INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

RENAL ISCHEMIA RESULTS in cellular necrosis in the S3 segment of the proximal tubule and compromises renal function (4, 35). In rodent models of renal ischemic injury, the resulting regenerative response is dependent on replenishing the population of proximal tubule cells by a wave of proliferation occurring 1-3 days postinjury (35). Evidence amassed from several laboratories has shown changes in the expression of several promitogenic growth factors in kidney during the process of recovery from acute ischemic renal injury. These include epidermal growth factor (EGF), heparin-binding EGF, insulin-like growth factor-I (IGF-I), hepatocyte growth factor (HGF), and fibroblast growth factors (8, 11-13, 22). The administration of one of three mitogenic growth factors (EGF, IGF-I, or HGF) to rats immediately following injury increases proximal tubule cell proliferation and hastens the recovery of renal function (9, 23, 24). These observations suggest that the production of mitogenic peptides in kidney mediates the early cellular events of renal regeneration following injury.

Despite the importance of cell proliferation, the restoration of renal tubular epithelial structure and function is contingent on a variety of other cellular events including migration along the proximal tubular basement membrane (TBM) and differentiation, hypertrophy, and apoptosis of hyperplastic proximal tubule cells (4, 35). These events are required for the proximal tubule to acquire normal morphology, cell-cell contacts, and transport capacity; complete structural and functional recovery may take 4-6 wk (4, 32, 35). The identities of the factors that mediate these nonproliferative events following ischemic injury are not well delineated.

Transforming growth factor- (TGF-) is a polypeptide growth factor thought to play an important role in wound healing and tissue regeneration (21). TGF- inhibits proliferation of renal proximal tubule cells in vitro and stimulates extracellular matrix (ECM) synthesis, cell clustering, tubulogenesis, and apoptosis (10, 25). We recently demonstrated the rapid and prolonged expression of TGF-1 mRNA and peptide primarily in regenerating proximal tubules for up to 14 days postischemic injury in rats (3). Thus, TGF- has the potential to mediate many of the important nonproliferative events that occur in the regenerating kidney.

A precise role for TGF- in the renal regenerative response remains to be defined. The following study was initiated to characterize the expression of several genes after ischemic injury, known to be potently induced by TGF- in renal tubular epithelial cells in vitro (18, 21, 33). These genes, plasminogen activator inhibitor-1 (PAI-1), tissue inhibitor of metalloprotease-1 (TIMP-1), ₁(IV) collagen, and the EIIIA splice variant of fibronectin, have been used as bioassays for TGF- activity (15-17, 31, 33) and are thought to play critical roles in tissue remodeling following injury. Therefore, this study aims to establish whether TGF- bioactivity is enhanced in kidney postischemic injury and to determine a possible role for TGF- in stimulating ECM synthesis during renal recovery.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Animals. Male Sprague-Dawley rats weighing ~250 g were housed with 12:12-h light-dark cycle and food and water available ad libitum. Acute renal failure was induced by 60 min of bilateral renal artery clamping exactly as described previously, with sham-operated controls included at each time point (23, 24). The extent of renal injury during the procedure was monitored such that the mean level of serum creatinine measured 24 h following injury fell into the range used in previous studies, i.e., 2.8-4.0 mg/dl (3, 23, 24).

In some studies, rats were administered a neutralizing antibody against TGF- (1D11, a gift from Genzyme Tissue Repair) or an equal amount of nonimmune mouse IgG as a control (Sigma, St. Louis, MO). The antibodies were administered via tail vein injection immediately following removal of renal artery clamps and once more 48 h following removal. Rats received 2.5-10 mg/kg body wt of antibody for each of the two injections. No differences were observed between doses. Data were pooled for statistical analysis.

Isolation of kidney tissue. At the indicated times, rats were anesthetized with ketamine and pentobarbital. Their kidneys were perfused with sterile PBS to remove all blood from the organs (3, 23, 24). Both kidneys were quickly excised and cut longitudinally, and one-half of each was frozen in liquid nitrogen and stored at 70°C. The remaining halves were prepared for immunohistochemical analysis and in situ hybridization studies (see below).

Ribonuclease protection assays. Total cellular RNA from whole kidney was obtained using the Ultraspec RNA kit (Biotecx, Houston, TX). Ribonuclease protection assays were carried out as described (3). cDNA probes cloned into pGEM-T corresponding to rat TIMP-1 (bp 118-333), the fibronectin-EIIIA splice variant (FN-EIIIA, bp 1034-1401), and ₁(IV) collagen (homologous to mouse bp 4693-5140) were provided by Dr. Jerry Morrissey, Washington University (15, 16). The rat PAI-1 cDNA was a gift from Dr. Tom Gelehrter, University of Michigan (38). Riboprobe templates for TIMP-1, FN-EIIIA, and ₁(IV) collagen were generated by digestion with Apa I and transcribed with SP6 polymerase in the presence of [³²P]CTP; the resulting antisense probes protected fragments of 216, 369, and 448 nt, respectively. The rat PAI-1 riboprobe template was generated by PCR amplification of the region 218-760 of the rat PAI-1 cDNA. The sense primer was 5'-GAATCAGGCCACCAACTT-3'; the antisense primer was 5'-GGATTTGTGGAACAGGCG-3'. The underlined regions refer to the bacterial promoter sequence of T7 and SP6 polymerase, respectively, allowing the generation of "sense" and "antisense" riboprobes from the purified product of the PCR amplification. To quantify expression of specific mRNAs, duplicate determinations of samples from three animals per group and three corresponding sham-operated control animals from each time point were run on the same gel. The intensity of the resulting signals was determined using a phosphorimager (Molecular Dynamics, Sunnyvale, CA). In all samples, the expression of each gene was corrected by dividing probe-specific signal by that obtained for the housekeeping gene cyclophilin (Ambion, Austin, TX). Comparison between gels was accomplished by normalizing data to the mean value of the respective sham-operated control animals (3). Statistical analysis was by Student's t-test for unpaired sample means; P < 0.05 for two-tailed analysis was considered significant.

In situ hybridization. Digoxigenin-labeled antisense and sense riboprobes were synthesized using the riboprobe templates described above. A BLAST search (1) of the GenBank database yielded no significant homology with other mRNAs. Tissues were fixed by immersion in Bouin's solution (Sigma) and embedded in paraffin. In situ hybridization was performed on 5-µm sections exactly as described previously with a probe concentration of 1-2 ng/ml and hybridization at 50°C for 12-18 h (3). Posthybridization washing and immunologic detection was performed exactly as described previously (3).

Immunohistochemistry. Localization of TIMP-1 was performed on Bouin's fixed, paraffin-embedded sections using rabbit anti-TIMP-1 IgG (5 µg/ml; Clontech, catalog no. AB770) incubated overnight at 4°C. Detection was by the streptavidin-biotin immunoperoxidase technique with aminoethyl carbozole as a substrate (Histostain SP kit, Zymed).

PAI-1 was localized on renal tissue that was snap frozen in OCT optimal cutting temperature compound in cold isopentane. Unfixed cryostat sections (7 µm) were stored at 20°C until use. Sections were fixed on slides in cold acetone for 10 min, subsequently rinsed in PBS, and blocked by sequential incubations in avidin, biotin (Zymed), and blocking buffer (PBS containing 20% normal goat serum, 0.3% BSA, and 0.3% Triton X-100). The primary antibody (rabbit anti-rat PAI-1, 5 µg/ml; American Diagnostica, Greenwich, CT) was applied overnight at 4°C. Following PBS washes, the primary antibody was localized by subsequent incubations in biotinylated goat anti-rabbit IgG (Zymed) and streptavidin-conjugated CY3 (Zymed, 1:500).

ECM glycoproteins were localized on frozen sections that were subjected to acetone fixation, PBS rinsing, and incubation for 1 h in blocking buffer (PBS containing 1% goat serum and 0.3% BSA). The primary antibody was applied overnight at 4°C in blocking buffer. The primary antibodies were mouse monoclonal anti-collagen type IV (1:500, Collaborative Research) and rabbit anti-human cellular fibronectin (Dako, 1:2,000), which recognizes all forms of fibronectin. Two different primary antibodies were used to localize the EIIIA splice variant of fibronectin: mouse monoclonal anti-cellular fibronectin (ED-A specific domain, Harlan, MAS 521, 1:100), and mouse monoclonal anti-cellular fibronectin (Sigma, clone FN-3E2, 3 µg/ml) (27). Tissues were washed and exposed to their species-appropriate secondary antibody conjugated to CY3 (1:500; Jackson Immunoresearch Laboratories, West Grove, PA). After extensive washing, coverslips were applied in glycerol/PBS (1:1) and visualized with epifluorescent illumination.

	RESULTS

Top Abstract Introduction Methods Results Discussion References

mRNA expression of ECM-associated genes following ischemic injury. We characterized the renal expression postischemia of genes known to be potently induced by TGF-. Figure 1 illustrates a series of ribonuclease protection assays performed on total RNA extracted from kidneys of sham-operated or postischemic rats at various times postsurgery. Two protease inhibitors, PAI-1 and TIMP-1, were expressed at low levels in kidneys obtained from sham-operated rats. There was a significant, approximately fourfold, increase in the mRNA expression of these genes as early as 12 h following reperfusion relative to levels observed in sham-operated controls. The expression of these genes peaked 24 h postischemia and remained significantly elevated above sham-operated control values for 5 days (TIMP-1) or 7 days PAI-1 (Fig. 1, A and B).

View larger version (100K): [in this window] [in a new window]   View larger version (27K): [in this window] [in a new window]   Fig. 1.   Ribonuclease protection assay of extracellular matrix-related genes in response to ischemic acute renal failure. A: plasminogen activator inhibitor (PAI-1), tissue inhibitor of metalloprotease-1 (TIMP-1), fibronectin (FN) EIIIA, and 1(IV) collagen mRNAs were detected in total cellular RNA originating from rat kidneys at various times (0.5-28 days) following sham operation or 60 min of bilateral renal ischemia (ARF). Rat cyclophilin mRNA was measured as an internal standard. B: protected fragments from ribonuclease protection assays were quantified with a phosphorimager. Levels for each mRNA were standardized to level of cyclophilin mRNA for each sample. Expression of mRNA from kidneys of rats rendered ischemic (bars) were normalized to levels originating from kidneys of sham-operated rats (solid line) for each time point. Maximum standard error of the mean (SEM) obtained in any sham-operated control group is indicated by the broken line. Data originate from 3 different rats per group at each time point and are expressed as means ± SE. * P < 0.05 vs. appropriate sham-operated control.

The mRNA expression of ECM-glycoprotein genes was also examined. Ribonuclease protection was performed using an antisense riboprobe specific for the EIIIA splice variant of the fibronectin gene (FN-EIIIA) and the ₁-chain of type IV collagen. Similar to the results obtained for PAI-1 and TIMP-1, the levels of mRNA for FN-EIIIA and ₁(IV) collagen were low in sham-operated rat kidneys. FN-EIIIA mRNA levels were significantly increased within 12 h of reperfusion, whereas significantly enhanced levels of ₁(IV) collagen mRNA were not detected for 3 days (Fig. 1, A and B). The mRNAs for FN-EIIIA and ₁(IV) collagen remained significantly elevated above levels seen in sham-operated controls for 7 and 28 days, respectively (Fig. 1, A and B).

Localization of ECM-related gene products in normal and postischemic kidneys. To localize the expression of ECM-related gene products in the postischemic regenerating kidney, both nonisotopic in situ hybridization and immunohistochemical analyses were performed. Figure 2 illustrates results obtained from in situ hybridization using an antisense probe for the rat PAI-1 gene. No signal was evident in either the cortex or outer medulla in kidneys from sham-operated control rats (Fig. 2A). However, when analyzed at either 3 or 7 days postischemic injury, PAI-1 mRNA was expressed in the renal outer medulla (Fig. 2B), most notably in regenerating cells of the proximal tubule (Fig. 2, C and D, arrowheads). Immunoreactivity of rat PAI-1 protein is shown in Fig. 3. PAI-1 immunoreactivity was present at low levels in glomeruli of sham-operated rats (Fig. 3D, arrow). In renal outer medulla, the PAI-1 signal was not distinguishable from that of nonimmune IgG control samples (Fig. 3, A vs. F). In contrast, PAI-1 immunoreactivity was prominent in the outer medulla at either 3 days (Fig. 3B) or 7 days (Fig. 3C) postischemic injury in sloughing necrotic cells, flattened-regenerating cells (arrowheads), and the underlying TBM. Ischemia did not appreciably alter the PAI-1 signal in the cortex (Fig. 3E).

View larger version (144K): [in this window] [in a new window]   Fig. 2.   Localization of renal PAI-1 mRNA by nonisotopic in situ hybridization. Rat PAI-1 mRNA was localized in kidneys from sham-operated rats and from rats rendered ischemic. A and B: cross sections through rat cortex (cx) and outer medulla (om) of sham-operated rat (A) or 72 h postischemic injury (B). C and D: higher magnifications through outer medulla both 3 days (C) and 7 days postinjury (D). Arrowheads show positive hybridization in regenerating proximal tubule cells. E: no hybridization was observed with equal concentration of "sense" control riboprobe. Magnification for A and B is shown in A. Magnification for C-E is shown in C.

View larger version (155K): [in this window] [in a new window]   Fig. 3.   Immunofluorescent localization of PAI-1. Frozen kidney sections (7 µm) fixed in acetone were probed using an antibody generated against rat PAI-1. Sections are through rat kidney outer medulla (A-C and F) or cortex (D and E) of sham-operated rat (A and D) and ischemic rat 3 days postinjury (B, E, and F) or 7 days postinjury (C). Positive immunoreactivity in the outer medullary regenerating proximal tubules is denoted by arrowheads (B and C). Positive immunoreactivity observed in glomerulus is denoted by arrows (D and E). White bar in F is 100 µm and applies to A-F for all panels.

TIMP-1 demonstrated a markedly different pattern of expression in kidney compared with PAI-1. Both in situ hybridization (Fig. 4A) and immunohistochemical analyses (Fig. 5C) demonstrated that TIMP-1 mRNA and protein are primarily localized in distal nephron segments. In renal cortex, TIMP-1 mRNA and protein were localized in distal tubules and collecting ducts (arrows, Fig. 5, C and D), whereas in the outer medulla, both TIMP-1 mRNA and protein were localized to the thick ascending limb of Henle (Fig. 4, A and B, and Fig. 5A). Renal ischemic injury enhanced TIMP-1 mRNA and protein expression primarily in thick ascending limb of Henle in the outer medulla (Fig. 4, C vs. D, and Fig. 5, A vs. B, arrows) as well as distal tubules and collecting ducts in the cortex (Fig. 5, C vs. D). Regenerating proximal tubules did not express detectable TIMP-1 mRNA or protein (Figs. 4D and 5B, arrowheads).

View larger version (148K): [in this window] [in a new window]   Fig. 4.   Localization of renal TIMP-1 mRNA by nonisotopic in situ hybridization. TIMP-1 mRNA was localized in kidneys from sham-operated rats and rats rendered ischemic using a digoxigenin-labeled antisense riboprobe. A and B: cross sections through renal outer medulla (om) and cortex (cx) of sham-operated rat (A) or a rat 3 days following induction of ischemic acute renal failure (B). C and D: higher magnification through outer medulla of sham-operated (C) and postischemic (D) rat kidneys. Hybridization is evident in cortical distal tubules and medullary thick ascending limbs (thin arrows in A) and collecting ducts (wide arrow in A). No hybridization is evident in regenerating proximal tubules (arrowheads in B and D). E: no hybridization signal was evident when using a sense control riboprobe (3 days postischemia). Magnification for A, B, and E is shown in E. Magnification for C and D is shown in D.

View larger version (118K): [in this window] [in a new window]   Fig. 5.   Immunohistochemical localization of TIMP-1. Five-micrometer sections of Bouin's-fixed kidneys were probed using an antibody generated against TIMP-1. In section through renal outer medulla (A) and cortex (C) of sham-operated rat, positive staining in the thick ascending limb of Henle (A, arrow) or distal tubule (C, arrow) is evident. In section through outer medulla (B) or cortex (D) 3 days postischemia, staining in thick ascending limb of Henle (thick arrow, B) or in cortical collecting duct (thick arrow, D) is evident. Positive staining was not detected in damaged proximal tubules of the outer medulla (B, arrowhead). No staining was evident when the antibody was replaced with an equal concentration of nonimmune rabbit IgG (3 days post ischemia, E).

₁(IV) Collagen mRNA was not prominent in cross sections of rat kidney from sham-operated rats (Fig. 6A) by in situ hybridization but was easily detected following injury (Fig. 6B). Expression of ₁(IV) collagen mRNA was not restricted to a distinct nephron segment but was most prominent in regenerating proximal tubules (Fig. 6, C and D, arrowheads). Immunofluorescent analysis of renal tissue following ischemic injury demonstrated that collagen type IV is abundantly present in the basement membranes of all nephron segments and in the glomerulus (Fig. 7). In the outer medulla, collagen type IV immunoreactivity was not remarkably altered within 24 h of ischemic injury (Fig. 7, A vs. B). However, in areas adjacent to regenerating proximal tubules 3 days and 7 days postinjury, collagen type IV immunoreactivity occupied more tubulointersti-tial space than was observed in sham-operated or 24 h postischemic rats (Fig. 7, C and D). Collagen type IV immunoreactivity was unaltered in cortex in response to ischemic injury (Fig. 7, E vs. F).

View larger version (150K): [in this window] [in a new window]   Fig. 6.   Localization of 1(IV) collagen mRNA by nonisotopic in situ hybridization. mRNA for 1(IV) collagen was localized in kidneys from sham-operated rats and rats rendered ischemic. A and B: cross sections through rat cortex (cx) and outer medulla (om) of sham-operated rat (A) or 72 h postischemic injury (B). C and D: higher magnification through outer medulla both 3 days (C) and 7 days (D) postinjury. Arrowheads show positive hybridization in regenerating proximal tubule cells. E: no hybridization was observed with equal concentration of sense control riboprobe. Magnification for A and B is shown in A. Magnification for C-E is shown in E.

View larger version (86K): [in this window] [in a new window]   Fig. 7.   Immunofluorescent localization of collagen type IV frozen kidney sections (7 µm) fixed in acetone were probed using an antibody generated against collagen type IV. Sections are through rat kidney outer medulla (A-D and G) or cortex (E and F) of sham-operated rat (A and E) or after various times following ischemic injury, i.e., 24 h (B), 72 h (C), or 7 days (D, F, and G). Arrowheads denote positive immunoreactivity in basement membrane of regenerating proximal tubules and increased collagen immunoreactivity in tubulointerstitial spaces. E: no immunoreactivity was observed when the primary antibody was replaced with nonimmune mouse IgG. White bar in G is 100 µm and applies to A-G.

Similar analysis was performed for the fibronectin gene. FN-EIIIA mRNA was undetectable in sham-operated control rat kidneys by in situ hybridization (Fig. 8A) but was potently expressed following ischemic injury (Fig. 8B). FN-EIIIA mRNA was localized primarily in outer medullary regenerating proximal tubule cells (Fig. 8, C and D, arrowheads). Immunofluorescence analysis was performed using an antibody specific for FN-EIIIA (Sigma Clone FN-3E2). In cortex, FN-EIIIA immunoreactivity was abundant in glomeruli but absent from TBMs (Fig. 9A). In comparison, TBMs in the outer medulla exhibited faint immunoreactivity (Fig. 9B). FN-EIIIA immunoreactivity was enhanced as early as 24 h postischemic injury (data not shown) but was most prominent after 7 days in the basement membrane of regenerating proximal tubules in the outer medulla (Fig. 9D). FN-EIIIA immunoreactivity was essentially unaltered in the cortex postischemia (Fig. 9C). Similar results were obtained using another antibody specific for FN-EIIIA (Harlan-MAS 521, data not shown). In contrast, total fibronectin immunoreactivity was prominently expressed in glomeruli and the TBM of all nephron segments; a slight increase in tubulointerstitial fibronectin was observed 7 days postinjury (data not shown).

View larger version (144K): [in this window] [in a new window]   Fig. 8.   Localization of the EIIIA splice variant of fibronectin mRNA by nonisotopic in situ hybridization. FN-EIIIA mRNA was localized in kidneys from sham-operated rats and rats rendered ischemic. A and B: cross sections through rat cortex (cx) and outer medulla (om) of sham-operated rat (A) or 72 h postischemic injury (B). C and D: higher magnifications through outer medulla both 3 days (C) and 7 days (D) postinjury. Arrowheads show positive hybridization in regenerating proximal tubule cells. E: no hybridization was observed with equal concentration of sense control riboprobe. Magnification for A and B is shown in A. Magnification for C-E is shown in E.

View larger version (83K): [in this window] [in a new window]   Fig. 9.   Immunofluorescent localization of EIIIA fibronectin. Frozen kidney sections (7 µm) fixed in acetone were probed using an antibody specific for the EIIIA splice variant of fibronectin (Sigma clone 3E2). Sections are through rat kidney cortex (A and C) or outer medulla (B and D) of sham-operated rat (A and B) or 7 days after ischemic injury (C and D). Arrows denote positive immunoreactivity in glomerulus; arrowhead denotes positive immunoreactivity in the basement membrane of regenerating proximal tubules. No immunoreactivity was observed when the primary antibody was replaced with nonimmune mouse IgG (not shown). White bar in C is 100 µm and applies to A-D.

Effect of TGF- antibody administration on the postischemic expression of ECM-associated genes. To determine whether TGF- activity mediates the expression of ECM-associated genes postischemic injury, rats were subjected to 60 min of bilateral renal ischemia or sham surgery. A TGF--neutralizing antibody (anti-TGF--IgG) or nonimmune mouse IgG (control IgG) were administered to rats via tail vein injection immediately following reperfusion and once again 48 h postreperfusion. Serum creatinine levels were 3.1 ± 0.3 (range 1.5-4.2) and 2.8 ± 0.3 (range 1.0-3.8) mg/dl 24 h following surgery for vehicle-treated and anti-TGF- treated rats, respectively, and were not statistically different at any time point evaluated. The mRNA expression of FN-EIIIA, ₁(IV) collagen, PAI-1, and TIMP-1 were all significantly enhanced in kidneys of postischemic rats 3 days postinjury relative to sham-operated controls. Importantly, the levels of FN-EIIIA, ₁(IV) collagen, PAI-1, and TIMP-1 mRNAs measured in kidneys of postischemic animals treated with anti-TGF- antibodies were significantly suppressed by 52%, 73%, 43%, and 27%, respectively, compared with postischemic animals that received the nonimmune IgG control (Fig. 10). These data suggest that TGF- activity postischemic injury participates in the activation of genes associated with ECM remodeling.

View larger version (33K): [in this window] [in a new window]   View larger version (19K): [in this window] [in a new window]   Fig. 10.   Effect of in vivo administration of a neutralizing antibody against transforming growth factor- (TGF-) on expression of extracellular matrix-associated genes following ischemic acute renal failure. Bilateral renal ischemia was induced in rats, and recovery was allowed for 72 h. Animals were treated twice with anti-TGF- antibody or nonimmune mouse IgG (control). Expression of extracellular matrix-associated genes was measured by ribonuclease protection assay. A: representative autoradiograms for sham-operated or postischemic (ARF) rats treated with control IgG or TGF--neutralizing antibody (T). B: data obtained from phosphorimager scans are expressed as means ± SE and are normalized (represented by solid horizontal lines) to levels obtained from sham-operated vehicle-treated controls (n = 6). There was no effect of antibody treatment on the expression of these genes in sham-operated animals (n = 3, data not shown). * P < 0.05 in postischemic antibody-treated vs. postischemic vehicle-treated controls by Student's two-tailed t-test.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

Renal ischemia results in increased proteolytic activity and cell lysis (4). This increased proteolytic activity has been hypothesized to mediate degradation of the underlying proximal TBM. Consistent with this hypothesis, Walker (34) demonstrated a profound decrease in immunofluorescent staining of laminin within 18 h of injury (34). Furthermore, immunofluorescent staining of laminin and fibronectin were enhanced within 3-5 days postinjury, suggesting a shift from a ECM-degrading environment to one that promotes ECM synthesis (34). Based on these observations, we hypothesized that the increased expression of TGF- in kidney postischemic injury contributes to the maintenance and/or remodeling of the proximal tubule basement membrane by enhancing synthesis of ECM. TGF- has been shown to increase the mRNA expression of PAI-1, TIMP-1, fibronectin, and ₁(IV) collagen in renal proximal tubule cells in culture (18, 21, 33). This study was undertaken to determine the possible relationship between the expression of TGF- and these ECM-related molecules following the induction of ischemic acute renal failure. The expression of each of these gene products was potently increased as early as 12 h to 3 days postinjury and remained significantly elevated for between 7 and 28 days. We have shown that the renal expression of TGF-1 is enhanced as early as 12 h and as late as 14 days postischemia (3). Therefore, there is a temporal correlation between the expression of these mRNAs and that reported for TGF-1 mRNA (3).

PAI-1 is thought to play a critical role in biological processes such as wound healing, embryogenesis, tumor invasion, and angiogenesis, where cells migrate across or through basement membranes (2, 19, 30). In kidney, PAI-1 mRNA expression is normally restricted to endothelial cells but is expressed in renal parenchyma in models of renal disease such as proliferative glomerulonephritis, lupus nephritis, and endotoxemia that are associated with increased TGF- expression (2, 17, 19, 28). These observations suggest that TGF- activity may increase PAI-1 expression in renal parenchyma. In situ hybridization and immunohistochemical data in this study are consistent with this suggestion. We were unable to detect PAI-1 expression in proximal tubule cells of kidneys of sham-operated rats. However, both PAI-1 mRNA and protein were clearly evident in regenerating proximal tubules in the outer medulla following ischemic injury. Thus, there is a strong spatial and temporal relationship between the renal expression of PAI-1 and that reported for TGF-1 in this model (3). We suggest that PAI-1 expression in the postischemic kidney suppresses fibrinolytic activity, resulting in the maintenance and/or build up of the proximal tubule basement membrane. In addition, PAI-1 may participate in coordinating migration of newly formed proximal tubule cells in a manner similar to that suggested for mesangial cells in Habu snake venom-induced glomerulonephritis (2).

Matrix metalloproteases (MMPs) are a family of zinc- and calcium-dependent proteases that include collagenases, gelatinases, and stomelysins that collectively act to degrade ECM (7). The balance between TIMPs vs. MMPs is important in maintaining ECM homeostasis (7). Like PAI-1, TIMP-1 is induced by TGF- in vitro (7) and is enhanced in the setting of renal fibrotic diseases (5, 36). However, in contrast to PAI-1, TIMP-1 was not expressed in damaged and/or regenerating proximal tubules. Rather, it was predominantly expressed in distal tubule, collecting ducts, and thick ascending limb of Henle. These results were unexpected in light of recent studies in which TIMP-1 mRNA expression was shown to be enhanced in human proximal tubular epithelial cells in response to hypoxia in vitro (26). It is possible that our techniques were not sufficiently sensitive to detect TIMP-1 expression in proximal tubules or that these discrepancies are the result of differences in the species or experimental models used. The role of TIMP-1 expression following ischemic injury is unclear. However, TIMP-1 may contribute a protective effect to the distal nephron.

We were also interested in characterizing the expression of glycoprotein constituents of the basement membrane under the control of TGF-. Type IV collagen comprises ~50% of the tubular epithelial basement membrane, and the mRNA of the ₁-chain is potently induced by TGF- in cultured proximal tubules (18). The turnover of collagen IV is slow under basal conditions, and the mRNA of the ₁-chain is expressed at low levels. Consistent with the findings of Walker (34), we were unable to detect any significant alteration in the collagen IV immunoreactivity in the TBM within 24 h of ischemic injury. Walker (34) speculated that the high content of collagen IV in the TBM precluded observing any detectable decrease by proteolysis by immunofluorescence analysis. However, ₁(IV) collagen mRNA is potently induced in tubular epithelial cells of the outer medulla following injury, and collagen type IV immunoreactivity occupies more tubulointerstitial space by 7 days postinjury. Thus, an alternative hypothesis is that TGF- activity in the outer medulla results in increased collagen IV synthesis and maintenance of collagen IV content in the TBM in the presence of a proteolytic milieu.

Fibronectins are high-molecular-weight glycoproteins thought to play important roles in cell adhesion, differentiation, and migration. Differential pre-mRNA splicing of fibronectin type III repeat domains results in the formation of two well-characterized splice variants referred to as EIIIA (EDA+ in human) and EIIIB (EDB+ in human) (29). FN-EIIIA mRNA expression is induced in proximal tubules following stimulation with TGF- in vitro (20, 33). In this study, the specific increase in FN-EIIIA mRNA in regenerating proximal tubules of the outer medulla demonstrated by in situ hybridization correlated closely with the deposition of FN-EIIIA in the TBM of regenerating proximal tubules detected by immunofluorescence with either of the two different FN-EIIIA-specific antibodies.

Several laboratories have demonstrated that FN-EIIIA+ mRNA is abundantly expressed during embryogenesis (27) and wound healing (6) compared with normal adult tissue. The FN-EIIIA domain is adjacent to the RGD binding domain of the fibronectin molecule that mediates cell attachment through 51 integrins. EIIIA+-containing fibronectins have altered cellular binding properties (37) and have been shown to affect differentiation of hepatic lipocytes to myofibroblasts following hepatic injury (14). Thus, its expression following renal ischemic injury suggests an important functional role for this glycoprotein in events such as migration and/or differentiation of the newly formed proximal tubule cells. We suggest TGF- influences redifferentiation of renal proximal tubule cells either directly or indirectly though the production of FN-EIIIA.

To provide direct evidence for the involvement of TGF- following the induction of ischemic acute renal failure, immunoneutralization experiments were carried out for 72 h postinjury resulting in ~30-80% inhibition in the expression of mRNAs investigated in this study. The suppression of mRNA induction was similar to that reported in diabetic mouse kidneys using the same antibody (31). These results suggest that TGF- participates in renal regeneration by enhancing the synthesis of genes associated with the ECM.

In this model of acute renal failure, serum creatinine levels return near baseline by 7 days postinjury despite the presence of obvious morphological abnormalities and high levels of TGF-1 and ECM-related mRNAs (3, 24). It is likely that TGF- immunoneutralization for only 72 h postinjury precluded observing physiologically significant effects that may be more apparent at later times during regeneration. Future studies will be aimed at determining the effect of immunoneutralization on the cellular events that occur later in the course of renal regeneration (e.g., hypertrophy, differentiation). We hypothesize that TGF- activity postischemia participates in renal regeneration by promoting ECM synthesis and providing newly formed proximal tubule cells a solid substrate for adhesion, migration, and exposure to ECM-based ligands that influence cellular behavior.