evidence for sustained renal hypoxia and transient hypoxia

Nephrol Dial Transplant (2008) 23: 1135–1143doi: 10.1093/ndt/gfm808Advance Access publication 29 November 2007

Original Article

Evidence for sustained renal hypoxia and transient hypoxia adaptationin experimental rhabdomyolysis-induced acute kidney injury

Christian Rosenberger1, Marina Goldfarb2, Ahuva Shina3, Sebastian Bachmann4, Ulrich Frei1, Kai-UweEckardt5, Thomas Schrader6, Seymour Rosen7 and Samuel N. Heyman3

1Department of Nephrology and Medical Intensive Care, Charite Universitaetsmedizin, Berlin, Germany, 2Nephrology Unit, BikurHolim Hospital, 3Hadassah University Hospital Mt. Scopus and the Hebrew Medical School, Jerusalem, Israel, 4Department ofAnatomy, Humboldt University, Berlin, 5Department of Nephrology and Hypertension, University of Erlangen-Nuremberg,Erlangen, Germany, 6Department of Pathology, Charite Universitatsmedizin, Berlin, Germany and 7Department of Pathology, BethIsrael Deaconess Medical Center and Harvard Medical School, Boston, MA, USA

AbstractBackground. Indirect evidence suggests that hypoxia con-tributes to the pathophysiology of rhabdomyolysis-inducedacute kidney injury (AKI). However, the cellular locationand kinetics of hypoxia, as well as potential hypoxia adap-tation are unclear.Methods. Rhabdomyolysis was induced in rats by IM glyc-erol (GLY) injection, which largely recapitulates the fullclinical syndrome. Additional rats received IV myoglobin(MYO), in order to assess the contribution of MYO per se.We performed immunohistochemistry for hypoxia markers[pimonidazole (PIM) adducts and hypoxia-inducible fac-tors (HIFs)] and the cell-protective HIF target gene hemeoxygenase-1 (HO-1). Furthermore, we sought a potentialnegative feedback loop to terminate HIF activation, drivenby HIF prolyl-hydroxylase-2 (PHD-2).Results. In GLY, progressive tubular injury, mainly of prox-imal tubules (PT), developed over time, but its extent washeterogeneous. PIM, HIFα and HO-1 were all absent in con-trols, but strongly positive in GLY, with a specific spatio-temporal pattern. In PT, (a) PIM was detectable throughoutthe study with a maximum at 6 h, (b) HIF was activatedonly at 3 h and (c) HO-1 and PHD-2 appeared at 6 h andpersisted at a lower level at 24 h. Apart from tubular castformation, MYO did not cause overt tissue damage, but ledto strong activation of HIFs, in a pattern similar to 3 h ofGLY.Conclusions. Our data suggest that renal hypoxia occurs inrhabdomyolysis, and that MYO, at least partly, contributesto hypoxia generation. Since in the most affected tubulestranscriptional hypoxia adaptation is transient and inhomo-geneous, pharmacologic HIF enhancement holds the poten-tial to improve outcome in rhabdomyolysis-induced AKI.

Correspondence and offprint requests to: Christian Rosenberger, Nephrol-ogy and Medical Intensive Care, Charite Universitaetsmedizin, Augusten-burger Platz 1, 13353 Berlin, Germany. Tel: +49-30-450553232; Fax:+49-30-79741921; E-mail: [email protected]

Keywords: acute renal failure; heme oxygenase-1; HIFprolyl hydroxylase-2; hypoxia-inducible factors;pimonidazole

Introduction

Rhabdomyolysis-induced acute renal injury (AKI), alsonamed pigment nephropathy or crush kidney injury, devel-ops following skeletal muscle trauma related to physical,thermal, ischemic, infective, metabolic or toxic causes, andis associated with high morbidity and mortality [1–4].

The pathophysiology of AKI in rhabdomyolysis is likelycomplex and incompletely understood [3]. Myoglobin(MYO) has been shown to bind to Tamm-Horsfall pro-tein, especially in an acidic milieu, forming tubular casts[5]. Cast-bound MYO can scavenge nitric oxide (NO) [6]and generate reactive oxygen species (ROS) [7–12], caus-ing tubular toxicity and vasoconstriction. Muscle swelling,intravascular volume depletion, reduced cardiac output [13]and lactic acidosis, all probably contribute to renal in-jury. In experimental models, renal cortical blood flow ismarkedly reduced [14]. The pathology in humans is re-markable for pigmented casts in distal tubules and collect-ing ducts with varying degrees of injury in other nephronsegments [15,16]. In animals, glycerol (GLY) injection intothe hind limbs produces renal changes resembling that ob-served in the human [17].

Using oxygen microelectrodes, we have previouslyshown an acute reduction of renal parenchymal oxygena-tion during the infusion of MYO [18]. Alterations in renalparenchymal oxygenation have not been assessed in themore clinically relevant GLY model.

Hypoxia adaptation is largely conferred through theso-called hypoxia-inducible transcription factors (HIFs)[19–21]. HIFs are heterodimers composed of a con-stitutive β-subunit and one of at least two different

C© The Author [2007]. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved.For Permissions, please e-mail: [email protected]

Dow

nloaded from https://academ

ic.oup.com/ndt/article/23/4/1135/1878886 by guest on 25 N

ovember 2021

1136 C. Rosenberger et al.

oxygen-dependent α-subunits (HIF-1α and -2α). HIF ismainly regulated by oxygen-dependent proteolysis of theα-subunits. The key enzymes of HIFα degradation are HIFprolyl-hydroxylases (PHDs) [22,23]. PHDs require molec-ular oxygen for substrate and their Km is within the rangeof ambient oxygen concentrations [24]. Therefore, PHDscan be considered cellular oxygen sensors. HIFs are ubiq-uitous, instantaneaously activated upon hypoxia and short-lived after reoxygenation. HIFα immunohistochemistry canserve to detect hypoxia within tissues. It is noteworthy thatnon-hypoxic HIF activation by various growth factors andcytokines has been demonstrated in vitro [25,26]. How-ever, since the latter is caused by enhanced HIFα tran-scription/translation, it likely takes longer to establish thanhypoxic HIF activation, which is due to impaired HIFαdegradation. Heme oxygenase-1 (HO-1) is one of the mostprominent HIF target genes with well-proven cell protec-tive properties [27]. Importantly, since it generates carbonmonoxide, a vasodilator, and biliverdin, a potent radicalscavenger, HO-1 has the potential to counterbalance at leastsome factors suspected to cause rhabdomyolysis-inducedAKI.

Pimonidazole (PIM) is a bioreductive agent which, afterin vivo delivery, binds to tissues at pO2 below 10 mmHg,where it can be visualized with the help of commerciallyavailable antibodies [28,29]. In previous studies from ourlab immunohistochemistry for HIFα or PIM was unde-tectable in normal kidneys, but strongly positive after var-ious hypoxic stimuli [30–33]. Thus, HIFα and PIM mostlikely detect those pathologic conditions in which oxygenavailability becomes limiting, ensuing either cell injury oradaptive responses [34].

The present study employs high amplification immuno-histochemical techniques to seek evidence of renal hypoxiain rhabdomyolysis-induced AKI, to define its cellular lo-cation, kinetics, and relationship to tubular injury, as wellas potential adaptive responses. Our investigation includesevents shortly after the initiation of AKI, when tubular in-jury is sublethal and potentially reversible—a time neverexamined in human material. Indeed, our work suggeststhat hypoxia contributes to tubular injury in a clinicallyrelevant model of rhabdomyolysis-induced AKI.

Methods

Animals and materials

Male Sprague Dawley (SD) rats (345 ± 6 g) were used,fed regular chow with free excess to water. Chemicals werepurchased from Sigma (St. Louis, MO, USA). Experimentswere conducted in accordance with the NIH Guide for theCare and Use of Laboratory Animals.

GLY model

Under ether anaesthesia rats were injected with 50% GLY,8 ml/kg, applied in divided volumes into four sites at thehind limb muscles. Dipyrone (50 mg/kg IP) was given forpain management.

MYO infusion model

As previously detailed [18], under ketamine anaesthesia(100 mg/kg) and following catheterization of the femoralvein, MYO was infused over 45 min (380 mg/kg, dissolvedin 5 ml saline).

Functional studies

Rats were kept in metabolic cages (Nalge, Rochester, NY,USA) for 24 h urine collections. Blood sampling from thefemoral vein was obtained under anaesthesia at baseline,and during sacrifice in the 24 h experiments. Plasma andurine samples were processed for the determination of cre-atinine, urea, sodium and potassium, and creatine phospho-kinase (in a Roche/Hitachi Modular

R©, Grenzach-Wyhlen,

Germany). Creatinine clearance, fractional sodium reab-sorption and potassium excretion were calculated.

Determination of renal morphology

Under pentobarbital anaesthesia (60 mg/kg), the kidneyswere perfusion-fixed selectively through the abdominalaorta with 1.25% glutaraldehyde (for plastic embeddedsemi-thin sections) or with 3% paraformaldehyde (forparaffin-embedded sections for immunohistochemistry),as previously detailed [31,33]. Semiquantitative assess-ment of cortical tubular damage was performed, using a96-field grid, superimposed on a representative low powerfigure (magnification 90×) of a 3-µm paraffin section(haematoxylin and eosin staining, comparable with Fig-ure 4A and B). To each field of the grid a tubular injuryscore was assigned, based on the most severe tubular dam-age observed therein: 0 = no significant tubular damage(corresponding with ‘P’ in Figure 1), 1 = moderate damage(‘P’ in Figure 1), 2 = severe damage (‘P’ in Figure 1).Field scores were added and divided by 96.

Immunohistochemistry

PIM (HypoxyProbe, Pharmacia International, Belmont,MA, USA), which binds to tissues with pO2 levels below10 mmHg [28], was injected in vivo (60 mg/kg IV)1 h before the kidneys were perfusion-fixed. The fol-lowing primary antibodies were used as previously re-ported [30–33,35,36]: mouse–anti-human HIF-1α (α67,Novus Biologicals, Littleton, CO, USA, 1:10 000),rabbit–anti-mouse HIF-2α (PM9, gift from PatrickMaxwell, Hammersmith Hospital, Imperial College, Lon-don, UK, 1:10 000), rabbit–anti-rat HO-1 (Stress-gen, Victoria, Canada, 1:60 000), rabbit–anti-mouseHIF prolyl-hydroxylase-2 (Novus Biologicals, Littleton,CO, USA, 1:10 000), mouse–anti-PIM (1:1000 Hy-poxyprobe; Natural Pharmacia International, Belmont,MA, USA). Immunostaining was assessed semiquantita-tively according to the following score: 0 = no signals,1+ = staining in <5% of tubular profiles or IC/EC, 2+ =5–20%, 3+ = 20–33%; 4+ = 33–50%.

Statistics

Data are represented as means ± SEM. Paired Student’s t-test was used for the comparison of functional parameters.

Dow



ovember 2021

Evidence for sustained renal hypoxia and transient hypoxia adaptation in experimental rhabdomyolysis-induced AKI 1137

Fig. 1. Fine morphology in rhabdomyolysis-induced AKI. AKI was in-duced in rats by IM GLY injection. Asterisk = medullary thick ascend-ing limb (mTAL); C = cast; CD = collecting duct; DT = distal tubule(discrimination between distal nephron segments was impossible in thismaterial); P = proximal tubule (largely intact); P = moderately injured P;P = severely injured P; S3 = S3 portion of the proximal tubule; T = thinlimb; VB = vascular bundles. Proximal tubular injury increases from 3 h(A) to 6 h (C) and 24 h (E). At 24 h in the outer medulla tubular injurymainly occurs in S3 (B), whereas in the inner stripe mTAL are filled withcasts (D). In the papilla morphology is preserved, apart from tubular casts,which at 24 h mainly occur in thin limbs (F). Magnification 800×.

ANOVA with post hoc Student’s t-test was applied forbetween-group comparisons of morphology and immuno-histochemistry. Statistical significance was set at P < 0.05.

Results

A total of 42 animals were included into the study: 5 un-treated controls, 6 with IV MYO infusion, sacrificed at 3 h,and 31 with IM GLY injection (10, 6 and 15 for the 3, 6and 24 h time points, respectively). Rhabdomyolysis wasconfirmed by an increase in serum creatine phosphokinasein a subset of three animals at 3 h after GLY injection:1210, 1560 and 970 units/l, compared with baseline levelsof 44, 62 and 34 units/l. A 33% mortality was noted onlyin GLY-treated animals kept for 24 h.

Functional data: acute renal failure in GLY

At 24 h GLY led to a significant reduction in creatinineclearance, fractional sodium reabsorption, and to a signifi-cantly enhanced serum creatinine, serum urea and fractionalexcretion of potassium (Table 1).

Table 1. Functional data

Baseline Day 1 P value

Urine volume (ml/h) 0.27 ± 0.03 0.45 ± 0.14 NSPCr (µmol/l) 51 ± 3 177 ± 14 0.01ClCr (ml/min) 0.66 ± 0.05 0.12 ± 0.04 0.01ClCr (ml/min/100 g) 0.20 ± 0.02 0.03 ± 0.01 0.01PUrea (mmol/l) 8.2 ± 0.4 42.6 ± 2.6 0.01TRNa (%) 99.27 ± 0.08 97.57 ± 0.45 0.05FEK (%) 9.0 ± 1.5 71.5 ± 8.0 0.01

Functional parameters in rats subjected to Rh-AK at baseline (day 0) andat 24 h later (n = 10).FEK = fractional excretion of potassium; ClCr = creatinine clearance;PUrea = plasma urea; PCr = plasma creatinine; TRNa = fractional sodiumreabsorption.

Morphologic findings: proximal tubular injury in GLY,minimal structural alterations in MYO

In GLY, the major morphologic injury developed in proxi-mal tubules (PT), with the formation of casts in the distalnephron, observations in line with previous studies [15–17]. PT injury progressed over time (Table 2, Figure 1),mainly appearing in the cortex (mostly S2 segments) andoccasionally in the outer stripe (S3 segments). Alterationsranged from mild (some loss of brush border, cytoplas-mic vacuoles, but generally preserved epithelial structure(‘P’ in Figure 1) to severe (complete destruction of tubulararchitecture and eventually necrois, ‘P’ in Figure 1). Per-fusion fixation allowed us to discriminate between opentubules (presumably belonging to filtering nephrons) andcollapsed tubules (presumably non-filtering nephrons). Ingeneral, morphology was heterogeneous within the samesection, with groups of open and collapsed tubules side byside, but the number of collapsed tubules increased overtime, suggesting a reduced number of filtering nephrons(Figure 2A and B). In distal nephron segments, morphologywas relatively well preserved throughout the study, albeittubular casts and cellular debris were seen (Figure 1). InMYO, tubular damage was minimal (Figure 3), althoughtubular casts and capillary congestion were comparable to3 h of GLY.

PIM-adducts are detectable in GLY but not in MYO

To test if profound renal hypoxia occurs in GLY, andwhether its location predicts tubular injury, we searchedfor PIM-adducts, which accumulate in tissues at oxygentensions below 10 mmHg [28]. Indeed, while control ani-mals all were negative for PIM, in GLY, strongest stainingoccurred in PT (mostly S2 and S3) and preceded the de-velopment of tubular damage, suggesting hypoxic genesisof injury (Figures 2, 4 and Table 2). Throughout the study,there was a gradient of hypoxia from the superficial cor-tex towards the medulla, which is in contrast with physi-ologic renal oxygen gradients and with those described incontrast media-induced AKI [30], suggesting a differentpathophysiology in GLY. Within the cortex, PIM signalswere likely nephron-restricted, being clustered in stripes oftubules from the renal capsule towards the medulla (Figure2C and D). Tubules with the strongest PIM were open and

Dow



ovember 2021


Table 2. Renal morphology and immunohistochemistry: data are presented, stratified into the various experimental groups, different time points andkidney regions; the cell-specific intensity of staining is also underscored for each region

Glycerol injection

Myoglobin infusion 3 h (n = 6) 3 h (n = 7) 6 h (n = 6) 24 h (n = 10)

Damage scorea 0 0.3 ± 0.2 PT 0.7 ± 0.3∗∗∗ PT 1.1 ± 0.1∗∗∗ PTb

PIMb Cor 0 1.3 ± 0.3∗ PT > CD 3.3 ± 0.3∗ PT > CD 1.7 ± 0.3∗ PT > CDOS 0 1.0 ± 0.4∗∗ PT > CD 2.7 ± 0.3∗ PT > CD 0.7 ± 0.3 PT > CDIS 0 1.3 ± 0.3∗ TAL > CD 1.7 ± 0.3∗ TAL 0.3 ± 0.3Pap 0 1.0 ± 0.4∗∗ IC 1.0 ± 0.0∗∗ IC 0.3 ± 0.3

HIF-1αb Cor 2.0 ± 0.3∗ PT > CD > IC 2.3 ± 0.2∗ PT > CD > IC 0.2 ± 0.2 CD 0.8 ± 0.2∗∗∗ CDOS 1.8 ± 0.2∗ PT > CD > IC = EC 2.0 ± 0.2∗ PT > CD > IC 0 0IS 1.0 ± 0.3∗ CD > TAL > IC 1.7 ± 0.2∗ CD > TAL > IC 0 0Pap 0.8 ± 0.3∗∗ CD = IC 3.4 ± 0.2∗ CD = IC 0.2 ± 0.2 CD = IC 0

HIF-2αb Cor 2.0 ± 0.3∗ IC > EC = G 2.3 ± 0.2∗ IC > EC 0 0.7 ± 0.2∗∗∗ IC > ECOS 1.7 ± 0.2∗ IC > EC 2.0 ± 0.2∗ IC > EC 0 1.3 ± 0.3∗ IC > ECIS 0.7 ± 0.2∗∗∗ IC > EC 1.9 ± 0.3∗ IC > EC 0 0Pap 0.8 ± 0.3∗∗∗ IC > EC 3.3 ± 0.4∗ IC > EC 0 0

HO-1b Cor 0 0 3.2 ± 0.3∗ PT > CD 1.7 ± 0.2∗ PTOS 0 0 2.5 ± 0.2∗ PT > CD 1.3 ± 0.2∗ PTIS 0 0 1.7 ± 0.2∗ CD > TAL 0.3 ± 0.2Pap 0 0 1.5 ± 0.2∗ CD > IC 0.2 ± 0.2

Cor = cortex; CD = collecting duct; EC = endothelial cell; G = glomerulus; IC = interstitial cell; IS = inner stripe; OS = outer stripe; Pap = Papilla;PT = proximal tubule.aTubular damage was more severe in the cortex (S2) than in the outer stripe (S3); tubular injury score: 0 = no damage, 1 = epithelium at least partiallypreserved, 2 = epithelium completely damaged.bSemiquantitative staining: 0 = no signals, 1+ = staining in <5% of tubular profiles or IC/EC, 2+ = 5–20%, 3+ = 20–33%; 4+ = 33–50%.∗P < 0.001, ∗∗P < 0.01, ∗∗∗P < 0.05 versus five control kidneys that were all immunonegative for all parameters (ANOVA).

had no major damage, suggesting that hypoxia developedin filtering nephrons. To test if MYO per se had inducedrenal hypoxia in GLY, PIM was investigated 3 h after thebeginning of a 45-min MYO infusion, but no signals weredetected (with the rare exception of some faint signals incortical PT, not shown).

Early activation of HIFα in GLY and MYO

Results of HIFα immunohistochemistry are outlined forGLY (Figure 5 and Table 2) and MYO (Figure 3 and Ta-ble 2). Control animals were negative for either HIFα iso-form. With occasional exceptions (intensity of papillaryimmunostaining), the two main isoforms, HIF-1α and -2α,were upregulated in a largely comparable pattern at 3 hof GLY and at 3 h of MYO, suggesting that MYO perse induces regional renal hypoxia and contributes to hy-poxia generation in GLY. This study largely confirms thepreviously described cell-type specificity of the HIF iso-forms [30–33]: HIF-1α mainly in tubules and in papillaryinterstitial cells, and HIF-2α exclusively in endothelial andinterstitial cells. But exceptionally, in both GLY and MYO,additional HIF-1α occurred in tubulo-interstitial capillaries(mostly in medullary rays and in the outer medulla, arrowin Figure 3E), pointing to a complex pathophysiology ofGLY/MYO. This study confirms that CD and the papillarycells hold a high potential for hypoxia adaptation. By con-trast, in PT, the most severely injured tubular segment, HIF,was only detectable at 3 h, when morphologic changes weremoderate. Later on, as severe injury developed, PT becamenegative for HIF, suggesting defective/exhausted hypoxiaadaptation. Interestingly, at 6 h PIM staining was maximal

(Figure 2C), but HIFα was negative for all renal cell types(Table 2).

Expression of the cell-protective HIF target gene HO-1after 6 h of GLY

To test whether HIF activation was accompanied by upreg-ulation of HIF target genes, we performed immunohisto-chemistry for HO-1. In controls and at 3 h after GLY, norenal HO-1 was detectable (not shown), but strong tubularHO-1 signals occurred at 6 h after GLY (Figures 2E and6A–D, Table 2) in cell types with previous HIF-1α acti-vation (compare Figure 5), suggesting hypoxic induction:cortical PT (Figures 2E and 6A), S3 in the outer stripe(Figure 6B) and CD in all renal zones (Figure 6B–D). At24 h after GLY, HO-1 signals were less abundant and in-tense (Figure 2F), which is consistent with previous mRNAdata in the same model [37]. Interestingly, at the time pointof maximum HO-1 (6 h) HIF was undetectable (Table 2).Moreover, careful assessment of parallel sections showedthat in individual tubules coincident staining for HO-1, HIFand PIM were extremely rare, if ever detectable (comparerectangles in Figure 2). Taken together, such data suggest asequential rather than concomitant expression of PIM, HIFand HO-1 in the nephron.

The HIF target gene PHD-2 is activated after 6 h of GLY

To test if a negative feedback loop could be responsiblefor the lack of HIF expression at 6 and 24 h of GLY, weperformed immunohistochemistry for PHD-2, which is themain renal isoform of the key HIF degradation enzyme [38],

Dow



ovember 2021


Fig. 2. Morphology, and immunohistochemistry for pimonidazoleadducts and the cell-protective gene heme oxygenase-1 at 6 and 24 hof rhabdomyolysis-induced AKI in the renal cortex (low power). AKI wasinduced by IM GLY injection. A, C, E and B, D, F are parallel sections.A and B are H&E stainings. Rectangles mark corresponding regions. A =arcuate artery; halfed arrow indicates S1 segments of proximal tubules;G = glomerulus; HO-1 = heme oxygenase-1; V = arcuate vein; PIM = pi-monidazole. Poor illumination at the corners of panels (C) to (F) is due toinsufficiency of the photographic device, and was partially corrected elec-tronically. Morphology is quite well preserved at 6 h (A), but seriously al-tered at 24 h, with many of the tubules collapsed (suggesting non-filteringnephrons) and filled with casts (B). PIM is more pronounced at 6 h (C)than at 24 h (D). The same applies for HO-1 (E, F). At 6 h PIM signalslocate in groups of tubules forming stripes from the renal capsule towardsthe medulla, suggesting that they belong to the same nephron (white rect-angle in C). These deeply hypoxic nephrons obviously correspond withfiltering glomeruli (open tubules in A) and have limited histologic damage(A). Appearance of HO-1 as well, suggests a nephron-specific pattern andincludes S1 segments (E). But, there is little, if any, overlap between PIM(C) and HO-1 (E) expressing tubules. Magnification: 90×.

and itself an HIF target gene. In controls, PHD was detectedin virtually all cells of the papilla and inner stripe, in distaltubules of the outer stripe and medullary rays (not shown),as well as in distal tubules of the labyrinth (Figure 7C).This expression pattern is in line with previous mRNA andimmunohistochemical [38] data. At 6 h of GLY, additionalnuclear immunostaining appeared in glomeruli and in PTof the labyrinth (Figure 7A and B), medullary rays (Figure7A) and outer stripe (not shown). Thus, de novo PHD-2activation clearly occurred in formerly HIF-positive celltypes, which is consistent with a negative feedback loop tostop HIF activation. This staining pattern persisted at 24,but less intensely and less abundantly in PT.

Discussion

This is a proof of concept study providing immunohis-tochemical evidence for hypoxia and hypoxia adaptation

Fig. 3. Morphology and HIFs at 3 h of MYO infusion. MYO was infusedIV for 45 min under ketamin anaesthesia. Kidneys were perfusion-fixedat 3 h after the start of MYO infusion. Arrow = endothelial cell (EC);arrowhead = interstitial cell (IC); asterisk = medullary thick ascendinglimb (mTAL); CD = collecting duct; DT = distal tubule (discrimina-tion between distal nephron segments was impossible in this material);PT = proximal tubule; S3 = S3 portion of the proximal tubule; T = thinlimb; VB = vascular bundles. Histology partly resembled that at 3 h afterIM GLY injection, since tubular casts and capillary congestion were de-tectable in all renal zones (A, D, G, J). Moreover, HIFs largely paralleledthe findings at 3 h of GLY (compare Figure 5): HIF-1α in the cortex in PT,DT and CD (B), in the outer stripe in S3 and CD (E), in the inner stripein CD and mTAL (H), and in the papilla in CD and IC (K); HIF-2α in ICand EC of all renal zones (C, F, I, L). However, in striking contrast withGLY, no major tubular damage occurred. Moreover, HIF activation slightlydiffered from GLY, in that EC occasionally were positive for HIF-1α (E),glomerules (C) and vascular bundles (I) were positive for HIF-2α, and thepapilla was less intensely stained (K, L). Magnification: 600×, except forA, D, G, J (250×).

in rhabdomyolysis-induced AKI. Our data suggest that(a) renal hypoxia and hypoxia adaptation occur early inexperimental rhabdomyolysis-induced AKI caused by IMGLY injection; (b) MYO per se contributes to hypoxia gen-eration; (c) the extent of tubular hypoxia relates to celldamage and (d) insufficient hypoxia adaptation in proxi-mal tubular cells may explain their marked injury in theGLY injection model.

Dow



ovember 2021


Fig. 4. Immunohistochemistry for PIM adducts at 3 h of rhabdomyolysis-induced Rh-AKI. AKI was induced by IM GLY injection. Arrowhead =interstitial cell (IC); asterisk = medullary thick ascending limb (mTAL);CD = collecting duct; PT = proximal tubule; S3 = S3-portion of theproximal tubule. In (D) the renal pelvis and parapelvine portions of thekidney were artificially covered with grey for a better distinction of thepapilla. Deep hypoxia was detectable in the superficial cortex (A, PTand CD), in S3 of the outer stripe (B), in mTAL and CD of the innerstripe (C), as well as in the interstitial compartment of the papilla (D, E).Magnification: A–D: 600×; E: ×90×.

Theoretically, the validity of hypoxia detection tech-niques used herein may be questioned, but a considerableamount of evidence backs the concept that, indeed, in thisstudy PIM and HIF immunostaining correspond with cel-lular hypoxia. First, elegant studies have proven that PIMbinding to tissues is highly reliable at a pO2 below 10 mmHgand largely independent of pH and redox state [28,29]. Sec-ond, HIF activation was quite rapid (within 3 h), withinthe time frame previously reported for hypoxic inductionin whole animals [32]. To the best of our knowledge, non-hypoxic HIF activation has not been demonstrated in wholeanimals, so far, but most likely would require more time toestablish than hypoxic HIF activation. In cell cultures, hy-poxic HIF activation is instantaneous [39] and based on im-paired HIFα degradation, whereas non-hypoxic HIF upreg-ulation has been shown after hours, and relies on increasedHIF-1α synthesis via activation of the phosphatidylinos-itol 3-kinase (PI3K) or mitogen-activated protein kinase(MAPK) pathways ([26,40] and references therein).

Fig. 5. Immunohistochemistry for HIFs at 3 h of rhabdomyolysis-inducedAKI. AKI was induced by IM GLY injection. Arrow = endothelial cell(EC); arrowhead = interstitial cell (IC); asterisk = medullary thick ascend-ing limb (mTAL); CD = collecting duct; DT = distal tubule (discrimi-nation between distal nephron segments was impossible in this material);PT = proximal tubule; S3 = S3 portion of the proximal tubule; T = thinlimb; VB = vascular bundles. HIFα occurred in all renal zones, HIF-1α

mainly in tubular profiles and in papillary IC and HIF-2α exclusively inEC/IC. Some HIF-2α signals clearly located in EC, but in most cases dis-tinction between IC/EC was impossible. Tubular HIF-1α mainly locatedin PT and CD of the superficial cortex (A), in S3 of the outer stripe (C),in CD and mTAL of the inner stripe (E), and in CD of the papilla (G).Magnifications: A–F and inset in H: 400×; G and H: 90×.

PIM staining suggests a more severe/protracted renal hy-poxia in GLY when compared with MYO, which is in linewith much less tubular damage [18] in the latter. To ourknowledge, the two experimental models have not beencompared directly, so far. Theoretically, the extent andkinetics of myoglobinuria should be different. Moreover,cofactors like intravascular volume depletion and acido-sis likely aggravate hypoxia in GLY. It is noteworthy thatsome authors [5] have shown that acidosis enhances thetubulotoxicity of MYO. Jiang et al. have shown that HIFactivation is half maximal at 1.5–2% oxygen [41], whichcorresponds with 11–15 mmHg, whereas Gross et al. havereported that nitroimidazole binding to tissue occurs below

Dow



ovember 2021


Fig. 6. Immunohistochemistry for the cell-protective HIF target gene HO-1 at 6 h of rhabdomyolysis-induced AKI (higher power). AKI was inducedby IM GLY injection. CD = collecting duct; G = glomerulus, PT =proximal tubule (probably S2); S1 = S1 portion of the proximal tubule;S3 = S3 segment of the proximal tubule. In this model at 6 h, HO-1induction is maximal, and predominantly locates in nephron segmentswhich had upregulated HIF-1α at 3 h (compare Figure 2): in the cortex(A) in PT including S1, in the outer stripe in S3 and CD (B), in the innerstripe (C) and in the papilla (D) in CD. Magnification: 250×.

10 mmHg [28]. We believe that in MYO, renal hypoxia wasmore moderate, below the threshold for PIM detection.

We show that the cell-protective HIF target gene HO-1appears de novo and that HIF precedes HO-1 by severalhours in the same cell types, which is consistent with hy-poxic upregulation of HO-1. The kinetics of HO-1—withhighest expression at 6 h after induction of injury—are inline with a previous mRNA study [37]. Interestingly, in thecortex of rats with rhabdomyolysis-induced AKI Ishizukaet al. have only found HO-1 immunostaining in what theycalled ‘distal tubules’ [42], which at least from the pho-tographs, in our opinion were collecting ducts. The factthat, in addition, we detected HO-1 in PT may reflect ahigher sensitivity of our staining technique. Obviously, atleast some PT are able to mount an adaptational responseto hypoxia.

We report considerable variation of tubular damage, PIM,HIF and HO-1 among different nephrons, as well as betweendifferent nephron segments. Nephron segment-specific HIFactivation has been demonstrated in several previous stud-ies with different hypoxic stimuli [30–33]. As to differencesbetween nephrons, the most likely explanation is that fil-tering and non-filtering nephrons exist side by side, as hasbeen revealed by micropuncture studies [43,44]. We havepreviously shown that functioning freshly engrafted humankidney transplants activate HIF, whereas non-functioninggrafts are HIF-negative [36]. Possibly, glomerular filtrationand energy consumption for tubular salt reabsorption areprerequisites for the development of tubular hypoxia andHIF activation.

Fig. 7. Immunohistochemistry for the HIF target gene HIF prolyl-hydroxylase-2 at 6 h of rhabdomyolysis-induced AKI. AKI was inducedby IM GLY injection. Halfed arrows point to medullary rays; DT = ‘distaltubules’, which were not further characterized, potentially distal convo-luted tubules, connecting tubules, thick ascending limbs or collectingducts; G = glomerulus; PT = proximal tubules; (B) is a higher power ofthe region outlined by the rectangle in (A). In the cortex of control kidneys(C) cytoplasmic signals are detectable in ‘distal tubules’, but proximaltubules and glomeruli are negative. In AKI (A, B), ‘distal tubules’ remainpositive, but additional nuclear signals appear in glomeruli and in proximaltubules. Magnification: A: 100×, B, C: 250×.

Despite of the proximity of PIM and HIF signals, therewas virtually no colocation within individual cells. This isin line with previous [36] and [30] reports that HIF activa-tion is maximal at moderate hypoxia, whereas PIM adductsoccur at more severe hypoxia. Thus, both methods of hy-poxia detection are complementary rather than redundant.As to the missing overlap between HIF and HO-1, HIF isshort-lived and its peak activity precedes that of its targetgenes by several hours. It is noteworthy that PHD-2 is anHIF target gene capable of shutting off HIF activation [45].Such a negative feedback loop might explain the lack ofHIF in PT at later stages of Rh-AKI.

In conclusion, in experimental rhabdomyolysis-inducedAKI, location and kinetics of overt cell injury on the one

Dow



ovember 2021


hand and immunohistochemical markers on the other handsuggest that hypoxia occurs within the kidney and thatit contributes to cell damage. Transcriptional adaptationto hypoxia seems to be limited to certain cells and to arelatively short period of time. Therefore, extending suchadaptational responses might be a reasonable therapeuticapproach in rhabdomyolysis.

Acknowledgements. This work, formerly presented in part as an abstract(the 2007 ISN World Congress of Nephrology, Rio de Janeiro, Brazil),was supported by the Russell Berrie Foundation and D-Cure, DiabetesCare in Israel, by the Open European Nephrology Science Centre, and bythe Harvard Medical Faculty Physicians at Beth Israel Deaconess MedicalCenter, Boston, MA.

Conflict of interest statement. None declared.

References

1. Huerta-Alardın AL, Joseph Varon J, Marik PE. Bench-to-bedside re-view: Rhabdomyolysis—an overview for clinicians. Crit. Care 2005;9: 158–169

2. Malinoski DJ, Slater MS, Mullins RJ. Crush injury and rhabdomyol-ysis. Crit Care Clin 2004; 20: 171–192

3. Vanholder R, Sever MS, Erek E et al. Rhabdomyolysis. J Am SocNephrol 2000; 11: 1553–1561

4. Abassi ZA, Hoffman A, Better OS. Acute renal failure complicatingmuscle crush injury. Semin Nephrol 1998; 18: 558–565

5. Clyne DH, Kant KS, Pesce AJ et al. Nephrotoxicity of low molecularweight serum proteins: physicochemical interactions between myo-globin, hemoglobin, bence-jones proteins and tamm-horsfall muco-protein. Curr Probl Clin Biochem 1979; 9: 299–308

6. Flogel U, Merx MW, Godecke A et al. Myoglobin: a scavenger ofbioactive NO. Proc Natl Acad Sci USA 2001; 98: 735–740

7. Moore KP, Holt SG, Patel RP et al. A causative role for redox cyclingof myoglobin and ist inhibition by alkalinization in the pathogenesisand treatment of rhabdomyolysis-induced renal failure. J Biol Chem1998; 273: 31731–31737

8. Shah SV, Walker PD. Evidence suggesting a role for hydroxyl radicalin glycerol-induced acute renal failure. Am J Physiol (Renal FluidElectrolyte Physiol) 1988; 255: F438–443

9. Nath KA, Balla J, Croatt AJ et al. Heme protein-mediated renal injury:a protective role for 21-aminosteroids in vitro and in vivo. Kidney Int1995; 47: 592–602

10. Paller MS. Hemoglobin- and myoglobin-induced acute renal failurein rats. Role of iron in nephrotoxicity. Am J Physiol (Renal FluidElectrolyte Physiol) 1988; 255: F539–F544

11. Salahudeen AK, Wang C, Bigler SA et al. Synergistic renal pro-tection by combining alkaline-diuresis with lipid peroxidation in-hibitors in rhabdomyolysis: possible interaction between oxidant andnon-oxidant mechanisms. Nephrol Dial Transplant 1996; 11: 635–642

12. Rassaf T, Flogel U, Drexhage C et al. Nitrite reductase function ofdeoxymyoglobin: oxygen sensor and regulator of cardiac energeticsand function. Circ Res 2007; 100: 1749–1754

13. Hsu CH, Kurtz TW, Waldinger TP. Cardiac output and renal bloodflow in glycerol-induced acute renal failure in the rat. Circ Res 1977;40: 178–182

14. Ayer G, Grandchamp A, Wyler T et al. Intrarenal hemodynamics inglycerol-induced myohemoglobinuric acute renal failure in the rat.Circ Res 1971; 29: 128–135

15. Bywaters EGL, Beall D. Crush injuries with impairment of renalfunction. Br Med J 1941; 1: 427–432

16. Herrera GA. Myoglobin and the kidney, an overview. UltrastructPathol 1994; 18: 113–117

17. Abdul-Ezz SR, Walker PD, Shah SV. Role of glutathione in an animalmodel of myoglobinuric acute renal failure. Proc Nati Acad Sci USA1991; 88: 9833–9837

18. Heyman SN, Rosen S, Fuchs S et al. Myoglobinuric acute renal failurein the rat: a role for medullary hypoperfusion, hypoxia, and tubularobstruction. J Am Soc Nephrol 1996; 7: 1066–1074

19. Maxwell PH. HIF-1’s relationship to oxygen, simple yet sophisticated.Cell Cycle 2004; 3: 156–159

20. Metzen E, Ratcliffe PJ. HIF hydroxylation and cellular oxygen sens-ing. Biol Chem 2004; 385: 223–230

21. Semenza GL. Hydroxylation of HIF-1: oxygen sensing at the molec-ular level. Physiology 2004; 19: 176–182

22. Ivan M, Kondo K, Yang H et al. HIFalpha targeted for VHL-mediateddestruction by proline hydroxylation: implications for O2 sensing.Science 2001; 292: 464–468

23. Jaakkola P, Mole DR, Tian YM et al. Targeting of HIF-alpha to thevon Hippel-Lindau ubiquitylation complex by O2-regulated prolylhydroxylation. Science 2001; 292: 468–472

24. Epstein ACR, Gleadle JM, McNeill LA et al. C. elegans EGL-9 andmammalian homologs define a family of dioxygenases that regulateHIF by prolyl hydroxylation. Cell 2001; 107: 43–54

25. Haddad JJ, Harb HL. Cytokines and the regulation of the hypoxia-inducible factor (HIF)-1α. Int Immunopharmacol 2005; 5: 461–483

26. Hellwig-Burgel T, Stiehl DP, Wagner AE et al. Review: hypoxia-inducible factor-1 (HIF-1): a novel transcription factor in immunereactions. J Interferon Cytokine Res 2005; 25: 297–310

27. Otterbein LE, Soares MP, Yamashita A et al. Heme oxygenase-1:unleashing the protective properties of heme. Trends Immunol 2003;24: 449–455

28. Gross MW, Karbach U, Groebe K et al. Calibration of misonida-zole labeling by simultaneous measurements of oxygen tension andlabeling density in multicellular spheroids. Int J Cancer 1995; 61:567–573

29. Arteel GE, Thurmann RG, Raleigh JA. Reductive metabolism of thehypoxia marker pimonidazole is regulated by oxygen tension inde-pendent of the pyridine nucleotide redox state. Eur J Biochem 1998;253: 743–750

30. Rosenberger C, Heyman SN, Rosen S et al. Up-regulation of HIF inexperimental acute renal failure: evidence for a protective transcrip-tional response to hypoxia. Kidney Int 2005; 67: 531–542

31. Rosenberger C, Griethe W, Gruber G et al. Cellular responses tohypoxia after renal segmental infarction. Kidney Int 2003; 64: 874–886

32. Rosenberger C, Mandriota S, Jurgensen JS et al. Expression ofhypoxia-inducible factor-1alpha and -2alpha in hypoxic and ischemicrat kidneys. J Am Soc Nephrol 2002; 13: 1721–1732

33. Goldfarb M, Rosenberger C, Abassi Z et al. Acute-on-chronic renalfailure in the rat: Functional compensation and hypoxia tolerance. AmJ Nephrol 2006; 26: 22–33

34. Hochachka PW, Buck LT, Doll CJ et al. Unifying theory of hypoxiatolerance: molecular/metabolic defense and rescue mechanisms forsurviving oxygen lack. Proc Natl Acad Sci USA 1996; 93: 9493–9498

35. Rosenberger C, Pratschke J, Rudolph B et al. Immunohistochemicaldetection of hypoxia-inducible factor-1alpha in human renal allograftbiopsies. J Am Soc Nephrol 2007; 18: 343–351

36. Rosenberger C, Rosen S, Shina A et al. Hypoxia inducible factors andtubular cell survival in isolated perfused kidneys. Kidney Int 2006;70: 60–70

37. Nath KA, Balla G, Vercellotti GM et al. Induction of heme oxygenaseis a rapid, protective response in rhabdomyolysis in the rat. J ClinInvest 1992; 90: 267–270

38. Li N, Yi F, Sundy CM et al. Expression and actions of HIF prolyl-4-hydroxylase in the rat kidneys. Am J Physiol Renal Physiol 2007;292: F207–F216

39. Jewell UR, Kvietikova I, Scheid A et al. Induction of HIF-1alpha inresponse to hypoxia is instantaneous. FASEB J 2001; 15: 1312–1314

40. Semenza GL. Targeting HIF-1 for cancer therapy. Nat Rev Cancer2003; 3: 721–732

41. Jiang BH, Semenza GL, Bauer C et al. Hypoxia-inducible factor-1levels vary exponentially over a physiologically relevant range of O2tension. Am J Physiol (Cell Physiol 40) 1996; 271: C1172–C1180

Dow



ovember 2021


42. Ishizuka S, Nagashima Y, Numata M et al. Regulation and immuno-histochemical analysis of stress protein heme oxygenase-1 in rat kid-ney with myoglobinuric acute renal failure. Biochem Biophys Res Com1997; 240: 93–98

43. Oken DE, Arce MT, Wilson DR. Glycerol-induced hemoglobinuricacute renal failure in the rat. I. Micropuncture study of the developmentof oliguria. J Clin Invest 1966; 45: 724–736

44. Oken DE, Dibona GF, McDonald FD. Micropuncture studies of therecovery phase of myohemoglobinuric acute renal failure in the rat. JClin Invest 1970; 49: 730–738

45. D’Angelo G, Duplan E, Boyer N et al. Hypoxia up-regulates pro-lyl hydroxylase activity: a feedback mechanism that limits HIF-1responses during reoxygenation. J Biol Chem 2003; 278: 38183–38187

Received for publication: 21.8.07Accepted in revised form: 16.10.07

Dow



ovember 2021

evidence for sustained renal hypoxia and transient hypoxia

Documents