|Home | About | Journals | Submit | Contact Us | Français|
Acute kidney injury (AKI) and acute lung injury (ALI) represent serious, complex clinical problems. The combination of AKI and ALI drastically decreases survival. However, detailed knowledge about the interactions between these two organs is scarce.
We used two different models of AKI together with P. aeruginosa inhalation to study kidney-lung cross-talk in mice during AKI and bacterial pneumonia. AKI was induced by folic-acid injections or by myohemoglobinuria following i.m. injection of glycerol. To characterize pneumonia, we measured O2-saturation, colony-forming units in lung homogenates, and neutrophil (PMN) recruitment. Plasma creatinine and cystatin C concentrations served to quantify AKI. We also examined lung and kidney histology as well as PMN transmigration and F-actin polymerization. Sub-groups of mice received anti-PMN-antibody or platelet-depleting serum to assess the role of PMN and platelets, respectively.
AKI by itself did not cause clinically-relevant ALI. P. aeruginosa-induced pneumonia was PMN-dependent, whereas pneumonia-induced AKI was platelet-dependent. AKI attenuated pulmonary PMN recruitment during pneumonia and worsened pneumonia. Mice with AKI had lower O2-saturations and greater bacterial load than mice without it. PMN from mice with FA-induced AKI also had impaired transmigration and F-actin polymerization in vitro.
Our data demonstrate clinically-relevant kidney-lung interactions during AKI and pneumonia, that depend both PMN and platelets.
Both acute kidney injury (AKI) and acute lung injury (ALI) remain frequent and serious challenges in critically-ill patients. Survival decreases drastically when AKI and ALI occur together.1,2 The development of AKI in the setting of ALI carries an in-hospital mortality rate of 58%, compared to 28% in ALI patients without AKI.2 Our current understanding is that kidney-lung cross-talk plays a major role in worsening outcomes. However, knowledge regarding the underlying mechanisms is incomplete at best and at times even controversial. Experimental studies have lead investigators to propose various mechanisms, including release of soluble inflammatory mediators, apoptosis, and leukocyte recruitment.
AKI leads to a systemic release of various chemokines and cytokines as well as increased expression of pro-inflammatory genes in healthy lungs.3,4 AKI further activates the proapoptotic transcriptome5 and subsequently causes endothelial cell apotosis, even in healthy lung tissue.6 The fundamental role of platelets and leukocytes in ALI, and in particular neutrophils, has been well established in different experimental models.7,8 However, the relevance of neutrophils or platelets for kidney-lung interactions remain unknown despite recent research efforts. Rat models of both bilateral renal ischemia-reperfusion and bilateral nephrectomy have demonstrated significant recruitment of neutrophils into healthy lungs.3,9 Contrary to these findings, we have shown that AKI protects from HCl-induced ALI by attenuating neutrophil recruitment into inflamed lungs.10 We have also provided evidence that uremic neutrophils, but not normal neutrophils in a uremic milieu, fail to recruit into the inflamed lung.10
Several clinical and experimental studies have revealed deleterious effects of ALI on renal function. However, most mechanisms identified to date are related to mechanical ventilation rather than to ALI itself. Lung-protective ventilation, as opposed to conventional ventilation, did not only reduce mortality from severe ALI but also the rate of AKI.11,12 The changes in intra-thoracic pressure associated with mechanical ventilation alters systemic and subsequently renal hemodynamics, leading to decrease in glomerular filtration rate, free water clearance, urinary Na-excretion, and creatinine clearance.13–15 Moreover, mechanical ventilation can induce so-called biotrauma and subsequent systemic release of pro-inflammatory cytokines, which correlate with the degree of remote organ failure.12,16 Mechanical ventilation can further lead to apoptosis of renal epithelial cells.17 However, little is known about the effects of ALI itself on remote organ function and the underlying mechanisms. In a recent clinical study with more than 1800 patients hospitalized for community acquired pneumonia, 34% of the patients developed AKI.18 AKI was associated with a significant increase in the risk of death out to one year. Patients who developed AKI with or without sepsis displayed significantly higher pro-inflammatory markers and coagulation activation.
The paucity of data regarding the direct effects of ALI on remote organs as well as the controversy surrounding the effects of AKI on pulmonary functions prompted us to conduct the following study. We have used two murine models of AKI together with bacterial ALI (P. aeruginosa pneumonia) to further unravel kidney-lung interactions during AKI and pneumonia as well as their underlying mechanisms (Fig. 1A).
Both i.p. injection of folic-acid and i.m. injection of glycerol, via subsequent myohemoglobinuria,19,20 cause AKI within 24h after injection as evidenced by significant and corresponding increases plasma creatinine (Fig. 1B) and cystatin C (Fig. 1C). The degree of AKI in both models is equivalent to the clinical stage RIFLE-R/I.21
Inhalation of P. aeruginosa the leads to a significant increase in circulating interleukin 6 (IL-6), indicating substantial inflammation (Fig. 1D). Plasma IL-6 levels during pneumonia are also significantly elevated with pre-existing AKI (Fig. 1E). The preexistence of AKI does not affect the plasma levels of IL-6 during pneumonia. Mice with FA-induced AKI display a non-significant trend towards increased IL-6 plasma levels (Fig. 1E).
Inhalation of P. aeruginosa leads to clinically-relevant pneumonia and pulmonary inflammation in mice at 24h, as evidenced by decreased SpO2, isolation of P. aeruginosa CFUs from lung homogenates, and pulmonary recruitment of neutrophils (Fig. 2A–C). Neutrophil depletion prior to induction of pneumonia results in more severe pneumonia with worse oxygenation and higher bacterial load (Fig. 2D+E).
Similar to our models of renal ischemia-reperfusion and bilateral nephrectomy, FA-induced AKI itself (group 2: AKI+sham vs. group 1: control+sham) does not impair oxygenation (Fig. 3A), does not lead to significant recruitment of neutrophils into the lung (Fig 3E), and does not cause overt histological changes in the lung (Fig. 4C).
Pre-existing AKI, however, significantly worsens bacterial pneumonia. Mice with FA-induced AKI demonstrate significantly reduced oxygenation (Fig. 3A) and higher bacterial load after P. aeruginosa inhalation than mice without AKI (Fig. 3C). Moreover, FA-induced AKI impairs pulmonary neutrophil recruitment during bacterial inflammation, as indicated by reduced MPO in lung homogenates (Fig. 3E). The effects of AKI appear to be independent of the model used. After inhalation of P. aeruginosa, glycerol-induced myohemoglobinuric AKI also further decreases SpO2 (Fig. 3B), increases the pulmonary bacterial load (Fig. 3D), and impairs neutrophil recruitment (Fig. 3F).
As both oxygenation and bacterial load after P. aeruginosa inhalation are neutrophil-dependent, the data provide evidence for neutrophil-dependent deterioration of pneumonia during AKI. Interestingly, histological evidence of pulmonary inflammation, e.g. interstitial edema and cell infiltration, appears to be less pronounced with pre-existing AKI (Fig. 4G, group 4: AKI+ pneumonia) than without it (Fig. 4E, group 3: control+pneumonia). The amount of P. aeruginosa-positive blood cultures was not statistically different between mice with and without pre-existing FA-induced AKI (group 3: pneumonia+control = 4/20 vs. group 4: AKI+pneumonia = 5/20).
Neutrophils from mice with FA-induced AKI reveal a markedly reduced in vitro transmigration towards fMLP compared to neutrophils from control mice (Fig. 5A). In absence of fMLP stimulation, neutrophils from mice with FA-induced AKI do not migrate through the transwell similarly to neutrophils from control mice (data not shown). This suggests that neutrophils from mice with FA-induced AKI are not in an activated state, which could have rendered them unresponsive to the stimulus.
To determine whether the abnormal migration of FA-treated neutrophils was related to abnormal cytoskeleton rearrangement, we examined F-actin reorganization in neutrophils stimulated with fMLP on fibrinogen-coated slides. Neutrophils from control mice display one short and narrow lamellipodia of F-actin at one pole of the cell (Fig. 5B). In contrast, neutrophils from mice with FA-induced AKI show a prominent and large protrusion of F-actin (Fig. 5B), as evidenced by a significant increased in surface area of F-actin compared to that of neutrophils from control mice (Fig. 5C). In addition, some neutrophils from mice with FA-induced AKI cells exhibited 2 protrusions of F-actin.
P. aeruginosa pneumonia induces AKI, equivalent to the clinical stage RIFLE-R 21, as demonstrated by significant increases in plasma creatinine (Fig. 6A) and cystatin C (Fig. 6B). The significant rise in urinary neutrophil gelatinase-associated lipocalin (NGAL) provides strong evidence for direct tubular injury during pneumonia (Fig. 6C). Moreover, unchanged pulse distension22,23 during pneumonia makes significant changes in systemic hemodynamics following inhalation of P. aeruginosa unlikely (Fig. 6D), further substantiating the notion of direct kidney injury in our pneumonia model.
Contrary to our findings in the lung, pneumonia does not lead to neutrophil influx into the kidneys, as indicated by histology (Fig. 4F) and unchanged renal MPO (Fig. 6E). Moreover, pre-treatment with a neutrophil-depleting antibody does not affect the development of pneumonia-induced AKI (Fig. 6F).
PAS-staining of kidney sections shows brush-border defects of tubular epithelial cells in both pneumonia-induced AKI (Fig. 7A) and FA-induced AKI (Fig. 7B). Whereas TUNNEL-staining reveals only very mild signs of apoptosis in pneumonia-induced AKI (Fig. 6C), there is evidence for overt apoptosis in FA-induced AKI (Fig. 6D).
Prompted by recent findings about pronounced hemostasis activation in patients with AKI following CAP18, we next examined the role of platelets in pneumonia-induced AKI. Mice pre-treated with anti-platelet serum demonstrate lower SpO2 (Fig. 8A) and higher pulmonary bacterial load (Fig. 8B) after P. aeruginosa inhalation than mice pre-treated with control serum, indicating overall worsening of pneumonia. However, platelet-depleted mice display lower plasma creatinine concentrations after P. aeruginosa inhalation, indicating a better preserved renal function.
Both AKI and ALI significantly reduce survival, particularly when they occur together. Nonetheless, only limited and at times even controversial information is available regarding kidney-lung cross-talk under pathological conditions. Moreover, there are no data regarding the mechanisms of kidney-lung interactions during bacterial ALI and AKI. A better understanding is of utmost importance, as pneumonia represents the most frequent cause of sepsis which in turn is the leading cause of AKI in critically-ill patients.24,25
Using a murine two-hit model of AKI and bacterial pneumonia, we show that FA-induced AKI itself has no clinically overt effect on healthy lungs, as indicated by unchanged oxygenation 48h after induction of AKI. Similarly, lung histology is largely unaffected by FA-induced AKI, although there is a slight trend toward increased lung MPO-activity, indicating mild neutrophil influx. Preexisting AKI, however, attenuates pulmonary neutrophil recruitment during pneumonia but does not affect plasma IL-6 levels during pneumonia. AKI appears to increase bacterial load during pneumonia, to further impair oxygenation, and to ultimately worsen pneumonia. We were able to reproduce these findings in a clinically relevant model of myohemoglobinuric AKI and to thereby further substantiate the general concept of AKI-impaired neutrophil recruitment into the inflamed lung.
As outlined before, there is experimental evidence to substantiate the concept of AKI-induced inflammation in the healthy lung.3,9,27–30 We have also observed a non-significant, slight increase in pulmonary neutrophil content in mice with isolated AKI (group 2). We observed similar changes for plasma IL-6 level after induction of AKI. We observed similar non-significant changes for plasma IL-6 level after induction of AKI. This is in contrast to other mouse models of AKI, such as ischemia-reperfusion and bilateral nephrectomy. However, these models require major abdominal or retroperitoneal surgery which by themselves could explain the elevated levels of IL-6.3,4
In models of bilateral nephrectomy and renal ischemia-reperfusion, we could not find signs of impaired oxygenation or significant pulmonary edema.10 To the contrary, both our previous study10 and our current study suggest that AKI exerts a strong anti-inflammatory and clinically relevant effect on the injured lung. We have previously demonstrated in two different murine models of AKI combined with aseptic ALI that uremic neutrophils fail to recruit into the inflamed lung and thereby protect from ALI. 10 These effects appeared to depend solely on uremic neutrophils and occurred in both normal and uremic milieus. 10 In our current study, we provide further evidence to support the idea of AKI-induced inhibition of pulmonary neutrophil recruitment. Pre-existing AKI attenuates neutrophil recruitment into the lung after inhalation of P. aeruginosa. Similar to neutrophil-depleted mice, mice with pre-existing AKI display worse oxygenation and higher bacterial load after inhalation of P. aeruginosa than mice without AKI. Moreover, neutrophils from mice with FA-induced AKI also demonstrated significantly impaired transmigration and F-actin polymerization in vitro. Combining our previous results and these findings, we postulate that AKI-induced inhibition of neutrophil recruitment gives rise to clinically relevant aggravation of bacterial ALI with worse oxygenation and higher bacterial load.
Although our previous findings suggest that the observed effects depend directly on neutrophils rather than soluble factors, we could not find changes in the expression of neutrophil surface molecules, e.g. L-selectin (data not shown). We therefore focused on intracellular events controlling neutrophil recruitment. Directed neutrophil migration depends on the coordinated rearrangement of cytoskeleton structures that provide both protrusion and contraction. Actin filaments polymerize at the leading edge, generating a protrusive leading edge that pushes the cell forward. Contraction occurs through the assembly of actomyosin filaments along the main body and at the tail, suppressing lateral protrusions but enabling protrusions at the leading edge.31–33 Our data show that neutrophils from mice with FA-induced AKI have a substantial transmigration defect. Moreover, these neutrophils cannot limit F-actin formation. As a single and restricted lamellipodia is essential for directed migration,32,34, the defective migration of neutrophils from mice with FA-induced AKI seen in our transwell assays may be attributed to abnormal cytoskeleton regulation.
Available data together allow us to hypothesize that the effect of AKI on pulmonary neutrophil recruitment depends on the condition of the lung. Under non-inflammatory conditions, mechanisms causing AKI or AKI itself appear to have remote, pro-inflammatory effects on the lung that can enhance neutrophil recruitment. AKI, nonetheless, does not promote neutrophil recruitment during pulmonary inflammation, which inherently triggers neutrophil recruitment into the lung. Here, AKI exerts anti-inflammatory effects toward neutrophils and ultimately inhibits neutrophil recruitment. Histological signs of less pulmonary inflammation in mice with pneumonia plus AKI compared to those with only pneumonia further add to the notion of AKI-induced anti-inflammatory effects. Neutrophil recruitment itself can trigger excessive inflammatory responses in the affected tissue. Thus, a lack of neutrophil recruitment can lead to attenuated tissue inflammation.35 Moreover, these findings also allow us to hypothesize that pulmonary damage through P. aeruginosa itself, e.g. via exotoxins,36 plays a greater role under theses conditions than damage caused by neutrophil-dependent pulmonary inflammation per se.
Findings from a recent study about kidney-lung interaction during AKI and ventilation-induced lung injury also support a differential concept regarding pulmonary neutrophil recruitment during AKI.37 In the setting of high tidal volume ventilation, which in itself causes lung inflammation, AKI impaired pulmonary neutrophil recruitment into the lung. Under less inflammatory conditions, e.g., low tidal volume ventilation or spontaneous breathing, these effects were not seen.
Finally, as neutrophils and their recruitment represent the body’s the first-line defense against infections35, our experimental findings also have the potential to explain several clinical observations. For example, patients with AKI more frequently develop bacteremia and poor outcome in the setting of peritonitis, hematologic malignancies, and after cardiac surgery than patients without AKI.38–41
Various mechanisms have been attributed to ALI-induced AKI. Mostly, mechanisms of injury have been considered a consequence of mechanical ventilation, including a reduction in cardiac output, redistribution of renal blood flow, and stimulation of hormonal and sympathetic pathways.13 Recent animal data suggest that maintenance of stable hemodynamics and alveolar ventilation are crucial in preserving renal function in the setting of ALI.42 Mechanical ventilation also induces apoptosis in tubular epithelial cells17 as well as a so-called biotrauma.12,16
We here provide further evidence that bacterial ALI (pneumonia) itself, without underlying mechanical ventilation, can cause remote organ injury. Here, pneumonia leads to AKI within 24h, as evidenced by corresponding changes in plasma creatinine and cystatin C concentrations. Pneumonia-induced AKI appears to be due to direct tubular damage rather than pre-renal factors, as evidenced by the significantly increased NGAL concentrations and stable pulse distension. Moreover, direct bacterial damage to the kidney is also unlikely, as only a small number of mice had blood cultures positive for P. aeruginosa, and we could not find histological evidence of bacterial invasion into the kidneys.
Previously unrecognized as mediators of remote organ failure, here, platelets have emerged as crucial for the development of pneumonia-induced AKI. Despite worsening pneumonia, platelet-depleted mice showed nearly normal plasma creatinine concentrations. Platelet and platelet-derived microparticles represent well-known prognostic factors for the severity of septic organ failure.43,44 Nonetheless, whether platelets themselves or platelet-derived microparticles or both are responsible for this effect, requires further studies,
Our study reveals clinically relevant kidney-lung interactions in a murine model of bacterial pneumonia combined with AKI. Our data further substantiate the notion of AKI-induced inhibition of neutrophil recruitment, which translates into higher bacterial load and worse oxygenation in the setting of bacterial pneumonia. Impaired neutrophil recruitment during AKI seems, at least partially, to be due to altered cytoskeleton arrangement. On the other side, bacterial pneumonia, causes AKI in a platelet-dependent manner.
The Animal Care and Use Committee of the University of Pittsburgh approved all animal experiments. We used 10 to 12-week-old wild-type C57BL/6 mice. Mice were housed in a barrier facility under specific pathogen–free conditions.
Intraperitoneal injections of FA (450mg/kg dissolved in NaHCO3) served to induce AKI (time = 0h, Fig. 1).45,46 We have identified this dose of FA in pilot studies, such that plasma creatinine levels rose significantly but without FA leading to severe illness or death. Control animals received an equivalent volume of NaHCO3 i.p. We measured plasma creatinine, and cystatin C to assess renal function, using commercially available assays. We also used commercially available assays to measure plasma levels of IL-6.
Following previously published models of glycerol-induced myohemoglobinuric AKI19,20,47, we injected mice i.m. with glycerol (5ml/kg or normal saline (control). We measured plasma creatinine and cystatin C to assess renal function, using commercially available assays.
We used inhalation of P. aeruginosa strain UI-18 (PA-7) to induce pneumonia.7
For each induction cycle, a total of 4 mice was placed in an inhalation chamber48. 2 mice had undergone AKI-induction 24h prior, the other 2 mice had received control treatments 24h earlier (time = 24h, Fig. 1). 5ml of bacteria suspension (1014 CFU/ml in NaCl 0.9%) or 5ml of NaCl 0.9% (sham) were aerosolized over 30min, after which the mice were removed from the chamber. 24h after inhalation (time = 48h, Fig. 1), we measured oxygen saturation and pulse distension in unanesthetized mice, using a small-animal pulse oximeter.22,23 Mice were then anesthetized to obtain blood samples via puncture of the inferior vena cava and to harvest lungs and kidneys for further analyses.
Induction of AKI and subsequent pneumonia together with respective control experiments resulted in a maximum of 4 different experimental groups (Fig. 1): Mice with control injection and sham inhalation (group 1), mice with AKI and sham inhalation (group 2), mice with control injection and pneumonia (group 3), and mice with AKI and pneumonia (group 4).
A subgroup of mice received either neutrophil-depleting antibody (α-GR-1, clone RB6–8C5) or anti-platelet serum (Accurate Chemicals, Westbury, NY) prior to inhalation to assess the role of neutrophils and platelets in our pneumonia model, respectively.8,10 In previous studies, this sufficiently depleted circulating neutrophils or platelets but did not affect other cell counts.10,49 When compared to a control antibody, injection of a neutrophil depleting antibody significantly decreased the number of circulating neutrophils (mean±SD: 136±86/μl vs.2000±217/μl) but did not affect the number mononuclear cells. Similar, pre-treatment with anti-platelet serum significantly decreased the number of circulating platelets (mean±SD: 313±47 × 1000/μl vs. 681±65 × 1000/μl) without altering white cell counts.
To determine global neutrophil content in both kidneys and lungs, we measured tissue myeloperoxidase activity (MPO) according to our previously published protocol.10,49 We have shown before49 that changes in MPO represent an excellent indicator of global tissue neutrophil recruitment.
Lungs were inflation-fixed by means of tracheal instillation of 2x Zinc fixative. Kidneys and lungs were then excised and embedded in paraffin.
5μm tissue sections were cut and stained with hematoxylin and eosin under standard techniques.
Slides were digested with 20ug/mL Proteinase K and quenched with H2O2. Slides were then stained with ApoTag Peroxidase Kit according to the manufacturer’s instructions. Finally, slides were stained with DAB, washed, counterstained with Aqueous Hemotoxylin and blued in Scott’s Tap Water Substitute.
Slides were soaked in Periodic acid and rinsed in dH2O. Slides were then soaked in Schiff Reagent, washed in running tap water and counterstained with Gill No. 3 Hematoxylin.
At the time of sacrifice, blood samples were collected, and lungs were removed aseptically and placed in 3 ml of sterile saline each. The lungs were the homogenized with a tissue homogenizer under a vented hood. Serial 1:10 dilutions of both blood samples and lung homogenates were made. 100μl of each dilution were plated on Pseudomonas isolation agar plates, the plates were incubated for 18h at 37°C, and then the number of colonies were counted.
Neutrophils were isolated from bone marrow cells by Percoll gradient as previously described.50
As described recently 50, migration was evaluated using 3 μm transwell chambers coated with fibrinogen in chemotaxis assay. Migration without fMLP or toward fMLP was allowed for 3 hours. The migrated cells in the bottom well were counted with a hemocytometer.
To examine F-actin structures, we followed our recently published protocol. 50 Briefly, neutrophils (5 × 104) were prestimulated with fMLP and placed on fibrinogen-coated slides for 0 and 10 minutes at 37°C. The cells were fixed, permeabilized with 0.1% Triton X-100, and stained with rhodamine-labeled phalloidin.
Z series of fluorescence images were captured using a Leica DMI6000 fluorescence microscope at 63/1.3 NA objective, with ORCA-ER C4742-95 camera driven by Openlab software. Z series were analyzed by deconvolution using Volocity software. Quantifications were performed using region measurement in Openlab software.50
All data are presented as median (inter-quartile range). Statistical analysis included Shapiro-Wilks test for normality, one-way ANOVA, Kruskal-Wallis one-way analysis of variance, post-hoc Student-Newman-Keuls test, t-test, Mann-Whitney-U test, and Fisher’s exact test (p<0.05 was considered statistically significant). N=5–14 mice for all experiments.
Supported for this work was provided by NIH grants 5K08GM081459-03 (to KS), 5R01DK070910 (JAK), 5R01HL080926 (KS and JAK), T32 HLL07820 (to SC), and KL2 RR024154 (RM).
The authors have no financial relationships to disclose.