THP binds non-specifically to a broad spectrum of protein molecules [11,16]. To confirm this property, microwells were coated with 100 μL of 20 μg/mL BSA, human IgG, human recombinant TNF-α, IFN-γ, IL-6, or IL-1β and incubated at 37 °C for 2 h and 4 °C overnight. THP (100 μL at 10 μg/mL) was then added to the microwells and incubated at 37 °C for 2 h. HRP-conjugated anti-uromucoid antibodies were added to detect THP binding to different proteins. We thus confirmed THP purified from normal human urine non-specifically bound different protein molecules with diverse capacities (Figure 1). Maximum binding occurred between THP and TNF-α; minimum binding occurred between THP and IFN-γ.

To verify binding between THP and TNF-α, Western blots (2 to 16 μg/mL THP; Figure 2A) and ELISA (0.5 to 6 μg/mL TNF-α; Figure 2B) were performed.

Binding between THP and TNF-α was clearly dose-dependent. Moonen et al. [30] argued that native TNF-α and IL-1β do not bind THP because TNF-α in liquid phase does not competitively inhibit THP binding to TNF-α-coated microtiter plates. However, Hession et al. [27] argued that THP is a renal ligand for cytokines in vivo. Why THP only reacts with denatured, not native, TNF-α, and whether TNF-α and IL-1β are denatured in vivo requires further investigation.

It is believed that the binding affinity between two proteins is less than antigen-antibody or ligand-receptor interactions. We estimated the THP-TNF-α binding affinity as described by Katanick et al. [31] with some modifications. The specific binding vs. TNF-α concentrations in two experiments are plotted in Figure 3A (Experiment 1) and 3C (Experiment 2).

The Y transformed curves derived from the specific binding data are plotted in 3B (Experiment 1) and 3D (Experiment 2). B_max and K_d values were calculated by Scatchard plot analysis shown in Figure 3. The K_d values of the two experiments were 1.368 × 10⁻⁶ M (Figure 3B) and 1.647 × 10⁻⁶ M (Figure 3D). However, Rhodes et al. [32] demonstrated two binding affinities between sheep THP and IgG; a high-affinity K_d of 10⁻⁸–10⁻⁹ M and a low-binding affinity K_d of 10⁻⁶–10⁻⁷ M. Although we did not measure the K_d of human THP-IgG binding, it was 20% lower than human THP-TNF-α binding as shown in the ELISA results (Figure 1). Experiments to measure the binding affinity of different mammalian THPs with human TNF-α are underway.

Sherblom et al. [14] demonstrated a lectin-like interaction between human recombinant TNF-α and uromodulin. We hypothesized that THP molecules may play dual roles in protein binding as their high carbohydrate content (25%–30%) are targets of lectin-like domains of TNF-α. Conversely, some THP domain structures such as the zona pellucina (ZP) domain may possess adhesive molecule-like or even lectin-like properties and bind to carbohydrate components in TNF-α molecules. We assessed the carbohydrate compositions of THP and THP-binding molecules. Lectin-binding ELISA method was performed with 5 lectins with specific sugar moieties to detect the sugar compositions of BSA, IgG, TNF-α, IFN-γ, and THP (Figure 4). IFN-γ contained minimal glycomoieties, consistent with its minimal binding to THP. We noted that BSA, IgG, and TNF-α contain abundant β(1,4)-GlcNAc oligomers and GlcNAc/branched mannose. These glycomoieties are also present in THP. Muchmore et al. [33] found that high-mannose glycopeptides [Man5(6)GluNAc₂-Asn] THP interact with recombinant TNF-α and IL-1β. It is possible the high-mannose glycans in human THP are carried by Asn251 [21,23]. IL-2 also exhibits lectin-like properties specific for high-mannose glycopeptides capable of binding THP [13,33]. Lucas et al. [34] demonstrated that trypanosome-TNF-α interaction was inhibited by N,N′-diacetylchitobiose. This domain structure also possessed lectin-like affinity for TNF-α. These results suggest that sugar-lectin-sugar interactions are one of the mechanisms for THP-protein binding.

THP contains complex carbohydrate side chains around the protein core structure; thus, it is necessary to determine whether the side chain glycomoiety or protein-core structure mediates THP-TNF-α binding. We used carbohydrate-degrading enzymes (neuraminidase, β-galactosidase), protein-degrading enzyme (proteinase K), and glycoprotein-degrading enzymes (O-sialoglycoprotein endopeptidase, carboxy- peptidase Y) to digest intact THP as shown in Figure 5A. Digested THP products were then immunoblotted with TNF-α (1° binder) and HRP-anti-TNF antibody (2° binder). As shown in Figure 5B, only β-galactosidase-digested THP products reacted with TNF-α on Western blots. Thus, galactoside residues in the THP side chain prevent THP-TNF-α binding. Because the amount of THP (4 μg) used in Figure 5 was insufficient for reacting with TNF-α as shown in Figure 2, it is difficult to conclude whether the protein-core structure is involved in THP-TNF-α binding. We suggest the ZP domain responsible for polymerization of THP into supra-molecular structures mediates homologous THP-THP binding and heterogeneous THP-TNF-α binding. Genetic manipulation to delete the ZP domain is in progress.

Siaα(2,3)Gal/GalNAc, Siaα(2,6)Gal/GalNAc, β(1,4)-GalNAc oligomers, mannose residues, and GlcNAc/branched mannose are present in THP and TNF-α (Figure 4). To determine whether these common monosaccharides (sialic acid, mannoside, glucosamine, and galactosamine) are also involved in THP-TNF-α binding, we preincubated each monosaccharide with either TNF-α or THP and then added the other molecule. As demonstrated in Figure 6, mannose, GlcNAc, and GalNAc significantly suppressed THP-TNF-α binding. These results are consistent with Sherblom et al. [13,34] in which high mannose-containing glycopeptides such as Man5GlcNAc2-R and Man6Glc-NAc2-R mediate THP-TNF-α binding through sugar-lectin-sugar interactions. In contrast, sialic acid did not influence protein-protein binding, as reported by Parsons et al. [7] and Huang et al. [35]. These results support our central theme that both sugar-lectin and sugar-protein interactions between cognate sites in THP and TNF-α mediate THP-TNF-α binding.

Purified human IgG, BSA, neuraminidase, β-galactosidase, proteinase K, and carboxypeptidase Y were purchased from Sigma-Aldrich (St. Louis, MO, USA). O-sialoglycoprotein endopeptidase was purchased from Cedar Lane Laboratories (Burlington, NC, USA). Biotin-conjugated lectins specific to unique carbohydrate moieties, including MMA [Siaα(2,3)Gal/GalNAc], SNA-1 [Siaα(2,6)Gal/ GalNAc], GNA [mannose], DSA [β(1,4)-GlcNAc oligomers], and ConA [GlcNAc/branched mannose] were obtained from E-Y Labs (San Mateo, CA, USA). Human recombinant TNF-α, IFN-γ, IL-6, and IL-1β were obtained from R&D Systems (Minneapolis, MN, USA). Mouse monoclonal anti-human TNF-α antibody and HRP-conjugated anti-TNF-α antibody were purchased from R&D Systems. α-Methyl-D-mannoside (α-MDM), N-acetyl-D-glycosamine (GlcNAc), N-acetyl-D-galactosamine (GalNAc) and N-acetylneuraminic acid (sialic acid, SA) were purchased from Sigma.

Twenty-four-hour urine was collected in a clean glass bottle from a normal individual. We followed the purification procedures described in our previous report [20]. THP purity and relative molecular weight were detected by 10% SDS-PAGE. Identification of THP was confirmed by Western blotting with anti-uromucoid antibody (The Binding Site Ltd, University of Birmingham Research Institute, Birmingham, UK). This study was approved by the IRB and Medical Ethics Committee, Taipei Veterans General Hospital, Taiwan (VGH IRB NO: 94-07-27A). All participants provided signed, informed consent.

THP digestion was performed as described by Sherblom et al. [14]. All enzyme digestions were performed at 37 °C for 16–24 h. Final enzyme concentrations were: neuraminidase (10 U/mL) in 50 mM sodium acetate, pH 5.0; β-galactosidase (0.05 U/mL) in 50 mM sodium acetate, pH 5.0; O-sialoglycoprotein endopeptidase (50 μg/mL) in PBS, pH 7.2; carboxypeptidase Y (enzyme:substrate 1:10) in 0.2 M pyridine-acetate buffer, pH 7.2; proteinase K (0.5 μg/mL) in 10 mM Tris buffer, pH 7.5 with 1 mM MgCl2. The enzyme-cleaved THP products were then heated at 65 °C for 60 min to inactivate residual enzyme. The digested products were intensively dialyzed against alkaline distilled water, pH 9.0 for 24 h, changing the dialysate every 2 h to remove the products smaller than 10 kDa. The digested products were lyophilized and stored at −20 °C.

A Scatchard plot was used to analyze the binding affinities of THP and TNF-α. We followed the method described by Kananick et al. [31] with modifications: radioisotope was replaced by HRP-conjugated-TNF-α in the ELISA. Briefly, recombinant human TNF-α (0, 500, 1,000, 1,500, 2,000, and 2,500 pg/mL) were added to THP-coated microwells and incubated at room temperature for 2 h with continuous shaking. The mixture was spun at 450 g for 20 min. Unbound THF-α in the supernatant and thrice-washed aspirates were collected. Bound TNF-α in the microwells was lysed with 200 µL lysis buffer containing 50 mM borate, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, and 25 mM phenylmethylsulphonyl fluoride, pH 8.0. Bound and unbound TNF-α were measured by ELISA. A Scatchard plot was drawn and analyzed by the “Prism” statistical program provided by GraphPad Software [36] to calculate the dissociation constants (K_d). K_d is defined as the ratio of unbound and bound molecules at equilibrium: K_d = [A] × [B]/[AB]. Thus, small K_d indicates high-affinity interactions and large K_d values indicate low-affinity interactions.

Results are reported as mean ± S.D. Continuous variables were assessed by non-parametric Wilcoxon rank-sum test using commercially available software (Stata/SE8.0 Windows). Statistical significance was indicated by p < 0.05.