Extension of recombinant human RANTES by the retention of the initiating methionine produces a potent antagonist.

Extension of recombinant human RANTES by a single residue at the amino terminus is sufficient to produce a potent and selective antagonist. RANTES is a proinflammatory cytokine that promotes cell accumulation and activation in chronic inflammatory diseases. When mature RANTES was expressed heterologously in Escherichia coli, the amino-terminal initiating methionine was not removed by the endogenous amino peptidases. This methionylated protein was fully folded but completely inactive in RANTES bioassays of calcium mobilization and chemotaxis of the promonocytic cell line THP-1. However, when assayed as an antagonist of both RANTES and macrophage inflammatory polypeptide-1α (MIP-1α) in these assays, the methionylated RANTES (Met-RANTES) inhibited the actions of both chemokines. T cell chemotaxis was similarly inhibited. The antagonistic effect was selective since Met-RANTES had no effect on interleukin-8- or monocyte chemotractant protein-1-induced responses in these cells. Met-RANTES can compete with both [I]RANTES and [I]MIP-1α binding to THP-1 cells or to stably transfected HEK cells recombinantly expressing their common receptor, CC-CKR-1. These data show that the integrity of the amino terminus of RANTES is crucial to receptor binding and cellular activation.

RANTES is a member of a large family of cytokines, known as chemokines, which have the ability to recruit and activate a wide variety of proinflammatory cell types (1). They are small polypeptides of 8 -10 kDa and have been further classified into CXC or CC chemokines based on the spacings of the cysteine residues proximal to the amino terminus. CXC chemokines primarily activate neutrophils, whereas CC chemokines have effects on several leucocyte cell types. RANTES is a CC chemokine, and in vitro it can produce chemotaxis and activation of monocytes, eosinophils, and T cells, particularly CD4 ϩ CD45RO ϩ (memory) T cells (2), but not neutrophils. These results imply a role for RANTES in diseases such as allergen induced late phase skin reactions or in allergic asthma. This hypothesis is strengthened by the fact that large amounts of RANTES are found in nasal polyp tissues, which are rich in infiltrating eosinophils (3). In addition, injection of RANTES into dog skin has been shown to induce a large eosinophilic infiltrate in vivo (4), and migration of human T lymphocytes was observed on injection of human RANTES into a human/severe combined immune deficiency mouse model (5).
MIP-1␣ shares an overlapping cell-type specificity with RANTES in vitro (6,7) and has been shown to elicit an inflammatory response mediated through mast cell degranulation in vivo (8). A common receptor for these two CC chemokines has been cloned (9,10) and is a member of the seven transmembrane G-protein linked receptor family. Recombinant expression of the receptor has shown that it can transduce a functional response on stimulation by both chemokines.
We report the purification of human RANTES expressed heterologously in Escherichia coli. In this system, the protein retains its initiating methionine residue, which renders it inactive as an agonist, while enabling it to antagonize effects induced both by RANTES and MIP-1␣. It is able to compete for binding of both the radiolabeled ligands on THP-1 cells and to the recombinant RANTES/MIP-1␣ receptor with high affinity.

EXPERIMENTAL PROCEDURES
Protein Expression and Purification-The cDNA coding for RANTES was cloned from a human peripheral blood monocyte GT11 cDNA library by PCR using primers based on the published sequence (2). The resultant PCR product was subcloned into the E. coli expression vector pT7-7 (11) and transformed into E. coli strain BL21(DE3). High level expression of T7 polymerase and RANTES was induced by addition of isopropyl-1-thio-␤-D-galactopyranoside to the medium.
The protein was purified from inclusion bodies by gel filtration on a Sephacryl S-300 HR column equilibrated in 0.1 M Tris-HCl buffer, pH 8.0, containing 6 M guanidine HCl and 1 mM dithiothreitol. Renaturation was carried out by a 20-fold dilution into 0.1 M Tris-HCl buffer, pH 8.0, containing 1 mM oxidized glutathione and 0.1 mM reduced glutathione and stirring overnight at 4°C. The renatured protein was concentrated by adjusting the pH to 4.5 with acetic acid, and loading onto a HiLoad SP 26/10 column equilibrated in 50 mM sodium acetate buffer, pH 4.5. The protein was eluted with a gradient of 0.6 -2.0 M NaCl in the same buffer. The Met-RANTES 1 -containing fractions were dialyzed extensively against 1% acetic acid, and then against 0.1% trifluoroacetic acid and lyophilized. Recombinant full-length human RANTES and MIP-1␣ were expressed, purified, and renatured from E. coli (12).
Analytical Methods-SDS-PAGE was carried out on 4 -20% acrylamide mini-gels (Novex) according to the manufacturer's instructions, and the proteins were visualized by staining with Coomassie Brilliant Blue R-250. Purified RANTES was quantified by the extinction coefficient of A 1 cm 0.1% ϭ 1.6 at 280 nm calculated from the amino acid sequence. Amino acid analysis was carried out by gas phase hydrolysis under vacuum in a nitrogen atmosphere for 24 h in 6 M HCl containing 1 mg/ml phenol at 112°C. The resultant amino acids were quantified using a Beckman 6300 system using norleucine as an internal standard. Protein sequence was obtained with a model 477A protein sequencer (Applied Biosystems) using on line quantification of amino acid phenylthiohydantoin derivatives with a model 120A phenylthiohydantoin-derivative analyzer. Electrospray ionization mass spectroscopy (ESI-MS) was carried out on a Trio 2000 instrument (VG Biotech, Altrincham, UK). * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Samples from reverse phase HPLC (trifluoroacetic acid/acetonitrile system) were dried and then dissolved in methanol/water/acetic acid (49.5/ 49.5/1 (v/v)) and infused at a rate of 2 ml/min. Spectra were co-added by repetitively scanning until an acceptable signal to noise ratio was obtained (3-4 min). Circular dichroism spectra were obtained using a Jasco J600 spectropolarimeter. 1 H NMR spectroscopy was according to Chung et al. (13).
Bioassays-THP-1 cell chemotaxis was carried out using 96-well micro-Boyden chambers (Neuro-Probe, Cabin John, MD) fitted with 5-m filters. 5.6 ϫ 10 5 cells in 200 l of medium (RPMI 1640 containing 0.01 M HEPES, 10% heat-inactivated fetal calf serum, 2 mM L-glutamine, and 0.005% gentamicin) were placed in the upper chamber. 370 l of the medium described above, but without fetal calf serum, containing the ligand and appropriate dilutions of Met-RANTES, were placed in the lower chamber. After 60 min of incubation at 37°C under 5% CO 2 , the cells were removed from the upper wells, and 200 l of phosphate-buffered saline containing 20 M EDTA added to detach the cells bound to the filter. After 30 min of incubation at 4°C, the plate was centrifuged at 1800 ϫ g for 10 min, and the supernatants were removed from the lower wells. The number of cells that had migrated were measured by the Cell Titer 96 nonradioactive cell proliferation assay (Promega), which monitors the conversion of tetrazolium blue into its formazan product. T cell chemotaxis was performed according to Fincham et al. (14), and calcium mobilization was performed in THP-1 cells according to Tsien et al. (15) with the modifications described previously (16). The data obtained were fitted using Grafit 3.01 software (17) to a four-parameter logistic equation. The IC 50 is defined as the concentration of antagonist giving half the maximal inhibition. Typical agonist concentrations were 5ϫ the EC 50 value determined for induction of the response.
Recombinant Receptor Expression-MIP-1␣/RANTES (CC-CKR1) receptor was cloned by reverse transcriptase PCR from the human eosinophilic cell line EOL-3 using specific primers based on the published cDNA sequence (9). CC-CKR-1 cDNA was subcloned into the mammalian cell expression vector pcDNA1 neo (Invitrogen) and stably expressed in the HEK 293 cell line as described elsewhere (12).
Receptor Binding Assays-Equilibrium competition binding assays were carried out according to Van Riper et al. (18) but including 0.02% sodium azide in the binding buffers. The data obtained were fitted using Grafit (17) Enzymic Digestion-0.27 ml of the leucine aminopeptidase ammonium sulfate suspension (Sigma) was brought into solution by adding 1.15 ml of 38 mM sodium phosphate buffer, pH 7.2, containing 9.5 mM MgCl 2 , 1.9 mM phenylmethylsulfonyl fluoride, and 1.9 mM benzamidine-HCl. The solution was centrifugally concentrated in a Centricon 10 cell to 0.35 ml, a further 1.15 ml of buffer-inhibitor solution was added, and the volume was again reduced to 0.35 ml. 0.8 mg of Met-RANTES was dissolved in H 2 O and adjusted to 2 mg/ml in a phosphate-buffered saline solution (8 mM Na 2 HPO 4 , 1.5 mM KH 2 PO 4 , 14 mM NaCl, 0.3 mM KCl, pH 7.2), containing 0.45 M sodium phosphate buffer, pH 7.2, 5 mM MgCl 2 , 2.4 mM phenylmethylsulfonyl fluoride, and 2.4 mM benzamidine-HCl. 0.08 mg of the leucine amino peptidase solution described above was added, and the solution was adjusted to 0.1% sodium lauryl sarcosinate. Incubation was carried out at 23°C, and samples were removed at the times indicated in Fig. 1. The amount of methionine removed was analyzed by isocratic elution on reverse phase HPLC.

RESULTS
Protein Characterization-Very high levels of recombinant RANTES were obtained when the cDNA coding for the mature form of the protein was expressed under control of the T7 RNA polymerase system, although the protein would not express in E. coli when driven by either the p L or the Trp promoter. The protein was purified from inclusion body material, renatured, and shown to be refolded by both NMR (13) and CD spectroscopy (results not shown). Electrospray ionization mass spectroscopy analysis of the purified protein gave a mass of 7,978 as opposed to the expected mass of 7,847 for the oxidized mature protein. Amino terminal sequence analysis showed that the initiating methionine had been retained, accounting for the additional mass of 131. We have therefore called the protein Met-RANTES. Unlike naturally isolated (19) or full-length recombinant RANTES (2), Met-RANTES was unable to cause chemotaxis or calcium mobilization in the promonocytic cell line THP-1, even when using concentrations as high as 10 Ϫ6 M.
Comparison of the NMR spectra of the methionylated and nonmethionylated forms of RANTES showed that no major conformational change had been induced by addition of the amino-terminal amino acid. In both cases, the three-dimensional location of the amino terminus of RANTES is partially disordered in solution, similar to the results described for another CC chemokine MIP-1␤ (20). Since the only difference between active RANTES and our Met-RANTES was the initiating methionine, we postulated that this residue was responsible for the observed difference in activity between the two proteins. This was tested by treating Met-RANTES with CNBr to remove the methionine. Activity was observed even without a further renaturation following the harsh treatment in 70% formic acid. Enzymic removal of the amino-terminal methionine using leucine aminopeptidase (Fig. 1) also activated the protein. These results further confirm that the initial lack of activity in Met-RANTES is not due to misfolding of the recombinant methionylated protein.
Antagonistic Properties-Since it was correctly folded, we next attempted to antagonize the function of RANTES in chemotaxis assays using the methionylated protein (Fig. 2a). A concentration of 3.5 nM RANTES was used as the chemotactic stimulus for THP-1 cells, 5 times the EC 50 value for RANTES in this assay. The Met-RANTES clearly inhibits the chemotactic response under these conditions, with an IC 50 value of 6 nM, which is a 1.7-fold molar excess. Since RANTES and MIP-1␣ share a common receptor, (9, 10), we also studied the effect of Met-RANTES against MIP-1␣ induced chemotaxis of THP-1 cells. Again, nanomolar inhibition was seen, with an IC 50 value of 0.49 nM, using 1 nM MIP-1␣ (5 ϫ EC 50 ) as a stimulus (Fig.  2a). RANTES also induces migration of T cells (2), and Met-RANTES was able to antagonize the response to 23 nM RAN-TES (5 ϫ EC 50 ) with an IC 50 of 6 nM (Fig. 2b), which is the same concentration as the EC 50 for RANTES-induced response. As with the THP-1 cells, Met-RANTES antagonized the response induced by 0.5 nM MIP-1␣ (5 ϫ EC 50 ), with an IC 50 of 6 nM, which is 60-fold higher than the EC 50 for MIP-1␣-induced lymphocyte chemotaxis (Fig. 2b). No antagonism of MCP-1induced chemotaxis of THP-1 cells or IL-8 induced chemotaxis of T cells was observed, in keeping with the known receptor selectivity of the MIP-1␣/RANTES receptor, CC-CKR1 (9, 10).
As a second bioassay, we studied the ability of chemokines to mobilize calcium in THP-1 cells (Fig. 3). Here again, Met-RANTES was capable of antagonizing the effects of RANTES. The concentration required for half-maximal inhibition of the calcium mobilized by 66 nM RANTES (5 ϫ EC 50 ) was 88 nM. Although Met-RANTES was capable of inhibiting the mobilization of calcium by 23 nM (5 ϫ EC 50 ) MIP-1␣, an inhibitory effect was only seen at concentrations above 100 nM, and the IC 50 of 1.2 M is 260-fold higher than the EC 50 value (4.6 nM) of the agonist-induced response. It was again unable to antagonize the calcium mobilization produced as a result of stimulation with either MCP-1 or IL-8, which are known to act at distinct receptors (21)(22)(23) Receptor Binding-Equilibrium competition binding studies using the promonocytic cell line, THP-1 (Fig. 4a)   tified (9). We therefore tested the ability of Met-RANTES to compete with [ 125 I]RANTES and [ 125 I]MIP-1␣ on HEK 293 cells stably transfected with the CC-CKR-1 receptor (Fig. 4b) (Fig. 4c). DISCUSSION The retention of the initiating methionine in recombinant proteins expressed heterologously in E. coli, particularly those produced in inclusion bodies, is not uncommon. This is not always the case, as the endogenous E. coli MAP (24) often removes the Met residue. The retention of the initiating Met does not impair bioactivity in the case of recombinant cytokines such as granulocyte-macrophage colony-stimulating factor (25), interleukin-2 (26), and interleukin-5 (27). However some 5-10-fold shifts in receptor binding affinity have been seen in certain cases such as hirudin (28) and interleukin-1␤ (29).
The retention of methionine in recombinant RANTES produces a protein that shows no agonist activity, despite the fact that it is correctly folded (13). Furthermore, it acts as a functional antagonist. The addition of a single amino acid to the amino terminus of RANTES creates an antagonist that is almost equipotent: a 1.7-fold molar excess is required for halfmaximal inhibition of RANTES-induced chemotaxis of THP-1 cells, whereas the RANTES-induced T cell chemotaxis is inhibited with an IC 50 equal to the EC 50 for the response. Similarly, the calcium mobilization induced by RANTES is inhibited to 50% with an equimolar concentration to that used to induce the response. The same potency is observed for the inhibition of chemotaxis induced by MIP-1␣, both in THP-1 and T cells. Interestingly, however, the IC 50 value for the inhibition of the MIP-1␣ induced calcium mobilization in THP-1 cells is 260-fold higher than the EC 50 . Met-RANTES therefore antagonizes RANTES and MIP-1␣ with similar potency in the chemotaxis response but shows a clear difference in the other assay, calcium mobilization. This suggests that either there is an additional receptor for these chemokines or that there are two different signaling pathways for these two responses. A third possibility would be that different ligands evoke distinct signaling pathways at the same receptor.
The extension of RANTES by a single amino acid at the amino terminus thus produces a more potent antagonist than those described for other chemokines, which have been produced by amino-terminal deletions. Data from mutagenesis studies of the CXC chemokine interleukin-8 shows that the amino-terminal region is crucial for signaling, and deletion of five amino acids produces an antagonist (35,36). However, a 30-fold molar excess is required for half-maximal inhibition of elastase release, and the reduction of neutrophil chemotaxis is not complete at 10 Ϫ6 M (36). Residues 9 -76 of the CC chemokine, monocyte chemotractant protein-1 (MCP-1) produce an antagonist that inhibits THP-1 cell chemotaxis with an IC 50 of 20 nM induced by 5 nM MCP-1 (37). Truncation of residues 2-8 from the amino terminus of MCP-1 produces an antagonist (38), which causes half-maximal inhibition of monocyte chemotaxis at a ratio of 75:1. This variant has been suggested to act as a dominant negative repressor of the active form of the ligand (39). However, the Met-RANTES antagonist appears to act as a competitive inhibitor on the shared MIP-1␣/RANTES receptor since, first, direct binding of [ 125 I]Met-RANTES has been demonstrated, and second, it can compete with both che-mokines for binding.
Single amino acid changes have produced antagonists of other cytokines, such as interleukin-4 (32), interleukin-6 (33), and granulocyte-macrophage colony-stimulating factor (34). In the cases of these cytokines, they bind and signal through heterodimeric receptor complexes. Detailed studies have shown that the side chains involved in the initial binding between the ligand and its receptor, and the side chains involved in causing a signaling response are spatially separated. A mutant becomes an antagonist if its region interacting with the signaling part of the receptor can be eliminated without altering the binding region. To date, no mutagenesis data is available on RANTES, indicating whether the residues involved in binding and signaling differ. Chemokines, and other chemotactic peptides such as C5a, bind and signal through a seven-transmembrane G-protein coupled receptor. C5a interacts with its receptor through two distinct sites (30). Binding of C5a to the amino terminus of the receptor is proposed to produce a conformational change in the ligand, allowing it to properly interact with the second site at the carboxyl terminus to achieve functional activation (31). Such a dramatic change in bioactivity by the addition of a single amino acid emphasizes the importance of the amino terminus of RANTES. The extension of the sequence by one residue has removed the ability to form a functional interaction with the region of the receptor involved in G-protein-mediated signal transduction. However, since the overall structure of the protein is unaffected, receptor binding is unimpaired, producing a highly potent antagonist. In vivo studies with this readily available molecule will be extremely valuable in determining whether blocking the RANTES receptor(s) involved in the recruitment of leucocytes to inflammatory sites will relieve chronic inflammation.