TECHNICAL FIELD OF INVENTION

This invention relates to processes for producing uromodulin, a glycoprotein having a molecular weight of 85 kilo daltons. More particu¬ larly, this invention relates to the isolation and characterization of uromodulin from crude urine and the purification of uromodulin to homogeneity as assessed by reduced or unreduced sodium dodecyl sulfate-polyacrylamide gel (SDS-PAGE). Uromodulin itself, or the carbohydrate portion thereof, is useful as an immuncteuppressive agent and an anti- inflammatory agent.

BACKGROUND ART

During pregnancy, the histoincompatible placental unit is protected from maternal immuno- surveillance for reasons which to date remain un¬ explained (G. Chaout, "The Riddle of the Foetal Allograft", Ann. Immunol. (Inst. Pasteur, 135D, p. 301 (1984)). Various compounds associated with pregnancy have been proposed as the immunosuppres¬ sive agent responsible for this protection. These compounds include human chorionic gonadotropin, alpha fetoprotein, human placental lactogen, preg¬ nancy associated plasma protein A and SP-1. Upon further analysis, however, none of these compounds

has been identified as a useful immunoregulatory agent (A.V. Muchmore and R.M. Blaese, "Immunoregu¬ latory Properties of Fractions From Human Pregnancy Urine: Evidence that Human Chorionic Gonadotropin Is Not Responsible", J. Immunol. , 118, p. 881 (1977); S.F. Contractor and H. Davies, "Effect of Human Chorionic Somatomammatropin and Human Chorionic Gonadotropin on Phytohemagglutinin Induced Lymphocyte Transformation", Nature (New Biol.), 243, p. 284 (1973); S. Yachin, "The Immunosuppressive Properties of Alpha Fetoprotein: A Brief Overview", Ann. NY Acad Sci. , 417, p. 105 (1983); W.H. Stimson, "Are Pregnancy - Associated Serum Proteins Responsible for the Inhibition of Lymphocyte Transformation by Pregnancy Serum?", Clin. Exp. Immunology, 40, p. 157-60 (1980); C. Cerni et al.," Immunosuppression By Human Placental Lactogen and the Pregnancy Specific Beta-1 Glycoprotein (SP-1)", Arch. Gynakol, 223, p. 1 (1977); J.A. Mclntyre et al., "Immunolo- gical Studies of the Human Placenta: Functional and Morphologic Analysis of PAPP-A" _> Immunology, 44, p. 577 (1981)).

To date therefore, altύw gh the importance in other immunosuppressr e applications and therapy of the compounds responsible.for the immunologic protection of the placenta rom maternal immuno- surveillance has been-recognized, no such compounds have been isolated, purified or characterized.

DISCLOSURE OF THE INVENTION This invention solves the problems referred to above by providing processes for producing and purifying to homogeneity uromodulin, a glycoprotein which has immunosuppressive and anti-inflammatory activity. Uromodulin is an 85 kilo dalton glycopro- tein having about a 30% carbohydrate content. The carbohydrate portion of uromodulin is also charac-

terized by immunosuppressive and anti-inflammatory activity.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1, Panel A depicts in tabular form the purification of uromodulin. Figure 1, Panel B depicts the elution pattern, bioactivity and immuno- reactive profiles of uromodulin purified according to one embodiment to this invention.

Figure 2 depicts several SDS-PAGE analyses of fractions of uromodulin obtained at various frac- tionation steps of the process of this invention. Lane A depicts standard molecular weight markers. Lane B depicts uromodulin after concentration of peak 1 from Figure 1, Panel B. Lane C depicts final purification of uromodulin after elution from an isoelectric focusing gel.

Figure 3 ^' depicts a Wescern Blot analysis of Con-A fractionated pregnancy urine. Lane A depicts molecular size standards stained with Amido Black. Lanes B and C depict the reactivit of rabbit serum used as a probe against a crude fraction of preg¬ nancy urine. Lane D depicts the protein silver stain of 1 μl of the crude pregnancy urine fraction.

Figure 4, Panel I depicts the dose-response function of uromodulin from a single donor added at the initiation of culture to human peripheral blood mononuclear cells stimulated with tetanus toxoid and subsequently harvested. Figure 4, -Panel III depicts the effect of uromodulin on the generation of spon- taneous monocyte-mediated cytotoxicity..

Figure 5 displays in graphic form the specificity of uromodulin for interleukϋi-l.

Figure 6 depicts the binding activity of antisera specific for uromodulin to interleukin-1. Figure 7, Panel A depicts the kinetics of binding of uromodulin to TNF. Figure 7, Panel B depicts the inhibition by Con-A of IL-1 binding to

uromodulin. Figure 7, Panel C depicts the failure of Con-A to compete with TNF for binding to uromo¬ dulin. Figure 7, Panel D depicts the inhibition by wheat germ agglutinin of TNF binding to uromodulin. Figure .8 depicts an SDS-PAGE analysis of reduced and unreduced uromodulin.

Figure 9 depicts the ability of EA4, an anti-uromodulin antibody; to recognize intact vs. modified uromodulin. Figure 10 depicts the biological activity of intact vs. modified uromodulin.

Figure 11 depicts the biological activity of pronase digested uromodulin.

Figure 12 depicts the biological activity of N-glycanase digested uromodulin.

Figure 13 depicts the inhibition by N-glycanase digestion of the ability of uromodulin to bind to interleukin-1.

DETAILED DESCRIPTION OF THE INVENTION This invention relates to processes for isolating and purifying uromodulin from crude human urine. According to this invention, uromodulin may be purified to homogeneity as assessed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE).

Generally, one embodiment of this process comprises the steps of contacting crude urine with an affinity column containing a lectin which recog¬ nizes mannose, eluting uromodulin from the column and dialyzing small molecular weight contaminants, i.e. contaminants having a molecular weight below about 10 kilo daltons, from the uromodulin. According to another embodiment, the process of this invention further comprises an isoelectric focusing purification step.

Uromodulin may be isolated from any human urine, as well as from pregnancy urine. In addition,

it may be isolated from non-human pregnancy or non- pregnancy urine.

This invention also relates to the uromod¬ ulin produced according to the above-described pro- cesses. Uromodulin is an 85 kilo dalton glycopro¬ tein having about a 30% carbohydrate content. It is a single peptide having intra-chain disulfide lin¬ kages. We believe that the active portion of uromodulin comprises the N-linked sugars. Uromodulin is characterized by immunological and immuno-suppressant activities. For example, uromodulin suppresses antigen-specific T cell pro¬ liferation in vitro at concentrations as low as 100 pM. It also is a potent inhibitor of spontaneous monocyte cytotoxicity, acting at concentrations as low as 10 ^" M. In addition, uromodulin specifically binds and inhibits interleukin-1 (IL-1) and tumor necrosis factor (TNF). IL-1 and TNF are known to be mediators of immunoresponses such as inflammation. Accordingly, uromodulin's activity of binding and inhibiting IL-1 and TNF indicates its utility as an immunosuppressant or anti-inflammatory agent. Without being bound by theory, we believe that these activities are attributable to the N-linked gly- ^• cosylation of the uromodulin molecule. Uromodulin is, therefore, advantageously useful in immunothera¬ peutic methods and compositions for treating mammals, including humans. Uromodulin itself, as well as the carbohydrate portions derived therefrom, are parti- cularly useful as immunosuppressive or anti-inflam¬ matory agents.

Uromodulin or uromodulin derivatives prepared according to the processes of this invention may be employed in a conventional manner in immunotherapeutic and anti-inflammatory methods and compositions.

Such methods of treatment and their dosage levels and requirements are well-recognized in the art and

may be chosen by those of skill in the art from available methods and techniques. For example, uromodulin or its carbohydrate derivatives may be combined with a pharmaceutically acceptable adjuvant for administration to a patient in an amount effec¬ tive to provide immunosuppressant or anti-inflammatory effects and accordingly to lessen the severity of the target disease or symptoms. The dosage and treatment regimens will depend upon factors such as the patient's health status, the severity and course of symptoms and the judgment of the treating physician. In order that the invention herein described may be more fully understood, the following examples are set forth. It should be understood that these examples are for illustrative purposes only and are not be construed as limiting this invention in any manner.

EXAMPLES A. Preparation Of Uromodulin According to one embodiment of this inven¬ tion, we purii3ti.ed uromodulin from crude human ~ ^~ s.eg- nancy urine as follows. We collected and stored at -20°C unfractionated crude human urine from individual donors at 20-40 weeks of gestation. We passed s±y . . liters of urine over a lectin affinity column, preferably a concanavalin-A Sepharose column, (Pharmacia, Uppsala, Sweden) with a total bed volume of 200 ml. Although the use of Con A is preferred as described above _. other lectins which recognize annose, such as lentil lectin, may also be employed. We then washed the column with four bed volumes of phosphate-buffered saline (PBS). This resulted in the removal of 90% of the starting protein, while retaining 80-100% of the uromodulin. See Figure 1, Panel A.

We then eluted the uromodulin from the column with two bed volumes of 250 πiM methyl manno- pyranoside in PBS and dialyzed the small molecular weight contaminants from the uromodulin for 48 hours at 4°C against three changes of fifty volumes of distilled water. Although we used methyl mannopyra- noside to elude the column, any mannose, pyranoside or saccharide which competes with the lectin in the affinity column may also be used. We then lyophilized and resuspended the dialysate in 15 ml PBS. As an alternative to these lyophilization and resuspension steps, other conventional procedures may be used to concentrate the uromodulin to an appropriate volume to be placed on a molecular sizing column. We separated the resuspended material on a molecular sizing column with a large exclusion pore size, preferably a 2.5 x 90 cm Fractogel 55S (Merck, Rahway, New Jersey U.S.A.; exclusion limit 500 to 750 kilo daltons) by eluting the column with PBS. This resulted in resolution of uromodulin from the small molecular weight contaminants. At this stage, the only consistent contaminant on SDS-PAGE was a 30 kD material that co-purified with the uromodulin. The uromodulin obtained after this step was substan- tially homogeneous on 12.5% SDS-]?AGE. Uromodulin migrated as the first peak, resulting in a 95% pure ^*■ yield. See Figure 1, Panel B and Figure 2, Lane 2.

According to an alternate embodiment of this invention, the uromodulin obtained by the above-described process may be further purified by isoelectric focusing to resolve the 30 kD contaminant. For example, we pooled and dialyzed the first peak resolved above against distilled water, then resus¬ pended the dialysate in 0.01 M phosphate buffer at pH 7.0. We then loaded the resuspended dialysate onto preabsorbent wicks and focused the loaded wicks on precast isoelectric focusing (IEF) gels, pH 4.5 to 9.0 (LKB, Broma, Sweden). We collected the area

directly under the wick and eluted it with distilled water. We then concentrated the eluate in filters, preferably Centricon filters (Amicon, Danvers, Massachusetts, U.S.A.) with a cut-off of 30 kilo daltons. A molecular sieving high-performance liquid chromatography (ΞPLC) column, preferably TCK 3000 (Bio-Rad, Richmond, California, U.S.A.), was used when necessary, to remove any remaining focusing buffers. Since uromodulin fails to aggregate in isoelectric gels, while contaminants migrate freely, the above-described process yielded a broad single 85 kilo dalton band on unreduced 12.5% SDS-PAGE. See Figure 2, Lane C. Under reducing conditions with dithiothreitol, SDS-PAGE of uromodulin yielded a single band at 95 kilo daltons, suggesting that uromodulin is._a single peptide with intra-chain disulfide linkages.

B. Assays Of Purified Uromodulin At each fractionation step described above, we assayed the biological activity of the purified material by measuring the inhibition of tetanus toxoid induced ". cell proliferation of normal human peripheral DIOO I Tnonuclear cells using a modifica- tion of the assay described in A. V. Muchmore J.M. Decker ^" and R.M.-Blaese, " Evidence That Specific 01igosaccharides Block Early Events Necessary for the Expression of Antigen-Specific Proliferation by Human Lymphocytes", J. Immunol., 125, pp. 1306-11 (1980).

In this assay, we centrifuged heparinized human blood at 200XG for 5 minutes in a RT 6000 (Sorvall) and collected the buffy coat layer of cells. We allowed the cells to settle in PBS for 30 minutes to remove clusters of platelets and then washed the cells two more times in PBS. We resus¬ pended the cells in RPMI 1640, with added penicillin,

strepto ycin and glutamine. Subsequently, we incu¬ bated 2 x 10 viable cells in 10% autologous plasma with optimized concentrations of tetanus toxoid (Massachusetts Department Of Public Health) and added various concentrations of uromodulin or derivatives of uromodulin to a final volume of 0.2 ml.

The cultures were then pulsed with 0.5 microcurie of

3 H thym dine for six hours on the sixth day and

3 . . . counted for H thymidine incorporation. Using this assay, we determined that puri¬ fied uromodulin blocks in vitro antigen specific T cell proliferation to recall tetanus toxoid antigens at concentrations as low as 100 pM.

We also used our purified uromodulin as an immunogen to raise a heteroantibodies in rabbits. Subsequently, we used this antiserum as a probe in Western blot analysis of crude pregnancy urine.

The crude fractions of pregnancy urine were separated on 12.5% SDS-PAGE and transferred to nitrocellulose (Schleicher and Schuell protocols, S and S, Keene, New Hampshire, U.S.A.). Bound anti¬ body on the nitrocellulose was detected using a biotinylated goat antiserum to rabbit IgG, followed by an avidin-biotin-horseradish-peroxidase complex reagent (Vector, B rliι_.g_ιa_ι_e, California, U.S.A.). This antibody fails to bind significantly to any protein found in normal human serum having a mole¬ cular weight of less than 150 kilo daltons. It did bind weakly in an area consistent with human IgG, which could not be removed by absorption of our antisera with immobilized human serum. This suggested that human antibody was recognizing the rabbit IgG. See Figure 3.

Figure 3, Lanes B and C depict the reac- tivity of the rabbit antiserum. Lane A represents molecular size standards stained with Amido black. Land D represents protein silver stain of 1 μl of the crude urine fraction. The analysis exhibited a

single major band at 85 kilo daltons, with two minor bands seen only on overloaded gels.

As further evidence that the rabbit anti¬ serum was recognizing an immunosuppressive molecule, we conjugated the antiserum to cyanogen bromide-acti¬ vated Sepharose (Pharmacia, Uppsala, Sweden) and used it as a solid-phase immunoabsorbent. We adsorbed crude human pregnancy urine to this column, then washed it extensively and eluted it with 0.1 M glycine buffer (pH = 2.8). The eluate had a molecular weight of 85 kilo daltons on SDS-PAGE and exhibited immuno¬ suppressive activity in vitro to tetanus toxoid (infra, page 10-11).

We also conjugated a purified immunoglob- ulin G (IgG) fraction of this antiserum to alkaline phosphatase and developed a sensitive direct guati- tative enzyme-linked immunosorbent assay (ELISA) for uromodulin. The assay consisted of a sandwich assay such as that described by A. Voller et al. in Manual Of Clinical Immunology, pp. 359-71 (1980). See Fig¬ ure 1, Panel A.

We then examined the in vitro bioactivity of the purified uromodulin obtained by the process set forth above. We measured the dose-response func- tion of uromodulin added at the initiation of culture to human peripheral blood mononuclear cells stimulated with tetanus toxoid and harvested as described in E. S. Kleinerman et al., "Defective Monocyte Killing in Patients With Malignancies and Restoration of Function During Chemotherapy", Lancet, 2(8204), pp. 1102-05 (1980). The results shown are expressed as mean CPM ± 1 SE of tritiated thymidine uptake added on day 6 of cultures run in triplicate. We obtained similiar results from 12 different measure- ents representing five separate batches of uromodulin from different donors. The mean CPM of 12 control cultures was 37,736 ± 7,029; cultures with 3.5 10 -9 M

uromodulin exhibited 14,448 ± 4,170. Inhibition ranged from 42% to 91% [t(ll) = 5.89, P = 0.0002]. Using our assay, which ^• measures the inhibition of antigen-specific T cell proliferation, we found that uromodulin exhibits a broad dose-response curve with acti .vi.ty demonstrable from 10-9 to 10-11 M. See

Figure 4, Panel I. Additionally, uromodulin exhibited no apparent antigen specificity, as it inhibited the proliferation of the following antigens: tetanus toxoid, streptokinase-streptodornase and Candida. We found that addition of uromodulin only 12 hours later resulted in failure of inhibition. Furthermore, uromodulin had no effect on cell viability even after seven days of culture in vitro. Thus, uromodulin was immunosuppressive only if added at the initiation of a 6-day culture. See Figure 4, Panel III.

We also added uromodulin to in vitro assays of B-cell and monocyte function. First, we assayed B-cell function with a sensitive reverse-hemolytic plaque assay which measures the total number of anti¬ body-secreting cells after polyclonal stimulation with pokeweed mitogen. We enumerated the plaque- forming cells on day 7 according to Ξ. Kirchner et al., "Polyclonal Immunoglobulin Secretion by Human B Lymphocytes Exposed to Epstein - Barr Virus In Vitro", J. Immunol., 122, p. 1310-13 (1979), and concluded that uromodulin did not affect this assay of B-cell function.

We also used uromodulin in an assay which measures the in vitro development of spontaneous monocyte-mediated cytotoxicity in humans. This assay is regulated by suppressor cells and is sensitive to a number of exogenous agents capable of modulating monocyte function (E. S. Kleinerman et al., "Defec- tive Monocyte Killing in Patients With Malignancies and Restoration of Function During Chemotherapy", Lancet, 2(8204), pp. 1102-05 (1980); E. S. Kleinerman, L.A. Zwelling, R. Schwartz and A.V. Muchmore, "Effect

of L-Phenylalanine Mustard, Adriamycin, Actinomycin D, and 4'-(9-acridinyl-amino)methanesulfon-m-anisidide on Naturally Occurring Human Spontaneous Monocyte- Mediated Cytotoxicity", Cancer Res. , 42, pp. 1692-95 (1982)). We added purified uromodulin obtained by the process set forth above in varying concentrations at the initiation of culture. After 6 days, we re¬ suspended the cells and counted them for viability using trypan blue exclusion (viability ranged from 85% to 95% with no difference noted between treated and untreated cultures). We added two hundred thou¬ sand viable cells to microtiter dishes and assayed cytotoxi .ci.ty m. tri.pli.cate usi.ng 51Cr labeled chicken red blood cell targets. Our results are expressed as mean % 51Cr release ± 1 SE of cultures run at a

1:1 target to effector cell ratio. ~*See Figure.4,

Panel III. We found uromodulin to _be a potent inhibitor of spontaneous monocyte toxicity when added at the beginning of culture, acting at con- centrations as low as 10~ M..

These data suggest that the primary site of action of uromodulin is likely to be ^' at the mono¬ cyte or T cell level and that uromodulin dees not acting by nonspecifically blocki y cellular division or by blocking IL-2 function, as theε._ two mechanisms of action would also inhibit proliferation at later stages as well as the initiation..of culture.

These results demonstrate that uromodulin blocks in vitro generation of spontaneous monocyte mediated cytotoxicity. Without being bound by theory, we believe that uromodulin blocks early events re¬ quired for a normal in vitro antigen specific T cell proliferation response. Because uromodulin acts early in the sequence of events required for a T cell proliferation response and is ineffective later in culture, we believe that uromodulin does not block proliferating cells.

ELISA With Uromodulin Against IL-1 Coated Plates

IL-1 is a 15 kilo dalton macrophage-derived protein which effects a variety of immunostimulatory and inflammatory, responses and which is believed to mediate the first step in such responses. IL-1 induces fever, causes the release of collagenase by synovial cells, induces prostaglandin synthesis by fibroblasts and acts as a co- itagen. In addition, IL-1 has been also found to mediate acute phase reactant synthesis by hepatocytes and IL-2 synthesis by thymocytes. All of these activities suggest that IL-1 may play a central role in the regulation of diverse inflammatory host responses/ and in the regulation of. cellular differentiation and proli¬ feration. We believe that uromodulin is the first isolated and characterized substrate for IL-1. As shown in the following assays, uromodulin specifi¬ cally inhibits IL-1 by binding to it. This binding activity indicates that uromodulin is useful as an immunosuppressive or anti-inflammatory agent.

Using the purified uromodulin obtained by the process set forth above and a SP-2 fusion partner, we produced a series of monoclonal antibodies raised against uromodulin using the method described in R.H. Kennett, Monoclonal Antibodies, pp. 365-80 (R.H. Kennett, T.J. McKearn & B. Bechtol ed. 198Θ . We then coated 96 well ELISA plates (Immulon, _ Dynatech, Virginia, U.S.A.) overnight with 1 μg/ml recombinant murine IL-1 having a specific activity of 5 x 10 U/mg of protein (generous gift of Dr. Peter Lomedico, Hoffman-LaRoche, Nutley, New Jersey, U.S.A.) in pH 9.6 carbonate buffer. This recombinant IL-1 may also be prepared according to the method described in P. Lomedico et al.,

"Cloning And Expression Of Murine IL-1 In E.Coli", Nature, 312, p. 418 (1984). We next washed these plates three times with phosphate-buffered saline

with added Tween 20. We added uromodulin at various concentrations and allowed it to incubate for 2 hours at room temperature. We added 2 μl/ml of monoclonal anti-uromodulin EA4, a monoclonal anti- body developed using uromodulin as an .L-mmunogen, to some wells to demonstrate competition. We then washed these plates with phosphate-buffered Tween 20 and allowed a 1:400 dilution of rabbit heteroantisera directed against uromodulin to incubate for an addi- tional two hours. We then washed the plates 3 times. The presence of bound uromodulin was detected using an unmodified uromodulin specific antiserum and bound rabbit IgG was detected using an alkaline phosphatase modified goat anti-rabbit antiserum (Sigma, St. Louis, Missouri, U.S.A.), as described in A. Voller,

D. Bidwell and A. Bartlett", "Enzyme Linked Immuno- sorbent Assay", Manual of Clinical Immunology, p. 359 (N.R. Rose & H. Friedman ed. 1980). Changes in OD were measured at 405 nm using an appropriate alkaline phosphatase substrate. As shown in Figure 5, uromo¬ dulin binds with high affinity to IL-1.

We next coated Immulon plates (Dynatech, Virginia) with 2 μg/ml of recombinant IL-1 in car¬ bonate buffer (pH = 9.6) for 18 hours. We then washed αes plates three times in phosphate-buffered saline with 1% Tween-20 (PBS Tween). We then added various mixtures of uromodulin with or without com¬ peting substances such as carbohydrates, uromodulin derivatives or other unrelated glycoproteins, and incubated the plates for 2 hours at room temperature. We then-washed the plates three times and added a 1:400 dilution of monospecific rabbit anti-uromodulin for one hour. We washed the plates another three times. We detected the presence of bound rabbit antisera by using a solid phase purified alkaline phosphatase modified goat anti<-rabbit i munoglobulin. We found that uromodulin specifically binds to IL-1 with a K _d of 3 x lθ ^"10 M.

We next coated 96 well Immunlon microtiter plates (Dynatech, Alexandria, Virginia, U.S.A.) with 2 μg/ml recombinant murine IL-1 in carbonate buffer (pH = 9.6) and incubated them overnight (eighteen hours) at 4°C. We then washed the plates 3 times with PBS Tween.

We added various concentrations of uromo¬ dulin, as prepared in above, to some of the wells and incubated the plates for 2 hours at room tempera- ture. We then washed the plates with Tween-20 and added one of three antisera specific for uromodulin, i.e., 5155 (rabbit IgG, 1:800 dilution), CG7 (mono¬ clonal IgM, 10 μg/ml) or EA4 (monoclonal- IgG, 10 μg/ml) ^" to the wells and incubated the plates for another 1 hour at room temperature. Subsequently, we washed the plates 3 times with Tween-20 and devel¬ oped them with an alkaline phosphatase modified goat anti-rabbit or goat antimouse antiserum (Sigma, St. Louis, Missouri, U.S.A.) using Sigma 104 sub- strate. We then analyzed the plates spectrophoto- metrically at 405 nm using a Dynatech (Virginia) microplate reader. The results are depicted in Figure 6A, in which background absorbance (.060 to .15 OD units) consisting of all reagents has been subtracted out. As shown in that figure, 5155 detected high affinity binding of uromodulin to IL-1 coated plates and could measure levels as low as 10~ M. CG7 was much less efficient than the rabbit heteroantisera but could also detect bound uromodulin. EA4 failed to detect bound uromodulin under all conditions, suggesting that EA4 and IL-1 might bind to identical or nearby epitopes. No significant binding was observed in the absence of IL-1. A similar assay to that described above was also run, with the exception that 5 μg of 5155, . EA4 or anti-HLA (IgGl monoclonal, Cappel, West Chester, Pennsylvania, U.S.A.) were added with

uromodulin and allowed to incubate for 2 hours. The plates were washed and probed for bound uromodulin with 5155 and further developed for bound rabbit immunoglobulin as above. The results, depicted in Figure 6B, show that EA4 but not CG7 or anti-HLA (an unrelated monoclonal IgG) competes with the binding of uromodulin to IL-1 coated plates. Further spe¬ cificity controls run under similar conditions demonstrated that purified uromodulin failed to bind to uncoated ELISA plates or ELISA plates coated with other growth factors, such as IL-2, transferrin or insulin, or unrelated proteins, bovine serum albumin or fetal calf serum. Additionally, in analogous assays, recombinant murine IL-1 specifically bound to uromodulin-coated plates using IL-1 specific rabbit antiserum.

Inhibitory Effect Of Uromodulin In An Assay Specific For IL-1 Induced Cell Proliferation

The following demonstrates that uromodulin inhibits the activity of IL-1 in a standard mouse thymocyte comitagenic assay for IL-1 at concentra¬ tion of 10 ^"9 to 10 ^"10 (3.7 ng = 4.3 x lθ ^"10 M). In this assay, we incubated 10 C3Ξ/HEJ thymocytes from mice less than 6 weeks old in RPMI 1640 with 10% fetal bovine serum, with or without 1 μg of PHA (Burroughs Wellcome, North Carolina, U.S.A.). Various concentrations of uromodulin or human chorionic gonadotropin (run as a control) were added at the initiation of culture. We then added ultrapure human IL-1 (5U) (Genzyme, Boston,

Massachusetts, U.S.A.) to all cultures except where noted. After 3 days, the cultures were pulsed with

3 0.5 microcuries of H thymidine for 6 hours and counted. The results, shown below, represent the mean of triplicate determinations:

Nanograms added Uromodulin (CPM) HCG (CPM)

100 2021 12323

33 4234 15530

11 4201 14531

3.7 5694 14612

CPM - no stimulant - 252

CPM - PHA alone - 2467

CPM - PHA + 5U IL-1 - 13461

As shown in this table, uromodulin inhibits the standard mouse thymocyte co-mitogenic assay for IL-1 activity.

ELISA With Uromodulin Against TNF-Coated Plates

Tumor necrosis factor (TNF) is produced by macrophages and mononuclear phagocytes and is selectively cytotoxic or cytostatic in vitro for a broad range of animal and human tumor cells (K. Haranaka and N. Satomi, "Note: Cytotoxic Activity of Tumor Necrosis Factor (TNF) on Human Cancer Cells in Vitro," Japan J. Exp. Med., 51, pp. 191-94 (1981)). As shown in the following assays, uromodulin specif¬ ically binds to TNF at a site different from that at which uromodulin binds to IL-1. We believe that uromodulin is the first compound to be isolated and characterized to which both IL-1 and TNF bind. This binding activity indicates that uromodulin is useful as an immunosuppressant and anti-inflammatory agent.

We followed the procedures set forth supra for the ELISA against IL-1 coated plates using recombinant TNF (Biogen, Cambridge, Massachusetts, U.S.A. - specific activity 1 million U/mg) alone or in the presence of Con-A, or wheat germ agglutinin in place of the IL-1. The results of these assays are depicted in Figure 7. Panel A depicts the

kinetics of binding of uromodulin to TNF. Panel B depicts in graphic form the inhibition by Con-A, a lectin, of IL-1 binding to uromodulin. Panel C depicts in graphic form the failure of Con-A to compete with TNF for binding to uromodulin. Panel D depicts in graphic form the inhibition by wheat germ agglutinin (WGA), another lectin, of TNF binding to uromodulin. WGA blocks the binding of TNF to uromodulin by covering the carbohydrate portion of the molecule. These data demonstrate that both IL-1 and TNF specifically bind to uromodulin and that IL-1 and TNF bind to uromodulin at different sites on the carbohydrate portion of the molecule.

C. The Role Of The Carbohydrate Portion Of Uromodulin In The

Biological Activity Of Uromodulin

The following examples demonstrate the important role that the carbohydrate portion of uro¬ modulin plays in the activity of uromodulin. In these examples, the purified uromodulin starting material was prepared from the urine of women in the second and.'lhird trimester of pregnancy as described supra,, but it was not subjected to iso¬ electric focusing. . Instead, approximately 12 liters of first void morning ^■ T ^I ~ ~- was passed over 600 cc of immobilized Con A Sepharose (Pharmacia, Uppsala, Sweden) in a large scintered glass funnel. We washed away any unbound material_wi _' th four bed volumes of PBS and eluted the bound material with two bed volumes of 250 mM alpha methyl mannose in PBS. We then dialyzed the eluate against four changes of distilled water and lyophilized the material. Subsequently, we resuspended the material in PBS and separated it on a 1.2 x 120 cm Fractogel 55S column (Merck, Rahway, New Jersey, U.S.A.). We then collected, dialyzed and lyophilized the void volume.

This procedure yielded uromodulin which was substantially homogeneous on 12.5% SDS-PAGE, carried out as described in U.K. Laemmli, "Cleavage of Structural Proteins During the Assembly of the Head of the Bacteriophage T4", Nature, 227, p. 680 (1970) in which gels were stained with a sensitive silver stain according to CR. Merril, R.C. Switzer and A. Van Keuren, "Trace Polypeptides in Cellular Extracts and Human Body Fluids Detected by Two Dimensional Electrophoresis and a Highly Sensitive Silver Stain", Proc. Natl. Acad. Sci. USA, 76, pp. 4335 (1979). A 12.5% SDS-PAGE analysis of reduced and unreduced uromodulin is shown in Figure 8. The apparent increase in molecular size of the reduced uromodulin (Lane A) suggests that uromodulin exists as a single polypeptide with intra-chain disulfide bonds.

Carbohydrate Content Of Uromodulin

In the following examples, we chemically modified our purified uromodulin to determine the role of its carbohydrate portion in the biological activity of the molecule.

First, we assessed the presence of sialic acid in the purified uromodulin prepared above by methods using purified neuraminidase and H ₂ S0 ₄ to hydrolyze that sugar from the intact glycoprotein. The total sialic acid released was measured using a sensitive method based on thiobarbituric acid, as described in K.S. Hammon and Papermaster "Flurometric Assay of Sialic Acid in the Picomole Range, A Modification of the Thiobarbituric Acid Assay", Anal. Bio. Chem. , 74, p. 292 (1976). Both methods revealed the presence of large quantities of sialic acid, with the H-S0 ₄ method yielding estimates of 7-10% w:w sialic. acid. We then measured the total amount of carbohydrate following acid hydrolysis and measurement of sugars using the anthrone method

(R.G. Spiro, "Analysis of Sugars Found in Glyco- proteins", Methods in Enzymology, Vol. Ill, pp. 1-5 (E.F. Neufeld & V. Ginsburg ed. 1966)). By this method, we determined that uromodulin is approxi- mately 30% w/w sugar and rich in sialic acid. We then assessed the role of the carbohydrate portion of uromodulin in the biological activity of the molecule. To do this, we prepared various carbo¬ hydrate derivates of uromodulin. We obtained a succinylated derivative of uromodulin by treating 1 ml of purified uromodulin (1 mg/ml) in 0.01 M sodium bicarbonate buffer of pH 8.2 with a 50 molar excess of an acetone solution of succinic anhydride (100 mg/ml). As the addition of succinic anhydride caused the pH of the reaction mixture to drop, we added sodium hydroxide to main¬ tain the pH of the solution at 8.2. We allowed the reaction mixture to stand at room temperature for two hours, then dialyzed the mixture for 18 hours against 0.01 M ammonium bicarbonate buffer to obtain the succinylated derivative of uromodulin in which the native protein configuration had been altered (pH = 7.5).

We next prepared a reduced and carboxy- methylated derivative of uromodulin by dialyzing 1 ml of purified uromodulin (1 mg/ml) overnight against 100 volumes of 6 M guanidine HCL containing 0.01 M dithiothreitol. We then covered the reduced denatured glycoprotein with aluminum foil and added a 50 molar excess of sodium iodoacetate. We allowed this reaction to proceed in the dark for two hours. We separated the modified protein from the low molecular weight reactants by passing it over a column of Sephadex G-25 (Pharmacia, Uppsala, Sweden) . This treatment resulted in a derivative of uromo¬ dulin in which internal sulfylhydryl bonds were broken, leading to unfolding of the native protein.

We then analyzed the succinylated and re¬ duced and carboxymethylated uromodulin derivatives prepared above to confirm that their protein structure had been altered. First, we compared each derivative to unmodified uromodulin for the ability to bind the ^' ΞA4 monoclonal antibody which is protein specific, and requires native configuration for binding.

To carry out this comparison, we used a variation of the ELISA assay described supra. We coated 5 μg/ml of EA4 in carbonate buffer (pH = 9.6) onto Immunlon icrotiter plates and incubated the plates overnight at 4°C. We washed the plates with Tween-20 and then incubated them with either a modi¬ fied uromodulin or intact uromodulin for another 2 hours at room temperature. We washed the plates . with and then detected the presence of bound uromod¬ ulin using a 1:400 dilution of our rabbit anti-uro¬ modulin described supra. The presence of bound rabbit antiserum was detected using an alkaline phosphatase modified rabbit antiserum (Sigma, St. Louis, Missouri, U.S.A.). The results are shown in Figure 9, in which background OD in the presence of all reagents except uromodulin (0.090) was subtracted from all values. As demonstrated in that figure, succinylation and reduction and carboxymethylation essentially negated the ability of EA4 to recognize the modified forms of uromodulin demonstrating that EA4 specifically binds to the N-linked carbohydrate moiety. Next, we compared the biological activity of these modified uromodulin derivatives to that of unmodified uromodulin in our standard T cell pro¬ liferation assay. The results of the assay, shown in Figure 10, represent triplicate determinations expressed as a percentage of control values with no uromodulin (12,540 + 2,100 CPM (ISE)) with background CPM in the absence of tetanus toxoid (140 CPM). As

demonstrated in that figure, the uromodulin deriva¬ tives retained the in vitro biological activity of unmodified uromodulin.

We then predigested pronase (10 mg/ml) in 0.01 M Tris HCL buffer (pH = 8.0) containing 0.01 M calcium chloride at 37°C for two hours to destroy any glycohydrolases present in the pronase prepara¬ tion. Although we used pronase in this example, any proteolytic enzyme useful to digest proteins to yield carbohydrates may also be employed. Following this procedure, no intact protein of any size could be detected on 12.5% silver stained SDS-PAGE. We then added 25 μl of pre-digested pronase to 1 ml of purified uromodulin (1 mg/ml) in the same buffer, and heated the mixture to 60°C. We incubated the mixture for six hours, then added a second 25 μl li uot of pronase. We added a third 25 μl aliquot after 22 hours of incubation. We terminated the reaction after 30 hours by boiling it for 3 minutes'. We removed the precipitate by centrifugation at

10,000XG. We placed the supernatant on a 0.7 x 50 cm Bio-Gel P-4 column (Bio-Rad, Richmond, California, U.S.A.) and eluted it with 0.01 M ammonium bicar¬ bonate buffer (pH = 7.5). Eluted carboh ^y drate was detected using the anthrone method as de?~rlbed by Spiro supra. Protein was estimated by meas'-.riiig absorbance at 280 nm.

We pooled the first peak from this column, which contained the majority of the carbohydrate;, - and exhaustively lyophilized it to remove the vola¬ tile buffer. ^■ -•We also collected control peaks to run in parallel. These fractions were then resuspended in distilled water to their original starting volume: and various dilutions were tested in our T cell pro- liferation assay for biological activity. The results of this assay are shown in Figure 11, in which results are expressed as CPM ± 1 SE of triplicate determina¬ tions and in which no added uromodulin gave 74,400

CPM ± 8500. As demonstrated in that figure, the carbohydrate peak contained all of the biological activity of these various fractions. Furthermore, the carbohydrate peak was actually more active than intact uromodulin, based on its ability to induce greater inhibition of T cell proliferation.

Since pronase digestion yielded all avail¬ able carbohydrate moieties of uromodulin, we attempted to selectively digest the molecule to release only the N-linked carbohydrate moieties based on the suppo¬ sition that the N-linked carbohydrates constitute the biologically active portion of uromodulin. We digested 1 ml of purified uromodulin (1 mg/ml) in 0.2 M sodium phosphate buffer (pH = 8.8) containing 5 mM EDTA with 2.6 units of N-glycanase, an enzyme specific for N asparagine-linked oligosaccharides (Genzyme, Boston, Massachusetts, U.S.A.). Although we used N-glycanase in this example, any endoglycosidase effective to digest proteins to yield N-linked carbohydrates may also be employed. We incubated the reaction mixture at 37°C for 48 hours', then separated the reaction mixture on a 0.7 x 50 cm Bio-Gel P-4 column (Bio-Rad, Richmond, California, U.S.A.). We estimated protein by its emission intensity at 330 nm and carbohydrate content was . estimated by the anthrone procedure supra. N-glycanase treatment decreased the apparent mole¬ cular weight of uromodulin on 12.5% SDS-PAGE by approximately 10 kD. We tested the N-glycanase digested uromod- . ulin for in vitro bioactivity in our T cell prolifera¬ tion assay. The results, shown in Figure 12, are expressed as CPM + 1 SE of triplicate determinations. As demonstrated in that figure, this modified uro- modulin was at least as active as its undigested counterpart.

We then analyzed the ability of N-glycanase digested uromodulin to bind to IL-1 coated plates in a variation of our previously described ELISA assay. In this modified assay, we coated microplate wells with 1 μg/ml of recombinant IL-1 in carbonate buffer (pH = 9.6). After washing the plates, we then added various concentrations of unmodified uromodulin or N-glycanase digested derivatives in PBS Tween to the washed IL-1 coated plates and incubated them for 2 hours at room temperature. We washed and developed the plates with our uromodulin specific rabbit heteroantiserum. The results are shown in Figure 13, in which background absorbance in the absence of uromodulin (0-08) was subtracted from all values and results are expressed at 405 nm.

As demonstrated in Figure 13, the N-glyca¬ nase digested uromodulin failed to. bind efficiently to IL-1 coated plates, thus suggesting that the carbohydrate portion of uromodulin is responsible for the binding of uromodulin to IL-1.

This conclusion was further supported .by the following ELISA assay. In this assay, we took 500 ng of αr^rnodulin and digested it with N-glycanase. We ti __._. ^" lyophilized the digested uromodulin to dry- ness _tnd resuspended it in a 40:40:20 suspension of isobutanol, -ethanol and water. After a 1 hour extrac¬ tion period, we centrifuged the precipitate at 10,000XG, lyophilized the soluble material and re¬ suspended i ^in water. Various quantities of this extract or a control preparation of ovalbumin were added along with 40 ng of uromodulin to IL-1 coated microtiter plates incubated, washed and developed as described supra.-■

As shown in the table below, the isobutanol: ethanol:water extract of N-glycanase treated uro¬ modulin totally blocked binding of unmodified uro-

modulin, further evidencing that IL-1 binds to uromodulin via N-linked sugars.

OP 405 n *

40 ng uromodulin 485 40 ng uromodulin

+ 5 μl extract 86

40 ng uromodulin

+ 2.5 μl extract 181

40 ng uromodulin + 1.2 μl extract 121

40 ng uromodulin

+ 0.6 μl extract 244

40 ng uromodulin

+ 5 μg ovalbumin 513 40 ng uromodulin

+ 5 μg N-glycanase digest of ovalbumin 490

While we have hereinbefore described a number of embodiments of this invention, it is apparent that our basic constructions can be altered to provide other embodiments which utilize the pro¬ cesses and compositions of this invention. There¬ fore, it will be appreciated that the scope of this invention is to be defined by the claims appended hereto rather than by the specific embodiments which have been presented hereinbefore by way of example.

* A decrease in OD 405 corresponds to inhibition of binding of uromodulin to the IL-1 coated plate.