INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

THE CD44 FAMILY of cell surface glycoprotein receptors is widely expressed in embryonic, normal adult, and neoplastic tissues. CD44s serve as adhesion molecules in cell-cell and cell-substrate interactions that mediate processes such as cell migration during organogenesis or wound repair and metastasis (2, 11). Hyaluronic acid, a ubiquitous polysaccharide synthesized by fibroblasts, chondrocytes, and mesothelial cells, is the major ligand for CD44 (2, 11, 19). However, CD44 can bind other ligands, including osteopontin (20, 24, 25), a secreted phosphoprotein expressed at many epithelial cell surfaces in communication with the outside environment including the nephron (16, 17).

Little is known about the role of CD44 in the kidney. In kidneys of normal rats, CD44 has been reported to be undetectable (26) or detectable in intrinsic glomerular cells, parietal epithelial cells of Bowman's capsule, medullary tubules, and occasional cortical tubules (thick ascending limb of Henle's loop and distal tubules; see Refs. 6 and 12). Under normal conditions in humans, it is undetectable (22) or barely detectable in the distal tubule (9, 21) by immunohistochemistry. However, consistent with its role as a mediator of cancer cell proliferation and migration, CD44 is easily detected in the cell membranes of malignant but not benign renal cell tumors in humans (22).

CD44 is expressed de novo by proximal tubular cells in kidneys of kdkd mice, which develop a progressive T cell-mediated tubulointerstitial nephritis. Osteopontin is coexpressed in injured tubules, and hyaluronic acid accumulates in the interstitial space, particularly in areas of tubular injury (19). It has been suggested that the interaction of CD44 with these ligands could participate in the tubulointerstitial inflammatory response in kdkd mice (19).

De novo CD44 and ligand expression at wound margins accompanies cellular proliferation and migration that effect repair of injured vascular endothelial (5) and epithelial (15) tissues. Recovery of renal function after ischemic acute renal failure is dependent upon a process of reepithelialization effected via cellular proliferation and migration across portions of the denuded proximal tubule basement membrane. To determine whether de novo expression of CD44 and ligands in the regenerating proximal tubule might play a role in the process of recovery postischemic injury, we characterized the expression and localization of CD44 and two of its ligands, hyaluronic acid and osteopontin, in kidneys of rats that underwent renal ischemia and in kidneys of sham-operated controls. Our data are consistent with CD44-ligand interactions playing a role in the process of proximal tubule regeneration after renal ischemia.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animals. Male Sprague-Dawley rats (Harlan, Indianapolis, IN) weighing ~250 g were housed in a 12:12-h light-dark cycle with food and water available ad libitum.

Induction of acute renal failure. Acute renal failure was induced by 60 min of bilateral renal artery clamping, as described previously (1). To control for the extent of renal injury during the procedure, animals were selected such that the level of serum creatinine measured 24 h after injury fell into the range used in previous studies (1, 16). Levels of serum creatinine measured 24 h postischemia (means ± SE) were 4.2 ± 0.3, 3.2 ± 0.5, 3.2 ± 0.5, 3.0 ± 0.3, and 3.5 ± 0.1 mg/dl in groups of rats used to obtain kidneys 1, 2, 3, 5, and 7 days postischemia, respectively. Levels were 2.6 ± 0.1 mg/dl (n = 3 rats) 12 h postinjury in the group of rats that was killed at this time. Levels of serum creatinine in sham-operated rats were measured at the time of death and averaged 0.6 ± 0.01 mg/dl (n = 27). These levels did not differ significantly from group to group. Levels of serum creatinine at the time of death in rats that had been rendered ischemic were 2.0 ± 1.2, 1.9 ± 0.9, 0.8 ± 0.1, and 0.7 ± 0.1 mg/dl at 2, 3, 5, and 7 days postinjury, respectively.

Northern blot analysis. RNA obtained from kidneys of sham-operated rats and rats rendered ischemic as before (16) were used in RNA blot hybridizations. Total RNA (20 µg) was fractionated in 2.2 M formaldehyde-1.2% agarose gels and was transferred onto Zeta Probe membranes (Bio-Rad, Hercules, CA). Hybridization and washing conditions were as previously reported (16). A specific cDNA probe for CD44 was generated from a 1228-bp clone that was kindly provided by Dr. M. Jain (Beth Israel Hospital, Harvard Medical School, Boston, MA; see Ref. 5).

With the use of the PCR, T3 and T7 RNA polymerase promoter sequences were added to an amplified CD44 cDNA. The specific T3- and T7-linked oligonucleotides were 5'-AATTAACCCTCACTAAAGGGTATATCCTCCTCGCATCCA-3' and 5'-TAATACGACTCACTATAGGGTCCTGTCTTCCACTGTTCC-3', respectively. The synthesis of a 667-bp PCR product was confirmed using ethidium bromide-agarose gel electrophoresis. Restriction enzyme digest demonstrated cDNA fragments of the predicted size for CD44. The CD44 PCR product was radiolabeled with [³²P]dCTP (Amersham, Arlington Heights, IL) using a random prime labeling kit (Stratagene, La Jolla, CA).

To determine that equivalent amounts of RNA were present in each lane, an ethidium bromide-agarose gel containing aliquots of the same RNA used for Northern blots was photographed under ultraviolet light as before (1).

In situ hybridization. In situ hybridization was performed on tissue sections originating in kidneys from sham-operated rats and rats rendered ischemic as previously described (16). Briefly, kidneys were fixed in Bouin's fixative and were processed for paraffin sectioning. Sections of 5 µm were cut and mounted on super frost Plus slides (Fisher Scientific, Pittsburgh, PA).

A 667-bp CD44 cDNA fragment was used as the template to generate sense and antisense probes. The transcription probes were labeled with digoxigenin-11-dUTP (Boehringer Mannheim, Indianapolis, IN) using the RNA transcription kit (Stratagene) in the presence of RNasin (Promega). One hundred microliters of hybridization buffer (16) containing the CD44 probe were heated to 80°C for 5 min and applied on tissue sections, covered with a coverslip, and incubated in a humid chamber at 50°C overnight. Slides were washed at 60°C in 50% formamide, treated with RNase, and rinsed in 0.1× saline sodium citrate at 50°C. Sections were blocked with 2% normal sheep serum and were incubated with the anti-digoxigenin antibody (Boehringer-Mannheim). Color development was with nitro blue tetrazolium and bromochloroindolyl phosphate in the presence of levamisole, and the reaction was stopped when staining was judged to be satisfactory. The tissues were mounted in aqueous medium and were photographed under light microscopy.

Histochemistry. Immunohistochemistry was performed as previously described (16) on 5 µm Bouin's-fixed paraffin-embedded sections. After deparaffinization in xylene, tissue was pretreated with 0.6% H₂O₂ in 80% methanol, and sequential incubations were carried out in solutions containing avidin and biotin (Zymed, San Francisco, CA), respectively.

The primary osteopontin antibody MOIIIB10 was obtained from the developmental studies hybridoma bank maintained by the Department of Pharmacology and Molecular Sciences, Johns Hopkins University School of Medicine (Baltimore, MD) and the Department of Biological Sciences, University of Iowa (Iowa City, IA). The bank is maintained under contract NO1-HD-6-2915 from the National Institute of Child Health and Human Development. The osteopontin antibody was applied in the 1/10th concentration of the blocking buffer overnight at 4°C. The primary CD44 antibody was mouse anti-rat monoclonal antibody OX49 (Pharmingen, San Diego, CA) and was used at a concentration of 1:50. As controls, nonimmune cellular supernatant obtained from the same source as MOIIIB10 was substituted for MOIIIB10, and mouse IgG₁ was substituted for OX49. Positive staining was not observed in control experiments.

The biotinylated proteoglycan (b-PG) fragments used to localize hyaluronic acid were kindly provided by Dr. C. Underhill (Georgetown University, Washington, DC). Tissue sections were incubated with 8 µg/ml of b-PG in 10% goat serum and 90% PBS overnight at 4°C. Immunohistochemistry was performed as described above. As a control, tissue sections were incubated in b-PG that had been absorbed with 0.1 mg/ml hyaluronic acid.

Physiological studies. Levels of serum creatinine were measured as described previously (10).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

To determine whether CD44 is expressed in rat kidney after ischemic injury, we first performed a Northern analysis of RNA extracted from kidneys of sham-operated rats or rats with acute renal injury rendered ischemic 1, 3, 5, or 7 days before death (Fig. 1). A heterogeneous mixture of mRNAs was present in kidneys from rats rendered ischemic but not in kidneys from sham-operated rats. The sizes of transcripts (4.5, 3.3, 2.0, and 1.6 kb) were identical to those reported elsewhere (5) in rat vascular smooth muscle cells. Multiple transcripts could have been generated by alternative splicing or by polyadenylation (5).

View larger version (74K): [in this window] [in a new window]   Fig. 1.   Northern blot of CD44 mRNA originating from kidneys of sham-operated rats (S) or rats with acute renal injury (A) rendered ischemic 1, 3, 5, or 7 days before extraction. Four transcripts are visible at 4.5, 3.3, 2.0, and 1.6 kb. An ethidium bromide-stained gel with 28S and 18S bands demarcated is shown to verify equivalent loading.

To delineate the localization of CD44 mRNA in kidney postischemia, we performed in situ hybridization using a CD44-specific antisense probe or a sense probe and sections of kidneys from sham-operated rats and rats rendered ischemic (Fig. 2). Little or no CD44 mRNA is expressed in kidneys from sham-operated rats (Fig. 2A). No staining is observed when sense riboprobe is substituted for the antisense probe (Fig. 2B). However, CD44 mRNA is expressed in regenerating proximal tubules in the inner cortex by 1 day postinjury (Fig. 2C) and continues to be expressed on days 3 (Fig. 2D), 5 (Fig. 2E), and 7 postischemia (Fig. 2F).

View larger version (190K): [in this window] [in a new window]   Fig. 2.   Localization of renal CD44 mRNA by nonisotopic in situ hybridization. CD44 mRNA in kidneys from rats subjected to sham surgery (S) 1 day before death or rendered ischemic 1-7 days before death (ARF) was localized using a digoxigenin-labeled antisense riboprobe or a sense riboprobe where indicated. A, 1 day S; B, 1 day ARF, sense riboprobe; C, 1 day ARF; D, 3 day ARF; E, 5 day ARF; F, 7 day ARF. Arrowheads show proximal tubules (C-F). Sections are representative of >5 experiments. Magnification is shown in A.

To determine whether changes in levels of CD44 peptide accompany the changes in CD44 mRNA, we performed immunohistochemistry using anti-CD44 antibody. There is little or no CD44 staining in kidneys originating from sham-operated rats (Fig. 3A) or in rats rendered ischemic 12 h postinjury (Fig. 3B). However, CD44 is expressed in proximal tubules by 1 day postinjury (Fig. 3, C and E) and at 5 days postinjury (Fig. 3, D and F). Enlargement of Fig. 3C shows that CD44 is expressed along the basal and lateral membranes of regenerating tubules (Fig. 3E).

View larger version (157K): [in this window] [in a new window]   Fig. 3.   Immunohistochemical localization of CD44 peptide in kidneys. CD44 peptide in kidneys from rats subjected to sham surgery (S) or rendered ischemic (ARF) 12 h, 1 day, or 5 days before death was localized using an antibody generated against CD44. A, 12 h S; B, 12 h ARF; C, 1 day ARF; D, 5 day ARF; E, 1 day ARF; F, 5 day ARF. Arrowheads show basal and lateral membranes of proximal tubular cells (E). Sections are representative of >5 experiments. Magnifications are shown for A-D and E and F.

The data shown in Figs. 1-3 establish that CD44 is upregulated in regenerating proximal tubules after ischemic injury. To shed light on the identities of possible ligands for CD44 in this setting, we localized hyaluronic acid and osteopontin in sections of kidneys obtained from rats rendered ischemic and from sham-operated controls.

It has been reported previously that hyaluronic acid is localized in rat kidney to the renal papilla in which it surrounds the interstitial cells (4). In kidney sections from sham-operated controls, incubation with absorbed b-PG results in negative staining (Fig. 4A). The use of b-PG shows that hyaluronic acid is expressed in the renal papilla of kidney from sham-operated controls as previously reported (Fig. 4B; see Ref. 4).

View larger version (174K): [in this window] [in a new window]   Fig. 4.   Localization of hyaluronic acid in kidneys. Hyaluronic acid in kidneys from rats subjected to sham surgery (S) or rendered ischemic 1-5 days before death (1-5 day ARF) was localized using biotinylated proteoglycan (b-PG) or a control (absorbed b-PG) when indicated. A, 1 day S control; B, 1 day S; C, 1 day ARF control; D, 1 day S; E, 1 day ARF; F, 3 day ARF; G, 5 day ARF. Arrowheads delineate margins of proximal tubule epithelial cells (E-G). Sections are representative of >5 experiments. Magnifications are shown for A and B (B) and for C-G (G).

In kidney sections obtained 1 day postischemia, incubation with absorbed b-PG results in negative staining (Fig. 4C). No hyaluronic acid staining is observed in cortex of kidneys from sham-operated rats (Fig. 4D). By 1 day postischemia (Fig. 4E) and at 3 (Fig. 4F) and 5 (Fig. 4G) days postischemia, hyaluronic acid is present in the renal cortex and is localized in the interstitium between regenerating proximal tubules (Fig. 4, E-G) but not in the tubules themselves. The margins of proximal tubule cells can be delineated in Fig. 4, E-G.

We have shown previously that the expression of osteopontin in the distal tubule and medullary thick ascending limb of Henle's loop is enhanced within 1 day of inducing acute ischemic renal injury in the rat (16). In addition, the coexpression of osteopontin and osteopontin mRNA is observed in regenerating proximal tubules at 5 days after ischemia (16). To better delineate the time course of enhanced osteopontin expression in the regenerating proximal tubules, we performed immunohistochemistry for osteopontin in kidneys obtained from rats 12 h and 1, 3, 5, and 7 days after injury and after sham surgery (Fig. 5).

View larger version (201K): [in this window] [in a new window]   Fig. 5.   Immunohistochemical localization of osteopontin peptide in kidneys. Osteopontin peptide in kidneys from rats subjected to sham surgery (S) or rendered ischemic (ARF) 12 h-7 days before death was localized using an antibody generated against osteopontin or, when indicated, a control antibody. A, 1 day ARF stained with control antibody; B, 1 day S; C, 12 h ARF; D, 1 day ARF; E, 3 day ARF; F, 5 day ARF; G, 7 day ARF. Arrowheads show proximal tubules (C-G). d, Distal tubule; p, papillary proliferation. Sections are representative of >5 experiments. Magnification is shown in F.

As in kidneys from normal rats (16), the use of control antibody resulted in no staining (Fig. 5A), and no staining for osteopontin was observed in proximal tubules in kidneys obtained from sham-operated rats (Fig. 5B). Enhanced staining in distal tubules but not in damaged proximal tubules was observed beginning as soon as 12 h after injury (Fig. 5C) and at 1 day (Fig. 5D). By 3 days postischemia, cells lining regenerating tubules stained for osteopontin (Fig. 5E), as was the case at 5 days (Fig. 5F). Papillary proliferations that are observed within regenerating proximal tubules (1) were clearly stained at 7 days (Fig. 5G).

Figure 6 shows serial sections of renal cortex originating from rats rendered ischemic 3 days (A and B) or 7 days (C and D) before death. The identical regenerating tubules are represented in Fig. 6, A and B or C and D. Coexpression of osteopontin (Fig. 6, A and C) and CD44 (Fig. 6, B and D) can be demonstrated in some regenerating tubules. Osteopontin and CD44 are present in some of the same cells. As previously reported (16), osteopontin is expressed in distal tubules of regenerating kidneys (Fig. 6C). However, CD44 is not coexpressed at this site (Fig. 6D).

View larger version (114K): [in this window] [in a new window]   Fig. 6.   Serial sections of rat kidneys obtained from a rat rendered ischemic 3 days (A and B) or 7 days (C and D) before death stained for osteopontin (A and C) or CD44 (B and D). +, Regenerating tubule that stains for both peptides. Arrowheads show cells that stain for both peptides. Sections are representative of 5 experiments. Magnifications are shown for A and B (A) and for C and D (C).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

CD44-ligand interactions are thought to play important roles in tissue development and repair through the mediation of cell binding to endothelium or to the extracellular matrix (11). One source of evidence for such involvement in these processes is the localization of CD44 at sites of active cell proliferation and migration.

In 10-wk-old human embryos, CD44 is found in diving cells of the epidermis, trachea, lung, thyroid gland, and mesonephric ducts (21). In newborn rats, CD44 has been identified in basal layers of the epidermis, hair follicles, the lower parts of crypts in colon mucosa, and ductal epithelia of pancreatic glands (26).

CD44 is normally present in cell membranes of all layers of the stratifying epithelium in the developed human palate except for the stratum corneum. Within 1 day after experimental full-thickness wounding of the palate mucosa, a fibrin clot occupies the wound space, and, at the margins, sheets of epithelial cells begin to migrate into the wound. The first migrating epithelial cells are weakly CD44 positive. However, by 3 days postinjury, they are markedly positive. By day 1, the lateral sides of the wound bed stain positive for the CD44 ligand, hyaluronic acid. CD44 in the plasma membrane of migrating cells and hyaluronic acid in the wound margin and connective tissue matrix are coexpressed during all tissue repair stages, suggesting that the CD44-matrix-hyaluronic acid interaction plays an important role in reepithelialization (15).

Similar to events in the palate mucosa, CD44 mRNAs are upregulated in rat carotid arteries after balloon-induced injury. CD44 protein expression is greatest at the luminal edge of the growing neointima. CD44-expressing smooth muscle cells proliferate actively after injury, and hyaluronic acid expression increases at the same time throughout the neointima (5).

Osteopontin is normally expressed at low levels in the renal distal tubule and medullary thick ascending limb of Henle's loop (16). Within 1 day after acute ischemic injury in the rat, its expression is enhanced at these sites (7, 16). Later, during the process of regeneration, osteopontin is expressed in regenerating proximal tubules (16). The importance of osteopontin expression in the process of recovery is underscored by the observation that transgenic mice unable to express osteopontin do not recover normally from acute ischemic injury (13). Consistent with a wider role of osteopontin in the process of wound healing is the finding that wound repair in mice lacking a functional osteopontin gene is abnormal (8).

Ligands for osteopontin in addition to CD44 include integrins, primarily _v3, to which it binds via an arginine-glycine-aspartate (RGD) motif (17). The integrins _v1 and _v5 also serve as osteopontin receptors on vascular smooth muscle cells (17).

In the kidney, binding sites for RGD are normally detected along the basolateral membrane of proximal tubule cells. Within 1 day after ischemic injury, sites are increased along the basolateral membrane and can be detected also along the apical membrane, in peritubular capillaries, and on desquamated cells within tubular lumens (18). RGD binding sites within regenerating proximal tubules and in desquamated cells colocalize with the ₁ integrin subunit. Binding sites in the vasculature colocalize with the _v-subunit (18).

The administration of cyclic RGD peptides, RGD_DFV and RGD_DFLG, to rats after acute renal ischemia ameliorates the injury (14). To gain insight into the mechanism by which the amelioration occurs, effects of the peptides on cell-cell and cell-matrix adhesion were delineated in vitro, using cultured BS-C-1 cells. Each peptide was shown to block cell-cell adhesion. However, although RGD_DFV was a potent inhibitor of cell-matrix adhesion, RGD_DFLG was not. It was suggested that the beneficial action of the cyclic RGD peptides results predominantly from their ability to inhibit the cell-cell aggregation of sloughed desquamated cells (14). We (16) and others (14, 18) have suggested that the expression of RGD-containing osteopontin at distal nephron sites postischemia could serve such a function after acute renal injury.

The action of RGD_DFV to inhibit cell-matrix adhesion in vitro did not preclude a salutary action of this peptide on the course of acute renal failure in vivo (14). The absence of a negative action of RGD_DFV postischemia, coupled with the observation that neither the _v nor ₁ integrin subunits are expressed in regenerating proximal tubule cells after injury (18), suggest that RGD-ligand interactions may not be essential for the process of reepithelialization in the proximal tubule. If such is the case, and osteopontin plays a role in kidney repair after ischemic injury, as is indicated by impaired recovery in transgenic mice lacking osteopontin (13), its actions must be mediated via binding to a non-RGD-requiring (nonintegrin) receptor. Our data are consistent with CD44 being the nonintegrin osteopontin receptor.

The expression of CD44 in cells of regenerating proximal tubules is observed within 1 day after injury (Figs. 2 and 3), as is the synthesis of hyaluronic acid in the peritubular interstitium (Fig. 4). In contrast, osteopontin cannot be detected in regenerating tubules until later (3 days postischemia; Fig. 5). Although CD44 and hyaluronic acid are not both localized in regenerating proximal tubule cells, CD44 and osteopontin are coexpressed in at least some cells (Fig. 6). Such a temporal and spatial discordance between the synthesis of hyaluronic acid and osteopontin could reflect different roles for ligation of CD44 by each after ischemic injury. For example, CD44-hyaluonic acid may mediate cell-matrix attachment and CD44-osteopontin coexpression could, in some way, regulate cell migration as it has been proposed to do in malignant tumor cells (24).

Perspectives

Our studies do not assess directly a role for CD44 ligation by either hyaluronic acid or osteopontin in the process of recovery after injury, nor do they shed light on the mechanisms by which ligation promotes repair. However, as is the case in injured mucosa (15) and vascular smooth muscle (5), our findings of de novo CD44 and ligand expression postischemia in the kidney are consistent with a key role for CD44-ligand interactions in the process of relining of the partially denuded basement membrane by proliferating, migrating epithelial cells.

We have shown that enhanced renal expression of one CD44 ligand, osteopontin, occurs in association with the amelioration of acute renal failure that results from the administration of insulin-like growth factor I to rats before induction of ischemia (10, 16). Our demonstration of CD44 expression in the kidney postischemia sheds light on a possible mechanism for the beneficial action of insulin-like growth factor I in the rat (10) and perhaps in humans (3). A more thorough understanding of CD44-ligand interactions after renal injury could provide useful insights into additional therapeutic modalities.