Biochemical Journal

Research article

Contribution of leptin receptor N-linked glycans to leptin binding

Yuichi Kamikubo, Claudia Dellas, David J. Loskutoff, James P. Quigley, Zaverio M. Ruggeri

Abstract

The extracellular domain of the human leptin receptor (Ob-R) contains 20 potential N-glycosylation sites whose role in leptin binding remains to be elucidated. We found that a mammalian cell-expressed sOb-R (soluble Ob-R) fragment (residues 22–839 of the extracellular domain) bound leptin with a dissociation constant of 1.8 nM. This binding was inhibited by Con A (concanavalin A) or wheatgerm agglutinin. Treatment of sOb-R with peptide N-glycosidase F reduced leptin binding by ∼80% concurrently with N-linked glycan removal. The human megakaryoblastic cell line, MEG-01, expresses two forms of the Ob-R, of approx. 170 and 130 kDa molecular mass. Endo H (endoglycosidase H) treatment and cell culture with α-glucosidase inhibitors demonstrated that N-linked glycans are of the complex mature type in the 170 kDa form and of the high-mannose type in the 130 kDa form. Both isoforms bound leptin, but not after peptide N-glycosidase F treatment. An insect-cell-expressed sOb-R fragment, consisting of the Ig (immunoglobulin), CRH2 (second cytokine receptor homology) and FNIII (fibronectin type III) domains, bound leptin with affinity similar to that of the entire extracellular domain, but this function was abolished after N-linked glycan removal. The same treatment had no effect on the leptin-binding activity of the isolated CRH2 domain. Our findings show that N-linked glycans within Ig and/or FNIII domains regulate Ob-R function, but are not involved in essential interactions with the ligand.

  • class I cytokine receptor
  • human megakaryoblastic cell line
  • insect cell
  • leptin receptor (Ob-R)
  • N-linked glycosylation
  • peptide N-glycosidase F

INTRODUCTION

Leptin, the product of the ob gene, is a 16 kDa cytokine-like hormone secreted into the circulation mainly by adipocytes [1]. It regulates food intake and energy expenditure through activation of a hypothalamic receptor [2], and its plasma levels correlate with body fat energy stores [3]. Leptin has been shown to induce proliferation, differentiation and activation of haemopoietic cells [4], as well as influencing wound healing [5,6], angiogenesis [7], immune and inflammatory responses [8]. Moreover, studies in vitro and in animal models have suggested that leptin, through its receptor, contributes to the association between obesity and cardiovascular diseases by accelerating platelet aggregation and arterial thrombosis [913]. The leptin receptor (Ob-R) is the product of the db (diabetes) gene and a member of the class I cytokine receptor superfamily [14]. The gene is alternatively spliced to produce five polypeptides with identical extracellular and transmembrane domains but with different length C-terminal cytoplasmic domains. The molecular mass (without carbohydrate) of the long isoform (Ob-Rb) is ∼130 kDa, whereas the short ones are ∼100 kDa (Ob-Ra, Ob-Rc and Ob-Rd) or ∼90 kDa (Ob-Re) [2,15,16]. Ob-Ra is the most abundant isoform and is expressed in most peripheral tissues, whereas the long Ob-Rb is the only one fully capable of leptin-induced signalling through cytoplasmic domain association with JAK2 (Janus kinase 2) [1720]. The common extracellular portion includes the CRH1 [first CRH (cytokine receptor homology)] domain, an Ig (immunoglobulin) domain, the CRH2 (second CRH) domain and two FNIII (fibronectin type III) domains (Figure 1). The high-affinity leptin-binding site has been localized to the CRH2 domain [2123]. The Ig and FNIII domains are critically involved in Ob-R activation [24], whereas the role of the CRH1 domain remains unknown.

Figure 1 Schematic representations of human Ob-R

The domain structure of the long isoform of the human Ob-R is shown with its extracellular (residues 22–839), transmembrane and intracellular domains. The extracellular domain includes two CRH domains, an Ig-like domain and two FNIII domains. Each CRH domain is formed by two FNIII-like subdomains; in the case of the CRH2 domain, which contains the high-affinity leptin-binding site, these correspond to residues 428–536 and 537–635. Ob-R contains 20 potential N-linked glycosylation sites (indicated by hexagons) in the extracellular domain.

Structure–function relationships within Ob-R are not well understood; in particular, it remains to be established whether N-linked glycans are required for leptin binding and receptor activation and/or are responsible for the structural and functional heterogeneity of Ob-R in different cells. The extracellular domain of human Ob-R contains 20 potential N-linked glycosylation sites defined by the consensus sequence Asn-Xaa-Ser/Thr, where Xaa is any amino acid except proline (Figure 1). Previous studies have shown that recombinant Ob-R expressed in mammalian cells is extensively N-glycosylated [1921,25], since the apparent molecular mass of each isomer is higher than that calculated for the corresponding polypeptide chain (Table 1). Moreover, Haniu et al. [25] found that 18 of the 20 potential N-linked glycosylation sites are occupied by glycans in rsOb-R [recombinant sOb-R (soluble Ob-R)] expressed by CHO (Chinese-hamster ovary) cells. To date, however, it is not known whether these glycans are required for leptin binding. On the other hand, N-linked glycans have been shown to contribute variably to ligand binding in several class I cytokine receptors [2630]. Thus we undertook the present study in order to clarify the extent to which the degree of N-linked glycosylation influences Ob-R receptor function. By using specific lectins, glycosidic enzymes and inhibitors of N-linked glycosylation, we found that N-linked glycans in Ob-R are required for leptin binding, but not through direct interactions with the ligand.

View this table:
Table 1 Apparent molecular mass of Ob-R isomers expressed in mammalian cells

EXPERIMENTAL

Materials

All chemicals were of the highest analytical grade. BSA, 2-mercaptoethanol, Con A (concanavalin A), biotin-conjugated Protein G, tunicamycin, anti-β-actin mAb (monoclonal antibody), N-acetyl-D-glucosamine and α-methyl D-mannoside were obtained from Sigma Chemical Co. (St. Louis, MO, U.S.A.). Streptavidin–AP (alkaline phosphatase) and PNPP (p-nitrophenyl phosphate) were from Zymed Laboratories (South San Francisco, CA, U.S.A.). HRP (horseradish peroxidase)-conjugated goat anti-mouse IgG and HRP-conjugated goat anti-rabbit IgG were from Bio-Rad Laboratories (Hercules, CA, U.S.A.). Biotin-conjugated goat anti-mouse IgG was from MP Biomedicals (Solon, OH, U.S.A.). MAL II (Maackia amurensis lectin II) and biotinylated Con A were from Vector Laboratories (Burlingame, CA, U.S.A.). SNA (Sambucus nigra agglutinin), WGA (wheatgerm agglutinin), recombinant Protein G, CAS (castanospermine) and DNM (1-deoxynojirimycin) were from EMD Biosciences (La Jolla, CA, U.S.A.). RPMI 1640 medium, sodium pyruvate, L-glutamate, penicillin and streptomycin were from Gibco BRL (Invitrogen, Carlsbad, CA, U.S.A.). Fetal bovine serum was from Hyclone (Logan, UT, U.S.A.). A protease inhibitor mixture and endo H (endoglycosidase H) were from Roche Diagnostics (Indianapolis, IN, U.S.A.). Recombinant peptide N-glycosidase F (N-glycanase) and sialidase A were from Prozyme (San Leandro, CA, U.S.A.). A Coomassie Brilliant Blue G-250 dye solution (GelCode Blue Stain Reagent) was from Pierce (Rockford, IL, U.S.A.). PVDF membrane (Immobilon-P) and film (BioMax MR film) for Western blotting were from Millipore (Bedford, MA, U.S.A.) and Kodak (Rochester, NY, U.S.A.) respectively. Recombinant human leptin expressed in Escherichia coli was purchased from PeproTech (Rock Hill, NJ, U.S.A.); its concentration was determined by measuring the absorbance at 280 nm by using a molar absorption coefficient (ϵ) of 14060 M−1·cm−1. Leptin was biotinylated by using NHS (N-hydroxysuccinimide)–PEO4 (polyethylene oxide)–biotin (Pierce) according to the manufacturer's instructions. Biotin-conjugated WGA, SNA and MAL II lectins were also prepared by using NHS–PEO4–biotin. The anti-human Ob-R polyclonal antibody H-300, prepared by injecting into rabbits a fragment of Ob-R (residues 541–840) expressed in E. coli, and mAb B-3, directed against a peptide mapping at the C-terminus of Ob-Ra, were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, U.S.A.). Anti-V5 epitope and anti-His6 (hexahistidine) mAbs were purchased from Invitrogen. Human rsOb-R expressed in the mouse myeloma cell line NS0 was from R&D Systems (Minneapolis, MN, U.S.A.). In this chimaeric protein, residues 22–839 of the Ob-R extracellular domain are fused with the human IgG1 Fc domain and a C-terminal His6 tag, yielding a predicted mass of 120742 Da. The protein forms disulfide-linked homodimers. The concentration of rsOb-R was determined by using a BCA (bicinchoninic acid) assay kit (Pierce) and BSA as a protein standard.

Recombinant expression and purification of sOb-R fragments containing the CRH2 domain

We used Drosophila S2 (Schneider 2) cells to express two fragments of rsOb-R, residues 428–635 or 330–839. To express rsOb-R428–635, consisting only of the CRH2 domain with the leptin-binding site, a vector was constructed by first generating a PCR product by using Ob-Ra cDNA as a template and two primers (5′-ATAAGATCTATTGATGTCAATATCAATATC-3′ and 5′-TATGGGCCCATCATGACAACTGTGTAGGCTGG-3′). The product of this reaction was subcloned into pCR2.1 (Invitrogen), after which the DNA fragment was excised with the restriction enzymes BglII and ApaI and purified. The expression vector pMT/Bip/V5-His (Invitrogen) was digested with the same enzymes and, after purification, was ligated with the PCR fragment. The construct was verified by DNA sequence analysis (DNA Core Laboratory, Department of Molecular and Experimental Medicine, The Scripps Research Institute, La Jolla, CA, U.S.A.). The Ob-R fragment was expressed in S2 cells, stably transfected by using the calcium phosphate procedure (Drosophila expression kit from Invitrogen), as a fusion protein containing the V5 epitope and a His6 tag, yielding a predicted mass of 27186 Da. The cells were co-transfected with the vector pCoHygro (Invitrogen), which carries the gene for hygromycin B resistance, which allowed selection of stably transformed cells after 2–3 weeks of culture in a medium containing 300–500 μg/ml hygromycin B. S2 cells were maintained in a complete medium with 10% (v/v) heat-inactivated fetal bovine serum at a constant temperature of 23 °C. To induce expression, the cells were switched to a protein-free medium (insect XPRESS; Cambrex, East Rutherford, NJ, U.S.A.) and copper sulfate was added to a final concentration of 580 μM. After culture for 3 days at 23 °C, conditioned medium was harvested and used for purification of rsOb-R428–635. To achieve this, conditioned medium was applied to a Chelating-Sepharose Fast Flow column (2 cm×4 cm; Amersham Biosciences, Piscataway, NJ, U.S.A.) that was pre-equilibrated with 20 mM Hepes buffer (pH 7.4) containing 0.15 M NaCl [HBS (Hepes-buffered saline)]. After washing with HBS containing 10 mM imidazole, bound protein was eluted with HBS containing 50 mM imidazole and subsequently dialysed against HBS. The concentration of rsOb-R428–635 was determined by measuring the absorbance at 280 nm by using a molar absorption coefficient (ϵ) of 51380 M−1·cm−1. To express rsOb-R330–839 (containing the Ig, CRH2 and two FNIII domains), a plasmid vector was constructed by first generating a PCR product by using the Ob-Ra cDNA as a template and two primers (5′-ATAAGATCTACACAAGATGTCATATAC-3′ and 5′-TATGGGCCCATCACTCTGGTGTTTTTCAATATC-3′). This fragment was also expressed as a fusion protein containing the V5 epitope and a His6 tag, yielding a predicted mass of 62398 Da. The concentration of rsOb-R330–839 was determined by measuring the absorbance at 280 nm by using a molar absorption coefficient (ϵ) of 129105 M−1·cm−1.

Cell culture and preparation of cell lysate

Megakaryoblastic MEG-01 cells were cultured in RPMI 1640 medium supplemented with 10% (v/v) fetal bovine serum, 1 mM sodium pyruvate, 2 mM L-glutamate, 10 mM Hepes, 100 units/ml penicillin and 100 μg/ml streptomycin and maintained at 37 °C in a 5% CO2/air atmosphere. To inhibit N-linked glycosylation, MEG-01 cells [(1.6–4.0)×106] were cultured in the same medium containing tunicamycin (3 or 300 ng/ml), CAS (1 mM) or DNM (1 mM) for 24 h. To prepare the cell lysate, non-adherent cells in suspension were centrifuged at 350 g for 5 min at 4 °C and then washed twice by resuspension in ice-cold D-PBS [Dulbecco's PBS (Invitrogen, Carlsbad, CA, U.S.A.) without calcium and magnesium ions] followed by centrifugation (350 g for 5 min at 4 °C). After washing, the cells were incubated with a lysis buffer consisting of D-PBS containing 1% Triton X-100, 2 mM EDTA and a protease inhibitor mixture for 1 h at 4 °C. After incubation, the cell lysate was centrifuged at 10000 g for 10 min at 4 °C, and the resulting supernatant was subjected to Ob-R expression analysis. Protein concentration of the cell lysate was determined by the BCA protein assay.

Determination of the leptin-binding activity of Ob-R

The leptin-binding activity of rsOb-R, containing both Fc and His6 tags, was determined in solid-phase assays by using microtitre plates coated with either Protein G, which binds the Fc tag, or Ni2+ [Ni-NTA (Ni2+-nitrilotriacetate) HisSorb plate; Qiagen, Valencia, CA, U.S.A.], which binds the His6 tag. In one assay, 100 μl of Protein G [1 μg/ml in carbonate buffer (15 mM sodium carbonate and 35 mM sodium bicarbonate), pH 9.6] was adsorbed on the surface of microtitre-plate wells by overnight incubation at 4 °C. The wells were then blocked with 1% BSA for 1 h at 37 °C before adding various concentrations of rsOb-R diluted in 25 mM Tris/HCl (pH 7.4) containing 0.15 M NaCl, 0.1% Tween 20 and 0.1% BSA. After incubation for 1 h at 37 °C, the wells were washed four times with the above Tris buffer; then biotinylated leptin was added and the plates were incubated for 2 h at room temperature (25 °C). After this, the wells were washed again, and bound leptin was detected by using streptavidin–AP and a chromogenic substrate (PNPP) for AP. The concentration of bound leptin was determined kinetically by monitoring the change in the absorbance at 405 nm [mOD (milli-absorbance units/min)]. The data points were fitted to a hyperbola by using GraphPad Prism software 4.0 (GraphPad, San Diego, CA, U.S.A.), yielding a saturation binding curve from which the Kd (dissociation constant) of the interaction between rsOb-R and leptin could be calculated. In the other assay, rsOb-R was incubated for 1 h at 37 °C in microtitre wells coated with Ni2+. The wells were then washed four times with TBS-T [Tris-buffered saline-Tween 20; 25 mM Tris/HCl (pH 7.5) containing 0.1 M NaCl and 0.1% Tween 20] before adding biotinylated leptin diluted in TBS-T containing 0.1% BSA for 1 h at 37 °C. Bound leptin was determined by adding streptavidin–AP followed by PNPP, stopping the reaction with 0.75 M NaOH and reading the absorbance at 405 nm. To examine the effects of lectins with different carbohydrate-binding specificities, these were incubated at varying concentrations with rsOb-R for 1 h at 37 °C before measuring the binding of biotinylated leptin.

The leptin-binding activity of rsOb-R fragments was determined in a solid-phase assay using microtitre plates coated with anti-V5 tag mAb, which binds to a V5 epitope in the fragments. To prepare anti-V5-coated microtitre plates, 100 μl of the mAb (2 μg/ml in carbonate buffer, pH 9.6) was adsorbed on the surface of the plate wells by overnight incubation at 4 °C. The wells were then blocked with 1% BSA for 1 h at 37 °C, after which various concentrations of rsOb-R fragments, diluted in 25 mM Tris/HCl (pH 7.4) containing 0.15 M NaCl, 0.1% Tween 20 and 0.1% BSA, were added into the wells. After incubation for 1 h at room temperature, the wells were washed four times with Tris buffer, biotinylated leptin was added and the plates were incubated for 2 h at room temperature. Bound leptin was measured as described above.

The leptin-binding activity of Ob-R was evaluated also with a ‘pull-down’ assay using a leptin-conjugated gel prepared by coupling 400 μg of human leptin with 1 ml of CNBr-activated Sepharose 4 Fast Flow (Amersham Biosciences) according to the manufacturer's instructions. For the assay, 50 μl of the leptin-conjugated gel equilibrated with 10 mM PBS (pH 7.4) containing 0.1% Triton X-100 (PBS-TX) was mixed with 200 μl of rsOb-R (1 μg/ml in PBS-TX) for 1 h at 4 °C. After incubation, the mixture was centrifuged at 1300 g for 2 min at 4 °C, and the gel was washed with 1 ml of 10 mM sodium phosphate buffer (pH 7.4) containing 1 M NaCl and 0.1% Triton X-100, followed by 1 ml of PBS-TX. Unbound rsOb-R was recovered in the supernatant and kept at −20 °C until used; bound rsOb-R was eluted by resuspending the sedimented gel into 150 μl of loading buffer for SDS/PAGE (Bio-Rad) and boiling for 5 min. Unbound and eluted rsOb-R were analysed by SDS/PAGE followed by Western blotting with the polyclonal antibody H-300. To test the leptin-binding activity of Ob-R expressed in MEG-01 cells, cell lysate (100 μg of protein) was mixed with 10 μl of leptin-conjugated gel equilibrated with 50 mM Tris/HCl buffer (pH 7.4) containing 0.1 M NaCl, 0.1% Triton X-100, 2 mM EDTA and a protease inhibitor mixture. After 1 h at 4 °C, the mixture was centrifuged at 1300 g for 2 min at 4 °C, and the gel was washed with 100 μl of 50 mM Tris/HCl buffer (pH 7.4) containing 1 M NaCl, 0.1% Triton X-100 and 0.5 mM EDTA. After a final wash with 100 μl of 50 mM Tris/HCl buffer (pH 7.4) containing 0.1 M NaCl, 0.1% Triton X-100 and 0.5 mM EDTA, bound Ob-R was eluted from the gel, boiled with 30 μl of SDS/PAGE loading buffer for 5 min and then analysed by Western blotting with the anti-Ob-R mAb B-3. The amount of rsOb-R bound to leptin was expressed as a percentage of the total amount applied to the leptin-conjugated gel (the sum of bound and unbound rsOb-R), as determined by densitometric analysis of the corresponding bands after Western blotting (see below).

Binding of Protein G, lectins or anti-V5 mAb to rsOb-R

The receptor, diluted in TBS-T-BSA [TBS-T containing 0.1% BSA], was first immobilized by incubation in microtitre-plate wells coated with Ni2+ for 1 h at 37 °C. After four washes with TBS-T, 200 μl of biotin-conjugated Protein G (2 μg/ml in TBS-T-BSA), biotin-conjugated Con A (0.2 μg/ml in 10 mM Hepes buffer, pH 7.5, containing 0.15 M NaCl, 0.01 mM MnCl2, 0.1 mM CaCl2 and 0.1% BSA), biotin-conjugated WGA (2 μg/ml in TBS-T-BSA), biotin-conjugated SNA (1 μg/ml in TBS-T-BSA), or biotin-conjugated MAL II (20 μg/ml in TBS-T-BSA) were added into the wells and the plates were incubated for 1 h at 37 °C. Bound Protein G, Con A, WGA, SNA and MAL II were detected by using streptavidin–AP and PNPP. For binding of the anti-V5 mAb to rsOb-R fragments, 200 μl of the antibody (0.5 μg/ml in TBS-T-BSA) was added into the wells and the plates were incubated for 1 h at 37 °C. The bound anti-V5 mAb was detected by incubating with biotin-labelled anti-mouse IgG (1000-fold dilution in TBS-T-BSA) for 1 h at 37 °C, followed by incubation with streptavidin–AP and PNPP.

Digestion with peptide N-glycosidase F and endo H

N-linked glycans were removed under non-denaturing conditions by treatment with recombinant peptide N-glycosidase F (N-glycanase) or endo H. For N-glycanase digestion of rsOb-R, the enzyme was added at a 1:24 weight ratio in 22 mM PBS (pH 7.5) containing 0.07% Triton X-100, 1.8 mM EDTA and a protease inhibitor mixture, and incubated for up to 24 h at 37 °C. The deglycosylated rsOb-R was analysed by Western blotting as well as Con A- and leptin-binding assays. The molecular mass of N-glycanase-treated rsOb-R was determined by SDS/PAGE under reducing conditions, followed by protein staining with Coomassie Brilliant Blue G-250 dye solution (GelCode Blue Stain Reagent). The N-linked glycans of two rsOb-R fragments expressed in S2 cells were removed with the same method, but incubation of rsOb-R428–635 with Triton X-100 for 24 h at 37 °C resulted in drastic protein loss, presumably due to decreased solubility and/or protein aggregation (results not shown). Thus Triton X-100 was omitted from the reaction mixture with rsOb-R428–635. The deglycosylated rsOb-R fragments were analysed by Con A- and leptin-binding assays, as well as Western blotting using anti-V5 mAb. The N-linked glycans of Ob-R expressed in MEG-01 cells were removed by incubating cell lysates (30–150 μg of protein) with N-glycanase (0.5 μg) in 22 mM PBS (pH 7.5) containing 0.9% Triton X-100, 1.8 mM EDTA and a protease inhibitor mixture for 24 h at 37 °C. The deglycosylated MEG-01 Ob-R was analysed by Western blotting and leptin-binding assays. As a control, rsOb-R, rsOb-R fragments or cell lysates were incubated in the same buffer but without N-glycanase. For endo H digestion of Ob-R in MEG-01 cells, the enzyme (1 m-unit) was added to cell lysate (5 μg of protein) in 96 mM sodium acetate buffer (pH 5.7) containing 0.4% Triton X-100, 1.5 mM EDTA and a protease inhibitor mixture, and incubated for 24 h at 37 °C.

Sialidase digestion

Sialic acid residues were removed by incubating rsOb-R with sialidase A (1:2 weight ratio) in 56 mM PBS (pH 7.0) containing 0.07% Triton X-100, 1.8 mM EDTA and a protease inhibitor mixture, for 24 h at 37 °C.

Western blotting

Samples were boiled with loading buffer containing 4% (v/v) 2-mercaptoethanol for 5 min and then subjected to SDS/PAGE (5, 8, 14 or 4–20% gels) by the method of Laemmli [31]. Protein bands were transferred on to a PVDF membrane that was then blocked at 4 °C overnight with 5% (w/v) non-fat dried skimmed milk in TBS-T. The membrane was probed for 2 h at room temperature (25 °C) with anti-Ob-R polyclonal antibody H-300 (2 μg/ml), anti-His6 tag mAb (1 μg/ml), anti-V5 mAb (0.2 μg/ml) or anti-Ob-R mAb B-3 (1 μg/ml), followed by incubation with HRP-conjugated goat anti-rabbit IgG (2000-fold dilution) or HRP-conjugated goat anti-mouse IgG (2000-fold dilution) for 30 min. Immunoreactivity was detected with a SuperSignal West Pico Chemiluminescent kit (Pierce) by exposing the membrane to a Kodak BioMax MR film. The intensity of reactive bands was quantified by densitometry followed by image analysis using NIH (National Institutes of Health) Image software (http://rsb.info.nih.gov/nih-image/).

RESULTS AND DISCUSSION

Effect of lectins on leptin binding to rsOb-R

We first determined that Con A (which binds to α-linked mannose) and WGA (which binds to GlcNAc) [32,33] interact with rsOb-R expressed in mouse NS0 cells, and then established that the two lectins, at the highest concentration tested, inhibit leptin binding to the receptor by ∼50% (Figure 2A) or 70% (Figure 2C) respectively, with an effect specifically neutralized by α-methyl D-mannoside (Figure 2B) and N-acetyl-D-glucosamine (Figure 2D). The binding of WGA did not inhibit that of Con A (results not shown), and the two together caused more inhibition than either alone (Figure 2C). Thus the two lectins can interact concurrently with different N-linked glycans near the ligand-binding site and interfere independently with receptor function, possibly by causing conformational changes.

Figure 2 Effect of lectins on the leptin-binding activity of rsOb-R

Microtitre-plate wells coated with Ni2+ were used to immobilize rsOb-R (1 nM) through the His6 tag. This was followed by increasing concentrations of Con A (A) or WGA (C, □), as indicated, or by increasing concentrations of WGA with a constant amount Con A (6.3 μg/ml) of (C, ■); or by Con A (12.5 μg/ml) mixed with increasing concentrations of α-methyl D-mannoside (B); or by WGA (50 μg/ml) mixed with increasing concentrations of N-acetyl-D-glucosamine (D). After 1 h at 37 °C, biotin-conjugated leptin (1.6 nM) was added to the wells for an additional 1 h at 37 °C. Leptin bound to rsOb-R was detected as described in the Experimental section and expressed as a percentage of that bound in the absence of lectins. Results are the average of two separate experiments for each experimental point tested.

Effect of removing N-linked glycans on leptin binding to rsOb-R

Treatment with N-glycanase, which releases N-linked glycans from glycoproteins by cleaving the link between GlcNAc and asparagine residues, progressively reduced the molecular mass of rsOb-R. As visualized by immunoblotting, after 24 h the original band of 170 kDa was converted into a broad band of ∼130 kDa (Figure 3A), close to the calculated rsOb-R polypeptide mass (120742 Da). Corresponding bands and no smaller fragments could be seen after SDS/PAGE of untreated as well as N-glycanase-treated rsOb-R samples (Figure 3B), suggesting that the reduction in molecular mass was not caused by proteolysis. Treatment with N-glycanase for 9 or 24 h reduced leptin binding to rsOb-R by ∼40% and ∼80% (Figure 3C) and concurrently reduced Con A binding by ∼50% and ∼90% (Figure 3C) respectively. These results support the conclusion that the 130 kDa band represents deglycosylated rsOb-R lacking ∼40 kDa of N-linked glycans that influence leptin binding. In contrast with decreased leptin and Con A binding, Protein G binding to rsOb-R was increased after N-glycanase treatment (Figure 3C), probably as a consequence of N-linked glycan removal from the Fc tag.

Figure 3 N-glycanase digestion of rsOb-R

(A) Representative Western blot with an anti-His6 tag mAb of rsOb-R incubated with (+) or without (−) N-glycanase for varying periods of time at 37 °C and then subjected to SDS/4–20% PAGE under reducing conditions. (B) Protein staining with Coomassie Brilliant Blue G-250 of N-glycanase-treated rsOb-R after SDS/4–20% PAGE under reducing conditions. Lane 1, control rsOb-R incubated without N-glycanase; lane 2, rsOb-R incubated with N-glycanase for 24 h at 37 °C; lane 3, prestained molecular-mass standards. (C) Binding of leptin (1.6 nM), Con A (0.2 μg/ml) or Protein G (2 μg/ml) to rsOb-R incubated with N-glycanase for the indicated time periods and then immobilized on to microtitre plates coated with Ni2+. Results are means±S.D. for three independent experiments.

In additional experiments using a solid-phase assay, we confirmed that untreated Ob-R bound leptin in a dose-dependent manner, but the N-glycanase-treated receptor did not (Figure 4A). Moreover, mixing of varying concentrations of N-glycanase-treated Ob-R with leptin before addition to untreated rsOb-R immobilized on to Protein G-coated microtitre plates caused only a slight inhibition of leptin binding (Figure 4B), indicating that the released N-linked glycans have no significant direct interaction with leptin. Finally, addition of increasing concentrations of leptin to control or N-glycanase-treated immobilized rsOb-R resulted in a saturable binding to the former, but essentially no binding to the latter (Figure 4C). The Kd of leptin binding to untreated rsOb-R, as calculated from three independently performed saturation curves, was 1.80±0.37 (S.D.) nM, comparable with previously reported Kd values of ∼1 nM [2123]. The decreased binding function of N-glycanase-treated rsOb-R was confirmed with a ‘pull-down’ assay using leptin-conjugated gels. In this case, ∼80% of control rsOb-R bound to the leptin gel, but only ∼20% of N-glycanase-treated rsOb-R did. Altogether, these results indicate that N-linked glycans in rsOb-R account for ∼40 kDa of the mass and play a role in leptin binding.

Figure 4 Effect of removal of N-linked glycans on the leptin-binding activity of rsOb-R

(A) Protein G coated on microtitre plates was used to immobilize rsOb-R (○) or N-glycanase-treated rsOb-R (24 h incubation; ●) added at the indicated concentrations. After incubation and washing, biotin-conjugated leptin (5 nM) was added and incubated for 2 h at room temperature. Results are the average binding observed in two separate experiments. (B) RsOb-R (0.8 nM) was immobilized on to microtitre plates coated with Protein G, followed by varying concentrations of N-glycanase-treated rsOb-R mixed with biotin-conjugated leptin. Binding of the latter was expressed as a percentage of that measured in the absence of N-glycanase-treated rsOb-R. Results are the average normalized binding observed in two separate experiments. (C) RsOb-R (○) or N-glycanase-treated rsOb-R (●), each used at 0.8 nM, was immobilized on to microtitre plates coated with Protein G, after which biotinylated leptin was added at the indicated concentrations for 2 h at room temperature. Data were fitted to a hyperbola (corresponding to a one-site binding model) by using GraphPad Prism. The Figure illustrates a saturation binding curve with rsOb-R, and the lack of binding to N-glycanase-treated rsOb-R.

Sialic acids in rsOb-R are not necessary for leptin binding

Sialic acids are found at the outermost end of N- and O-linked glycans. In some cases, distinct sialic acids are recognized by cell membrane receptors (e.g. sialyl-LewisX by selectins). Binding assays demonstrated that rsOb-R reacts with SNA, which binds to α-2,6-linked sialic acid, but not with MAL II, which binds to α-2,3-linked sialic acid, indicating that rsOb-R includes sialic acids bound in α-2,6 linkage (Figure 5A). Treatment of rsOb-R with sialidase A, which cleaves all non-reducing terminal sialic acids and branched sialic acids, abolished SNA binding by 3 h, whereas leptin binding was not decreased even after 24 h (Figure 5B). Thus sialic acids attached to rsOb-R are not required for leptin binding.

Figure 5 Effect of sialic acid removal on leptin or α-linked sialic acid-binding lectin to rsOb-R

(A) Binding of biotin-conjugated lectins to rsOb-R. After rsOb-R (1 nM) was immobilized on to microtitre-plate wells coated with Ni2+, biotin-conjugated SNA (an α-2,6-linked sialic acid-binding lectin; 1 μg/ml), MAL II (an α-2,3-linked sialic acid-binding lectin; 20 μg/ml), Con A (0.2 μg/ml) or WGA (2 μg/ml) were added into the wells and the plates were incubated for 1 h at 37 °C. Bound biotin-conjugated lectins were detected using streptavidin–AP and PNPP. Results are the average of two separate experiments. OD405, absorbance at 405 nm. (B) RsOb-R (1 nM) treated with sialidase for the indicated time periods was immobilized on to microtitre-plate wells coated with Ni2+ before adding SNA (1 μg/ml) or leptin (1.6 nM). Black bars, SNA binding; white bars, leptin binding. Results are the average of two separate experiments. Results shown are binding as a percentage of that measured with untreated rsOb-R added to the wells at the same concentration.

Characterization of Ob-R expressed in MEG-01 cells

Previous studies have shown that human megakaryoblastic MEG-01 cells contain mRNA for both short and long Ob-R isoforms [4] and bind leptin with a Kd of ∼4 nM [12], but the expressed receptor has not been fully characterized. By Western blotting, we identified two Ob-R species in MEG-01 cell lysate of approx. 170 and 130 kDa (Figure 6A, lane 1). The former was resistant to treatment with endo H, which hydrolyses high-mannose (including Man4–9GlcNAc2) and hybrid-type N-linked glycans [34], whereas the latter was converted into a 100 kDa species (Figure 6A, lane 2), consistent with the calculated polypeptide mass of Ob-Ra (102490 Da). This indicates that the 130 kDa Ob-R of MEG-01 cells is a short form of the receptor containing high-mannose and/or hybrid-type N-linked glycans, whereas the 170 kDa species contains mature N-linked glycans of the complex type resistant to endo H digestion. We then examined the effect of treating the cells with DNM, which blocks trimming of N-linked glycans by α-glucosidases I and II after attachment of the precursor oligosaccharide (Glc3Man9GlcNAc2) to the nascent polypeptide [35]. Glycoproteins synthesized under these conditions lack the terminal sugar moieties of mature carbohydrate chains [36] and may be misfolded, since the presence of non-trimmed N-linked glycans prevents association with the chaperone, calnexin, which contributes to folding in the endoplasmic reticulum [37]. MEG-01 cells treated with DNM for 24 h exhibited an increased expression of 130 kDa Ob-R, with no effect on the 170 kDa species (Figure 6A, lane 3). A second α-glucosidase inhibitor, CAS, gave comparable results (results not shown). Endo H digestion of DNM-treated cell lysate converted the 130 kDa Ob-R form into the 100 kDa species (Figure 6A, lane 4), in agreement with the conclusion that the former contains only N-linked glycans of the mannosidic type. Since our results demonstrate that the 130 kDa Ob-R form has full leptin-binding activity (see below), it is apparent that processing of mannosidic-type glycans to allow association with calnexin is not essential to achieve functionality of the receptor. The fact that cells treated with α-glucosidase inhibitors continued to synthesize the 170 kDa Ob-R, which contains mature N-linked glycans resistant to endo H digestion, may be explained by the activity of an endo α-mannosidase directly acting on N-linked glycan precursors [38]. On the other hand, culture of MEG-01 cells for 24 h in a medium containing 300 ng/ml tunicamycin, an inhibitor of the first step in the formation of N-glycosidic protein–carbohydrate linkages [35], resulted in a marked reduction of both the 170 and 130 kDa Ob-R species (Figure 6B). Altogether, these results indicate that N-linked glycans are required for normal Ob-R expression in MEG-01 cells, but may be effective regardless of whether the mature or mannosidic type is present.

Figure 6 Characterization of Ob-R expressed in MEG-01 cells

(A) The lysate of cells cultured in either the absence (control, lanes 1 and 2) or presence (DNM-treated, lanes 3 and 4) of the α-glucosidase inhibitor, DNM, and then treated (+) or not (−) with endo H, was subjected to SDS/8% PAGE under reducing conditions. Ob-R in the lysate was detected by Western blotting (WB) using the anti-Ob-R mAb B-3. Note the presence of two Ob-R isoforms with a molecular mass of 170 and 130 kDa respectively. Endo H digestion had no effect on the 170 kDa isoform, whereas the 130 kDa isoform was converted into an ∼100 kDa species. (B) Western-blot analysis (after SDS/4–20% PAGE under reducing conditions) of lysates obtained from cells cultured in the presence of two different concentrations of tunicamycin compared with the lysate from untreated cells. The 170 and 130 kDa Ob-R isoforms were barely detectable in cells cultured with 300 ng/ml tunicamycin, whereas expression of β-actin (bottom panel) was not altered, indicating the specificity of the effect. Positions of molecular-mass markers are shown on the left (sizes in kDa).

Effect of removal of N-linked glycans on the leptin-binding activity of Ob-R expressed in MEG-01 cells

Next, we examined the contribution of N-linked glycans to the function of Ob-R expressed in MEG-01 cells. Both the 170 and 130 kDa Ob-R isoforms bound to a leptin-conjugated gel (Figure 7A, lanes 1 and 2) and both, in particular the 130 kDa isoform, bound also when derived from cells treated with DNM or CAS (Figure 7A, lanes 3–6), suggesting that trimming of N-linked glycans in Ob-R has no major influence on leptin binding. On treatment of cell lysates with N-glycanase for 24 h at 37 °C, both the 170 and 130 kDa species were reduced to 100 kDa, and the latter exhibited a greatly reduced ability to interact with leptin (Figure 7B, lane 4). This indicates that the 170 and 130 kDa species are both short Ob-R isoforms (Ob-Ra) containing 40 and 23% carbohydrate by mass, either of complex or high-mannose structure respectively. The two Ob-R species expressed by MEG-01 cells, therefore, appear to be the product of the same mRNA, but differ in their post-translational carbohydrate processing. Of note, binding of the 130 kDa isoform to leptin was considerably lower after incubation of cell lysates for 24 h at 37 °C (Figure 7B, lane 2) compared with cell lysate treated in the same manner but not incubated (Figure 7A, lane 2), suggesting that the short Ob-Ra form containing high-mannose type N-linked glycans may be thermodynamically labile in a buffer containing Triton X-100. Altogether, these results provide further evidence that carbohydrate chains, whether of a high-mannose or complex type, play a role in maintaining the ligand-binding function of Ob-Ra. Such a concept should apply to all Ob-R isoforms, since all have the same extracytoplasmic structure.

Figure 7 Effect of removal of N-linked glycans on leptin binding to Ob-R expressed in MEG-01 cells

Leptin binding to Ob-R in a cell lysate was evaluated using the ‘pull-down’ assay with leptin-conjugated gel followed by Western-blot analysis of the bound species using the anti-Ob-R mAb B-3. (A) Ob-R in the total cell lysates (lanes 1, 3 and 5) compared with leptin-bound Ob-R (lanes 2, 4 and 6). Cells were cultured either in the absence (control) or in the presence of DNM or CAS, as indicated. (B) Ob-R in the total cell lysate (lanes 1 and 3) compared with leptin-bound Ob-R (lanes 2 and 4). The cell lysate used was either left untreated or pre-incubated with N-glycanase for 24 h at 37 °C, as indicated. mOD/min, milli-absorbance units/min.

In spite of the proven ability to bind leptin, the contribution of the two MEG-01-expressed Ob-Ra isoforms to cellular activities remains to be defined. Only the long Ob-Rb form, with its full-length 302-residue intracytoplasmic domain, is believed to be capable of leptin-induced signalling through activation of JAK2 and STAT3 (signal transducer and activator of transcription 3). The short isoforms of Ob-R, including Ob-Ra, are capable of activating, albeit weakly, gene expression and signal transduction in cultured cells through different pathways, but whether they have any function in vivo remains to be established. Nakata et al. [39] previously demonstrated that leptin potentiates the increase in cytosolic Ca2+ induced by ADP stimulation of MEG-01 cells, a function that according to these authors could be explained by the expression in the same cells of the Ob-Rb. Our studies, however, have identified only two forms of Ob-Ra in MEG-01 cell lysates, differing in carbohydrate content. Biotinylation of membrane surface proteins by using NHS–PEO4–biotin confirmed that two forms of Ob-Ra are expressed on MEG-01 cells (results not shown). Thus any functional effect of leptin in these cells should be mediated by the short Ob-R isoforms. It remains to be established whether the same conclusion can be extended to platelets, in which leptin has been shown to potentiate ADP-induced aggregation [1012], but whose repertoire of Ob-R isoforms has not been definitively established.

Role of N-linked glycans in leptin binding to rsOb-R expressed in Drosophila S2 cells

To determine which domain(s) contain the N-linked glycosylation site(s) required for leptin binding, we evaluated the function of two rsOb-R fragments expressed in Drosophila S2 cells. Addition of increasing concentrations of leptin to immobilized rsOb-R330–839 (comprising Ig, CRH2 and two FNIII domains, with a total of 11 potentially glycosylated asparagine residues) resulted in a saturable interaction (Figure 8A). The corresponding Kd calculated from three separate binding isotherms was 3.51±0.52 (S.D.) nM. Treatment of the fragment with N-glycanase for 24 h reduced the molecular mass from 70 to 60 kDa (Figure 8B) and abolished binding of both Con A and leptin (Figure 8C). Thus the insect-cell-expressed Ob-R fragment comprising extracytoplasmic residues 330–839 contains ∼10 kDa of carbohydrate by mass that is required for leptin binding. N-linked glycans in insect cells are of the paucimannosidic type, with chains that include one to three mannose residues either with or without a fucose residue [40,41], thus considerably different from the mature glycans of mammalian cells. Nonetheless, as shown by the similar Kd values measured for leptin binding to mammalian- or insect-cell-derived Ob-R fragments (1.80±0.37 and 3.51±0.52 nM respectively), either type can support receptor function. These results, therefore, validate the use of recombinant Ob-R obtained in insect cells to evaluate structure–function relationships.

Figure 8 Effect of N-linked glycan removal on leptin binding to rsOb-R fragments expressed in S2 cells

(A) The rsOb-R fragments (rsOb-R428–635, ●; rsOb-R330–839, ○; each at 0.5 nM) were immobilized on to microtitre plates coated with anti-V5 mAb, which binds to the V5 tag in the fragments. Biotinylated leptin was added at the indicated concentrations, and the plates were incubated for 2 h at room temperature. Bound leptin was detected using streptavidin–AP and PNPP. mOD/min, milli-absorbance units/min. Data were fitted to a hyperbola (corresponding to a one-site binding model) by using GraphPad Prism. (B) Western-blot (WB) analysis with anti-V5 mAb of rsOb-R fragments incubated with or without N-glycanase for 24 h at 37 °C and then subjected to SDS/4–20% PAGE (for rsOb-R330–839) or SDS/14% PAGE (for rsOb-R428–635) under non-reducing conditions. Lanes 1 and 2, rsOb-R330–839 incubated without or with N-glycanase respectively; lanes 3 and 4, rsOb-R428–635 incubated without or with N-glycanase respectively (note, in each case, the small amounts of what appear to be dimeric fragments). Positions of molecular-mass markers are shown on the left (sizes of kDa). (C) Untreated and N-glycanase-treated rsOb-R fragments were immobilized on to microtitre-plate wells coated with Ni2+ for measuring the binding of Con A and anti-V5 mAb. In the coating solutions, rsOb-R428–635 was at a concentration of 10 nM (Con A binding) and 2 nM (anti-V5 mAb binding); the corresponding concentrations of rsOb-R330–839 were 1 and 0.25 nM. Biotinylated Con A was added at 0.2 μg/ml; anti-V5 mAb was added at 0.5 μg/ml, and this was followed, after incubation and washing of the wells, by a biotinylated anti-mouse IgG. Bound proteins were detected by adding to the wells streptavidin–AP, followed by PNPP. To measure leptin binding, untreated and N-glycanase-treated rsOb-R fragments (each at 1 nM) were immobilized on to microtitre plates coated with anti-V5 mAb. Biotinylated leptin was added at 3.1 nM, and bound leptin was detected using streptavidin–AP and PNPP. Binding to N-glycanase-treated rsOb-R fragments is shown as a percentage of that to each untreated fragment under the same conditions. Results are the average of two separate experiments. The binding of anti-V5 mAb provides a measure of the presence of the fragments in the wells. Note that removal of N-linked glycans may decrease solubility of the fragments, particularly rsOb-R330–839, as suggested by the decreased binding of anti-V5 mAb to ∼50% of the control.

We then examined whether the three potentially glycosylated asparagine residues in the Ob-R CRH2 domain, which contains the ligand interaction site (Figure 1), directly contribute to leptin binding. To this end, we expressed the domain (rsOb-R428–635) in S2 cells. Addition of increasing leptin concentrations to the fragment immobilized on to microtitre plates resulted in a saturable interaction (Figure 8A). The corresponding Kd, calculated from three independently performed binding isotherms, was 2.52±0.87 (S.D.) nM. Treatment of the fragment with N-glycanase slightly reduced the molecular mass (Figure 8B) and abolished Con A binding, but had no effect on leptin binding (Figure 8C), in agreement with the knowledge that non-glycosylated CRH2 expressed in bacteria can bind leptin [22]. The contrasting results obtained with the Ob-R fragment consisting only of the CRH2 domain, as opposed to that containing also Ig and FNIII domains, led to the conclusion that N-linked glycans are not directly involved in ligand interactions. Rather, it appears that carbohydrate chains in the domains flanking the CRH2 domain (Figure 1) influence receptor function indirectly. Preliminary results indicate that N-linked glycans in both Ig and FNIII domains are involved in modulating Ob-R activity, but more experimental work will be required to define with precision which among eight potential glycosylation sites are important, as well as the structural bases through which carbohydrate can influence leptin binding.

Proposed mechanism for the role of N-linked glycans in leptin binding to its receptor

Several studies on class I cytokine receptors have provided variable results concerning the role of N-linked glycans in ligand binding (Table 2). For example, mutagenesis study on the β-subunit of the GM-CSF (granulocyte/macrophage colony-stimulating factor) receptor revealed that one N-linked glycosylation site within the CRH domain and two other N-linked glycosylation sites in its close proximity were required for the formation of a high-affinity complex with the cytokine [27]. In contrast, N-linked glycans in the CRH domains of other cytokines receptors, including the IL-6 (interleukin-6) receptor, the signal-transducing receptor [gp130 (glucoprotein 130)] and the prolactin receptor, were shown not to be essential for ligand binding [2830]. Our present results provide evidence for an indirect contribution of N-linked glycans to Ob-R function, which can only be appreciated in the context of the intact receptor or larger fragments, since glycans within the ligand-binding domain itself appear to have no effect. Thus carbohydrate may have a structural role; for example, it may be required to maintain an active conformation of the ligand-interaction site in Ob-R. Accordingly, it is known that N-linked glycans may make the tertiary structure of a folded protein more thermodynamically stable than that of the corresponding non-glycosylated form [42]. Three-dimensional structural analyses have documented that core GlcNAc and mannose residues in carbohydrate side chains can interact directly with residues in the polypeptide backbone, thus decreasing flexibility and stabilizing protein structure [43,44]. The fact that different Ob-R isoforms exhibit similar leptin-binding activity regardless of high-mannose or complex-type carbohydrate structure supports the concept that N-linked glycans may exert their structural influence through interactions between core glycans and specific polypeptide residues. Future studies will directly address these questions.

View this table:
Table 2 N-linked glycan contribution to ligand binding in class I cytokine receptors

Acknowledgments

We thank James C. Paulson (The Scripps Research Institute) for helpful discussions and a critical reading of this paper. This work was supported by grant numbers HL-75736 and HL42846 from the National Institutes of Health (Bethesda, MD, U.S.A.).

Abbreviations: AP, alkaline phosphatase; BCA, bicinchoninic acid; CAS, castanospermine; CHO cell, Chinese-hamster ovary cell; Con A, concanavalin A; CRH, cytokine receptor homology; CRH1, first CRH; CRH2, second CRH; DNM, 1-deoxynojirimycin; D-PBS, Dulbecco's PBS without calcium and magnesium ions; endo H, endoglycosidase H; FNIII, fibronectin type III; GM-CSF, granulocyte/macrophage colony-stimulating factor; HBS, Hepes-buffered saline; His6, hexahistidine; HRP, horseradish peroxidase; Ig, immunoglobulin; IL, interleukin; JAK2, Janus kinase 2; mAb, monoclonal antibody; MAL II, Maackia amurensis lectin II; NHS, N-hydroxysuccinimide; PBS-TX, PBS with Triton X-100; PEO4, polyethylene oxide; Ob-R, leptin receptor; PNPP, p-nitrophenyl phosphate; sOb-R, soluble Ob-R; rsOb-R, recombinant sOb-R; S2, Schneider 2; SNA, Sambucus nigra agglutinin; TBS-T, Tris-buffered saline with Tween 20; WGA, wheatgerm agglutinin

References

View Abstract