Severe Fibronectin-Deposit Renal Glomerular Disease in Mice Lacking Uteroglobin

See allHide authors and affiliations

Science  30 May 1997:
Vol. 276, Issue 5317, pp. 1408-1412
DOI: 10.1126/science.276.5317.1408


Despite myriads of biological activities ascribed to uteroglobin (UG), a steroid-inducible secreted protein, its physiological functions are unknown. Mice in which the uteroglobin gene was disrupted had severe renal disease that was associated with massive glomerular deposition of predominantly multimeric fibronectin (Fn). The molecular mechanism that normally prevents Fn deposition appears to involve high-affinity binding of UG with Fn to form Fn-UG heteromers that counteract Fn self-aggregation, which is required for abnormal tissue deposition. Thus, UG is essential for maintaining normal renal function in mice, which raises the possibility that an analogous pathogenic mechanism may underlie genetic Fn-deposit human glomerular disease.

Blastokinin (1) or UG (2) is a steroid-inducible, evolutionarily conserved, homodimeric secreted protein with many biological activities including the ability to inhibit soluble phospholipase A2 (sPLA2) activity, inflammation, and chemotaxis of neutrophils and monocytes (3). UG binds with high affinity to specific putative receptor sites on several cell types (4, 5) and through this pathway inhibits cellular invasion of the extracellular matrix (4). Although UG was first discovered in the rabbit uterus, it is also expressed in numerous extrauterine tissues (6) and has been detected in the blood (7) and urine (8), but it is not expressed in the kidneys (6). Depending on the tissue of origin or its interaction with xenobiotics (for example, progesterone, retinol, and polychlorinated biphenyls), UG has been given several names (5, 9). The tissue-specific expression of the gene encoding UG is regulated by steroid hormones (3). The nonsteroid hormone prolactin further enhances steroid-induced UG gene expression (10), and the proinflammatory cytokine interferon-γ stimulates UG production in the murine lungs (11), suggesting a potential role of UG in the regulation of immunological inflammatory processes. In addition, the inhibition of the recognition of embryonic and sperm antigens by lymphocytes and the inhibition of chemotactic peptide (formyl-Met-Leu-Phe)-induced monocyte and neutrophil chemotaxis by UG suggest that UG has immunomodulatory properties (12). Human UG (hUG) is encoded by a single-copy gene on chromosome 11q12.3-13.1 (13), a region to which a number of candidate disease genes have been mapped. Despite more than three decades of intense investigations, which uncovered several important biological properties of UG in vitro, the in vivo functions, until now, remained obscure.

To understand the physiological roles of UG, we generated UG-deficient (UG−/−) mice by gene-targeting in embryonic stem (ES) cells. The UG gene from the 129/SVJ mouse strain (14) was used in the gene-targeting construct (Fig. 1A) that was introduced into ES R1 cells (15) by electroporation. Gancyclovir and G-418 counter-selection of the electroporated cells yielded 156 clones. Southern (DNA) blot analysis identified a 5.1-kb Hind III fragment of the wild-type UG allele and an additional 8.2-kb Hind III fragment resulting from homologous recombination in three out of the 156 clones analyzed (Fig. 1B). These ES R1 clones were injected into C57BL/6 blastocysts (16), generating two different lines of mice, each of which descended from an independent chimeric founder. Heterozygous (UG+/–) offspring carrying the targeted UG gene locus were mated, and the genotypes of the progeny were analyzed by polymerase chain reaction (PCR) (Fig. 1C) and Southern blot analyses (Fig. 1D).

Figure 1

Targeting of the UG locus. (A) Diagram of the UG gene locus, targeting construct, and resulting UG targeted locus. B, Bam HI; E, Eco RI; H, Hind III. (B) Southern blot analyses of the targeted ES R1 cell clones; wt, wild type. (C) Representative PCR analyses of genomic DNA from tail biopsies of offspring. The genotypes and their corresponding PCR products are as follows: UG+/+, 304 base pairs (bp); UG+/–, 304 and 667 bp; and UG−/−, 667 bp. (D) Southern blot of mouse tail genomic DNA. (E) RT-PCR analyses of total RNA extracted from the lung tissues of littermates with UG+/+, UG+/–, and UG−/− genotypes. A 273-bp RT-PCR product was detectable in the lungs of UG+/+ and UG+/– mice but lacking from those of UG−/−mice. (F) Protein immunoblot analysis. Proteins (30 μg of each) from lung lysates were resolved by electrophoresis on 4 to 20% gradient SDS-polyacrylamide gels under nonreducing conditions and immunoblotted with anti-UG. (G) Immunohistochemical localizationof UG in bronchiolar epithelial cells. The dark staining over the bronchiolar epithelial cells of a UG+/+ mouse (upper panel) indicates UG immunoreactivity. Note the absence of immunoreactivity in UG−/− mouse lungs (lower panel). Methods are described in (17-19, 31). Magnification ∼×100.

We tested the UG gene–targeted mice for expression of UG mRNA and UG protein in several organs including the lungs. Using reverse transcription–PCR (RT-PCR) (17), we detected UG mRNA from the lungs of UG+/+ and UG+/– but not of UG−/− mice (Fig. 1E). Immunoblot analyses (18) of UG protein in the lungs yielded corroborative results (Fig. 1F). Histopathological analyses (19) of the lungs of UG−/− mice lacked UG-specific immunostaining in bronchiolar epithelial cells (Fig. 1G). The prostate and the uteri of UG−/− but not of UG+/+ and UG+/–mice lacked UG mRNA and protein.

Of the 179 mice born to crosses of UG+/– mice, 46 (26%) were UG+/+, 90 (50%) were UG+/–, and 43 (24%) were UG−/−, indicating that the disrupted UG gene locus is inherited in a Mendelian fashion and that UG+/+, UG+/–, and UG−/− mice were equally viable at birth. UG−/− mice developed a progressive illness characterized by heavy proteinuria and hypocalcemia associated with profound weight loss. Histopathological examination (19) of affected UG−/− animals revealed a fulminant renal glomerular disease (Fig. 2). Compared with the glomeruli of the UG+/+ mice, those of UG−/− mice were hypocellular and had massive eosinophilic proteinaceous deposits. Heterozygotes had a milder form of the renal disease observed in UG−/− mice. The majority of the UG−/− mice that initially appeared to be healthy had focal glomerular deposits at 2 months of age. However, at about 10 months of age (late onset disease) many of these apparently healthy mice had extreme cachexia similar to that of the mice dying at 4 to 5 weeks of age (early onset disease). The histopathology of the kidneys of mice with late onset disease showed not only severe glomerulopathy as with early onset disease but also had marked fibrosis of the renal parenchyma and tubular hyperplasia (Fig. 2). Although the predominant pathology in the UG−/− mice was found in the kidneys, histopathological studies also uncovered focal areas of necrosis in the pancreas.

Figure 2

Severe renal glomerular disease in UG−/− mice. Hematoxylin and eosin staining of kidney sections from a UG+/+ mouse (A) and its UG−/− (B) littermate. (C) Kidney section of a 10-month-old mouse with severe parenchymal fibrosis. (D) A region of the same mouse kidney in (C), showing renal tubular hyperplasia. Magnification ∼×215; g, glomerulus; f, fibrosis; t, tubule. (E) Transmission electron microscopy of the glomerular deposit of a UG−/− mouse with severe renal disease. Original magnification ×6000. (F) The same sample as in (E), original magnification ×60,000, which shows the long striated fibrillar structures indicative of collagen (Col) and short diffuse ones consistent with Fn fibrils. (G) Fn immunofluorescence of a kidney section from a UG+/+ mouse stained with anti-Fn. (H) Fn immunofluorescence of a kidney section from a UG−/−mouse with severe renal disease. Mason’s trichrome staining of the kidney sec tions from UG+/+(I) and UG−/− (J) mice. The bluish staining over the glomeruli of the UG−/− mouse kidney section is collagen. Magnification ∼×215.

Because reactive amyloidosis may occur in response to inflammation and because UG has immunomodulatory and antiinflammatory properties, we stained kidney sections from UG+/+ and UG−/− mice with Congo red and examined them under polarized light. Amyloid proteins yield a positive birefringence in this test; however, the glomeruli of UG-null mice were negative. Immunofluorescence studies for the presence of immunoglobulin A (IgA), IgG, or IgM immunocomplexes in the glomeruli of UG−/−mice and immunohistochemical analyses for the presence of major amyloid proteins were also negative. Thus, the glomerular deposits of UG−/− mice contained neither amyloid proteins nor immunocomplexes.

We examined the kidney deposits of UG−/− mice by transmission electron microscopy (20). These deposits contained primarily two types of fibrillar structures: one type of long and striated fibrils, which were relatively infrequent, and another type of short and diffuse fibrils, which were more abundant (Fig. 2, E and F). Because extracellular matrix (ECM) proteins, such as collagen and fibronectin, produce similar fibrillar structures, the glomerular deposits in UG−/− mice may contain these proteins. We analyzed the glomerular deposits by immunofluorescence (21) using antibodies to murine Fn (anti-Fn). Whereas Fn-specific immunofluorescence in the renal glomeruli of wild-type mice was virtually undetectable (Fig. 2G), that in the glomeruli of UG−/− littermates was intense (Fig. 2H). When Masson’s trichrome staining was used, the glomeruli of UG+/+ mice were negative (Fig. 2I) and those of UG−/−(Fig. 2J) mice were positive, suggesting the presence of collagen in the glomerular deposits. Immunofluorescence, with antibodies specific for collagen I and collagen III, confirmed these results. Because Fn is known to interact with other ECM proteins, we also tested for the presence of laminin, vitronectin, and osteopontin in the glomeruli of UG+/+ and UG−/− mice by immunohistochemistry, the results of which were negative.

To determine whether excessive production of Fn could account for its deposition in the renal glomeruli, we assessed the relative amount of Fn mRNA in the kidneys, lungs, and liver of UG−/− and UG+/+ mice by RT-PCR and densitometry. The results indicate that relative amounts of Fn mRNA were essentially identical in both UG+/+ and UG−/− animals. Thus, overproduction of Fn mRNA was not a likely cause of Fn deposition in the glomeruli of UG−/−mice. We then compared the Fn protein in the plasma, kidneys, and liver of UG−/− and UG+/+ mice by SDS–polyacrylamide gel electrophoresis (SDS-PAGE) under reducing conditions and the protein immunoblotting. In the plasma, kidneys, and liver of wild-type mice, only 220-kD Fn species were detected; whereas the plasma and the liver lysates of UG−/− mice had the 220-kD Fn band, the kidney lysates contained another distinct, covalently linked, multimeric Fn band (Fig. 3A).

Figure 3

Detection of multimeric Fn in UG−/− mice and the effect of UG on Fn-Fn and Fn-collagen interactions. (A) Immunoprecipitation (18) and protein immunoblotting of Fn from plasma, kidney, and liver of UG+/+ and UG−/− mice. A multimeric Fn band (bold arrow) was detected only in the kidney lysates of UG−/− mice. Ori, origin; Std, Fn standard. (B) Equimolar concentrations of UG and Fn were incubated and then immunoprecipitated with anti-Fn, and the immunoprecipitates were resolved by SDS-PAGE under reducing conditions and detected by immunoblotting with anti-Fn or anti-UG. The immunoprecipitates contain both Fn (lane 2, upper panel) and UG (lane 2, lower panel). Lanes 1 of both panels represent Fn and UG standards. (C) Equimolar concentrations of 125I-UG and Fn were incubated at 4oC for 1 hour, and the resulting complex was resolved by electrophoresis on 6% nonreducing, nondenaturing polyacrylamide gels. Lane 1, Coomassie blue–stained Fn-UG heteromer; lane 2, its autoradiogram. (D) Immunoprecipitation of plasma (18) from UG+/+ and UG−/− mice with anti-Fn and immunoblotting with anti-Fn and anti-UG. Fn (upper panel); UG (lower panel). Std, standards for UG and Fn. (E) Affinity cross-linking of 125I-Fn with unlabeled Fn in the absence (lane 2) and presence of various amounts of UG (lanes 3 through 5). The intensity of the very high molecular weight, radioactive Fn band (lane 2) formed in the absence of UG is reduced in a dose-dependent manner. Lane 1, 125I-Fn with unlabeled Fn in the absence of UG and DSS. Open arrowhead, multimeric Fn; lower thin arrow, 220-kD Fn. (F) Affinity cross-linking of125I–collagen I (125I-Col I) with unlabeled Fn in the absence (lane 3) and presence (lane 4) of UG. Lane 1, Coomassie blue–stained collagen I; α1, α1 chain of collagen I and α2, α2 chain of collagen I. Lane 2, 125I-collagen I and unlabeled Fn in the absence of UG and DSS.

On the basis of current concepts, critical initial steps in Fn matrix assembly and fibrillogenesis, at least on the cell surface, are thought to involve integrin activation and Fn self-aggregation (22). Because UG is a potent inhibitor of sPLA2 (3), a key enzyme in the inflammatory pathway, the lack of UG in UG−/− mice may contribute to the development of glomerulonephritis, an inflammatory renal disease (23). Moreover, lysophosphatidic acid (LPA), a by-product of PLA2 hydrolysis of phosphatidic acid, causes integrin activation, Fn matrix assembly, and fibrillogenesis (22). Thus, we measured the specific activity (micromoles per minute per milligram of protein) (24) of serum PLA2 of UG−/− mice [36 ± 3.3 (SEM)], which was significantly higher (P < 0.05, Student’st test) than that of UG+/+ mice [18 ± 2.8 (SEM)]. These results raised the possibility that higher PLA2 activity may lead to increased LPA production and consequently promote integrin activation in UG−/− mice.

To further examine how UG may prevent Fn self-assembly, we determined whether it disrupts Fn-Fn interaction in vitro. We incubated equimolar concentrations of UG and Fn, immunoprecipitated with anti-Fn, resolved the immunoprecipitates by SDS-PAGE under reducing conditions, and protein immunoblotted (18) with either anti-Fn or anti-UG. Fn coimmunoprecipitated with UG (Fig. 3B). To confirm these results, we also incubated 125I-labeled UG with Fn and resolved the complexes by electrophoresis, using a 6% polyacrylamide gel under nondenaturing and nonreducing conditions (Fig. 3C). Detection of an Fn-UG heteromer suggested that soluble Fn may interact with UG. To delineate whether Fn-UG heteromerization takes place in vivo, we immunoprecipitated the plasma of UG+/+ and UG−/− mice with anti-Fn, which does not cross-react with UG (Fig. 3D). This antibody coprecipitated both Fn and UG from the plasma of UG+/+, but not from UG−/− mice, suggesting that Fn-UG heteromers are present in the plasma of UG+/+ mice.

To determine the specificity and affinity of UG binding to Fn, we incubated 125I-labeled Fn with unlabeled Fn in the presence and absence of UG and affinity cross-linked with disuccinimidyl suberate (DSS) (25). In the absence of UG,125I-Fn formed a high molecular weight, radioactive complex with unlabeled Fn, but in the presence of UG the formation of Fn-Fn aggregates was inhibited in a concentration-dependent manner (Fig. 3E). To determine whether there is any difference between the binding affinities of Fn for UG and that of Fn for itself, we did binding experiments in which 125I-Fn was incubated with unlabeled Fn (immobilized on multiwell plates) together with various concentrations of UG. In separate experiments, we also did binding studies of 125I-Fn with unlabeled immobilized Fn using various concentrations of unlabeled soluble Fn. The Scatchard analyses of the data from both of these binding experiments yielded straight lines with dissociation constants (K d) of 13 nM for UG binding to Fn and 176 nM for Fn binding to itself. These results suggest that, because of a relatively higher binding affinity of UG for Fn, UG may effectively counteract Fn self-aggregation. We also did affinity cross-linking experiments in which radio iodinated (125I)–collagen I was incubated with unlabeled Fn in the absence or presence of UG, as described above for Fn. The results indicate that UG counteracts the formation of high molecular weight125I-collagen–Fn aggregates (Fig. 3F).

To test whether UG protects the renal glomeruli from Fn accumulation, we administered soluble human Fn (hFn) alone or hFn mixed with equimolar concentrations of UG intravenously to UG+/+ and to apparently healthy UG−/− littermates (26). The rationale for injecting hFn was to be able to discriminate between endogenous murine Fn and administered hFn. The methods of intravenous administration and immunohistochemical detection of hFn in various tissues have been described (27). After 24 hours, histological sections of the kidneys were examined by immunofluorescence with a monoclonal antibody to hFn. Human Fn immunofluorescence in the glomeruli of wild-type mice injected with either a mixture of hFn and UG (1:1 molar ratio) or with hFn alone was similar (Fig. 4, A and B). However, although the UG−/− mice injected with a mixture of hFn and UG had little hFn-specific immunofluorescence in the glomeruli (Fig. 4C), those receiving hFn alone had higher intensity immunofluorescence (Fig.4D). Administration of a mixture of hFn and bovine serum albumin (BSA) had no protective effect. To determine whether this protection could be overcome by injecting larger quantities of hFn in UG+/+mice, we injected 1 mg of hFn per animal daily for three consecutive days (27). Although intravenous administration of hFn to UG+/+ mice at lower doses (500 μg per animal) was not effective in causing any appreciable glomerular deposition (Fig. 4A), the administration of higher doses (3 mg per animal) led to a significant accumulation (28). Thus, UG may prevent glomerular Fn deposition, and UG+/+ as opposed to UG−/− mice may have a higher threshold for the accumulation of soluble Fn.

Figure 4

Inhibition by UG of glomerular Fn deposition, in vitro matrix assembly, and fibrillogenesis. Kidney sections of (A) a wild-type mouse that received a mixture of equimolar concentrations of Fn and UG intravenously; (B) a UG+/+ mouse that received the same dose of Fn as in (A) but without UG; (C) an apparently healthy UG−/− mouse that received a mixture of Fn and UG; and (D) a UG−/− mouse that received Fn alone [same dose as in (C)], but without UG. (E) Fn fibrillogenesis by cultured cells (29) grown in medium supplemented with soluble hFn alone. (F) A cell culture identical to the one in (E) that was fed with medium containing a mixture of equimolar concentrations of soluble hFn and UG. Magnification (A to F) ∼×145; g, glomerulus.

To determine whether UG prevents Fn fibrillogenesis and matrix assembly in vitro, we cultured mouse embryonic fibroblasts in medium containing either soluble hFn alone or a mixture of equimolar concentrations of hFn and UG (29). The level of fibrillogenesis in cell cultures treated with hFn alone was much higher (Fig. 4E) compared with those that received a mixture of hFn and UG (Fig. 4F).

Renal diseases are a major cause of human morbidity and mortality, and glomerular diseases are one of the major causes of renal failure. A familial glomerulopathy, characterized by heavy deposition of predominantly fibronectin, has been described (30), although the mechanisms of pathogenesis of this disease remain unclear. The availability of the UG-knockout mouse model may allow us to understand the pathogenic mechanisms of ECM protein–deposit human glomerulopathies in general and the predominantly Fn-deposit hereditary glomerulopathy (30) in particular.

  • * To whom correspondence should be addressed: E-mail: mukherja{at}


View Abstract

Stay Connected to Science

Navigate This Article