Hypoxia-inducible factors and tubular cell survival in isolated perfused kidneys

original article

http://www.kidney-international.org & 2006 International Society of Nephrology

Hypoxia-inducible factors and tubular cell survival in isolated perfused kidneys C Rosenberger1, S Rosen2, A Shina3, W Bernhardt4, MS Wiesener4, U Frei1, K-U Eckardt4 and SN Heyman3 1

Nephrology and Medical Intensive Care, Charite´ University Clinic, Berlin, Germany; 2Department of Pathology, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, USA; 3Department of Medicine, Hadassah Hospital, Mt Scopus and the Hebrew University Medical School, Jerusalem, Israel and 4Department of Nephrology and Hypertension, University of Erlangen-Nuremberg, Erlangen, Germany

Adaptation to hypoxic environment is conferred through hypoxia-inducible transcription factors (HIFs). We have previously shown that the HIF system is transiently activated in vivo in radiocontrast-induced acute renal failure, associated with profound hypoxia in the renal medulla. Medullary thick ascending limbs (mTALs), the most affected nephron segments in this model, were virtually unable to mount an adaptive HIF response. Here, we study correlations between oxygenation, HIF activation, and cell viability in a related ex vivo model, the isolated perfused rat kidney (IPK). In IPKs perfused with cell-free oxygenated medium, severe medullary hypoxic damage developed, affecting 4279% of mTALs in the mid-inner stripe. HIF-1a tubular immunostaining was noted with a zonal and tubular pattern largely similar to our findings in vivo: in 3473% of collecting ducts (CDs) within the mid-inner stripe and extensively in the papillary tip, whereas mTALs were all HIF-negative. In IPKs supplemented with RBCs (improved oxygen supply), mTAL damage was totally prevented and CDs’ HIF expression was attenuated (2274%). By contrast, although measures designed to reduce medullary hypoxia by decreasing tubular reabsorptive activity (furosemide, ouabain, or high-albumin-non-filtering system) reduced mTAL damage, all paradoxically resulted in increased HIF expression in CDs (5174%), and 1773% of mTALs became immunostained as well. Our data confirm that CDs and mTALs have markedly different HIF responses, which correlate with their viability under hypoxic stress. mTALs transcriptional adaptation occurs within a narrow hypoxic range, and it appears that workload reduction can shift mTALs into this window of opportunity for HIF activation and survival. Kidney International (2006) 70, 60–70. doi:10.1038/sj.ki.5000395; published online 17 May 2006 KEYWORDS: hypoxia-inducible factors; isolated perfused kidney; thick ascending limb; furosemide; ouabain; pimonidazole

Correspondence: C Rosenberger, Nephrology and Medical Intensive Care, Charite´ University Clinic, Campus Virchow, Augustenburger Platz 1, D-13353 Berlin, Germany. E-mail: [email protected] Received 2 November 2005; revised 29 January 2006; accepted 31 January 2006; published online 17 May 2006 60

Medullary hypoxic damage plays an important role in the development of acute and chronic renal dysfunction (for reviews see Brezis and Rosen,1 Molitoris and Sutton,2 and Nangaku3). Indeed, medullary hypoxic acute tubular necrosis has been documented under experimental settings, predominantly involving medullary thick ascending limbs of Henle’s loop (mTALs) and S3 segments in the outer medulla.4,5 Medullary injury is characterized by a particular gradient of damage, most pronounced in the mid-interbundle zone, away from vasa recta (for review see Brezis and Rosen1). Aside from frank necrosis, a less severe and reversible tubular hypoxic injury also exists, manifested morphologically by simplification of brush border in proximal tubular cells5–7 and mitochondrial swelling and nuclear pyknosis in mTALs.8 The extent of cellular energy store depletion probably governs apoptotic/necrotic death in some of these cells,9 whereas others, subjected to a more moderate and sublethal hypoxic stress, may exert adaptive responses and survive.10 Hypoxia-inducible factors (HIFs) are ubiquitous master regulators of such hypoxic adaptation (for reviews see Maxwell11 and Semenza12). HIFs are heterodimers of a constitutive b-subunit and one of two alternative a-subunits (HIF-1a and HIF-2a), which are regulated by oxygendependent proteolysis. HIF target genes govern a wide spectrum of biologic processes, for example, metabolism, vascular tone, erythropoiesis, angiogenesis, cell cycle, scavenging of free radicals, inflammation, immune response, and tumor growth (for reviews see Wenger13 and Bracken et al.14). Erythropoietin15 (for review see Chatterjee16) and heme oxygenase-1 (for review see Takahashi et al.17), two wellrecognized HIF target genes, proved cell protective in experimental renal injury. However, the role of the HIF system in vivo and its possible impact on renal disease is still incompletely understood. Using in vivo models, we have shown that HIF isoforms appear in tubular, interstitial, and vascular endothelial cells in animals exposed to atmospheric hypoxia,18 following global renal ischemia,18 and at the margin of infarcted regions within the kidney.19 HIF isoforms specifically appear in the renal medulla in response to regional hypoxia associated with Kidney International (2006) 70, 60–70

original article

C Rosenberger et al.: Hypoxia-inducible factors and tubular cell survival

20

30

40

50

60

70

80

90

100

0.5 0.4 0.3 0.2 0.1 *

* P<0.01 vs. CTR ** P<0.05 vs. CTR *** P <0.01 vs. RBC, ouabain

1.25 1.00

*

0.75 0.50

40

Time (minutes) 50 60 70

80

90

100

* ** RBC CTR Furosemide Ouabain

* P <0.01 vs. CTR ** P <0.05 vs. CTR

250 FEK (%)

0.00

95 90 85 80 75 70 65 60 55 50

30

300

**

0.25

Flow (ml/min)

Urine Volume (ml/10min)

0.0

20

RBC CTR Furosemide Ouabain NFK TRNa (%)

GFR (ml/min)

0.6

*

90 80 *** 70 60 50 40 30 20 10 20 30

*

200 150 100 50

* *

0 20

40

50 60 70 Time (minutes)

80

90

100

Figure 1 | Changes in GFR, urine production, and perfusate flow in isolated kidneys perfused over 90 min. Experimental groups: CTR ¼ control cell-free oxygenated perfusate; furosemide and ouabain ¼ agents added to the cell-free perfusate; NFK ¼ non-filtering kidney, albumin added to cell-free perfusate in order to abolish glomerular filtration; RBC ¼ 5% red blood cells added to the perfusate. Hypoxic stress is maximal in the CTR group, attenuated in the furosemide, ouabain, and NFK kidneys, and prevented in the RBC group.

diabetes (Heyman et al., J Am Soc Nephrol 2004; 15: 468A (abstract)), inhibition of prostaglandin or nitric oxide synthesis, radiocontrast administration,20 or owing to chronic tubulointerstitial fibrosis.21 In these studies, HIF expression, detected immunohistochemically, colocalizes with pimonidazole adducts, a sensitive indicator of critically low regional pO2,19–21 and with the expression of downstream HIF response genes such as heme oxygenase-1.18–20 Interestingly, in these studies, medullary tubular HIF expression was observed almost exclusively in the hypoxiaresistant collecting ducts (CDs). By contrast, hypoxia-sensitive mTALs were all HIF-negative during severe hypoxic stress, induced by radiocontrast, combined with the inhibition of prostaglandin and nitric oxide synthesis.20 Furthermore, mTALs unexpectedly became HIF-positive following the administration of the loop diuretic furosemide,20 known to block oxygen consumption for tubular transport, to alleviate medullary hypoxia,22 and to prevent mTAL hypoxic injury.23,24 One possible explanation for these findings is that tubular capability to mount an HIF response may confer protection. In addition, activation of HIF in mTALs might be restricted to a relatively narrow range of moderate hypoxia, and reduced oxygen consumption by the inhibition of Kidney International (2006) 70, 60–70

30

40

50 60 70 Time (minutes)

80

90

100

Figure 2 | Changes in tubular sodium reabsorption (TRNa) and fractional potassium excretion (FEK) in isolated kidneys perfused over 90 min. Experimental groups: CTR ¼ control cell-free oxygenated perfusate; furosemide and ouabain ¼ agents added to the cell-free perfusate; NFK ¼ non-filtering kidney, albumin added to cell-free perfusate in order to abolish glomerular filtration; RBC ¼ 5% red blood cells added to the perfusate. Hypoxic stress is maximal in the CTR group, attenuated in the furosemide, ouabain, and NFK kidneys, and prevented in the RBC group.

transport activity may shift tubular cells from severe lethal hypoxia into the window of opportunity for HIF activation and cell survival. To further challenge this hypothesis, we explored the impact of modifying oxygen supply and tubular transport upon medullary tubular cell survival and HIF expression in isolated perfused rat kidneys (IPKs). Isolated kidneys perfused with cell-free oxygenated medium are a well-recognized model of hypoxic cell injury, as tubular damage closely follows known renal oxygen gradients,25 and associates with increased reduced form of cytochrome aa3.26 In the present study, kidneys subjected to cell-free oxygenated medium (severe hypoxia) were compared with those supplemented with red blood cells (near-normal oxygen supply) or with kidneys in which transport activity was reduced in the setting of cell-free perfusate (attenuation of severe hypoxia). RESULTS Kidney function

The determinants of kidney function, displayed in Figures 1 and 2, are generally in accordance with the findings of previous studies, using similar manipulations.23 Noteworthy is the total prevention of glomerular filtration and urine production in the non-filtering kidney (NFK) group, and the marked diuresis and natriuresis observed in kidneys subjected to furosemide or ouabain. The RBC group tended to 61

original article

C Rosenberger et al.: Hypoxia-inducible factors and tubular cell survival

RBC

Control

Ouabain

Albumin

b

c

d

e

f

g

h

i

j

k

l

Papilla

Mid-inner stripe

Outer stripe

a

Figure 3 | Hematoxylin and eosin staining of the medulla of IPKs. Damaged tubules (with either nuclear pyknosis or frank necrosis) are indicated by arrows. Cell-free IPK preparations serve as controls. RBC ¼ IPK supplemented with red blood cells, whereas all other experimental groups were cell-free. No major morphologic changes occur (a–d) in the outer stripe or (i–l) in the papilla. (f) In the mid-inner stripe, tubular damage is greatest in controls. Improved oxygenation with (e) RBC or (g) reduced oxygen consumption with furosemide (not shown), ouabain, or (h) in non-filtering kidneys all result in preserved morphology. Original magnification  100.

have a more preserved glomerular filtration rate (GFR), and lower perfusate flow and urine volumes, but these parameters did not differ statistically from the control (CTR) group. Tubular damage

As previously described,1,8 in IPKs, tubular damage (comprising nuclear pyknosis or frank necrosis) is most pronounced in the mid-inner stripe of the outer medulla, whereas in all other renal zones (including the cortex, data not shown), morphology is quite well preserved. Figure 3 shows lowpower magnifications from hematoxylin and eosin sections from the medulla. Tubular damage is pronounced in CTR, but attenuated with maneuvers that improve medullary oxygenation. Morphologic differences between the experimental groups were confirmed by semithin preparations (Figure 4), where again, improved oxygen supply or reduced oxygen consumption go together with tubular preservation. As expected, the two major tubular components of the mid-inner stripe, mTALs and CDs, showed a markedly different extent of damage, with CDs being virtually unharmed. mTAL damage 62

followed well-established renal oxygen gradients, being more pronounced with increasing distance from vascular bundles, thus clearly supporting a hypoxic injury (Figure 4b). Semiquantitative assessement (Figure 9, lower panel) revealed that differences were significant in each experimental group with respect to CTR. In the latter, 42% of mTALs in the mid-inner stripe became severely injured. By contrast, in IPKs supplemented with RBCs, medullary mTAL damage was completely prevented. Similarly, attenuation of reabsorptive activity with furosemide, ouabain, or in the non-filtering system resulted in reduction of mTAL damage (involving 16, 1, and 10% of mTALs, respectively). Evidence for hypoxia in the renal medulla Pimonidazole adducts. Strong accumulation of pimonida-

zole was detected in the medulla of cell-free perfused kidneys (controls), whereas the cortex was void of signals (Figure 5). The staining pattern of the medulla is in concordance with known renal oxygen gradients (for review see Brezis and Rosen1), and parallels our previous results in radiocontrastinduced acute renal failure.20 The papilla shows almost Kidney International (2006) 70, 60–70

original article

C Rosenberger et al.: Hypoxia-inducible factors and tubular cell survival

a

a

Cortex

VB G

*

RBC

G

G CD VB

b

b

Outer stripe

Control

* CD

CD

A

VB *

CD * *

c

c *

*

Furosemide

Inner stripe VB

CD *

* VB * *

d CD

d

Papilla

*

*

Ouabain

* * *

* *

e

VB

*

Albumin

*

* * *

Figure 4 | Semithin sections of the mid-inner stripe. Cell-free IPK preparations serve as controls. RBC ¼ IPK supplemented with red blood cells. Asterisks mark mTALs; CD ¼ collecting ducts; VB ¼ vascular bundles. (b) Nuclear pyknosis or frank necrosis occurs mainly in mTALs, and most markedly in controls, and with increasing distance from vascular bundles (b). Improved oxygenation with (a) RBC or (c) reduced oxygen consumption with furosemide, (d) ouabain, or (e) in non-filtering kidneys all reduce the extent of mTAL damage. Original magnification  1000.

homogeneous staining (Figure 5d). Vascular bundles are negative, but strong staining appears in the inter-bundle zone in the inner stripe of the outer medulla (Figure 5c). In addition, in the outer stripe of the outer medulla, pimonidazole accumulation is restricted to a zone away from vascular bundles (Figure 5b). Oxygen gradients are most impressive on cross-sections of the mid-inner stripe (Figure 6a): relatively Kidney International (2006) 70, 60–70

Figure 5 | Immunohistochemical detection of pimonidazole adducts in cell-free perfused IPKs. The left-hand part of (a–c) points to the papillary tip. A ¼ arcuate artery; G ¼ glomerulus; VB ¼ vascular bundles. (a) The cortex is void of signals, (b–d) whereas strong signals appear in the medulla. (b) In the outer stripe of the outer medulla, pimonidazole adducts are detected in clusters of tubules away from vascular bundles. (c) In the inner stripe of the outer medulla, almost the entire inter-bundle zone is positive, except for tubules located next to vascular bundles. (d) The papilla shows almost homogeneous staining. Original magnification  90.

small pimonidazole-negative islands (containing the vascular bundles together with few adjacent tubules) stand in sharp contrast with intense staining that spreads over almost the entire inter-bundle zone. Inhibition of mTAL transport activity by ouabain considerably reduces the area of pimonidazole staining, which is now confined to a relatively thin network away from vascular bundles (Figure 6). HIF-1a expression. As previously described in vivo,18–20 renal HIF-1a immunostaining was almost exclusively found in tubular segments. In the CTR group, HIF-1a immunostaining appeared with a zonal and tubular pattern largely similar to our findings in experimental acute renal failure:20 strongest and most abundant signals were detected in the papilla. In the outer stripe of the inner medulla, HIF-1a 63

original article

C Rosenberger et al.: Hypoxia-inducible factors and tubular cell survival

a

b

Control

Ouabain VB

VB

VB

VB VB

VB

VB

VB VB

VB

VB VB

VB

c

d

VB

*

*

VB

VB

VB * VB

VB

e

f

VB

* * VB

VB

VB

VB

*

VB

Figure 6 | Effect of tubular workload on immunohistochemical detection of pimonidazole adducts and HIF-1a in the mid-inner stripe of the outer medulla. (a, c) VB ¼ vascular bundles; asterisks ¼ HIF-positive mTALs; C/E and D/F are parallel sections; the left panels show cell-free IPKs (controls), the right panels show cell-free IPKs supplemented with ouabain to reduce tubular workload and oxygen consumption. Cell-free IPKs exhibit strong pimonidazole signals of almost the entire inter-bundle zone. (b, d) With ouabain, the area of pimonidazole-staining is clearly reduced. In cell-free IPKs, HIF CD staining occurs mainly in the vicinity of vascular bundles (approximately 1–2 tubular diameters from VBs). (e) The area immediately adjacent to VBs (lowest hypoxia), and the area most remote from VBs (deepest hypoxia) are virtually void of HIF signals. (f) Paradoxically, in cell-free IPKs supplemented with ouabain, CD HIF is increased, but shifted toward the area most remote from VBs (which now most likely correspond with moderate hypoxia. Moreover, some mTALs become positive as well. Original magnification: (a, b)  110; (c–f)  200.

signals were clustered at the base of medullary rays. In the mid-inner stripe of the outer medulla, 3573% of CDs were positive, whereas mTALs were nearly all HIF-negative (Figures 7 and 9). Parallel sections stained for pimonidazole (Figure 6c) and HIF-1a (Figure 6e) reveal that most HIF signals in CDs are clustered halfway from the vascular bundles (region of best oxygenation) to the mid-interbundle zone (region of poorest oxygenation), presumably corresponding to an intermediate degree of oxygenation. In the RBC group (improved oxygen supply), CD HIF-1a expression was attenuated, involving 2274% of CDs. By contrast, measures designed to reduce medullary hypoxic damage by decreasing tubular reabsorptive activity (furosemide, ouabain, or non-filtering kidney groups), all paradoxically resulted in increased HIF expression in CDs (5174% for the combined means of the three groups) and 1773% of mTALs became immunostained as well (Figures 7 and 9). 64

This correlated well with tissue preservation, as described above, suggesting a causal relationship between HIF activation and cellular preservation. It is noteworthy that, as shown in Figure 6f, HIF-1a signals, more abundant now in CDs and expressed de novo in mTALs, are somewhat displaced outwards from the vascular bundles to the midinterbundle zone, where hypoxia still exists (as shown with parallel sections stained for pimonidazole; Figure 6d). Papillary and outer stripe HIF-1a immunoreactivity was unaffected by the various manipulations (Figure 7). HIF-2a expression. Renal medullary HIF-2a immunostaining was predominantly observed in vascular endothelial and interstitial cells (Figure 8), which largely parallels our previous observations in vivo.20 In addition, in the IPKs, we detected strong HIF-2a staining of endothelial cells within the vasa recta (Figure 8), which is different from our in vivo data. Unlike the prominent inter-group variation of tubular Kidney International (2006) 70, 60–70

original article

C Rosenberger et al.: Hypoxia-inducible factors and tubular cell survival

RBC

Control

Outer stripe

a

Ouabain

b CD

CD

CD

S3

S3

d

S3

S3

CD CD

S3

S3

Albumin

c

CD

CD S3

S3

S3

S3

e

f

g

h

Mid-inner stripe

* CD

*

*

CD

CD

CD

*

CD

*

CD *

*

CD CD

i

j CD

CD Papilla

k CD

l CD CD CD

CD

Figure 7 | Expression of HIF-1a in the medulla. Cell-free IPK preparations serve as controls. RBC ¼ IPK supplemented with red blood cells. Asterisks ¼ mTALs; S3 ¼ S3 segments of proximal tubulues; CD ¼ collecting ducts; (a–d) In the outer stripe of the outer medulla, HIF-1a appears in CDs and in S3, with apparently no major difference between the experimental groups. (i–l) In the papilla, HIF-1a appears in CDs and in interstitial cells, again with no clear difference between groups. (f) By contrast, in the mid-inner stripe, expression is confined to tubular sections and with marked variation between experimental groups and tubular subtypes (e, improved oxygenation): deeply hypoxic control IPKs show signals exclusively in CDs. CD HIF-1a is attenuated in RBC. Reduced oxygen consumption with either furosemide (not shown), (g) ouabain, or (h) in non-filtering kidneys paradoxically increases CD HIF-1a, and some mTALs become stained as well. Original magnification  250.

HIF-1a expression, the various protocols did not grossly affect the extent of medullary HIF-2a expression.

phenomenon had no major impact on the experimental system.

Indication for hypoxia and hypoxic adaptation in the renal cortex

DISCUSSION

Cortical tubular immunostaining for HIF-1a was negative in all experimental groups, with the unforeseen exception of the RBC kidneys, which occasionally showed immunopositivity in some connecting tubules and CDs (Figure 10). Cortical HIF-2a was again negative in all groups with the exception of kidneys subjected to RBC supplementation. HIF-2a immunoreactivity was observed in some glomeruli (capillary endothelial cells and probably mesangial cells), as well as in interstitial and capillary endothelial cells of the tubulointerstitium (Figure 10). Distribution of signals was segmental rather than diffuse, which might be owing to regional alterations of microcirculation upon perfusion with RBCs. However, there was no detectable histologic alteration in areas of cortical HIF activation, suggesting that the observed Kidney International (2006) 70, 60–70

The discovery of the HIF system markedly expands our insight regarding the cellular adaptive response to hypoxia (for reviews see Maxwell,11 Semenza,12 Wenger,13 and Bracken et al.14). Renal parenchymal hypoxia is believed to play a central role in the development of acute tubular necrosis and in the progression of chronic tubulointerstitial diseases (for reviews see Brezis and Rosen,1 Molitoris and Sutton,2 and Nangaku3). Hence, the establishment of knowledge of the cellular location and kinetics of HIF during changes in renal oxygenation, as well as its relationship with cell vulnerability, is likely to contribute to a better understanding of renal pathophysiology. IPKs perfused with cell-free oxygenated medium (control ¼ CTR) exhibit severe tissue hypoxia, despite high perfusate pO2. Epstein et al.26 found that 25–40% of cytochrome aa3 65

original article

C Rosenberger et al.: Hypoxia-inducible factors and tubular cell survival

Medulla, control: HIF-2


Renal morphology and HIF-1
Score

a Outer stripe

HIF-Positive CD (%)

70

b

Vascular bundle

*

60 50 40 30

**

20 10 0

RBC

CTR

Furos. Ouab.

NFK

Inner stripe

VB

c

Inter-bundle zone

HIF-Positive mTALs (%)

30 *

25 20 15 10 5 0

RBC

CTR

Furos.

Ouab.

NFK

Papilla

d

OM Tubular Injury (%)

60 50 40 30 * 20 10 0

* RBC

CTR

Furos.

Ouab.

NFK

* P <0.01; ** P<0.04 vs. CTR, ANOVA

Figure 8 | Expression of HIF-2a in the medulla. VB ¼ vascular bundles. Only controls (cell-free pefusion, equivalent to greatest hypoxia) are depicted, as no major differences were detectable between experimental groups. Signals appear in interstitial and capillary endothelial cells. (a) In the outer stripe of the outer medulla, HIF-2a is clustered away from vascular bundles. (b) In the inner stripe, signals appear in the inter-bundle zone, (c) as well as in vascular bundles. (d) Strongest and most abundand staining occurs in the papilla. Original magnification  250.

appears in its reduced form under these conditions. Indeed, in the absence of oxygen binding hemoglobin, arterial O2 content is only 1.5%, as compared with 7.1% in RBCenriched medium.25 The addition of loop diuretics to the perfusate markedly reduces the levels of reduced cytochrome aa326 in cell-free IPKs, and these agents were also found to increase renal medullary pO2 from 16 to 35 mm Hg in vivo.22 However, to our knowledge, tissue oxygen tensions have not been assessed in the IPK, so far. The present work, by using pimonidazole adduct detection, is the first to clearly demonstrate tissue oxygen gradients in this ex vivo model. Furthermore, the IPK allows for a close control of renal function and tissue oxygenation with a variety of interventions.27 Some of the manipulations employed in the present study recapitulate our previous in vivo experiments20 (delivery of furosemide to block mTAL oxygen consumption), but others (general blockade of oxygen consumption 66

Figure 9 | Outer medullary morphology and HIF-1a expression in isolated kidneys perfused over 90 min: the extent of mTAL necrosis and of HIF expression in mTALs and CD in the mid-inner stripe are expressed as the percentage of affected tubules out of all tubules counted in this region. Experimental groups: CTR ¼ control cell-free oxygenated perfusate; furosemide and ouabain ¼ agents added to the cell-free perfusate; NFK ¼ nonfiltering kidney, albumin added to cell-free perfusate in order to abolish glomerular filtration; RBC ¼ 5% red blood cells added to the perfusate. Hypoxic stress is maximal in the CTR group, attenuated in the furosemide, ouabain, and NFK kidneys, and almost prevented in the RBC group.

through ouabain or the cessation of GFR with high perfusate albumin) clearly expand beyond whole animal studies. As previously reported by Brezis,23 we show that severe tissue hypoxia in the cell-free perfused kidney (CTR) leads to medullary hypoxic damage, which is virtually absent in RBCIPKs and is ameliorated by the attenuation of reabsorptive activity with loop diuretics, with ouabain, or in the nonfiltering kidney. Moreover, our pimonidazole-binding studies are a proof of principle, demonstrating for the first time that ouabain considerably ameliorates hypoxia in the mid-inner stripe of the outer medulla – the zone where oxygen consumption is the major determinant of tissue oxygen content. As pointed out in Table 1, despite different types of those manipulations, the pattern of both HIF activation and pimonidazole accumulation in the IPK is remarkably consistent and in concordance with our previous data in experimental radiocontrast-induced acute renal failure.20 Kidney International (2006) 70, 60–70

original article

C Rosenberger et al.: Hypoxia-inducible factors and tubular cell survival

Table 1 | Comparison of HIF and pimonidazole staining in experimental ARF and the IPK Radiocontrast-induced acute renal failurea

Cell-free perfused isolated kidney

PIMb

HIF-1a

HIF-2a

PIMb

HIF-1a

HIF-2a













+ + + 

+ +  

   +

+ + + 

+ +  

   +

Inner stripe of the outer medulla Collecting ducts Thick ascending limbs Thin limbs Capillary endothelial cells Vascular bundles

+ + +c c 

+  +  

   + 

+ + +c c 

+  +  

   + +

Papilla Collecting ducts Thin limbs Capillaries

+ + c

+ + 

  +

+ +d c,d

+ +d d

 d +d

Cortex Outer stripe of the outer medulla Proximal tubules (S3)b Collecting ductsb Thick ascending limbsb Capillary endothelial cellsb

ARF, acute renal failure; HIF, hypoxia-inducible factor; IPK, isolated perfused kidney; PIM, pimonidazole. Signal abundance or intensity is not presented. a According to Rosenberger et al.20 b Only in clusters and away from vascular bundles. c Discrimination between thin limbs and capillaries is sometimes impossible, and therefore, only a gross appreciation is provided. d No counter-staining with endothelial markers was performed in the IPK, but the staining pattern is very much like in radiocontrast-induced acute renal failure.20

Taken together, one must conclude that transcriptional adaptational response to hypoxia is quite different along the nephron segments. As illustrated in Figures 7 and 9, CDs are the predominant location of inner medullary HIF-1a throughout the experimental groups and possess a strong potential to activate HIF over a wide hypoxic range. Modest expression of HIF-1a in CDs is observed in the near-normally oxygenated medulla (RBC group). As anticipated, HIF-1a expression is almost doubled under severe hypoxia (cell-free perfusate). Unexpectedly, however, it further increases with measures that improve medullary oxygenation in the cell-free IPK. Although such paradoxical increase in HIF expression may suggest that other mechanisms besides hypoxia could have influenced HIF activity, careful assessment of signal distributions favors that HIF upregulation is mainly oxygendependent: Whereas in cell-free IPKs (controls), HIF principally locates halfway between vasa recta and the midinterbundle zone (presumably corresponding to an intermediate degree of oxygenation), during workload reduction HIF appears in the mid-interbundle zone (seemingly, now the area with an intermediate degree of hypoxia). Thus, in both experimental conditions, HIF positivity could occur at comparable absolute tissue oxygen tensions, which most likely correspond with moderate hypoxia. In sharp contrast with the CDs, mTALs seem to have a poor ability to upregulate HIF, which is restricted to a narrow and mild hypoxic range: HIF was absent in mTALs in the RBC group, as well as under conditions of severe hypoxia associated with cell-free IPK (control group). However, mTAL HIF expression appeared in cell-free IPKs when medullary oxygenation improved (with either furosemide, ouabain, or albumin), in parallel with an increase in CD-HIF. Kidney International (2006) 70, 60–70

Our results are in accordance with cell culture experiments showing that HIF activation diminishes with severe hypoxia and anoxia.28 The molecular mechanisms of this decrease are still unclear. One possible explanation is a crosstalk between the HIF and the p53 system.29 The latter is being activated in severe and protracted hypoxia, and is capable of turning off the HIF response. Thus, a switch from HIF to p53 could form a reasonable mechanism to induce apoptosis, once that hypoxia is incompatible with cell survival. The very restricted HIF response in mTALs is more difficult to explain. Superoxide anions have been shown to reduce HIF levels in hypoxic cell cultures, probably through the activation of proteasomal degradation of HIF.30 Interestingly, hypoxic mTALs have been shown to produce large amounts of superoxide anions via NAD(P)H oxidase (nicotinamide adenine dinucleotide (phosphate) (reduced form).31 Taken together, these data offer a possible explanation for the restricted HIF response in hypoxic mTALs, as well as for the increased HIF expression with furosemide, ouabain, or in NFK, as shown in the present study. Nevertheless, the influence of reactive oxygen species on HIF is complex,32 and a variety of other factors like growth hormones and cytokines were also shown to modulate HIF activation in vitro (for review see Rosenberger et al.33). The present study provides important clues regarding the biological role of HIF. We suggest that HIF constitues a survival factor under acute renal medullary hypoxia. Supporting this view, CDs easily activate HIF, and show no major signs of structural alteration despite severe hypoxia, whereas mTAL fail to mount an adaptative HIF response and develop overt sublethal/lethal injury. Moreover, maneuvers that increase renal medullary oxygenation also lead to both 67

original article

C Rosenberger et al.: Hypoxia-inducible factors and tubular cell survival

mTAL preservation and mTAL HIF upregulation. The wide spectrum of cell-protective HIF target genes seems to back this hypothesis. However, determination of the cause-andeffect relationship between HIF-1a activation and survival is yet to be defined: HIF might exert cellular protection and reverse sublethal injury, or vice versa: dying cells may be unable to generate HIF. Beyond the increasingly evident role of HIF in acute renal injury, some data also suggest a physiologic role of HIF in the renal medulla. Recently, Monotham et al.34 reported rare HIF-1a immunohistochemical staining in the inner medulla and deep inner stripe of control rats, which increased after 36 h of water deprivation. The literature is conflicting with respect to HIF-1a activation in normal kidneys, and there seems to exist a difference between species (for review see Rosenberger et al.33). The second major oxygen-dependent isoform HIF-2a paralleled the zonal distribution of HIF-1a. However, cellular location was distinct (strictly non-parenchymal), and no major variation was detectable between experimental groups. We have previously described a similar cellular disparity between HIF-1a and -2a in vivo,18–20 whereas in most cell cultures, both HIF isoforms are equally activated by hypoxia.35 This supports the view that the IPK, despite being an ex vivo model, preserves important features with renal injury models of native kidneys. The fact that HIF-2a was quite uniformly expressed in the medulla irrespective of experimental manipulations may have at least two explanations. First, variation of intracellular oxygen tension in interstitial cells may be less pronounced than in tubular cells, whose reabsorptive work was modulated in this study. Second, HIF-2a activation may have reached a plateau throughout the experimental conditions. The detection of both HIF-1a and HIF-2a in the cortex of kidneys supplemented with RBCs is unexpected, contradicts their absence in cortical tissue in intact animals18 and in cell-free perfused kidneys, and is not associated by any morphologic injury (Figure 10). Cortical HIF activation is segmental rather than diffuse, and confined to superficial nephrons. It could be that some regional cortical flow irregularities promoted hypoxia, and induced HIF in a small subset of nephrons. In conclusion, our findings using the IPK model suggest that medullary tubular cells respond to sublethal hypoxic stress by increased HIF-1a expression. By contrast, under conditions of lethal hypoxic stress, outer medullary HIF-1arelated adaptive cell responses diminish or disappear. On the one hand, the cell-free IPK is likely to induce more pronounced and more sustained mTAL hypoxia with subsequent extensive cellular damage than in clinical acute tubular necrosis. On the other hand, however, the marked histologic alterations form a strength of this model, as the effect of various protective manipulations becomes more evident. Hence, studying HIF expression in the IPK may be considered a somehow exaggerated, but highly reproducible and instructive model of renal injury and injury prevention.

Cortex: HIF-1
, HIF-2


68

a

b Control

Red blood cells

c

d Control

Red blood cells CNT CD CD

Figure 10 | Expression of HIF-1a and -2a in the cortex. CD ¼ collecting ducts; CNT ¼ connecting tubules; (controls: a, c) no signals were detectable in cell-free isolated perfused kidneys. Surprisingly, only in IPKs supplemented with red blood cells (improved oxygen supply), both HIF-a-isoforms were detectable in the cortex: (d) HIF-1a in CDs and in CNTs, and (b) HIF-2a in glomerular cells, as well as in interstitial and capillary endothelial cells of the tubulo-interstitium. Original magnification  300.

MATERIALS AND METHODS IPK model, general Studies were conducted in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals. All chemicals were purchased from Sigma Co. Male Sprague–Dawley rats (280–320 g), fed on regular chow and allowed free excess to water, were used for all experiments. IPK was performed according to the method of Ross et al.36 Perfusion media consisted of a Krebs–Ringer–Henseleit solution containing (in mM) Na 143, K 4.5, bicarbonate 24, Ca 2.5, Mg 1.2, phosphate 1.2, and bovine serum albumin at a concentration of 6.7 g/dl (pH 7.4). The perfusate was supplemented with a mixture of 20 amino acids, as detailed previously.37 Kidney International (2006) 70, 60–70

original article

C Rosenberger et al.: Hypoxia-inducible factors and tubular cell survival

All perfusions, gassed with 5% CO2 and 95% O2, were carried out in a temperature-controlled chamber (381C) for 90 min at a constant pressure of 100 mm Hg at the catheter tip. The rate of perfusate flow was monitored by a Brooks flow meter in line. Sequential urine collection and perfusate samples were obtained for clearance measurements every 10 min after a 20 min period of stabilization. GFR was estimated from the clearance of 3H-inulin. Filtration fraction was extrapolated from the perfusate flow and GFR. Fractional sodium reabsorption and potassium excretion were calculated from their concentrations in the perfusate and urine, and the inulin clearance. At the conclusion of the experiments, kidneys were perfusion-fixed with 3% paraformaldehyde in phosphatebuffered saline (pH 7.4), stored in cooled phosphate-buffered saline, and processed for paraffin embedding. IPK model – experimental design Kidneys were subjected to the following five protocols: Oxygen-deprived kidneys: Oxygenated cell-free perfusion medium (control, CTR, n ¼ 11). Well-oxygenated kidneys: Oxygenated medium supplemented with fresh washed human RBCs at a target hematocrit of 10% (RBC, n ¼ 4). Attenuation of hypoxic stress by the inhibition of tubular transport: Oxygenated cell-free medium was supplemented with the loop diuretic furosemide 103 M (n ¼ 5) or with the Na/K ATPase inhibitor ouabain 103 M (n ¼ 4). In an additional group, albumin content was raised in the oxygenated cell-free medium to 12 g/dl in order to increase the colloid osmotic pressure and to abolish glomerular filtration (non-filtering kidney, NFK, n ¼ 4). Determination of tissue hypoxia by pimonidazole binding: Pimonidazole is a 2-nitro-imidazole that reliably binds to tissue at oxygen tensions below 10 mm Hg,38 and hence can serve as an HIFindependent indicator of hypoxia. For this purpose, additional IPKs were treated with pimonidazole (from Natural Pharmacia International, Belmont, MA, USA) 60 mg/kg (original animal weight), dissolved in 3 ml of saline, and infused over 5 min through a threeway stopcock, placed right upstream of the arterial canula. Pimonidazole injection started at 40 min after the start of perfusion, in order to allow parallel functional studies, and kidneys were fixed with formalin at 60 min. Assessment of morphology and immunohistochemistry Two micron paraffin sections were processed for routine histology or immunohistochemistry. Hematoxylin and eosin staining was performed according to standard procedures. Counterstaining with Richardson’s reagent served for a better discrimination of interstitial compartments. Immunohistochemistry for HIF-1a and HIF-2a was performed, using primary antibodies as reported previously19: mouse antihuman HIF-1a (a 67, Novus Biologicals, Littleton, CO, USA, 1:10 000 – which was previously proven to cross-react with rat HIF-1a18) and rabbit anti-mouse HIF-2a (PM9, gift from Patrick Maxwell, Wellcome Trust Centre for Human Genetics, Oxford, UK, 1:10 000). Immunohistochemistry for pimonidazole adducts was conducted exactly as described previously.19 Pimonidazole delivered in vivo (see above) reliably binds to deeply hypoxic tissue, where it can be detected with the help of commercially available antibodies (mouse anti-pimonidazole, Hypoxyprobe, Natural Pharmacia International, Belmont, MA, USA). Regional renal infarcts served as positive controls,19 whereas the non-infarcted part of the same Kidney International (2006) 70, 60–70

section served as a negative control. It is noteworthy that, using this technique, we find that kidneys of control animals (not of control IPKs) are all negative for pimonidazole adducts.19,20 The extent of outer medullary damage, manifested morphologically by mTAL nuclear pyknosis, was carried out in hematoxylin and eosin-stained slides at the mid-inner stripe of the outer medulla, the region characterized by the most intense hypoxic mTAL injury both in vitro and in vivo. mTAL damage was expressed as the percentage tubular profiles containing pyknotic nuclei or frank necrosis out of total tubules counted. HIF-1a was predominantly expressed in medullary tubular segments. Assessment of the extent of HIF-1a expression was carried out separately for mTALs and CDs in the mid-inner stripe of the outer medulla, expressed as the percentage of stained tubules out of all tubules counted. The extent of papillary HIF-1a and cortical HIF-2a expression was perceptually rather than quantitatively assessed. Statistics Data are presented as the mean7s.e.m. One-way analysis of variance with the post hoc Newman–Keuls test was carried out for between-group comparisons of the extent of medullary tubular damage and HIF-1a expression. Two-way analysis of variance was carried out for the comparison of functional parameters. ACKNOWLEDGMENTS

The study was supported by the German Research Foundation (DFG, SFB 423, project A 14), by funds from Harvard Medical Faculty Physicians at Beth Israel Deaconess Medical Center, Boston, MA, USA, and by the Russell Berrie Foundation and D-Cure, Diabetes Care in Israel. REFERENCES 1. 2.

3.

4.

5.

6.

7.

8. 9.

10.

11. 12. 13.

Brezis M, Rosen S. Hypoxia of the renal medulla: its implications for disease. N Engl J Med 1995; 332: 647–655. Molitoris BA, Sutton TA. Endothelial injury and dysfunction: role in the extension phase of acute renal failure. Kidney Int 2004; 66: 496–499. Nangaku M. Hypoxia and tubulointerstitial injury: a final common pathway to end-stage renal failure. Nephron Exp Nephrol 2004; 98: e8–e12. Agmon Y, Peleg H, Greenfeld Z et al. Nitric oxide and prostanoids protect the renal outer medulla from radiocontrast toxicity in the rat. J Clin Invest 1994; 94: 1069–1075. Shanley PF, Rosen MD, Brezis M et al. Topography of focal proximal tubular necrosis after ischemia with reflow in the rat kidney. Am J Pathol 1986; 122: 462–468. Shanley PF, Brezis ML, Spokes K et al. Transport-dependent cell injury in the S3 segment of the proximal tubule. Kidney Int 1986; 29: 1033–1037. Heyman SN, Rosen S, Epstein FH et al. Loop diuretics reduce hypoxic damage to proximal tubules of the isolated perfused rat kidney. Kidney Int 1994; 45: 981–985. Heyman SN, Brezis M, Reubinoff CA et al. Acute renal failure with selective medullary injury in the rat. J Clin Invest 1988; 82: 401–412. Lieberthal W, Menza SA, Levine JS. Graded ATP depletion can cause necrosis or apoptosis of cultured mouse proximal tubular cells. Am J Physiol 1998; 274: F315–F327. Brezis M, Kopolovic J, Rosen S. Hypoxic injury to medullary thick ascending limbs in perfused rat kidneys: reversible and irreversible phases. J Electron Microsc Tech 1988; 9: 293–298. Maxwell PH. HIF-1: an oxgen response system with special relevance to the kidney. J Am Soc Nephrol 2003; 14: 2712–2722. Semenza GL. Hydroxylation of HIF-1: oxygen sensing at the molecular level. Physiology 2004; 19: 176–182. Wenger RH. Cellular adaptation to hypoxia: O2-sensing protein hydroxylases, hypoxia-inducible transcription factors, and O2-regulated gene expression. FASEB J 2002; 16: 1151–1162.

69

original article

C Rosenberger et al.: Hypoxia-inducible factors and tubular cell survival

14.

26.

15.

16. 17. 18.

19. 20.

21.

22. 23.

24.

25.

70

Bracken CP, Whitelaw ML, Peet DJ. The hypoxia-inducible factors: key transcriptional regulators of hypoxic responses. Cell Mol Life Sci 2003; 60: 1376–1393. Bahlmann FH, Song R, Boehm SM et al. Low-dose therapy with the long-acting erythropoietin analogue darbepoetin alpha persistently activates endothelial Akt and attenuates progressive organ failure. Circulation 2004; 110: 1006–1012. Chatterjee PK. Pleiotropic renal actions of erythropoietin. Lancet 2005; 365: 1890–1892. Takahashi T, Morita K, Akagi R, Sassa S. Protective role of heme oxygenase-1 in renal ischemia. Antioxid Redox Signal 2004; 6: 867–877. Rosenberger C, Mandriota S, Juergensen JS et al. Expression of hypoxia-inducible factor-1alpha and -2alpha in hypoxic and ischemic rat kidneys. J Am Soc Nephrol 2002; 13: 1721–1732. Rosenberger C, Griethe W, Gruber G et al. Cellular responses to hypoxia after renal segmental infarction. Kidney Int 3003; 64: 874–886. Rosenberger C, Heyman SN, Rosen S et al. Up-regulation of HIF in experimental acute renal failure: evidence for a protective transcriptional response to hypoxia. Kidney Int 2005; 67: 531–542. Goldfarb M, Rosenberger C, Abassi Z et al. Acute-on-chronic renal failure in the rat: functional compensation and hypoxia tolerance. Am J Nephrol 2006; 26: 22–33. Brezis M, Agmon Y, Epstein FH. Determinants of intrarenal oxygenation. I. Effect of diuretics. Am J Physiol 1994; 267: F1059–F1062. Brezis M, Rosen S, Silva P, Epstein HF. Transport activity modifies thick ascending limb damage in the isolated perfused kidney. Kidney Int 1984; 25: 65–72. Heyman SN, Brezis M, Greenfeld Z, Rosen S. Protective role of furosemide and saline in radiocontrast-induced acute renal failure in the rat. Am J Kidney Dis 1989; 14: 377–385. Brezis M, Rosen S, Silva P, Epstein FH. Selective vulnerability of the medullary thick ascending limb to anoxia in the isolated perfused rat kidney. J Clin Invest 1984; 73: 182–190.

27.

28.

29. 30.

31.

32. 33. 34. 35.

36. 37.

38.

Epstein FH, Balaban RS, Ross RD. Redox state of cytochrome aa3 in isolated perfused rat kidney. Am J Physiol 1982; 243: F356–F363. Heyman SN, Brezis M, Rosen S. The isolated perfused rat kidney model in experimental renal injury. In: Bennett WM, DeBroe ME, Porter GA, Verpooten GA (eds). Clinical Nephrotoxins: Renal Injury from Drugs and Chemicals. Kluwer Academic Publishers: Dordrecht, 1998, pp 77–82. Jiang BH, Semenza GL, Bauer C, Marti HH. Hypoxia-inducible factor-1 levels vary exponentially over a physiologically relevant range of O2 tension. Am J Physiol 1996; 271: C1172–C1180. Schmidt T, Zou J, Bru¨ne B. HIF-1 and p53: communication of transcription factors under hypoxia. J Cell Mol Med 2004; 8: 423–431. Yang ZZ, Zhang AY, Yi FX et al. Redox regulation of HIF-1a levels and HO-1 expression in renal medullary interstitial cells. Am J Physiol Renal Physiol 2003; 284: F1207–F1215. Li N, Yi FX, Spurrier JL et al. Production of superoxide through NADH oxidase in thick ascending limb of Henle’s loop in rat kidney. Am J Physiol Renal Physiol 2002; 282: F111–F1119. Zou AP, Cowley Jr AW. Reactive oxygen species and molecular regulation of renal oxygenation. Acta Physiol Scand 2003; 179: 233–241. Rosenberger C, Rosen S, Heyman SN. Current understanding of HIF in renal disease. Kidney Blood Press Res 2005; 28: 325–340. Monotham K, Tanaka T, Ohse T et al. A biologic role of HIF-1 in the renal medulla. Kidney Int 2005; 67: 1428–1439. Wiesener MS, Turley H, Allen WE et al. Induction of endothelial PAS domain protein-1 by hypoxia: characterization and comparison with hypoxia-inducible factor-1alpha. Blood 1998; 92: 2260–2268. Ross BD, Epstein FH, Leaf A. Sodium reabsorption in the perfused rat kidney. Am J Physiol 1973; 225: 1165–1171. Epstein FH, Brosnan JT, Tange JD, Ross BD. Improved function with amino acids in the isolated perfused kidney. Am J Physiol 1982; 243: F284–F292. Franko AJ, Chapman JD. Binding of 14C-misonidazole to hypoxic cells in V79 spheroids. Br J Cancer 1982; 45: 694–699.

Kidney International (2006) 70, 60–70