Identification and Characterization of Two Distinct Truncated Forms of gp130 and a Soluble Form of Leukemia Inhibitory Factor Receptor α-Chain in Normal Human Urine and Plasma*

Leukemia inhibitory factor (LIF) is a polyfunctional cytokine known to require at least two distinct receptor components (LIF receptor α-chain and gp130) in order to form a high affinity, functional receptor complex. In this report, we present evidence that there are two distinct truncated forms of gp130 in normal human urine and plasma: a large form with a molecular weight of approximately 100,000, which is similar to a previously described form of soluble gp130 in human serum, and a previously undescribed small form with a molecular weight of approximately 50,000. Using a panel of monoclonal antibodies raised against the extracellular domain of human gp130, we were able to show that the small form of the urinary gp130 probably contained only the hemopoietin domain. Both forms of gp130 bound LIF specifically and were capable of forming heterotrimeric complexes with soluble human LIF receptor α-chain in the presence of human LIF. In addition to the soluble forms of gp130, a soluble form of LIF receptor α-chain was also detected in human urine and plasma.

Leukemia inhibitory factor (LIF) 1 is a polyfunctional cytokine that can act on a wide range of cell types including osteoblasts, hepatocytes, adipocytes, neurons, embryonal stem cells, and megakaryocytes (1). LIF exerts its multiple biological functions through a specific cell surface receptor system, which consists of at least two membrane-bound glycoproteins, the LIF-binding chain (LIFR␣) and gp130. LIF binds first to LIFR␣ with low affinity (2) and then to gp130 to form a high affinity functional receptor complex leading to activation of downstream signal transduction pathways (3)(4)(5)(6). Both LIFR␣ and gp130 are members of the hemopoietin or cytokine type I family of receptors (7,8). The extracellular domains of members of this receptor family share common structural features including hemopoietin domains characterized by four conserved cysteine residues and a WSXWS motif and three fibronectin type III (FN III) modules (7,8). The membrane-bound gp130 was initially defined as the signal transducer of the interleukin-6 (IL-6) receptor system (9,10) and has been shown subsequently to also be a component of the functional receptor complexes of ciliary neurotrophic factor (CNTF) (4), oncostatin-M (OSM) (3,11), cardiotrophin-1 (12,13), and interleukin (IL)-11 (IL-11) (14 -16).
In addition to the cell membrane-anchored forms of LIFR␣ and gp130, it has been reported that naturally occurring soluble forms of these receptor molecules are present in biological fluids and may act as natural inhibitors of LIF activity (17)(18)(19). We and others have shown previously that a soluble form of the mouse LIFR␣ with a molecular weight (M r ) of approximately 90,000 -150,000 occurs at high levels in normal mouse serum and is elevated dramatically during pregnancy (17,18). Recently, we have provided evidence that the soluble form of mouse LIFR␣ probably arises from an alternative splicing event of the LIFR␣ mRNA (20). Despite the high levels of soluble LIFR␣ in mouse serum, its analogue was not detected in human serum (17).
In contrast, a soluble form of gp130 with a M r of 90,000 -110,000 has been found in human serum (19). Although gp130 functions as the high affinity converting and signaling subunit in the receptor complexes for IL-6, LIF, OSM, CNTF, cardiotrophin-1, and IL-11, OSM was the only cytokine in this family initially demonstrated to bind to membrane-bound gp130 (3), and subsequently, it has been shown that OSM can bind directly to the soluble form of gp130 with low affinity (21,22). We and others have recently shown that the soluble form of gp130 was able to bind not only directly and specifically to OSM but also to LIF (22,23). Using biosensor technology, we were able to determine that the interaction between hLIF and soluble human gp130 was of low affinity, with an equilibrium dissociation constant of approximately 44 nM (23). This low affinity interaction could explain previous failures in detecting direct binding of LIF to the membrane-bound form of gp130.
In this study, we present evidence that there are two distinct truncated forms of gp130 in normal human urine and plasma: a large form with a M r of approximately 100,000, which is similar to that previously described (19), and a previously undescribed small form with a M r of approximately 50,000. Both forms bound LIF specifically and were capable of forming heterotrimeric complexes with soluble hLIFR␣ in the presence of hLIF. In addition to the soluble forms of gp130, a soluble form of LIFR␣ was also detected in human urine and plasma.

EXPERIMENTAL PROCEDURES
Reagents-Escherichia coli-expressed hLIF (a gift from Sandoz Pharmaceutical Co., Hanover, Switzerland) was radioiodinated using a modified iodine monochloride method (24). Anti-human gp130 monoclonal antibodies (mAbs), AM64, GPX22, and GPZ35, which were raised against Chinese hamster ovary cell-expressed extracellular domain of human gp130, were prepared as described previously (10,25). A goat anti-human LIFR␣ polyclonal antibody raised against the extracellular domain of human LIFR␣ was purchased from R & D Systems.
Expression and Purification of Soluble Human LIFR␣ and gp130 in Pichia pastoris-A soluble form of human gp130 (shgp130), which consists of the Ig-like domain, hemopoietin domain, and three FN III modules, was expressed in the methylotropic yeast P. pastoris with a FLAG TM epitope tag (DYKDDDDK) at its N terminus and purified on an anti-FLAG M2 affinity column by elution with FLAG peptide as described previously (23). A short form of soluble human gp130 (sshgp130) was made identically as shgp130 except that the construct lacked all three FN III modules. Protein quantitation for the purified samples was performed by amino acid analysis. To make a soluble form of hLIFR␣, a cDNA encoding the hLIFR␣ (26) was altered at its 5Ј-end to encode an XhoI site and an in-frame 12CA5 epitope (YPYDVPDYA) (27). The sequence at the N terminus of the recombinant LIFR␣ was GAPYPYDVPDYA. The 3Ј-end was modified to encode an XbaI site and a stop codon was introduced after position 536 (2) so that the recombinant LIFR␣ only contained the two hemopoietin domains and the intervening Ig-like domain. The cDNA was subsequently cloned into the yeast expression vector pPIC9 and expressed in P. pastoris as described (23). The protein was partially purified by gel filtration chromatography and quantified by Scatchard analysis of hLIF binding isotherms (17).
Human Urine Collection-Male and female normal human urine was collected from volunteers after informed consent, and 0.02% (v/v) Tween 20 and 0.02% (w/v) sodium azide were added. Small scale urine concentration was carried out using a Centriprep-10 (Amicon), and large scale concentration was performed using the Sartorius EasyFlow Device with a cellulose triacetate membrane (molecular weight cut-off of 20,000). Any precipitating materials in urine occurring before or after concentration were removed by centrifugation.
Cross-linking Assay-Aliquots of samples were incubated with 125 I-hLIF in the absence or presence of unlabeled hLIF or antibodies in a final volume of 15 l for at least 1 h at 4°C. Then 5 l of a 12 mM solution of the bifunctional cross-linker BS 3 (Pierce) in 20 mM phosphate-buffered saline (pH 7.0) was added, and the mixtures were incubated for 30 min at 4°C. Samples were mixed with 7 l of 4-fold concentrated SDS sample buffer and analyzed by either 7.5 or 10% SDS-PAGE under nonreducing conditions. The gels were dried and visualized by either autoradiography or PhosphorImager analysis (Molecular Dynamics).
Affinity Chromatography-The rhLIF affinity column was prepared by covalently coupling 1 mg of E. coli-derived rhLIF to 1 ml of Affi-Gel 10 (Bio-Rad) according to the manufacturer's instructions. Normal human urine samples were concentrated 100-fold as described above and incubated with 1 ml of hLIF-Affi-Gel 10 resin for 3-5 h at 4°C. After unbound proteins were removed by centrifugation, the hLIF affinity beads were washed with 16 ϫ 1 ml of PBS followed by additional washes with 8 ϫ 0.3 ml of 10-fold diluted Actisep elution medium (Sterogenes Bioseparations, CA). The bound protein was then eluted with 10 ϫ 0.3 ml of undiluted Actisep elution medium. The affinity column eluates were buffer-exchanged into PBS using NAP-5 columns (Amersham Pharmacia Biotech). Using the same procedures, LIF-binding proteins were enriched and partially purified on the hLIF affinity column from 50 ml of an outdated normal human plasma sample (obtained from the Royal Melbourne Hospital Blood Bank). Aliquots of buffer-exchanged fractions were analyzed for their ability to bind to 125 I-hLIF using the cross-linking protocol described above.
Immunoprecipitation-Aliquots (20 l) of the hLIF affinity column eluate, 0.2 g/ml shLIFR␣, or 2 g/ml shgp130 were incubated with 125 I-hLIF (800,000 cpm) in the absence or presence of 50 g/ml unlabeled hLIF in a final volume of 50 l for at least 1 h at 4°C. Then 10 l of a 12 mM BS 3 solution was added, and the mixtures were incubated for 30 min at 4°C. After adding 1 M Tris-HCl buffer (pH 7.5) to a final concentration of 50 mM, the cross-linking reactions were incubated for 40 min at room temperature. The cross-linked samples were then mixed with an anti-human LIFR␣ polyclonal antibody at a concentration of 50 g/ml. After a 30-min incubation at 4°C, the mixtures were added to 30 l of 50% (v/v) protein G-Sepharose gel slurry (Amersham Pharmacia Biotech) previously equilibrated in PBS and incubated for 30 min at 4°C. The samples were centrifuged, and the protein G-Sepharose beads were washed with 4 ϫ 0.5 ml of PBS. For elution, the beads were mixed with 30 l of 2ϫ concentrated SDS sample buffer. The supernatants were then analyzed by 7.5% SDS-PAGE under nonreducing conditions.
Protein Estimation-Protein concentrations of pure recombinant soluble receptors were determined by amino acid analysis on a Beckman 6300 high performance amino acid analyzer equipped with a model 7000 data analyzer (Beckman).
Inhibition of STAT-3 Phosphorylation in M1 Myeloid Cells-M1 cells (ϳ10 7 cells/sample) were stimulated for 5 min at 37°C with either 1 ng of hLIF or saline together with either soluble hLIF receptor or soluble hgp130 and then lysed in 50 mM Tris-HCl (pH 7.5) containing 150 mM NaCl, 2 mM EDTA, 1% Triton X-100, 2 mM NaF, 1 mM Na 3 VO 4 , and proteinase inhibitors. After pelleting insoluble material and protein standardization, approximately 100 g of total cellular proteins were subjected to 4 -15% acrylamide SDS-PAGE under reducing conditions and then transferred to a prewetted polyvinylidene difluoride membrane (PVDF-Plus, Micron Separations Inc.). After blocking, the membrane was incubated with an anti-phospho-STAT-3 polyclonal antibody (New England Biolabs), followed by incubation with a goat anti-rabbit Ig polyclonal antibody conjugated with horseradish peroxidase (DAKO, Denmark). The phosphorylated STAT-3 protein was visualized by radiography using the ECL system (Amersham Pharmacia Biotech). To check the quantity of protein loading, the same membranes were stripped with 0.1 M glycine-HCl, pH 3.0, for 30 -60 min and washed three times in PBS, 0.1% Tween 20 before reprobing with a rabbit polyclonal antibody to STAT-3 (K-15, Santa Cruz Biotechnology, Inc.).
Estimation of Soluble Human LIFR␣ Concentration in Plasma-To quantify the soluble LIFR␣, an aliquot of an outdated normal human plasma sample (obtained from the Royal Melbourne Hospital Blood Bank) was first precleared with protein G-Sepharose beads at a ratio of 1:0.2 (v/v) for 1 h at 4°C. The protein G-absorbed plasma was then incubated in the presence or absence of 2 g of a goat anti-human LIFR␣ polyclonal antibody for 1 h at 4°C, followed by the addition of 25 l of protein G-Sepharose beads and a 2-h incubation at 4°C. Immunoprecipitation of recombinant shLIFR␣ at various concentrations was performed in parallel except that the preabsorption step with protein G beads was not included. The immunocomplexes were washed with 3 ϫ 1 ml of PBS containing 0.02% (v/v) Tween 20 and eluted from the protein G beads by boiling in SDS sample buffer under reducing conditions for 5 min before being subjected to 7.5% acrylamide SDS-PAGE. The Western blotting was performed as described above except that the anti-human LIFR␣ polyclonal antibody and a rabbit anti-goat Ig polyclonal antibody conjugated with horseradish peroxidase (DAKO, Denmark) were used as the first and second antibodies, respectively.

Detection of LIF-binding Proteins in Normal Human Urine-
Human urine samples, collected from six healthy individuals (H1-H6) were concentrated and tested for soluble LIF-binding proteins by chemical cross-linking. Analysis of the cross-linking products by SDS-PAGE ( Fig. 1) indicated that 125 I-hLIF was cross-linked specifically to two species of proteins in all six samples with M r of approximately 100,000 (here referred to as the "large form") and 50,000 (here referred to as the "small form") after subtraction for the M r of the bound unglycosylated hLIF, respectively. The levels of the two hLIF-binding proteins varied in the six samples. This variation was likely to be due to the differences in protein content of these samples (data not shown).

The Two Species of the Urinary hLIF-binding Proteins Are
Not Precomplexed-To examine whether the two hLIF-binding proteins were part of a preformed complex in urine, concentrated human urine was fractionated on a Superdex 200 gel filtration column as shown in Fig. 2A, and fractions were then analyzed for 125 I-hLIF binding by chemical cross-linking. Analysis of column fractions 21, 23, 25, 27, 29, and 31 by SDS-PAGE ( Fig. 2B) after cross-linking showed that the two hLIF-binding proteins were completely separated from each other according to their sizes, suggesting that they do not exist in a preformed complex in human urine. The M r estimates of the two proteins by gel filtration were consistent with those obtained above. Also, it can be seen from Fig. 2B that there was a downward trend in M r for both the large and small forms of the LIFbinding proteins across the fractions being assayed. This may be due to differential glycosylation of the two proteins.
Identification of the Urinary LIF-binding Proteins as Soluble gp130 and LIFR␣-In mouse and human serum, the presence of soluble forms of the LIF receptor components, mouse LIFR␣ and human gp130, respectively, has been described (17)(18)(19). It has also been demonstrated that LIF can bind to gp130 directly (22,23), although the affinity was relatively low (23,28). To examine whether the detected 125 I-hLIF binding activity in human urine was due to the presence of these reported proteins, we first performed competitive cross-linking experiments, as shown in Fig. 3, in which 125 I-hLIF was mixed with increasing concentrations of unlabeled hLIF prior to crosslinking to a partially purified urinary LIF-binding protein sample (Fig. 3A). This was then compared with the same crosslinking to (a) a recombinant form of soluble human gp130 (shgp130; Fig. 3B), which consists of the Ig-like domain, hemopoietin domain, and three FN III modules with a FLAG TM epitope tag at the N terminus, (b) a recombinant short form of soluble human gp130 (sshgp130; Fig. 3C) identical to shgp130 except that the construct lacked the FN III modules, or (c) a recombinant form of soluble human LIFR␣ (shLIFR␣; Fig. 3D). Densitometric analyses of these data (not shown) revealed that half-maximal inhibition of 125 I-hLIF cross-linking to both the large and small forms of the two urinary binding proteins occurred at approximately 750 ng/ml unlabeled hLIF, similar to the hLIF concentrations required to inhibit 50% of the crosslinking to both shgp130 and sshgp130, whereas an approximately 3-fold smaller amount of hLIF was required to achieve the same inhibition for shLIFR␣. These results indicated that the relative binding affinities of 125 I-hLIF to the two forms of urinary binding proteins were similar to those of 125 I-hLIF to recombinant shgp130 and sshgp130, suggesting that the LIF binding activity in human urine might be due to the presence of soluble forms of gp130 with different truncations at the Cterminal ends.
To formally test this possibility, the urinary LIF-binding proteins were first purified by a hLIF affinity chromatography step as described under "Experimental Procedures." The eluates from the hLIF affinity column were then fractionated on a Superdex 200 gel filtration column to separate the large and small forms of the LIF-binding proteins (data not shown). The appropriate fractions were subsequently analyzed in crosslinking experiments, in which 125 I-hLIF was incubated with the purified large or small form of the urinary LIF-binding protein in the presence of anti-shgp130 mAbs prior to crosslinking. As shown in Fig. 4, cross-linking in the presence of mAb AM64 yielded a higher M r complex in addition to the 125 I-hLIF⅐large form complex (Fig. 4A, compare lanes 6 and 8) without a significant effect on 125 I-hLIF cross-linking to the small form (Fig. 4B, compare lanes 6 and 8), whereas the addition of mAb GPZ35 in the cross-linking mixture generated higher M r complexes in both cases ( Fig. 4A and B, lanes 10). In contrast, mAb GPX22 completely inhibited the cross-linking of 125 I-hLIF to the small form (Fig. 4B, compare lanes 6 and 9) but only partially inhibited the 125 I-hLIF cross-linking to the large form (Fig. 4A, compare lanes 6 and 9). For comparison, these mAbs were also added to the cross-linking mixtures of recombinant shgp130, sshgp130, and shLIFR␣. As before, mAb AM64 only affected shgp130 (Fig. 4A, lane 3) but not sshgp130 (Fig.  4B, lane 3) cross-linking to 125 I-hLIF. Unexpectedly, mAb GPZ35 did not significantly affect the cross-linking of 125 I-hLIF to either shgp130 (Fig. 4A, lane 5) or sshgp130 (Fig. 4B, lane 5) despite being able to produce a higher M r species when added to the cross-linking mixture of the small form (Fig. 4B, lane 10). This may be due to the possibility that mAb GPZ35 could not recognize Pichia-expressed gp130, suggesting that it recog- Human urine from six normal individuals, four males and two females (H1 to H6), was collected, and 0.02% (w/v) sodium azide and 0.02% (v/v) Tween 20 were added. The urine samples were concentrated 25-fold with Centriprep-10 (Amicon) and buffer-exchanged into PBS containing 0.02% (w/v) sodium azide and 0.02% (v/v) Tween 20 using NAP-5 columns (Amersham Pharmacia Biotech). Aliquots (10 l) of these samples were incubated with 125 I-hLIF (200,000 cpm) alone or in the presence of 50 g/ml of unlabeled hLIF and followed by cross-linking with BS 3 as described under "Experimental Procedures." The crosslinked products were analyzed by 10% SDS-PAGE under nonreducing conditions, and the gel was dried and autoradiographed. nizes a carbohydrate-related epitope. Alternatively, it may suggest that the FLAG peptide tagged at the N termini of the two recombinant Pichia-gp130 proteins somehow affects the GPZ35 binding. In the case of mAb GPX22, it almost completely inhibited the cross-linking of 125 I-hLIF to both shgp130 (Fig. 4A, lane 4) and sshgp130 (Fig. 4B, lane 4). As expected, these mAbs showed little effect on the cross-linking to shLIFR␣ (Fig. 4A,  lanes 13-15). By themselves, these mABs did not significantly cross-link to 125 I-hLIF (Fig. 4B, lanes 11-13).
Taken together, these results revealed that the small form of the urinary LIF-binding protein was a truncated form of gp130 probably containing only the hemopoietin domain but lacking all or almost all three FN III modules and that the large urinary LIF-binding form contained gp130. Meanwhile, two observations from the antibody analysis experiments also suggested that the LIF binding activity in the purified large form fraction was not entirely due to the presence of soluble gp130. First, mAB AM64 was able to almost completely shift the 125 I-hLIF⅐shgp130 complex to a higher M r species (Fig. 4A, lane  3) but only partially did so to the large form (Fig. 4A, lane 8). Second, mAb GPX22 could completely inhibit the cross-linking of 125 I-hLIF to the recombinant shgp130 (Fig. 4A, lane 4) but could only partially inhibit the 125 I-hLIF cross-linking to the large form (Fig. 4A, lane 9).
One possibility was that the purified large form fraction also contained the soluble LIFR␣ component. Evidence for this also came from a large scale purification (10 liters of human urine were used) of the urinary LIF-binding proteins using hLIF affinity chromatography described under "Experimental Proce-dures" (data not shown). When eluates from the affinity column were analyzed in cross-linking experiments, the fraction containing the most LIF-binding activity showed an extra higher M r species in addition to the normal large and small forms (Fig. 5, lane 8).
To test this formally, we reexamined the hLIF affinity column eluate and the purified large form LIF-binding fraction for the soluble hLIFR␣. As shown in Fig. 5, cross-linking of 125 I-hLIF to these samples, in the presence of an anti-human soluble LIFR␣ polyclonal antibody, generated higher M r complexes in addition to those seen when cross-linking was performed without the antibody (Fig. 5, compare lanes 8 and 10 and lanes  13 and 15, respectively). The appearance of these higher M r complexes with the anti-hLIFR␣ antibody coincided with the decreases in intensities of the bands corresponding to the 125 I-hLIF⅐large form complexes. Moreover, the extra higher M r species occurring in the eluate from the hLIF affinity column (Fig.  5, lane 8) almost completely disappeared upon the addition of the antibody (Fig. 5, lane 10), strongly suggesting that it was

FIG. 4. Analysis of the urinary hLIF-binding proteins by antihuman gp130 antibodies.
The urinary LIF-binding proteins were first purified by hLIF affinity chromatography as described under "Experimental Procedures." The eluates from the hLIF affinity column were then fractionated on a Superdex 200 gel filtration column to separate the large and small forms of the LIF-binding proteins (data not shown). Panel A, lanes 1-5, shgp130 (10 l of 0.5 g/ml) cross-linked to 125 I-hLIF (200,000 cpm) alone or in the presence of 25 g/ml unlabeled hLIF or 10 g/ml mAbs AM64, GPX22, or GPZ35, respectively; lanes 6 -10, purified large form hLIF-binding fraction (10 l) cross-linked to 125 I-hLIF (200,000 cpm) alone or in the presence of 25 g/ml unlabeled hLIF or 10 g/ml mAb AM64, GPX22, or GPZ35, respectively; lanes 11-15, shLIFR␣ (10 l of 0.05 g/ml) cross-linked to 125 I-hLIF (200,000 cpm) alone or in the presence of 25 g/ml unlabeled hLIF or 10 g/ml mAb AM64, GPX22, or GPZ35, respectively. The cross-linked products were analyzed by 7.5% SDS-PAGE under nonreducing conditions. Panel B, lanes 1-5, sshgp130 (10 l of 1 g/ml) cross-linked to 125 I-hLIF (200,000 cpm) alone or in the presence of 25 g/ml unlabeled hLIF or 10 g/ml mAb AM64, GPX22, or GPZ35, respectively; lanes 6 -10, purified small form hLIF-binding fraction (10 l) cross-linked to 125 I-hLIF (200,000 cpm) alone or in the presence of 25 g/ml unlabeled hLIF or 10 g/ml mAb AM64, GPX22, or GPZ35, respectively; lanes 11-13, 125 I-hLIF (200,000 cpm) cross-linking to 10 g/ml of mAb AM64, GPX22, or GPZ35, respectively. The cross-linked products were analyzed by 10% SDS-PAGE under nonreducing conditions. predominantly a form of soluble LIFR␣ with a M r higher than the soluble LIFR␣ that co-migrated with the large form of gp130. These were probably different glycosylated or truncated forms of the soluble LIFR␣. As a control, 125 I-hLIF was also cross-linked to a recombinant shLIFR␣ or a recombinant shgp130 in the presence of the same anti-human LIFR␣ antibody. As expected, the addition of the anti-human LIFR␣ antibody only yielded extra higher M r species in the cross-linking mixtures of shLIFR␣ (Fig. 5, compare lanes 1 and 3 and lanes  5 and 7). Together, these data indicated clearly the presence of a soluble LIFR␣ in the urinary samples.
These results were further confirmed in the immunoprecipitation experiment in which the resultant cross-linking product of 125 I-hLIF to the hLIF affinity column eluate was immunoprecipitated using the same anti-human LIFR␣ antibody (Fig.  5, lane 18). It was noted that the sizes of the 125 I-hLIF crosslinking proteins immunoprecipitated by the anti-hLIFR␣ antibody were similar to those observed when the antibody was added prior to cross-linking (Fig. 5, compare lanes 1, 3, and 16  and lanes 8, 10, and 18, respectively), suggesting that crosslinking of the antibody to the 125 I-hLIF⅐LIFR␣ complex continued to occur during immunoprecipitation.

Ternary Complex Formation of both the Large and Small
Forms of gp130 with shLIFR␣ in the Presence of hLIF-Both the large and small forms of the urinary gp130 as well as recombinant shgp130 and sshgp130 were analyzed in crosslinking experiments for their ability to form a ternary complex with shLIFR␣ in the presence of hLIF. As shown in Fig. 6, cross-linking of 125 I-hLIF to the purified large form or recombinant shgp130 in the presence of shLIFR␣ generated extra higher M r species (Fig. 6, lanes 7 and 3, respectively), the M r of which could be accounted for by summing up the M r of all three cross-linking components, suggesting the formation of a tripartite complex. The inability of the purified large form LIFbinding fraction by itself (Fig. 6, lane 6) to yield a higher M r complex corresponding to the ternary complex was probably due to insufficient LIFR␣ in the sample. When an extra amount of recombinant shLIFR␣ was added to the cross-linking mixture of this sample, the formation of a ternary complex became detectable (Fig. 6, lane 7). This was consistent with the above observation that the appearance of an extra high M r crosslinking species occurred only in the hLIF affinity column eluate containing the most LIF-binding activity. As above, when the cross-linking was performed with the small form of the urinary gp130 and recombinant sshgp130, both proteins were shown to be capable of forming ternary complexes with shLIFR␣ in the presence of hLIF (Fig. 6, lanes 9 and 5, respectively). The affinity of the ternary complex for hLIF was estimated by competing with unlabeled hLIF for the formation of the cross-linked radioactive complex. Fig. 7 shows the results of such an inhibition experiment using a mixture of shLIFR␣ and shgp130. In this experiment, cross-linked complexes of hLIF with either receptor alone could not be distinguished, but the ternary complex was clearly seen as a higher M r band (Fig. 7, indicated by the solid arrow). All cross-linked bands were inhibited by unlabeled hLIF, but the higher M r complex showed an IC 50 value 4 times lower than that for the lower M r complexes. This indicates a higher affinity for hLIF in the ternary complex, although it was not possible to derive accurate affinity constants from this type of data.
Evidence for a Soluble LIFR␣ in Human Plasma-It ap- peared that the soluble LIFR␣ in human urine was of low abundance, since the LIF binding activity was hardly detectable in unconcentrated urine (data not shown). This may be the reason that we were unable to detect the soluble LIFR␣ in human serum in our earlier study, which was based on detecting binding activity for LIF (17). To examine this possibility, a human plasma sample (50 ml) was loaded onto a hLIF affinity column (see "Experimental Procedures" for details), and the column eluates were then analyzed in cross-linking experiments for the presence of LIF binding activity. As shown in Fig.  8A, SDS-PAGE analysis of the cross-linked products of 125 I-hLIF to the five column fractions (eluates 1-5) revealed that there were at least three LIF-binding proteins present in human plasma with M r ranging from 50,000 to 130,000 (after subtraction for the M r of the bound 125 I-hLIF). Judging by the sizes of their cross-linking products with 125 I-hLIF, we assumed that two of these proteins, with M r of approximately 100,000 and 50,000, respectively, corresponded to the large and small forms of gp130 detected in human urine. In two of the eluates (eluates 2 and 3), there was an additional minor band (Fig. 8A, lanes 4 and 7, indicated by the open arrow), which was similar to the extra higher M r species previously observed in the hLIF affinity column eluate (Fig. 5, lane 8). Upon the addition of the anti-hLIFR␣ antibody prior to cross-linking, this minor band disappeared and was replaced by an increase in intensity of the band migrating at the top of the separating gel, indicated by the solid arrow (Fig. 8A, compare lanes 4 and  6 and lanes 7 and 9), suggesting that it corresponded to the cross-linked complex between 125 I-hLIF and the soluble LIFR␣. There was also a similar M r complex migrating at the top of the separating gel in the affinity column eluates, which predominantly occurred in eluate 1 (Fig. 8A, lane 1). The intensity of this complex was decreased upon the addition of unlabeled hLIF (compare lanes 1 and 2) but unaffected by the addition of the anti-hLIFR␣ antibody (compare lanes 1 and 3), probably indicating the presence of some aggregated form of gp130 in the sample. It is also worth noting that there was relatively more of the small form of gp130 in urine than in plasma.
Plasma shLIFR␣ levels were determined by immunoprecipitation with anti-hLIFR␣ followed by Western blotting with the same antibody (Fig. 8B). A specific band of ϳ140 kDa was detected in these experiments (lanes 2 and 4) and, by comparison with the intensity obtained for known amounts of the lower M r recombinant shLIFR␣ (lanes 6 -8), was estimated to be present in human plasma at ϳ10 ng/ml. Inhibitory Action of shLIFR␣ and shgp130 on LIF Biological Activities-LIF was originally defined by its ability to induce macrophage differentiation in the M1 myeloid leukemic cell line (1). One of the earliest detectable actions of LIF on M1 cells is the induction of tyrosine phosphorylation and activation of the signal transducer and activator of transcription, STAT-3 (29). Because of the small amounts of recombinant receptors available, we decided to test the inhibitory action of the soluble receptors on hLIF-induced STAT-3 activation in M1 cells, which occurs rapidly. Fig. 9 shows that shLIFR␣ and mixtures of shLIFR␣ and shgp130 were able to significantly inhibit tyrosine phosphorylation of STAT-3 but that, at the concentrations used, neither form of shgp130 alone nor a mixture of soluble human urinary LIFR␣ and gp130 could inhibit hLIFinduced STAT-3 phosphorylation.  1 and 2 and lanes 3 and 4 correspond to 200 and 500 l of human plasma, respectively, and lanes 5-8 correspond to 0.5 ml of 0, 1, 10, and 100 ng/ml recombinant shLIFR␣, respectively. (30,31). In the IL-6 receptor subfamily, these include IL-6 receptor ␣-chain (IL-6R␣) in human urine (32) and serum (33), LIFR␣ in mouse serum (17,18), CNTF receptor ␣-chain in human cerebrospinal fluid (34), and gp130 in human serum (19). Soluble receptors have been implicated in both enhancing and reducing the biological effects of their cognate ligands. For example, the complex of soluble IL-6R␣ and IL-6 is capable of interacting with cell surface gp130 to trigger a variety of biological responses (9), while a naturally occurring soluble form of gp130 in human serum has been implied to serve as a negative regulator in vivo of the signaling mediated by the IL-6⅐soluble IL-6R␣ complex (19), and soluble LIFR␣ in mouse serum also acts exclusively as an inhibitor of LIF signaling (17,18).
Here, we describe the presence in normal human urine and plasma of a soluble form of LIFR␣ and two distinct truncated forms of soluble gp130 (the large and small forms) that can bind hLIF specifically. To our knowledge, this is the first report describing a naturally occurring soluble form of LIFR␣ in human biological fluids as well as describing a small form of soluble gp130 capable of binding LIF directly and specifically. The large form of the urinary soluble gp130 with a M r of approximately 100,000 was similar to the previously described soluble form of gp130 in serum with a M r of 90,000 -110,000 (19). The previous detection of the soluble gp130 in serum was facilitated by the anti-human gp130 mAb AM64 (19). We have demonstrated in this study that mAb (AM64) could not recognize the soluble form of gp130 lacking FN III modules. This may be one of the reasons for the earlier inability (19) to detect the small form of gp130, which appeared to be also present in human plasma. Based on the similar sizes of the cross-linked complexes between 125 I-hLIF and sshgp130 and between 125 I-hLIF and the small form of urinary gp130, as well as the results obtained from the analyses with the anti-gp130 mAbs, we concluded that the small form of the urinary gp130 was likely to contain only the hemopoietin domain and was missing all or almost all of the three FN III modules.
Recently, an alternatively spliced mRNA encoding a soluble form of human gp130 was described from blood mononuclear cells (35). This transcript would encode a form of gp130 truncated within one base pair of the transmembrane domain and would correspond to the long form described here and elsewhere (19). Whether the short form is encoded by a separate transcript or by posttranslational processing of the longer form is currently unclear. It is of interest, however, that three alternate transcripts encoding soluble human LIFR␣ have been described from liver, placenta, and choriocarcinoma cell line (36). One possibility may be that the FN III domains of the long form allow it to be sequestered at tissue sites and that proteolytic cleavage to generate the short form could serve as an additional control point to regulate circulatory levels of bioactive forms of soluble gp130. Further experiments are required to address this issue.
The recombinant shLIFR␣ containing the two hemopoietin domains and the intervening Ig-like domain, but lacking all three FN III modules, was shown to be sufficient for hLIF binding, a finding consistent with previous results (28). In the presence of hLIF, this shLIFR␣ was also capable of forming ternary complexes with both the recombinant gp130 lacking all three FN III modules (sshgp130) and the naturally occurring small form of gp130, probably also lacking all three FN III modules. These findings were in agreement with a previous mutagenesis study of gp130, which demonstrated that only the membrane distal half of gp130, consisting of the Ig-like domain and hemopoietin domain, was responsible for the formation of a ternary complex with IL-6 and the IL-6R␣ (37). The exact roles of the three FN III modules in ternary complex formation of these hemopoietin receptors still remain to be determined. The ternary complexes of shLIFR␣ with both forms of the recombinant and the urinary gp130 in the presence of hLIF appeared to be heterotrimeric, which agreed with our previous finding that hLIF could form a cross-species heterotrimeric complex with soluble mouse LIFR␣ and human gp130 in solution (23) but differed from the IL-6⅐IL-6R␣⅐gp130 and CNTF⅐CNTF receptor ␣-chain⅐gp130⅐LIFR␣ complexes, which were hexameric (38 -40).
At the concentrations used, the shLIFR␣ displayed significant inhibition of hLIF-induced STAT-3 phosphorylation in M1 cells, but both forms of soluble gp130 were ineffective. Moreover, no significant increase in inhibition was observed when shLIFR␣ was mixed with either form of gp130.
These data are in agreement with previous data indicating that shLIFR␣ acts exclusively as a LIF antagonist (17,18) but contrast with previous reports that soluble gp130 inhibited the biological actions of IL-6 on Kaposi's sarcoma cells (41) as well as the proliferative actions of LIF and OSM on TF-1 erythroleukemia cells (22). The reason for these differences is probably the use in the latter study of a dimeric form of soluble gp130 at considerably higher concentrations than those used here (approximately 6 g/ml compared with 0.45 g/ml). This is probably also the reason that the soluble human urinary LIFR␣ and gp130 failed to inhibit STAT-3 phosphorylation in M1 cells stimulated by hLIF (Fig. 9). Nevertheless, taken together, all of these observations suggest that shLIFR␣ and shgp130 in serum and urine would serve to act as inhibitors of LIF, OSM, IL-6, CNTF, and IL-11 action. Although the concentrations of these receptors in normal serum appear insufficient to inhibit these cytokines, it is likely that some biological responses may elevate the concentrations of the receptors to a level that is able to suppress the action of these proinflammatory cytokines. For FIG. 9. Effects of soluble LIFR␣ and gp130 on hLIF-induced STAT-3 tyrosine phosphorylation in M1 myeloid cells. M1 cells were treated with either saline (Ϫ) or 1 ng of hLIF (ϩ), together with 0.05 g of shLIFR␣, 0.45 g of shgp130, 0.225 g of sshgp130, a mixture of 0.05 g of shLIFR␣ and 0.45 g of shgp130, a mixture of 0.05 g of shLIFR␣ and 0.225 g of sshgp130, or a mixture of soluble human urinary LIFR␣ and gp130, respectively, in a total volume of 1 ml. The total cellular proteins were analyzed by Western blotting as described under "Experimental Procedures." The upper panels show the level of phosphorylated STAT-3, and the lower panels show the level of total STAT-3. example, the concentration of shLIFR␣ detected in normal human plasma (ϳ10 ng/ml) was only 5-fold lower than the dose that gave significant inhibition of LIF-stimulated STAT-3 phosphorylation in M1 cells. Consequently, it will be of some interest to determine stimuli that result in increased secretion and circulatory levels of soluble hLIFR␣ and gp130.