INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

THE CELLULAR AND MOLECULAR responses of the kidney to ischemic insult are complex and incompletely understood (for review, see Refs. 15, 19, 63, 72). In the early phases after injury, the pathophysiological processes leading to cell damage and exfoliation predominate. Because the postischemic kidney has the ability to completely restore its structure and function, the second phase, which overlaps with the first, involves repair of the damaged kidney. In many instances, however, recovery is delayed or does not occur at all. Indeed, the mortality rate from renal failure is high (~40%) and has not changed significantly over 40 years (72). Therapeutic approaches that minimize injury and hasten recovery can be developed only when the cellular and molecular mechanisms of renal failure are more completely understood.

Recent attention has focused on the role of the integrins in the pathophysiology of acute renal failure (36, 54). The integrins are a superfamily of heterodimeric transmembrane glycoproteins found on every eukaryotic cell except erythrocytes; they are the major receptors for the extracellular matrix (ECM) and, along with cadherins, are believed to mediate cell-cell interactions (for review, see Ref. 31). Each integrin consists of unrelated - and -subunits whose association into noncovalent heterodimers is necessary for ligand binding and cell surface expression. The -chains can associate with one or more -subunits, forming receptors for laminins, collagens, fibronectin, and vitronectin.

In the kidney, integrins possessing the ₁-subunit are the most common. During development of the metanephros, ₁-integrins localize to all cell surfaces (37-39). As the developing nephron elongates and matures, ₁-integrins polarize to basal surfaces of tubular epithelia, where they interact with ECM components of the basement membrane, including laminin and type IV collagen. The surrounding interstitium is rich in fibronectin secreted by fibroblasts on which ₁-integrins are localized to all cell surfaces.

Genetic and biochemical studies implicate integrins containing the ₁-subunit in kidney epithelial differentiation and polarization. The ₈-, ₆-, and ₃-integrin subunits complex with ₁. Knockout of the ₈-subunit (51) impairs the ability of kidney mesenchymal cells to epithelialize, whereas antibodies recognizing ₆ (68) prevent polarization by tubular epithelia. Knockout of ₃ (40) or ₈ (51) decreases branching of the collecting duct system and, in the case of ₃ (40), results in the disruption of the glomerular basement membrane.

Just as integrins containing ₁ appear to mediate normal kidney development, they may also have a role in ischemic injury and repair. Using an in vitro model of kidney epithelial cells exposed to oxidative stress, Gailit et al. (21) demonstrated that the ₃-integrin subunit redistributed from basolateral to apical plasma membranes. On the basis of this observation, it was hypothesized (25, 54) that ₁-integrins likewise redistributed after injury from basal to apical cell surfaces of epithelial cells in vivo. According to this model, the decrease of ₁ on basal surfaces contributes to detachment of tubular epithelia (exfoliation) into the lumen, whereas the appearance of ₁ on apical plasma membranes of adherent epithelia and on surfaces of exfoliated cells mediates cell-cell adhesion. This hypothesis provided a potential mechanism to explain epithelial detachment (28, 44, 56, 60), tubular obstruction (3, 13, 45, 71), and backleak of glomerular filtrate that have been documented to occur following renal artery occlusion and reperfusion.

In addition to contributing to injury, integrins may also be important in repair. Because integrins mediate transmembrane signaling (for review, see Refs. 10, 17, 22, 30, 31), influencing diverse cell functions including cell behavior, differentiation, and gene expression, integrin-mediated cell-cell and cell-ECM interactions may also be critical to repair of the damaged epithelium. Tubular epithelial cells that remain attached alter their cytoskeletal organization and polarization (7, 48-50); they also dedifferentiate and proliferate (77). They may use integrins to spread and migrate over the underlying basement membrane to reestablish contacts with neighboring cells. Signals from these cell-matrix or cell-cell interactions may then initiate redifferentiation.

In this study, we used an in vivo model of unilateral ischemia in which rat kidneys are made ischemic by clamping the renal artery to investigate the role of ₁-integrins in ischemic injury and repair. We found that as early as 1-3 h after reperfusion, the ₁-integrin subunit is newly found on lateral surfaces of tubular epithelia most prone to ischemic injury. Damaged exfoliated cells obstructing tubular lumens do not appear to express ₁ even though an apical membrane marker, leucine aminopeptidase, and a basolateral membrane marker, Na⁺-K⁺-ATPase, are present on the plasma membranes of these cells. Surprisingly, tubular lumens are frequently filled with the ECM molecule fibronectin. During regeneration, ₁-integrins persist on lateral and basal surfaces of regenerating epithelia, with some cells expressing them apically as well. The data indicate that cell-cell and cell-ECM interactions mediated by ₁-integrins dynamically change during the injury phase, possibly contributing to epithelial detachment and tubular obstruction. During regeneration, ₁-integrins may direct the redifferentiation process.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Animals

Because one of the animal's kidneys is functionally intact, this model of unilateral ischemia results in only mild increases in plasma creatinine from 0.52 mg/dl before to 0.68 mg/dl after ischemia. Blood urea nitrogen also slightly increases from 17.5 mg/dl before to 20.5 mg/dl after ischemia (77). Glomerular filtration rate in the postischemic kidney, however, significantly decreases (4.2% of preischemic values) as measured by inulin clearance 1-3 h after reperfusion (43).

Preparation of Tissue

For cryosectioning, tissue pieces were equilibrated overnight at 4°C in 0.6 M sucrose-PBS(). Kidney pieces were then embedded in OCT medium (Miles Laboratories, Naperville, IL), frozen in liquid nitrogen, and sectioned using a Reichert Frigocut cryostat (Leica, Deerfield, IL). Five-micrometer sections were placed on SuperFrost Plus glass slides (Fisher Scientific, Boston, MA) and were either used immediately or stored at 20°C.

Immunohistochemistry

For immunostaining with antibody to Na⁺-K⁺-ATPase, sections were incubated for 5 min in 1% SDS immediately after PLP fixation (8). Controls for immunohistochemistry consisted of staining some sections with rabbit preimmune serum in place of the polyclonal primary antibody. Alternatively, the primary and/or secondary antibodies were omitted. In all instances, sections used for controls were of the identical time point and condition (ischemic, contralateral, or sham-operated kidney) used for primary and secondary antibody staining. Immunostaining was repeated a minimum of three times for each time point and condition on three sets of animals.

Antibodies

Initially, many different antibodies that recognize ₁ were tried. Many of the monoclonal and/or polyclonal antibodies either did not cross-react with rat or gave questionable immunostaining results. The rabbit polyclonal antibody generated against the integrin fibronectin receptor, ₅₁, used in this study was purified from human placenta (Telios Pharmaceuticals, San Diego, CA; GIBCO BRL, Grand Island, NY). This antibody, which was used to detect the ₁-integrin subunit in rat kidneys, was simultaneously screened by immunostaining Madin-Darby canine kidney (MDCK) cells, a distal tubule epithelial cell line, and immunoblotting and/or immunoprecipitating ₁-integrins from MDCK cell lysates. The staining pattern and repertoire of ₁-integrins expressed by MDCK are well established (55, 64, 80). The three different lots of polyclonal antibody recognizing ₅₁ used in this study immunoprecipitated from MDCK cell lysates a ₁-integrin profile identical to that obtained with an antibody generated against a 37-amino acid sequence in the cytoplasmic domain of ₁ (64). Because the ₅-subunit is not expressed by tubular epithelia in adult kidney (11, 37-39, 67) and the fibronectin receptor antibody recognizes an extracellular epitope of the ₁-integrin subunit, antibody staining with these antibodies identifies only the ₁-subunit in the kidney.

The rabbit anti-rat fibronectin polyclonal antibody (GIBCO BRL) was generated against rat plasma fibronectin. On Western blots it recognizes a band of ~220 kDa (data not shown), the molecular weight of fibronectin under reducing conditions. To verify the specificity of staining, competition experiments between the antibody and purified antigen were carried out. Purified rat fibronectin ranging in concentration up to 200 µg/ml was incubated for 1 h at 4°C with fibronectin antibody diluted 1:750, the concentration used to stain tissue sections. The antigen-antibody mixture was then added to the slides and incubated for 45-60 min at room temperature before being processed as described above. At a fibronectin concentration of 12.5 µg/ml, immunostaining was significantly diminished (data not shown); at 25 µg/ml, it was obliterated (see Fig. 10E).

Culture supernatants containing mouse monoclonal antibody against the ECM proteins, laminin (2E8) or type IV collagen (M3F7), were acquired from the Developmental Studies Hybridoma Bank; culture supernatants were used undiluted. Mouse monoclonal antibodies against the apical membrane marker, leucine aminopeptidase (1:800), and the basolateral membrane marker, Na⁺-K⁺-ATPase (1:100), were supplied by Drs. A. Quaroni (Cornell University, New York, NY) and D. Fambrough (Johns Hopkins University, Baltimore, MD), respectively.

Quantification of Cellular Localization of the ₁-Integrin Subunit

View this table: [in this window] [in a new window]   Table 2.   Cellular localization of 1-integrin subunit in outer stripe of outer medulla

Using representative images, the total number of tubules was counted, and the localization of ₁ to either basal or basal and lateral surfaces in the majority of cells was scored. Values were expressed as a percentage of the total number of tubules. Those tubules demonstrating intraluminal ₁ staining were also counted, and the value was expressed as a percentage of the number of tubules with luminal debris and/or cells.

Tubules sectioned tangentially were omitted because of the difficulty of imaging the luminal space. In addition, only discrete tubules either with or without exfoliated debris and/or cells were scored. At 48 h postischemia, in particular, when damage to tubular architecture is extensive, it was sometimes difficult to identify distinct tubules either because of the damage or because of the frequent presence of what appeared to be infiltrating inflammatory cells and/or fibroblasts. These tubules were not included.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Distribution of ₁-Integrin

In normal adult rat kidneys (see Fig. 1B) or contralateral controls (see Figs. 2E, 3D, 4C, 5C, and 5E) the ₁-integrin subunit localizes exclusively to basal cell surfaces of proximal and distal tubules in the cortex, outer stripe, and inner stripe, in agreement with observations in normal adult human kidneys (11, 37-39). After 40 min of renal ischemia induced by renal artery clamp in the absence of reperfusion (0 h postischemia), distribution of the ₁-subunit does not change. Basal plasma membranes of tubular epithelia, including those in the outer stripe of the outer medulla (Fig. 1A and Table 2), are strongly outlined, as are epithelia of the glomerulus and capillary tuft (data not shown).

View larger version (107K): [in this window] [in a new window]   Fig. 1.   Immunofluorescent staining of cryostat sections of postischemic, nonreperfused (A) and normal (B) adult rat kidneys reacted with polyclonal antibody to the 1-integrin subunit. In the outer stripe of the outer medulla, 1 localizes only to basal plasma membranes (arrows, A and B). Dashed lines in A outline tubular lumen. * Tubular profiles sectioned tangentially. Bar, 40 µm.

Distribution 1-3 h after reperfusion. Ischemic injury is most apparent in the S3 segment of proximal tubules in the outer stripe of the outer medulla (13, 23, 61), most likely due to hemodynamic factors leading to impaired reperfusion (4a). In this region, localization of the ₁-subunit changes 1 h after reperfusion (Fig. 2A and Table 2) compared with corresponding regions of contralateral control (Fig. 2E and Table 2) or normal adult kidneys (Fig. 1B). In S3 segments of control (arrowheads, Fig. 2E) or normal (Fig. 1B) kidneys, anti-₁ antibody exclusively stains basal plasma membranes; 97.1% of tubular profiles in the outer stripe of contralateral control kidneys show this distribution (Table 2). In contrast, by 1 h after reperfusion, the ₁-integrin subunit is detected on both basal and lateral surfaces (arrows, Fig. 2A) of some epithelia in S3 segments. Quantitation of ₁ localization indicates that 63.6% of tubules localize ₁ to both basal and lateral surfaces (Table 2). Epithelial cells are cuboidal in shape (Fig. 2B), apical microvilli are absent (Fig. 2B) and regions of tubulorrhexis representing severed basement membrane (brackets, Fig. 2B) are observed. Within a given tubule, basal and lateral plasma membrane staining of ₁ on some epithelia coexists with only basal membrane signal on other cells (Fig. 2A). In other tubules in the outer stripe, ₁ is expressed only basally (~36.4% of tubular profiles; Table 2).

View larger version (160K): [in this window] [in a new window]   Fig. 2.   Immunofluorescent staining of cryostat sections of postischemic (A, C, D) and contralateral nonischemic (E) kidneys taken 1 h after reperfusion and reacted with polyclonal antibody to the 1-integrin subunit. Corresponding Nomarski image of A is shown in B. In the outer stripe of the outer medulla (A, B), 1-integrin localizes to both basal and lateral surfaces of S3 proximal tubule segments (arrows). In other epithelia of the same tubule, 1 is also detected basally. Epithelia are cuboidal in shape and apical microvilli are absent (B). Cellular debris fill some tubular lumens (asterisks, B); 1 is not expressed by cellular debris (asterisks, A); it is also absent in regions where contact of luminal material with one another or with the adherent epithelium appears to be made (A, B). Bracketed area (A, B) highlights broken basement membrane. In the cortex (C) and inner stripe of the outer medulla (D), 1 localizes to basal plasma membranes, although infrequent lateral staining of proximal tubules in the cortex (arrow, C) is observed. In the outer stripe of control contralateral nonischemic kidney (E), 1 is found only on basal plasma membranes (arrowheads); microvillar brush border (E) stains nonspecifically. GL, glomerulus; TAL, thick ascending limb; CD, collecting duct. Bar, 40 µm.

Distribution 24-48 h after reperfusion. Although regeneration of the tubular epithelium commences immediately after injury (73), at 24 h postischemia injury predominates, based on disruption of normal tubular architecture. By 48 h after reperfusion, the injury phase appears to be complete and regeneration of the damaged kidney is the primary response. During this time in damaged S3 tubules, ₁ appears to localize primarily to basal plasma membranes of adherent epithelia (Fig. 3A). Quantitation of ₁ localization indicates that 85.8% of tubular profiles in the outer stripe express this integrin subunit only on basal cell surfaces (Table 2). The epithelial lining is sometimes difficult to identify because the adhering cells are flat (arrow, Fig. 3A; Ref. 32), filling in gaps of denuded basement membrane. The adherent epithelium thus is not detected in every section. In the remaining tubules (14.2%), however, basal and lateral staining of ₁ is observed (Table 2).

View larger version (190K): [in this window] [in a new window]   Fig. 3.   Immunofluorescent staining of cryostat sections of postischemic (A, C) and contralateral nonischemic control (D) kidneys taken 48 h after reperfusion and reacted with polyclonal antibody to 1-integrin subunit. Corresponding Nomarski image of A is shown in B. Amorphous material (asterisks, B) that fills tubular lumens of S3 segments in outer stripe of outer medulla is negative for 1 (asterisks, A), whereas the adherent epithelium localizes 1 primarily to basal plasma membranes (A); 1 is also not detected between the adherent epithelium and luminal material (arrows, A, B). At bottom right of B, a disrupted tubule appears to have been infiltrated by inflammatory cells and/or fibroblasts immunoreactive for 1 (A). In outer stripe of contralateral control kidney (D), 1 is found only on basal cell surfaces. In the cortex (C), distal tubules (DT) localize 1 to basal and lateral cell surfaces, whereas proximal tubules (PT) express it primarily on basal plasma membranes. Bar, 40 µm.

View larger version (77K): [in this window] [in a new window]   Fig. 4.   Immunofluorescent staining of postischemic (A) and nonischemic contralateral control (C) kidneys taken 24 h after reperfusion and incubated with polyclonal antibody to 1-integrin. Corresponding Nomarski image of A is shown in B. Appearance of 1 on basal surfaces of TAL of inner stripe of outer medulla is less distinct after ischemia (A) than in nonischemic contralateral control (C); 1 is also detected on basal and lateral surfaces of collecting tubules (CT). Bar, 40 µm.

Distribution 120 h after reperfusion. By 120 h after reperfusion, regeneration of the damaged kidney continues and in some areas appears to be complete. Flattened cells indicative of regeneration are observed, as are cuboidal cells characteristic of normal kidney epithelia. Among these cells, localization of the ₁-subunit varies. In some epithelia of regenerating S3 segments, strong localization of ₁ is observed on basal and lateral surfaces (short solid arrows, Fig. 5, A and B). Quantitation of its cellular localization indicates that 53.5% of tubules in the outer stripe of the outer medulla express ₁ both basally and laterally (Table 2). In either the same tubule or other tubules in the same region, ₁ is also found on apical plasma membranes (long solid arrows, Fig. 5, A and B). 22.2% of tubular profiles depolarize ₁ to all cell surfaces; basal, lateral, and apical localization is observed in ~5.8 cells/tubule (containing on average 23.9 cells; Table 2).

View larger version (156K): [in this window] [in a new window]   Fig. 5.   Immunofluorescent staining for 1-integrin in cryostat sections of postischemic (A, B, D) and nonischemic contralateral control (C, E) kidneys taken 120 h after reperfusion. In the outer stripe of the outer medulla of the postischemic kidney (A, B), some epithelial cells of S3 segments localize 1 to basal and lateral surfaces (short solid arrows), whereas others randomize it to apical, basal, and lateral plasma membranes (long solid arrows). Other epithelia show only a basal distribution (open arrows). Few exfoliated cells are found in tubular lumens. In these cells, 1 is either intracellular (A, B), absent (A), or randomized (A) on plasma membranes. Star in A highlights localization of 1-integrin between an exfoliated and an adherent tubular epithelial cell. In the outer stripe (C) of nonischemic contralateral kidneys, 1 is found exclusively on basal plasma membranes; the microvillar brush border in C stains nonspecifically. The inner stripe is not completely regenerated after reperfusion injury (D) compared with its corresponding contralateral control (E); 1 is basally localized (D), albeit weakly, on thick ascending limbs (TAL), and collecting tubules (CT) organize 1 basolaterally (D). Bar, 40 µm.

Distribution of Apical and Basolateral Membrane Markers

As expected, in normal adult or ischemic nonreperfused (Fig. 6A) rat kidneys, Na⁺-K⁺-ATPase localizes to basolateral surfaces of proximal and distal tubular epithelia, with distal tubules being more strongly reactive (33). In contrast, leucine aminopeptidase is found only on apical brush borders of proximal tubule cells (Fig. 6B; Ref. 12).

View larger version (146K): [in this window] [in a new window]   Fig. 6.   Immunofluorescent staining of cryostat sections of postischemic kidney reacted with monoclonal antibody to either Na+-K+-ATPase (A, C, E) or leucine aminopeptidase (B, D, F). In the absence of reperfusion (A, B) in the outer stripe of the outer medulla, the postischemic kidney localizes Na+-K+-ATPase (A) and leucine aminopeptidase (B) to basolateral surfaces and to apical microvilli of S3 segments, respectively. One hour after reperfusion, Na+-K+-ATPase intensely stains distal tubules and to a lesser extent S3 proximal tubules (C), where it localizes predominantly to basolateral surfaces; some staining of apical surfaces and what appear to be cytoplasmic vesicles is also seen (C). In contrast, aminopeptidase highlights the apical membrane, apical cytoplasm, and luminal space (D). Dashed lines in D mark basal surfaces of tubular epithelia. At 3 h after reperfusion, Na+-K+-ATPase is depolarized to all cell surfaces (E) in the majority of S3 proximal tubules. Punctate staining for leucine aminopeptidase (F) in apical cytoplasms predominates. Bar, 40 µm.

One hour after reperfusion, Na⁺-K⁺-ATPase (Fig. 6C) persists on basolateral surfaces; however, in some S3 segments it localizes to apical plasma membranes (Fig. 6C), in agreement with Molitoris and colleagues (48-50). The apical marker aminopeptidase is observed in a punctate pattern (Fig. 6D), possibly in the apical cytoplasm. Interestingly, cellular debris negative for ₁-integrin (asterisks, Fig. 2A) and Na⁺-K⁺-ATPase (data not shown) richly express aminopeptidase (Fig. 6D), consistent with previous studies demonstrating that cellular debris in tubular lumens at this time are composed of shed microvilli and/or apical membrane blebs (13, 73). As damage to the kidney increases after 3 h of reperfusion, Na⁺-K⁺-ATPase frequently localizes to all cell surfaces (Fig. 6E). In contrast, aminopeptidase has a strong punctate distribution in apical cytoplasms (Fig. 6F), reminiscent of endocytic vesicles; it also continues to be observed on cellular debris in the lumen (Fig. 6F).

In striking contrast to the lack of ₁ immunostaining in exfoliated cells at 24-48 h after reperfusion (Fig. 3A), S3 segment epithelial cells dislodged into tubular lumens randomly express both Na⁺-K⁺-ATPase (Fig. 7A) and aminopeptidase (Fig. 7B) on the plasma membrane. This observation indicates that the amorphous material congesting the lumen at this time consists of exfoliated cells and not only shed microvilli or other cellular debris seen earlier (1-3 h after reperfusion; Figs. 6, D and F). Nevertheless, as mentioned previously, these cells do not express ₁-integrin (Fig. 3A). Although epithelial cells adhering to basement membrane are not visible in every section at 24-48 h postischemia due to their flattened cell shape (see Fig. 3B), Na⁺-K⁺-ATPase and aminopeptidase are faintly detected on all cell surfaces of flattened epithelia adhering to basement membrane (arrows, Fig. 7, A and B). Dislodged cells, immunoreactive for Na⁺-K⁺-ATPase (Fig. 7, C and D) and aminopeptidase (Fig. 7, E and F), are also detected in tubular lumens of distal nephron segments in the inner stripe where Na⁺-K⁺-ATPase localizes to basolateral surfaces (Fig. 7C). Because of the aminopeptidase staining (Fig. 7E), these cells most likely are proximal in origin, having been propelled through the nephron to the distal segments.

View larger version (147K): [in this window] [in a new window]   Fig. 7.   Immunofluorescent staining of cryostat sections of postischemic kidney taken 48 h after reperfusion and reacted with monoclonal antibody to either Na+-K+-ATPase (A, C) or leucine aminopeptidase (B, E). Corresponding Nomarski images of C and E are shown in D and F, respectively. In S3 proximal tubules in the outer stripe of the outer medulla, Na+-K+-ATPase (A) and leucine aminopeptidase (B) strongly randomize to plasma membranes of exfoliated cells, whereas in adherent epithelia (arrows) weaker depolarized staining is observed. In the inner stripe, thick ascending limbs are immunoreactive for Na+-K+-ATPase (C, D), as is luminal material of dislodged cells, which also contain aminopeptidase (E, F). Bar, 40 µm.

At 120 h after reperfusion, localization of Na⁺-K⁺-ATPase and aminopeptidase varies among regenerating tubules. In some instances, Na⁺-K⁺-ATPase localizes to lateral and/or basal surfaces, with some cells expressing it apically (Fig. 8A). Aminopeptidase is detected primarily on apical surfaces (Fig. 8, C and D) but in some instances localizes to apical, lateral, and basal plasma membranes (Fig. 8C). Interestingly, Na⁺-K⁺-ATPase and aminopeptidase are weakly expressed by some tubules (Fig. 8, B-D). Intensity of staining also varies within a given segment (Fig. 8, A, C, and D).

View larger version (118K): [in this window] [in a new window]   Fig. 8.   Immunofluorescent staining of cryostat sections of postischemic kidney taken 120 h after reperfusion and reacted with monoclonal antibody to either Na+-K+-ATPase (A, B) or leucine aminopeptidase (C, D). Regenerating tubules in the outer stripe show heterogeneous staining. Some epithelia randomize Na+-K+-ATPase (A) and aminopeptidase (C) to all cell surfaces. Others localize Na+-K+-ATPase to lateral and/or basal plasma membranes (A, B) and distribute aminopeptidase to apical domains (C, D). Na+-K+-ATPase and aminopeptidase are weakly expressed along some regenerating tubules (B, D, respectively). Bar, 40 µm.

The data indicate that after ischemic-reperfusion injury, the adherent epithelium continues to express apical and basolateral membrane proteins; however, distribution is not polarized. The loss of epithelial polarity is observed by 1 h after reperfusion and persists to some extent 120 h later, when a spectrum of localization patterns and expression levels are observed, ranging from polarized to nonpolarized and faint to strong staining. In contrast to the absence of ₁ staining, amorphous material in tubular lumens at 24-48 h after reperfusion expresses apical and basolateral membrane markers, indicating that it consists of exfoliated cells. These proteins are uniformly found on all cell surfaces.

ECM Expression

Kidney sections from animals killed various times after reperfusion or animals made ischemic but not reperfused (0 h postischemia) were probed with monoclonal antibodies recognizing the major constituents of basement membrane: laminin and type IV collagen. In all cases, laminin and type IV collagen localized to tubular basement membranes (data not shown); however, as early as 1 h after reperfusion, immunostaining highlighted breaks in basement membranes (data not shown; also see Fig. 2B) not observed in corresponding contralateral controls (data not shown). Tubules with disrupted basement membranes were confined to the outer stripe. Tubulorrhexis appeared to be most prevalent at 24 h after reperfusion (Fig. 9A), declining thereafter (data not shown). Interestingly, in a few tubules, laminin localized to the luminal compartment (Fig. 9B). In contralateral controls (data not shown), tubulorrhexis was not observed and laminin was not found in the luminal space. Profiles of severed basement membranes were completely absent at 120 h postischemia (data not shown), indicating that by this time the basement membrane had regenerated.

View larger version (164K): [in this window] [in a new window]   Fig. 9.   Immunofluorescent staining of cryostat sections of postischemic kidney taken 24 h after reperfusion and reacted with monoclonal antibody to laminin. In the outer stripe, laminin outlines tubular basement membranes, highlighting regions of tubulorrhexis (arrows, A). Only rarely is laminin detected in tubular lumens (asterisk, B). Bar, 40 µm.

The existence of tubulorrhexis raised the possibility that interstitial fluid could enter the tubular lumen via plasma from congested capillaries. In addition, tubulorrhexis might promote interactions between epithelia at the wound edge and underlying interstitial matrix molecules.

To test this, the distribution of fibronectin, an important constituent of both plasma and the renal interstitium, was examined. In normal (see Fig. 10G) and contralateral control kidneys (data not shown) or kidneys that were made ischemic but not reperfused (0 h postischemia; data not shown), fibronectin outlines tubules, presumably localizing to the renal interstitium (1). This localization persists after ischemiareperfusion. However, in response to reperfusion injury, fibronectin also is detected in lumens of S3 segments and the distal nephron. In S3 segments, staining with anti-fibronectin antibody initially highlights regions of tubulorrhexis between 1 and 3 h after reperfusion (data not shown). By 24 h postischemia, hazy fibronectin staining of S3 segments and/or more proximal regions of the distal nephron is observed (single asterisks, Fig. 10A). Nomarski imaging shows that the hazy pattern of staining coincides with exfoliated cells and cellular material occluding the lumen (single asterisks, Fig. 10B). In a few instances, fibrils immunoreactive for fibronectin are detected between exfoliated cells (data not shown). By 120 h, fibronectin staining of luminal contents is no longer observed (Fig. 11A).

View larger version (134K): [in this window] [in a new window]   Fig. 10.   Immunofluorescent staining of cryostat sections of postischemic kidney taken 24 h after reperfusion (A, C, E) and reacted with polyclonal antibody to fibronectin. Corresponding Nomarski images of A and C are shown in B and D, respectively. Contralateral control kidney is shown in F; normal rat kidney is shown in G. In the outer stripe of the outer medulla, hazy fibronectin staining is seen in tubules containing exfoliated cells (single asterisks, A, B). Tubules may represent either the S3 segment or more proximal regions of the distal nephron. Some tubular lumens in which detached cells are not observed (double asterisk, B) contain fibronectin (double asterisk, A). In comparable regions in normal kidney (G), fibronectin outlines tubules and the renal interstitium. In the inner stripe of the outer medulla, strong fibronectin immunoreactivity is detected in luminal compartments of the distal nephron (C, D) but is not detected in contralateral control kidney (F). Staining of luminal fibronectin is obliterated (E) by preincubating anti-rat fibronectin antibody with purified rat fibronectin (25 µg/ml). Bar, 40 µm.

View larger version (111K): [in this window] [in a new window]   Fig. 11.   Immunofluorescent staining of cryostat sections of postischemic kidney taken 120 h after reperfusion and reacted with polyclonal antibody to fibronectin. Fibronectin is strongly expressed in the interstitium of the outer (A) and inner (B) stripes, suggesting an incipient fibrosis. However, it is absent from tubular spaces of proximal S3 (A) and distal nephron segments (B). Bar, 40 µm.

The majority of staining for fibronectin is confined to the lumen of the distal nephron after ischemic injury and reperfusion. At 3 h postischemia, fibronectin is detected in tubular lumens of distal nephrons running throughout the cortex (data not shown), outer stripe (data not shown), and inner stripe (data not shown). At 24 h postischemia, luminal fibronectin dramatically increases in distal nephrons of the inner stripe (Fig. 10C) but is also detected at significant levels in these segments in the cortex (data not shown) and outer stripe (double asterisk, Fig. 10A). The number of distal tubules with luminal fibronectin begins to decrease 48 h postischemia, but profiles are still found throughout the kidney (data not shown). By 120 h postischemia, distal luminal fibronectin is no longer observed (Fig. 11B). Localization of fibronectin in the lumen of the distal nephron is not detected in normal kidneys (data not shown), kidneys that were not reperfused (data not shown), or corresponding contralateral controls (Fig. 10F). In addition, fibronectin staining throughout the kidney is not detected when immunoreactive sites on the antibody are competed out with purified antigen (Fig. 10E) or sections are reacted with preimmune sera (data not shown; see MATERIALS AND METHODS).

Thus, after ischemic insult, the distribution of laminin and collagen IV to the basement membrane is largely unaltered, although the basement membrane itself may be severed. In contrast, fibronectin is detected in S3 segments of the proximal tubule and in substantial amounts in distal nephron segments.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Injury of rat kidneys by ischemia followed by reperfusion initiates changes in cell-cell and cell-ECM interactions and alterations in epithelial cell polarity (for review, see Refs. 19, 48). As we have shown here, one of the molecules most significantly affected is the ₁-integrin subunit, which is an element of receptors for collagens, laminins, and fibronectin in the kidney. Between 1 and 3 h after reperfusion, localization of ₁-integrin in adherent epithelia of proximal tubule S3 segments changes from exclusively basal to basal and lateral. Polarity begins to be lost, as evidenced by detection of the basolateral membrane marker Na⁺-K⁺-ATPase on the apical surface and internalization of the apical membrane marker leucine aminopeptidase. Cellular material, representing apical but not basolateral membrane debris, occupies tubular lumens. By 24-48 h after reperfusion, damage to the kidney increases. The basal lamina of the S3 segments is denuded by exfoliation of cells and is sometimes broken. Despite the fact that the Na⁺-K⁺-ATPase and aminopeptidase are weakly but uniformly distributed on both apical and basolateral cell surfaces, adherent epithelia of S3 segments continue to express ₁ primarily on basal and infrequently on lateral and apical surfaces. At the same time, depolarized exfoliated cells expressing both leucine aminopeptidase and the Na⁺-K⁺-ATPase fill the luminal space. These cells, however, do not express the ₁-subunit. Strikingly, large amounts of fibronectin are also observed in lumens of the distal nephron. At 120 h after reperfusion, repair from ischemic injury is evident. The ₁-subunit is again basally located in some regenerating epithelia; in others it is both basal and lateral or is also detected on the apical surface. Apical and basolateral membrane markers are polarized in some cells; in others they are weakly but uniformly expressed on all cell surfaces. Basement membranes of S3 segments are intact, and luminal fibronectin in distal nephron segments is no longer observed.

₁-Integrin in Renal Ischemic Injury and Repair

On the basis of the redistribution of ₃, the authors proposed a role for defective cell-cell and cell-ECM interactions in the pathophysiology of acute renal failure (26). Loss of integrin receptor on basal cell surfaces was hypothesized to cause exfoliation of epithelial cells into tubular lumens. Redistribution of integrins from basal to apical plasma membranes on remaining epithelia was also suggested to facilitate attachment of dislodged cells and fragments of basement membrane released during injury. Under these circumstances, ₁-integrins expressed by exfoliated cells were thought to promote cell-cell adhesion either directly or via bridging ECM molecules, culminating in tubular obstruction and, ultimately, renal failure.

In subsequent experiments using an in vivo rat model, Goligorsky and colleagues (62) observed ₁ on the apical surface of adherent cells and on the plasma membrane of exfoliated cells occupying the lumen. In contrast, our results reported here do not support a functional role for apical expression of ₁-integrins in renal ischemic injury. In our experiments, ₁-integrins were not seen on the apical surfaces of adherent epithelia at times soon after ischemia and only rarely on apical surfaces at later times when regeneration of the epithelium was underway (24-48 h postischemia). In addition, the lack of ₁ on exfoliated cells suggests that integrins neither mediate aggregation of exfoliated cells nor anchor them to the adherent epithelium.

Our findings are also inconsistent with a significant role for the basement membrane ECM molecules collagen and laminin in bridging cell interactions that could contribute to tubular occlusion. Although tubulorrhexis was observed, fragments or soluble components of the basement membrane, including laminin, were rarely seen in the tubular lumen. The absence of laminin in the luminal space further indicates that, within the sensitivity of our analysis, it is not secreted apically by the regenerating epithelium or randomly by the depolarized exfoliated cells. In contrast, fibronectin was abundantly expressed in luminal compartments. Fibronectin assembles into fibrils when bound by integrins on plasma membranes (76, 78, 79). Because fibronectin fibrils were not observed to outline exfoliated cells, we believe that it is unlikely to serve as a bridging molecule between adherent and exfoliated cells and to contribute in this way to tubular occlusion.

Model of ₁ in Renal Ischemic Injury and Repair

View larger version (25K): [in this window] [in a new window]   Fig. 12.   Model of 1-integrin in renal ischemic injury and repair. After ischemia and reperfusion, 1-integrins that are basally localized are deactivated, resulting in their disengagement from ligands in the basement membrane. Deactivated integrins are then free to diffuse to lateral surfaces. In distal tubules, where the degree of deactivation is not as severe, cells remain attached and basal localization of 1 is eventually recovered. In S3 segments, however, integrin deactivation results in exfoliation of some cells. Deactivated integrins, now devoid of ligand, are endocytosed and degraded. As the tubular epithelium regenerates, cells use 1 to flatten, spread, and migrate over denuded areas of basement membrane while dedifferentiating and depolarizing. New integrin - and -subunits may also be expressed to mediate attachment either to fibronectin, an interstitial matrix protein, or to other components of basement membrane newly expressed as part of the repair process. By transmitting a variety of intracellular signals, these integrins may be important in redifferentiation and repolarization. As a new epithelium reforms, 1-integrins are reactivated, ultimately returning to an exclusively basal location.

This model is consistent with our observations in the postischemic kidney. In both the S3 segment and the distal tubule in the cortex, we observed integrins on the lateral cell surfaces of attached cells early in the injury phase; in distal tubules in the inner stripe, we observed a decrease in ₁ levels on basal surfaces, consistent with its proposed degradation. In the distal tubule, where no exfoliation occurs, integrin localization returned to basal plasma membranes, suggesting that reactivation, whether by signaling mechanisms or resynthesis, resulted in reengagement of integrins with ligands of the basement membrane. In the S3 segment, considerable exfoliation was seen. No ₁-integrins were detected in the exfoliated cells, however, consistent with the proposal that degradation occurred. Because the antibodies used in our studies reacted with the extracellular region of the ₁-molecule, it is unlikely that the loss of immunoreactivity was due solely to degradation of the cytoplasmic part of ₁. Finally, our findings of a lack of polarity of both apical and basolateral markers as well as, in some cases, ₁ at later times after reperfusion are more indicative of a morphogenetic response to regeneration than an immediate reaction to injury.

Integrins other than those of the ₁ family may contribute to intraluminal obstruction and oliguria. Kidney epithelial cells may express integrins containing the ₃, ₅, or ₆ families in addition to ₁, either constitutively (37, 39, 58) or in response to injury (5, 74). Any or all of these could redistribute early on to the apical surfaces of attached cells and remain on the plasma membranes of exfoliated cells. Because fibronectin can bind to each of these integrin families, it could act as a bridging molecule. Although we do not favor this possibility in S3 segments because assembly of fibronectin into fibrils (via interactions with integrins) was not observed, our limited observations and analysis at this time do not permit us to exclude the possibility that fibronectin, present in apparently high concentrations in lumens of distal tubules, could be interacting with newly expressed integrins on apical surfaces of adherent epithelia. Also, because the interactions between these integrins and fibronectin are mediated by Arg-Gly-Asp (RGD) sequences in the fibronectin molecule, this scenario might help to explain the reported effectiveness of RGD peptides in ameliorating renal failure in vivo (25, 27, 53, 62). Clearly additional studies are required to address these points.

Integrin Activation and Deactivation

Activation of integrins is regulated by the cytoplasmic tails of both - and -subunits. Deletion of specific sequential motifs inactivates integrins, making them unable to bind ligand (17, 22, 35, 57). In contrast, elimination of the entire tail of the -subunit can render integrins constitutively active (17, 22, 35, 57). On the basis of these types of experiments, a hingelike model for regulation of integrin affinity has been proposed that postulates that the interaction between integrin - and -subunit cytoplasmic tails renders integrins unable to bind ligand until this interaction is relaxed by association with an "integrin-activator complex" (IAC) (22, 57).

Although integrin activation may be mediated by signal transduction pathways (so-called "inside-out signaling"), integrin deactivation may occur either reversibly through signal transduction and dissociation of postulated IACs or irreversibly by integrin degradation. In the kidney, one candidate signal transduction pathway is the one that leads to stress-activated protein kinase (SAPK; also called Jun kinase-1). SAPK is a mitogen-activated protein kinase that is closely related to the extracellular signal-regulated kinases (ERK) ERK1 and ERK2 but is activated by distinct mechanisms. After unilateral renal ischemia in the rat, SAPK activity increases within 5 min after reperfusion and remains elevated 90 min later (<