Lifr Antagonists

Assembly of LIF-R into Receptor Complexes

LIF-R binds several gp130 cytokines as a shared signal-transducing subunit. LIF, CNTF, and CT-1 all require LIF-R for signaling, and in some human cell types, OSM also utilizes LIF-R. All currently known LIF-R ligands signal by heterodimerizing LIF-R and gp130. In the simplest example, LIF forms a complex with LIF-R and gp130; the formation of complexes in solution suggests that the complexes are heterotrimers of LIF, LIF-R, and gp130 in a ratio of 1: 1: 1 (87). In a more complex example, CNTF must bind CNTFRa before binding to LIF-R and gp130 to assemble a hexameric complex (88).

gp130 Cytokines Bind to LIF-R Through Residues in Site 3

hLIF first binds to hLIF-R with low affinity (KD = 1 nM (89,90)) and then associates with gp130 in a higher affinity complex (KD = 10 pM (91)). Based on the crystalline structure of murine LIF, residues likely to be surface exposed on hLIF were subjected to alanine scanning mutagenesis to identify receptor-binding sites. Specific regions were chosen for analysis based on the site usage of GH and on previous analyses of mouse-human chimeras that had indicated a role for the C-D loop (92) and the D-helix (46) in species-specific binding to LIF-R. The alanine scanning mutagenesis demonstrated that the primary binding of LIF-R is mediated via site 3 of hLIF, including residues at the amino-terminus of the D-helix, carboxyl-terminus of the B-helix, and C-D loop. Site 3 is dominated by residues Phe-156 and Lys-159 at the top of the D-helix (93). These residues are spatially close together and have their side chains prominently exposed to the solvent in the mLIF structure (46). Substitution of either amino acid resulted in more than a 100-fold reduction in LIF-R binding and a substantial reduction in biological activity, measured in a proliferation assay of Ba/F3 cells stably transfected with hLIF-R and hgp130. The importance of Phe-156 and Lys-159 in site 3 is underlined by their conservation in all known ligands for LIF-R, including mLIF, hOSM, CNTF, and CT-1 (46,94). IL-6 and IL-11, cytokines that use site 3 but do not bind LIF-R, do not have phenylalanine and lysine in these positions.

In contrast to the profound effects of changes in site 3, there is evidence for at best weak binding of hLIF to LIF-R via site 1 at the carboxyl-terminus of the D-helix (93). A cocrystalline structure for LIF and LIF-R would be helpful in determining if LIF-R occupies site 1 on cytokines that do not bind ligand-specific receptors.

Creation of a LIF-R Antagonist by Disruption of Site 2 of hLIF

As the affinity of hLIF for gp130 is probably 100- to 1000-fold lower than the affinity for LIF-R, multiple simultaneous mutations were used in place of single mutations to identify the gp130-binding site (93). Both alanine substitutions and nonconservative substitutions in the beginning of the A- and C-helices (site 2) reduced gp130 binding to undetectable levels. Perhaps because of limitations of the gp130 binding assay, mutants that showed equal losses in gp130 binding showed varying amounts of biological activity in stimulating the proliferation of Ba/F3-hLIF-R/hgp130 cells. Activities ranged from a 250fold reduction in proliferation activity to a complete absence of agonist activity. In all cases, LIF-R binding was equivalent to wt hLIF (93). The preservation of LIF-R binding confirms the independence of sites involved in LIF-R and gp130 binding.

Several of the site 2 mutants were able to antagonize hLIF in proliferation assays on Ba/F3-hLIF-R/hgp130 cells. An hLIF site 2 mutant with no detectable agonist activity, hLIF-05, was chosen for further characterization (95). hLIF-05 contains the nonconservative mutations A117E, D120R, I121K, G124N, S127L, Q25L, S28E, Q32A, S36K (for the placement of the residues in the LIF structure, see Fig. 5). The antagonist is modestly potent on cells that are highly sensitive to LIF, requiring about 15 nM or a 10,000-fold excess over wt hLIF to inhibit by 50% the proliferation Ba/F3-hLIF-R/hgp130 cells. However, on HepG2 cells, a ratio of only 200-fold hLIF-05 to wt LIF was sufficient for 50% inhibition of hLIF (95). The modest potency is not surprising given that hLIF-05, which only binds LIF-R, must compete with wt LIF, which also assembles high-affinity LIF-R/gp130 complexes. Improving the potency of hLIF-05 by strengthening its binding to LIF-R has not yet been attempted.

As predicted for an antagonist that sequesters LIF-R, hLIF-05 inhibited the biological response to not just LIF, but to all known LIF-R ligands, including LIF, CNTF, CT-1, and OSM (95). hLIF-05 is best thought of as a broad

Figure 5 Schematic representation of hLIF, modeled on the mLIF crystalline structure, showing the distribution of substitutions in hLIF-05 (93). The substitutions in site 2 reduce binding to gp130. In the A-helix, the mutations include Q25L, S28E, Q32A, and S36K. In the C-helix, they include A117E, D120R, I121K, G124N, and S127L.

Site 3

Figure 5 Schematic representation of hLIF, modeled on the mLIF crystalline structure, showing the distribution of substitutions in hLIF-05 (93). The substitutions in site 2 reduce binding to gp130. In the A-helix, the mutations include Q25L, S28E, Q32A, and S36K. In the C-helix, they include A117E, D120R, I121K, G124N, and S127L.

LIF-R antagonist and not as an antagonist particular to hLIF. hLIF-05 was shown to be a competitive antagonist: An excess of wt ligand overcomes the antagonism by hLIF-05. The antagonism is specific, since hLIF-05 did not inhibit IL-6Ra-mediated responses and hLIF-05-inhibited hOSM responses only when they were mediated by LIF-R and not when they were mediated by the related OSM-R (95). The ability to discriminate between LIF-R and OSM-R may be useful in dissecting the behavior of hOSM in pathologies in which hOSM is elevated, such as rheumatoid arthritis (96).

To examine events closer to ligand binding, receptor phosphorylation was measured on cells that make all three receptor subunits, LIF-R, OSM-R, and gp130. Tyrosine phosphorylation of LIF-R and gp130 in response to hLIF was blocked by hLIF-05. The phosphorylation of LIF-R in response to OSM was also blocked. However, the phosphorylation of gp130 was not blocked by hLIF-05 in response to OSM, indicating that OSM-R was able to associate with gp130 in the presence of hLIF-05 (95). The inhibition of receptor phos-phorylation argues that hLIF-05 is acting at an early step in signal transduction and is consistent with the simple model that hLIF-05 works by sequestering LIF-R.


In a reversal of the normal family pattern, hOSM binds to gp130 with fairly high affinity (10 nM) but only weakly binds hLIF-R (93) and does not significantly bind hOSM-R (16). Binding of mOSM to the recently cloned mOSMR has not yet been quantified (18). Reducing the binding of hOSM to gp130 is unlikely to generate an effective antagonist for LIF-R or OSM-R, since the remaining binding to these subunits is so weak. Reducing the binding of OSM to LIF-R and OSM-R should generate a gp130 antagonist and not an antagonist of LIF-R and OSM-R. Consequently, although the placement of receptor binding sites on hOSM resembles the placement of binding sites on LIF (K.R. Hudson and J.K. Heath, personal communication, 1998) (56), one cannot make antagonists in the same fashion.


Assembly of CNTF Receptor Complexes

CNTF assembles receptor complexes by first binding to CNTFRa and then heterodimerizing gp130 and LIF-R (97). CNTF can activate LIF-R and gp130

without its ligand-specific receptor; however, formation of a CNTF-CNTFRa complex increases the ability of CNTF to activate LIF-R and gp130 by 10,000fold (98). In vivo, CNTF concentrations are likely to be too low to activate LIF-R and gp130 in the absence of CNTFRa. Hence neutralizing CNTFRa is probably sufficient to block CNTF responses in vivo.

Three receptor binding sites have been identified on CNTF. CNTF binds to CNTFRa with moderate affinity via site 1 on the A-B loop and the D-helix (99-102). CNTF binds to gp130 via residues in site 2 on the A-helix (99) and binds to LIF-R via site 3 at the boundary region of the C-D loop and the D-helix (103,104). A hexameric complex of two CNTF, two CNTFRa, one gp130, and one LIF-R has been proposed based on studies of the complex formation of the extracellular domains in solution (88).

Creation of CNTFRa Antagonist by Disruption of Site 3

As in the case of IL-6Ra antagonists, CNTFRa antagonists have been progressively refined. Site 3 was the first site to be targeted to develop CNTFRa antagonists. Inoue et al. substituted K155 with alanine in site 3 of hCNTF, abolishing the survival activity of hCNTF on chick dorsal root ganglion (DRG) neurons (103). The absence of agonist activity on chick DRG neurons could be demonstrated at concentrations as high as 20 |g/mL. On chick DRG neurons, K155A was a modestly potent antagonist, requiring 1000- to 10,000-fold excess to inhibit wt CNTF. However, on rat DRG neurons, K155A was a partial agonist (101). Introducing a bulkier substitution in the form of a tryptophan instead of an alanine (K155W) resulted in a lowering of the agonistic activity on rat cells but did not create an antagonist for rat cells. Therefore, to obtain an antagonist for rat cells, further modifications were required. The simultaneous substitution of both F152 and K155 was subsequently shown to broaden the species specificity. F152S/K155A and F152D/K155A antagonized CNTF on both chick and rat cells (101) (for the placement of the residues on the CNTF structure, see Fig. 6). The species specificity of the early CNTFRa antagonists must stem from subtle differences in how site 3 is engaged in the receptor complexes of the two species that are not apparent with wt hCNTF.

In a similar manner, Di Marco et al. enhanced the performance of the CNTF mutant, K155A, by simultaneously substituting F152 with alanine (104). Although a partial reduction in agonist activity occurred with the individual mutations, the combined mutations exhibited a complete absence of agonist activity on HepG2 cells. The reduction in agonist activity by F152A/ K155A occurred without changes in the binding to CNTFRa or in the binding of the F152A/K155A-CNTFRa complex to gp130. In contrast, LIF-R binding

Figure 6 Schematic representation of the hCNTF crystalline structure, showing the distribution of substitutions in the CNTFRa antagonist developed by Di Marco et al. (104). Substitutions in site 1 that enhance binding to CNTFRa include S166D and Q167H in the D-helix. Substitutions in site 3 that reduce binding to LIF-R include F152A and K155A at the N-terminus of the D-helix.

Figure 6 Schematic representation of the hCNTF crystalline structure, showing the distribution of substitutions in the CNTFRa antagonist developed by Di Marco et al. (104). Substitutions in site 1 that enhance binding to CNTFRa include S166D and Q167H in the D-helix. Substitutions in site 3 that reduce binding to LIF-R include F152A and K155A at the N-terminus of the D-helix.

by F152A/K155A was completely lost, indicating that site 3 of CNTF is a crucial site for LIF-R binding (104). The residues F152 and K155 in CNTF are homologous to the residues F156 and K159 that are crucial for the binding of LIF to LIF-R. The use of homologous residues to bind LIF-R suggests conservation in the way LIF-R ligands engage LIF-R. F152A/K155A was a moderately potent antagonist in assays on human IMR32 cells that express endogenous transmembrane CNTFRa (about 3000-fold excess required) but had no antagonistic activity on HepG2 cells when copresented with soluble CNTFRa (104). The preferential activity of the antagonist on cell lines that makes transmembrane receptor probably reflects the undirectional capture of CNTF by membrane-bound CNTFRa, which makes the formation of receptor complexes relatively insensitive to changes in CNTF affinity for LIF-R and gp130. The lack of sensitivity to changes in the affinity for LIF-R enables the antagonist to compete effectively with wt CNTF for CNTFRa.

When the mutations in site 3 were combined with mutations in site 1 (S166D/Q167H) that enhance the affinity of CNTF for CNTFRa by 30- to 50-fold, the enhanced antagonist was effective on HepG2 cells when copre-sented with soluble CNTFRa (104) (see Fig. 6). Combining site 1 and site 3 mutations also greatly improved the potency of the CNTFRa antagonist; CNTF responses were inhibited by 50% at a ratio of 30- to 100-fold excess over wt CNTF on IMR32 cells (and HepG2 cells) compared to the 3000-fold excess observed previously. On HepG2 cells, the EC50 of CNTF was shifted in the presence of the antagonist up to the value observed in the absence of CNTFRa, suggesting that the antagonist worked simply by competing for CNTFRa binding (104).

The observation that site 3 antagonists for CNTFRa behaved differently in assays dependent on membrane-bound CNTFRa than in assays dependent on soluble CNTFRa subsequently led to a more careful study of the specificity of CNTFRa antagonists. On cells with high concentrations of membrane-bound CNTFRa, such as IMR32 cells, the CNTF mutant with increased site 1 binding and decreased site 3 binding, F152A/K155A/S166D/Q167H, inhibited not only CNTF, but also LIF (102). The investigators suggested that the antagonist was able to trap gp130 in a nonproductive complex with CNTFRa. In contrast, on HepG2 cells, when soluble CNTFRa was made limiting, the antagonist inhibited CNTF, whereas sparing LIF and IL-6. The prediction is that in vivo, the antagonist will inhibit CNTF responses that are mediated by soluble CNTFRa, whereas acting as a general gp130 antagonist on cells that express high levels of CNTFRa.

A more specific CNTFRa antagonist could be created by simultaneously reducing the affinity of CNTF for both LIF-R and gp130. Existing mutations in site 2 of CNTF do not fully reduce gp130 binding, indicating a need for further delineation of site 2. Reducing the binding of the antagonist to gp130 would presumably widen the gap between the affinity of the antagonist for CNTFRa and the high affinity of wt CNTF for the receptor complex of CNTFRa, LIF-R, and gp130. To regain potency, the binding to CNTFRa may have to be further enhanced to bridge the gap in affinity.


The evidence that receptor antagonists work as competitive antagonists derives from the observation that the antagonists can be overcome by increasing con centrations of wt cytokine. Furthermore, the block in signal transduction can be observed as early as JAK activation, STAT activation, and receptor phos-phorylation. One prediction of this model is that if the wt cytokine is not present to activate the receptor, then the presence of the antagonist should not matter. An intriguing series of experiments on a GM-CSF receptor antagonist may suggest otherwise.

GM-CSF stimulates the function of mature neutrophils and eosinophils and promotes their viability as well as the viability of leukemic cells (105,106). The GM-CSF mutant, E21R, binds to the GM-CSF receptor a chain (GM-CSFR-a), a ligand-specific signal-transducing subunit, but does not bind to Pc, the common signaling chain shared with IL-3 and IL-5 receptor complexes (8). Surprisingly, E21R induces the apoptosis of hematopoietic cells, including eosinophils, in the absence of detectable GM-CSF (107,108). The E21R-triggered apoptosis can be recreated in Jurkat T cells that have been transfected with both GM-CSF a and Pc, indicating that both subunits are required. In the doubly transfected cells, treatment with E21R decreased the activity of the mitogen-activated protein kinase ERK (109). The mechanism is not yet understood. Distinguishing between an apoptotic effect secondary to blocking the GM-CSF receptor and a direct apoptotic effect has proven to be difficult. One possibility is that GM-CSF is produced by hematopoietic cells in an auto-crine fashion at levels below current detection but sufficient for cell survival. Another possibility is that since a portion of the GM-CSF a and Pc on myeloid cells has been shown to exist in preformed complexes in the absence of GM-CSF (110), the preformed complexes may provide a trickle of signal that is blocked by E21R (111). If receptor subunits prove to have activities that are independent from ligand-mediated oligomerization, receptor antagonists may cause unexpected effects.


Receptor antagonists can be used to dissect the hierarchical relationship among cytokines. For example, IL-6, tumor necrosis factor (TNF), IL-1, parathyroid hormone-related protein, and 1,25-dihyroxyvitamin D3 all induce human os-teoclast-like cell formation in human bone marrow culture. TNF and IL-1 induce IL-6, suggesting that their effects may be mediated by IL-6. Application of one of the IL-6Ra superantagonists, Sant5, revealed that the stimulation of osteoclast-like multinucleated cells (MNC) by IL-1 and TNF was de pendent on IL-6, but the stimulation by parathyroid hormone-related protein and 1,25-dihyroxyvitamin was not (112).

Inhibiting IL-6-induced osteoclast formation may be therapeutically beneficial in situations in which bone resorption is undesirably high such as postmenopausal osteoporosis. The inhibition of IL-6 would have to be maintained over a long period of time. Constitutively high levels of Sant5 were generated by transfecting Sant5 cDNA into a stromal cell line (PSV10), which normally expresses IL-6 and stimulates osteoclast formation. Sant5 expression inhibited the stimulatory effects of the transfected cells in cocultures with normal human bone marrow. Conditioned media from the transfected cells also inhibited the stimulatory effects of the parental cell line in similar cocul-tures (112). The successful expression of Sant5 in transfected cells suggests that gene therapy might be a promising method for delivering IL-6Ra antagonists. So far, the in vivo analysis of the superantagonists has been restricted by their species specificity. Although they are antagonists on human cells, the superantagonists are agonists on murine cells (112).

Another important use for receptor antagonists has been to determine the receptor usage of newly identified cytokines. For example, LIF-R antagonists were used to define the receptor usage of CT-1 when CT-1 was first characterized. A LIF-R antagonist was able to block the induction of c-fos by CT-1 in murine cardiac myotubes, indicating that CT-1 responses in heart were in fact dependent on LIF-R for signal transduction (113).

LIF-R antagonists have also been useful in dissecting the receptor dependence of activities exhibited by cells in culture. A LIF-R antagonist was used to demonstrate that LIF-R activation is required for the arrest of rod differentiation exerted by Muiller glial cells in cultures of mouse retinal cells (114). In situations where multiple LIF-R ligands may be present and the identities of the ligands are uncertain, LIF-R antagonists may be more useful than neutralizing antibodies for individual LIF-R ligands.


The in vivo applications of receptor antagonists for research purposes or as therapeutic agents present several challenges. Both wt cytokines and cytokines that have been modified to be receptor antagonists are rapidly cleared from the body. Hence, if the receptor antagonists are injected as a bolus, they must be repeatedly administered at frequent intervals. In several diseases, such as multiple myeloma, cytokine synthesis is continuously elevated; thus, for treat ment, the appropriate antagonist would need to be administered continuously and at high levels. Each of the examples of in vivo applications described below addresses the problem of applying a sustained level of receptor antagonist.

Another concern in the application of receptor antagonists is the possibility of eliciting an immune response. Repeated applications of antagonists over a long period may result in a neutralizing response to the antagonist if the amino acid substitutions create new antigenic regions on the cytokine. One way around the problems of both antigenicity and the need for continuous treatment might be gene therapy, which might also allow for the targeting of specific tissues. An alternative method to achieve long-term blockade of a cytokine might be to use antagonists deliberately to induce a neutralizing response to the wt cytokine. Both gene therapy approaches and vaccination approaches are discussed in the examples below.

Application of IL-4R Antagonists

IL-4 receptor antagonists are interesting molecules to clinicians because IL-4 has a central role in T-cell (Th2)-dominated diseases such as type 1 allergic responses. An antagonist based on murine IL-4 (Q116D/Y119D) disrupts binding to yc, whereas preserving binding to the IL-4R. In a variety of murine cell lines, the antagonist inhibited IL-4 responses by 50% at a 100- to 500fold molar excess of antagonist (6). Both IL-4 and IL-13 responses are inhibited, because the two receptor complexes share IL4-R as a subunit (115). During immunization with ovalbumin (OVA), BALB/c mice were administered the IL4R antagonist (given in two intraperitoneal injections per day for 8 days). The antagonist prevented immediate cutaneous hypersensitivity in response to intradermal rechallenge with antigen and prevented anaphylactic shock in response to intravenous rechallenge with antigen. Treatment completely inhibited the synthesis of OVA-specific IgE and IgG1. Interestingly, unlike in the IL4-/- mice or the IL-4R-/- mice, the antagonist also inhibited the production of IgG2a, IgG2b, and IgG3 (115). The different effects on immunoglobu-lin synthesis highlight the utility of applying receptor antagonists even when receptor knock outs are available for study. Explaining the differences between applying IL-4R antagonists and genetic deletion of IL4-R will enrich our understanding of IL4R in allergy.

GM-CSFR Antagonist in Juvenile Myelomonocytic Leukemia

The GM-CSFRa antagonist, E21R, induces apoptosis in a variety of hemato-poietic cells including juvenile myelomonocytic leukemia (JMML) cells

(116). Cells from JMML patients and normal donors were engrafted into im-munodeficient mice and a continuous supply of E21R was administered via minipump at the time of transplantation or 4 weeks later. At both time points, E21R profoundly reduced the JMML cell load in the mouse bone marrow, whereas sparing cells from normal donors (117). The sparing of normal cells, presumably because they are less sensitive to both GM-CSF and E21R, suggests that E21R might be a good therapeutic candidate for JMML.

Gene Therapy May Allow a Steady Supply of Receptor Antagonist

A recombinant adenovirus was constructed by inserting an IL-6 receptor su-perantagonist (Sant1, described above) under the control of a RSV promoter into a replication-incompetent adenoviral vector (118). Supernatants of cells infected in vitro contained active antagonist. After intravenous injection of the virus into mice, 1-2 ng/mL of antagonist could be detected in the serum. In vivo blockade of receptors could not be measured, because the IL-6 super-antagonist does not block murine IL-6 receptors. However, the antagonist was active, since serum from infected mice could inhibit wt IL-6 responses on human cell lines. (118).

Vaccination of Mice with an IL-6Ra Antagonist to Neutralize IL-6

Exploiting the potential antigenic nature of a receptor antagonist is the basis of an unusual approach to inhibiting cytokines. Vaccinating mice with an IL-6Ra antagonist, Sant1, in the presence of either complete Freund's adjuvant or aluminum hydroxide resulted in the production of autoantibodies against IL-6. When the procedure was carried out in transgenic mice that had been engineered to express high levels of hIL-6, a strong antibody response to both the receptor antagonist and wt hIL-6 was generated, suggesting that the vaccination overcame tolerance to the transgenic hIL-6. Although vaccination with wt IL-6 also elicited autoantibodies, it was much less potent than the antagonist. The antibodies elicited by vaccination completely masked the circulating IL-6. Mice injected with hIL-6 following vaccination with the receptor antagonist did not show a normal acute phase response in response to an injection of hIL6. (119).

Was this article helpful?

0 0
Arthritis Joint Pain

Arthritis Joint Pain

Arthritis is a general term which is commonly associated with a number of painful conditions affecting the joints and bones. The term arthritis literally translates to joint inflammation.

Get My Free Ebook

Post a comment