immune cell infiltrates into the renal interstitium and immune attack to parenchymal cells (Figure 2). While interstitial nephritis triggered by certain agents or drugs responds to cessation of the specific exposure, the other form can usually be well controlled by steroids or other global immunosuppressants that equally suppress innate and adaptive immunity inside and outside the kidney.

Immune complexes, complement and NETs drive glomerular AKI

Glomerular disorders can also cause AKI, e.g. rarely in anti-phospholipid antibody syndrome and more frequently in necrotizing or crescentic glomerulonephritis. The latter two involve either immune complex– or anti-neutrophilic cytoplasmic autoantibody–related microvascular injury [41], hence B cell–targeting therapies are effective (Figure 2) [42]. Glomerular injury involves neutrophil necrosis followed by the release of NETs and histones, which both elicit direct cytotoxic effects and support microvascular injury, and plasma leakage into Bowman’s space driving parietal epithelial cell hyperplasia, i.e. crescent formation [43]. Complement activation contributes to all forms of rapidly progressive glomerulonephritis and has become a promising target for therapeutic intervention [44]. The recruitment of monocyte/macrophages, mast cells and T cells are subsequent events that may contribute to long-term outcomes. Specific interventions are currently limited to the use of calcineurin inhibitors that inhibit T cell–mediated tissue injury and inflammation.

POST-RENAL AKI

Urine outflow obstruction can occur at every level of the urinary tract from the papilla to the urethra; only obstruction of both kidneys (when two functional kidneys are present) can cause AKI. Obstruction immediately increases interstitial pressure, which reduces the pressure gradient across the filtration barrier and hence reduces the glomerular filtration rate (Figure 2). Therefore the early phase of post-renal AKI is merely pressure gradient related and goes initially without structural changes. However, when obstruction persists, tubule degeneration starts from the glomerular–tubular junction and leads to diffuse tubular atrophy, interstitial fibrosis and persistence of atubular glomeruli [45]. This process is associated with a mild activation of innate immunity and recruitment of immune cells with predominately anti-inflammatory and pro-fibrotic phenotypes that contribute to the clearance of the atrophic tubules and a secondary wound-healing response in the interstitial compartment, a hallmark of AKI-related irreversible nephron loss or nephron loss in chronic kidney diseases (Figure 2). Sensing pressure directly via the mechanosensor PIEZO1 can drive innate immune cells to mount inflammatory responses [46], but whether that applies also to increase interstitial pressure inside the kidney is not yet known. However, targeting this immune response at a later stage is unable to sustain functional nephrons and hence improve renal function even when reducing interstitial fibrosis [47].

THE IMMUNE SYSTEM MODULATES FUNCTIONAL AND STRUCTURAL RECOVERY FROM AKI

Functional recovery after AKI involves diverse mechanisms including cast removal, vascular healing and endocycle-driven hypertrophy of remnant epithelial cells in injured as well as non-injured nephrons facing an increased workload substituting for those nephrons irreversibly lost during AKI [48]. In contrast, kidney regeneration, previously thought to be a widespread phenomenon, was recently found to be quite scarce and is mediated mainly by scattered tubular progenitors that can recover a few necrotic S3 tubule segments [48]. The activation state of peritubular macrophages has an important role in regulating this healing response [49]. Pro-inflammatory macrophage phenotypes suppress repair and promote further injury, while anti-inflammatory macrophage phenotypes endorse function recovery and regaining epithelial integrity [49]. Interestingly, tubular epithelial cells themselves induce such help from interstitial myeloid cells, e.g. by releasing TLR4 agonists that induce resident myeloid cells to secrete interleukin (IL)-22 that in turn endorse tubular integrity via the IL-22 receptor on tubular epithelial cells [50]. Thus the immune system plays an integral role not only in the injury phase of AKI, but also in the phase of function recovery, compensatory structural adaptation and nephron regeneration (Figure 3).

CONCLUSIONS AND FUTURE PERSPECTIVES

AKI is a clinical syndrome with a wide range of underlying disorders that is essential to acknowledge in experimental and clinical research aiming to understand the pathophysiology and finding cures for ‘AKI’. Only targeting the specific upstream trigger of each different form of AKI will likely have relevant effects. This is obvious from the many forms of AKI that can already be successfully treated, namely those with persistent triggers that can be controlled by, for example, complement inhibitors, immunosuppressants or simply by cessation of a harmful medication. Indeed, many episodes of mild pre-renal ischaemic AKI due to volume depletion progression to severe AKI can and is frequently successfully prevented by volume or vasopressor therapy. What remains a concern are those severe forms of ischaemic, toxic, septic or multi-causal AKI associated with late diagnosis, severe illness and complex comorbidities. In this setting, patient management is focused on overall survival and long-term outcomes, i.e. irreversible nephron loss. Thus, interfering with acute necroinflammation to limit nephron loss and specifically promoting kidney regeneration by activating those cells that contribute to re-epithelialization remain two possibilities. Pre-clinical studies convincingly indicate that this could be possible, but validation in humans is pending. A series of barriers for translating experimental insights into human cures remain [5, 6, 51]. The following aspects may be also considered (Box 1):

• early necroinflammation that can be targeted by either anti-inflammatory drugs, cell death inhibitors or both, most likely only in a preventive manner as the current diagnostic criteria for AKI do not match the window-of-opportunity for these drugs;

• suppressing persistent inflammation to facilitate the intrinsic capacity for the regeneration of lost tubular cells from local epithelial progenitor cells. If inflammation also modulates compensatory endocycle-driven hypertrophy of the remnant tubules is currently unknown. However, this mechanism mostly supports function recovery after AKI but not tissue regeneration; and

• if specific cytokine therapy, e.g. recombinant IL-22 or others could augment this process in a favorable manner is an attractive concept, but remains to be proven with clonal lineage tracing of renal progenitors.

• finally, understanding the functional contribution of resident dendritic cells to normal renal function is pending. Understanding their contribution to the kidneys role in keeping homoeostasis may also help to better define the signals that it needs to recover after injury.

Medical concepts develop in cycles of simplification and diversification. The current definition of ‘AKI’ has become extremely instrumental for many aspects of acute renal failure, but it hinders others. AKI is not an untreatable disease. AKI is a clinical syndrome caused by numerous different diseases with distinct pathophysiologies. Many of these can be treated with specific drugs, which is important to point out to withstand the therapeutic nihilism in ‘AKI’. Therapeutic nihilism is not even appropriate in the more prevalent forms of ischaemic and toxic tubular necrosis, because fluid therapy and early removal of causative agents very frequently prevent acceleration of necroinflammation-related kidney necrosis, conceptually like a catheter intervention for coronary thrombosis.

To conclude on the contribution of the immune system to AKI: resident and infiltrating immune cells contribute to all phases of AKI, mostly acting as amplifiers of spatial and temporal micromillieus. In a healthy kidney, they probably support the maintenance of homoeostasis. In the injury phase of tubular necrosis, they endorse necroinflammation as an erroneous attempt for host defence. In the resolution phase, they endorse the termination of necroinflammation to launch healing. In the repair phase, they clear dead cells and secrete pro-regeneratory and pro-fibrotic signals to augment and stabilize the remaining parenchyma. As such the contribution of the immune system is complex and context dependent. Therefore the use of global immunosuppressants may preferentially serve autoinflammatory, autoimmune or alloimmune forms of AKI, but not AKI in general. Dissecting the different causes and individualizing therapy in terms of individual cause and phase of AKI remains the challenge for the future.

FUNDING

H.J.A. was supported by the Deutsche Forschungsgemeinschaft (AN372/14-3, 16-2, 23-1, 24-1, 27-1). B.U.S. was supported by the Deutsche Forschungsgemeinschaft (Schr 1444/1-1; Projektnummer 360372040–SFB 1335-project 8) and European Research Council (ERC-2016-STG-715182).

CONFLICT OF INTEREST STATEMENT

All the authors declare no competing interests.

REFERENCES