Molecular differences in susceptibility of the kidney to sepsis-induced kidney injury

Sepsis is the most common cause of acute kidney injury (AKI) [1]. The expanding population of patients with sepsis and AKI, and the associated excess in mortality provide strong basis for basic research aimed at elucidating mechanisms underlying their complex pathophysiology. Both hemodynamic and non-hemodynamic pathogenic pathways have been shown to contribute to the development of septic AKI (S-AKI) [2, 3]. However, a more comprehensive understanding of the renal pathobiology in sepsis requires technologies of molecular biology. Unfortunately, several factors make the human research into cellular and molecular biology in critically ill patients problematic. In particular, difficult access to the human renal tissue and several confounding factors including disease state, the patient’s comorbidities and treatment history represent main impediments to the study of human S-AKI. Therefore, the use of clinically and biologically relevant animal models remains a cornerstone in the sepsis and AKI research. We have established large animal models of sepsis that allowed us to generate two distinct groups of septic animals, those with and without AKI [3]. The separation of animals to those with and without S-AKI is a unique tool to isolate and study factors discriminating AKI from non-AKI. Recently, we reported that early abnormal host response coupled with subsequent renal vasoconstriction despite apparent sepsis-induced systemic vasodilatation were major discriminating factors associated with the development of S-AKI in porcine models [3]. Herein, we extend our previous work and utilize genomic tools to identify differences in renal molecular patterns between those two distinct groups. Rather than utilizing genome-wide approach, we prospectively selected subset of genes most likely to be implicated in the key pathogenic pathways. These can be broadly clustered into the following major functional categories – inflammation/danger response (Toll like receptor 2, Toll like receptor 4, Tumor necrosis factor alpha), mitochondrial energy metabolism (Peroxisome Proliferator-Activated Receptor Gamma, Coactivator 1 Alpha); cellular stress response and vasoregulation (Heat shock protein HSP 90, Heat shock protein HSP 70; Heat shock protein HSP 27, Heat shock cognate 70 kDa protein, Hypoxia inducible factor 1, Inducible nitric oxide synthase, Cyclooxygenase-2, Angiotensin II Receptor, Type 2), and apoptosis (BCL2-Associated X Protein, Caspase 3).

From the set of experimental animals reported in the previous study [3], we have investigated 15 pigs, in which kidney tissue samples were available for further analyses: five sham- operated, time-, age- and weight-matched controls (control group), five septic animals without AKI (non-AKI group) and five animals developing septic AKI (AKI group). Thus, the present approach is distinct, but complementary to our previous work [3].

The relevant question of this exploratory study was whether renal tissue gene expression pattern in animals that develop septic AKI differs from pattern seen in septic subjects without AKI. We combined gene expression analysis with a classic large mammal model of sepsis. Since this modeling allowed us to generate septic animals both with and without AKI, the confounding effects of sepsis that are independent of AKI could be isolated and studied. Hence, a unique feature of this study is that we have not compared only AKI animals and control, healthy animals, as in previous studies in rodent models [4, 5], but extended this analysis to compare the gene expression in septic animals with and without AKI. The facts that AKI is affecting 40–50% of all septic patients [6] and that differences between septic patients with and without AKI are only poorly understood lend strong relevance and usefulness to our approach. In addition, by coupling this approach with traditional parameters of renal physiology, our study helped establish a picture of molecular-based differences between septic subjects with and without AKI. Our preliminary results show that the vulnerability of kidneys to develop AKI in sepsis is associated with early, exaggerated systemic inflammatory response coupled with different trajectories in tissue expression profile of genes representative of inflammatory, metabolic and apoptotic responses compared to AKI free septic animals.

In our study, septic animals that developed AKI had systemic hemodynamic phenotype (normo- or hyperdynamic) comparable to non-AKI subjects, but demonstrated higher systemic inflammatory response, an increase in renal vascular resistance and early fall in cortical microcirculatory blood flow. Our results are in excellent agreement with those from observational clinical studies [7‐10]. Indeed, Murugan et al. demonstrated that patients admitted with community acquired pneumonia who presented with or developed AKI had higher plasma levels of IL-6 and TNF-α than non-AKI patients [7]. Likewise, Payen et al. recently showed that the development of septic AKI strongly reflects the intensity of systemic cytokine levels [8]. However, the mechanistic link between systemic inflammation and renal injury remains to be demonstrated. In this context, our study further extends these clinical observations by demonstrating the temporal relationship between exaggerated systemic inflammatory stress, intra-renal hemodynamic alterations and significant changes in kidney gene expression in AKI subjects as opposed to septic subjects without AKI.

Rather than utilizing genome-wide approach, we focused on prospectively selected subset of genes most likely to be implicated in major pathogenic pathways. We observed different intrarenal expression of TLR-4, which was found to be significantly up-regulated in AKI animals only, a finding consistent with previous rodent data [5]. TLR-4 receptors are constitutively expressed at RNA level both on renal endothelial and epithelial tubular cells and represent pattern recognition receptors, which can be activated by both pathogen-associated molecular patterns (PAMPs) and endogenous ligands called damage-associated molecular patterns (DAMPs) [11, 12]. Upon activation by systemic or local danger signals, intrarenal TLR-4 drives local innate immune and inflammatory responses. Our results fit well to an attractive hypothesis that exposure of the endothelial and tubular epithelial cells to such a high amount of circulating cytokines delivered to these cells via both peritubular microcirculation and nephron filtration might trigger TLR-4-dependent propagation of microvascular and tubular injury [13]. Thus, targeting the ligand-TLR-4 interaction may represent a promising therapeutic target to tackle the endothelial and epithelial dysfunction in kidney injury [14‐16]. The up-regulation of renal TLR-4 might also be closely linked to the expression of the gene for cyclooxygenase-2 (PGHS-2), which was found up-regulated 14.6-fold compared to NON-AKI group in the current study. Indeed, TLR-4 dependence of cyclooxygenase-2 has been shown in rodent model of cecal ligation and puncture [17]. In the mammalian kidney, cyclooxygenase-2 is predominantly expressed in macula densa, medullary interstitial cells, cortical thick ascending limb and the endothelium of renal arterioles [18, 19]. Cyclooxygenase-2-derived prostanoids appear to play a critical role in regulating renal hemodynamics, tubuloglomerular feedback mechanism and medullary salt and water excretion [18, 19]. Our findings are consistent with previous experiment reporting increases in renocortical cyclooxygenase-2 mRNA and protein expression in a rat model of endotoxemia [20]. Moreover, in that study cyclooxygenase-2 inhibition ameliorated endotoxin-induced renal dysfunction [20]. Therefore, further delineation of downstream effects of cyclooxygenase-mediated renal physiology appears to be justifiable in septic AKI.

Another prominent change was that only animals with septic AKI had profoundly down-regulated expression of PGC-1 alpha, a critical regulator of mitochondrial metabolism and biogenesis. The importance of PGC-1 alfa in the pathogenesis of S-AKI has only recently been illustrated in endotoxemic mice, in which a substantial decline in the kidney expression of PGC-1 alfa was accompanied by biochemical and structural evidence of renal tubular mitochondrial dysfunction [21]. In addition, PGC-1 alpha knockout mice suffered from persistent kidney injury following endotoxemia, suggesting that sustained suppression of PGC-1 alfa induced by inflammatory stressors could play a role in the transition from adaptive alterations in cellular energy metabolism of tubular cells to injurious event in AKI [21]. Thus, the down-regulation of PGC-1 alpha described for the first time in our large animal model lends further credence to this concept. Potentially equally important is the finding of increased expression of the pro-apoptosis-related gene caspase-3 in AKI animals, because the putative role of apoptosis in the pathogenesis of AKI still remains controversial [22‐24]. Finally, pigs with septic AKI showed a marked increase in the expression of AGTR2. The rationale for looking at AGTR2 gene expression rested on the finding that angiotensin type 2 receptor-dependent pathways seem to participate in renal inflammatory processes and apoptosis in various types of acute renal injury [25].

Understanding the methodological limitations of our experimental approach is essential for any translational research. First, the design of the present study neither allows one to establish a causative link between observed genetic signature and kidney injury nor provide mechanistic insights of factors that regulate specific gene expression. Of note, both massive systemic inflammation and local hypoxia due to intra-renal microvascular derangements might influence renal tissue gene expression pattern. Hence, it is not possible to unambiguously determine the primacy and impact of inflammatory stress, renal circulatory alterations or both on genetic signature. The absence of significant increase in HIF-1A, a master regulator of adaptive cellular responses to hypoxia [26], might suggest that inflammation rather than substantial widespread tissue hypoxia is the primary and predominant event in early phase of septic AKI. Second, due to the small number of subjects, we were not powered to compare the effects of the two types of sepsis models on gene expression. Therefore, we cannot exclude model-specific differences [27]. On the other hand, clear-cut differences seen in a number of biologically relevant genes between AKI positive and AKI negative animals support the validity and usefulness of this pooling approach. Third, renal injury in sepsis is a dynamic process and a single snap shot analysis at one time period provides only limited perspective.

Furthermore, studies using tissue samples of the entire kidney might be confounded by tissue heterogeneity. In complex conditions such as sepsis and AKI alterations in renal genomic profile might be a consequence of changes in intrinsic renal tissue constituted by diverse cellular types and, to a variable degree, by infiltrating inflammatory cells. In addition, the finding of differentially expressed genes might not be unique to the kidneys. In support of this notion, Grigoryev et al. found striking similarity in transcription patterns of inflammatory genes in kidney and lung tissue in a murine model of ischemic AKI [28]. Finally, we studied young, otherwise healthy animals. Several risk factors like aging, hypertension, diabetes and chronic kidney disease are known to significantly increase the susceptibility to AKI in sepsis. Understanding the differences in renal genomic response in the presence of pre-existing comorbidities in comparison to healthy kidneys would be highly informative. Molecular genetic response in septic AKI is multifaceted, and, hence, number of relevant genes might have been missed.

We report the first attempt to distinguish subjects with septic AKI from those with sepsis alone based upon renal tissue gene expression in a large animal model. Combined gene expression analysis with a classic large mammal model of sepsis including integrative physiologic monitioring and complex intensive care of AKI animals may guide further research. Genes whose changes in expression levels were associated with the development of septic AKI are representative of local inflammatory and danger overstimulation (TLR-4 pathway) and linked to metabolic (PGC-1 alpha) and apoptotic (caspase 3) molecular responses. This genomic profile, albeit independent from systemic hemodynamics, is associated with remarkable systemic inflammation and different intra-renal hemodynamic phenotype compared with sepsis without AKI. Cyclooxygenase-2 derived lipid mediators and angiotensin type 2 receptor-dependent downstream mechanisms appear fruitful targets for future mechanistic studies. Despite the biological plausibility, our experimental data need to be treated as exploratory, thus requiring further validation.