Engineering the glycosaminoglycan-binding affinity, kinetics and ...

Protein Engineering, Design & Selection vol. 22 no. 6 pp. 367– 373, 2009 Published online May 4, 2009 doi:10.1093/protein/gzp013

Engineering the glycosaminoglycan-binding affinity, kinetics and oligomerization behavior of RANTES: a tool for generating chemokine-based glycosaminoglycan antagonists Barbara Brandner1, Angelika Rek2, Maria Diedrichs-Mo¨hring3, Gerhild Wildner3 and Andreas J. Kungl1,2,4 1

Institute of Pharmaceutical Sciences, University of Graz, Universita¨tsplatz 1, A-8010 Graz, Austria, 2ProtAffin Biotechnologie AG, Reininghausstrasse 13a, A-8020 Graz, Austria and 3Section of Immunbiology, Department of Ophthalmology, Clinic of Ludwig-Maximilians Universita¨t Mu¨nchen, Mu¨nchen, Germany

4

To whom correspondence should be addressed. E-mail: [email protected]

Binding to glycosaminoglycans (GAGs) is a necessary prerequisite for the biological activity of the proinflammatory chemokine RANTES in vivo. We have applied protein engineering methods to modulate equilibrium-binding affinity as well as binding kinetics of RANTES towards its GAG ligand which also altered the chemokine’s oligomerization behavior. Out of 10 mutants, A22K and H23K were chosen for further in vitro and in vivo characterization because their stability was comparable with wild-type (wt) RANTES. In chemical cross-linking experiments, A22K gave higher and H23K lower molecular weight aggregates compared with wtRANTES as shown on SDS – PAGE. All mutants contained an N-terminal methionine residue, a well-described G-protein-coupled receptor (GPCR) antagonistic modification, which resulted in the mutants’ inability to induce monocyte chemotaxis. In surface plasmon resonance experiments using immobilized heparan sulfate (HS) and physiological buffer conditions, Met-RANTES exhibited a significantly longer residual time on the GAG chip compared with the other RANTES variants. In Scatchard plot analysis, RANTES gave a bi-phasic, bell-shaped curve suggesting ‘creation’ of ligand-binding sites on the protein during HS interaction. This was not observed in the mutants’ Scatchard plots which gave Kd values of 317.5 and 44.5 nM for the A22K and H23K mutants, respectively. The mutants were subsequently tested for their inhibitory effect in a rat model of autoimmune uveitis where only H23K exhibited a transient improvement of the clinical disease score. H23K is therefore proposed to be a GPCRinactive GAG antagonist which displaces the wt chemokine from its natural HS-proteoglycan co-receptor. The protein engineering approach presented here opens new ways for the treatment of RANTES-related diseases. Keywords: heparan sulfate/chemokine/surface plasmon resonance/circular dichroism/protein engineering/uveitis

Introduction The cell-mediated immune response to inflammation depends on distinct patterns of leukocyte migration and

activation. Selective leukocyte-endothelial cell recognition is mediated by a sequential and coordinated release of inflammatory chemokines, proinflammatory cytokines and adhesion molecules from both resident and immigrating cells (Butcher, 1991). Selectin-mediated ‘rolling’ of the leukocytes followed by activation and integrin-mediated stable binding to the endothelium results in leukocyte extravasation into inflamed tissue. A stable haptotactic chemokine gradient is a prerequisite for this kind of directional migration and extravasation under shear stress in vitro (Rot, 1992). Chemokine immobilization is accomplished via binding to glycosaminoglycan (GAG) side chains of proteoglycans (PGs) (Rot, 1992; Witt and Lander, 1994), which is crucial for cell activation in vivo but not in vitro as has been demonstrated recently for RANTES and other chemokines (Ali et al., 2000; Middleton et al., 1997; Proudfoot et al., 2003). RANTES is an 8 kDa inflammatory chemokine of the CC subfamily. The protein attracts and activates monocytes, eosinophils, CD4þ/CD45ROþ (memory) T-cells, natural killer cells, basophils and dendritic cells, but not neutrophils, by binding to the G-protein-coupled receptors (GPCRs) CCR1, CCR3 and CCR5. The role of chemokines and chemokine receptors in disease has been reviewed extensively elsewhere (Gerard and Rollins, 2001; Johnson et al., 2004b; Power and Proudfoot, 2001; Ribeiro and Horuk, 2005). In short, CCR1 is a key player in transplant rejection (Gao et al., 2000), CCR3 plays a substantial but not fully clear role in asthma, whereas CCR5 was identified as a co-receptor for HIV-1 infection (Simmons et al., 1997). In addition, RANTES is associated with progressive glomerulonephritis, atherosclerosis, rheumatoid arthritis or delayed hypersensitivity. The N-terminus and the N-terminal loop of RANTES both were shown to be involved in receptor binding and activation (Blanpain et al., 2003; Pakianathan et al., 1997). N-terminal truncation mutants were synthesized, which retained affinity for RANTES receptors but did not induce chemotaxis, Ca2þ release or N-acetyl-b-D-glucosaminidase release (Gong et al., 1996). Heterologous expression of RANTES in Escherichia coli leads to the retention of the initiating N-terminal methionine, which converted the chemokine into a receptor antagonist with partial CCR5 agonistic activity, the so-called Met-RANTES (Proudfoot et al., 1996; Proudfoot et al., 1999). Alanine scanning mutagenesis identified the twodimensional binding epitope 44RKNR47 in the 40s loop as the principal GAG-binding domain of RANTES, which is a classical BBXB heparin-binding epitope also present in other chemokines and heparin-binding proteins (Proudfoot et al., 2001). Heparan sulfate proteoglycans (HSPGs), which

# The Author 2009. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: [email protected]

367

B.Brandner et al.

consist of a core protein and heparan sulfate (HS) side chains, are the most abundant class of cell surface PGs in mammals and act as co-receptors for diverse proteins such as growth factors, proteases, cytokines or chemokines (Park et al., 2000; Tumova et al., 2000; Witt and Lander, 1994). The transmembrane HSPGs syndecan-1 and syndecan-4 expressed by primary macrophages were identified recently as natural co-receptors for RANTES (Slimani et al., 2003). RANTES forms high molecular weight aggregates in a pH- and concentration-dependent manner (Czaplewski et al., 1999; Skelton et al., 1995). Experiments with the dimeric variant RANTES E66S and the tetrameric variant RANTES E26S demonstrated the impact of self-aggregation on biological function. Only wild-type (wt) RANTES but not the disaggregating variants could trigger the activation of the protein tyrosine kinase signal transduction pathway associated with T-cell activation (Appay et al., 1999). Experiments with the tetrameric variant RANTES E26A, the dimeric variants RANTES E66A and [44AANA47]-RANTES and the monomeric variant [Nme-Thr7] RANTES indicated that contrary to in vitro studies, in vivo RANTES has to oligomerize at least to a tetramer to be biologically active (Johnson et al., 2004a; Proudfoot et al., 2003). Major efforts have been put forward towards the development of low molecular weight receptor antagonists (Ribeiro and Horuk, 2005) in order to interfere with chemokine signaling pathways. Here we have identified RANTES mutants with high-GAG-binding affinity, modified oligomerization behavior and knocked-out GPCR activity which could potentially be applied as protein-based GAG antagonists in inflammatory disease settings. Experimental

Generation of chemokine mutants The RANTES mutant constructs were generated using two different PCR-based site-directed mutagenesis strategies. RANTES V49R/E66S, RANTES 15LSLA18 V49R/E66S, RANTES (3-68) V49K/E66S, RANTES (9-68) V49K/E66S and RANTES (9-68) V49R/E66S were amplified using the plasmid pRSET A containing the wtRANTES gene (a kind gift from Simi Ali, University of Newcastle upon Tyne, UK) as template. Amplification was performed in a Mastercycler Gradient (Eppendorf ) using a step-down PCR program. The mutant genes were gel-purified and cloned into the expression vector pET101/ D-TOPOw (Invitrogen Life Technologies) according to manufacturer’s instructions. All other mutations (A22K, H23K, T43K, N46K, N46R, Q48K, A22K/N46R, N46K/Q48K) were introduced employing the QuikChangew II site-directed mutagenesis method (Stratagene) according to manufacturer’s instructions. The employed mutagenesis primers were synthesized by Invitrogen Life Technologies. The pET101/D-TOPOw plasmids containing the RANTES mutant genes were transformed into the E. coli strain TOP 10 (Invitrogen). Prior to the expression, the correct sequences of the constructs were verified by DNA sequencing.

Protein expression and purification After transformation of the pET101/D-TOPOw RANTES mutant constructs into the BL21TM (DE3) Star E. coli strain 368

(Invitrogen), glycerin stocks were prepared, which served as direct precultures for protein expression in Erlenmeyer flasks. Cultures were grown under shaking at 378C in LB broth containing 100 mg/ml ampicillin to an OD600 of 0.8. Protein expression was induced by the addition of 0.5 mM isopropyl-b-D-thiogalactopyranoside. After 2 h, cells were harvested. The cell pellet from 1 l of fermentation was resuspended in 13 ml solution buffer (50 mM Tris, pH 8.0, 25% w/v sucrose, 1 mM NaEDTA, 10 mM DTT) and disrupted by sonication. After addition of 0.4 mg lysozyme, 20 mg DNAse I and 2 mmol MgCl2 per milliliter cell suspension, 13 ml lysis buffer (50 mM Tris pH 8.0, 1% Triton X-100, 1% sodium deoxycholate, 100 mM NaCl, 10 mM DTT) was added, and the suspension was incubated at room temperature for 1 h. After the addition of 7 mmol/ml NaEDTA, the suspension was frozen in liquid nitrogen, thawed and incubated with 3.5 mmol/ml MgCl2 until the viscosity decreased. NaEDTA was added to a concentration of 7 mmol/ml, and the suspension was centrifugated at 11 000g for 20 min at 48C. The pelleted inclusion bodies were sonicated in washing buffer with Triton X-100 (50 mM Tris – HCl, pH 8.0, 0.5% Triton X-100, 100 mM NaCl, 1 mM NaEDTA, 1 mM DTT), then in washing buffer without Triton X-100. The prepurified inclusion body pellets were solubilized in 2 ml 6 M guanidine, 50 mM Tris, pH 8.0, and 1 mM DTT per gram wet cell pellet under stirring for 30 min at 608C. After centrifugation, the solution was dialyzed against 1% acetic acid. The precipitant was spun down and the solution was lyophilized and resolubilized in 1.5 ml 6 M guanidine, 50 mM Tris, pH 8.0, 1 mM DTT per gram wet cell pellet. Protein refolding and purification by cation exchange chromatography with an SP Sepharose FF column (Amersham Biosciences) were done as described elsewhere (Proudfoot and Borlat, 2000). Purity (.98%) and identity of the RANTES mutant proteins were confirmed by silver-stained SDS– PAGE and nano-HPLC ESI-MS/MS, respectively. The RANTES mutants were stored as lyophylized powders for periods of .1 year without the loss of structure and activity. To obtain wtRANTES (wtRANTES, without the additional N-terminal methionine), the pRSET A/RANTES plasmid was used for transforming BL21TM (DE3) Star E. coli. RANTES was expressed as a fusion protein with an N-terminal His-tag and a MKKKWPR leader sequence directly preceding the RANTES protein and purified as described earlier. The purified RANTES fusion protein in 0.1% TFA was diluted to a final concentration of 1 mg/ml RANTES fusion protein in 100 mM Tris – HCl, pH 8.0, and the peptide tag was proteolytically cleaved by incubation with trypsin (sequencing grade from Sigma, 1:250, enzyme:substrate) at 378C for 3 h. Benzamidine was added to inactivate the protease, and the cleaved peptide was separated from uncleaved RANTES and the fusion tag by cation exchange chromatography. Purity (.95%) and biological activity of wtRANTES were confirmed by silver-stained SDS– PAGE and by chemotaxis assay, respectively.

Circular dichroism spectroscopy and analysis Circular dichroism (CD) spectra were recorded on a Jasco J-710 spectropolarimeter over a range of 195– 250 nm in PBS using a quartz cuvette with a path length of 1 mm. Protein spectra were collected with a response time of 1 s and a data point resolution of 0.2 nm. Four scans of 10 mM

Engineering RANTES– GAG interactions

protein solutions were averaged to obtain smooth data. All spectra were background-corrected and converted to units of mean residue ellipticity. The program SELCON (Sreerama and Woody, 1993) was used to estimate the secondary structure contents.

Surface plasmon resonance Binding of RANTES and the respective mutants to unfractionated HS was investigated on a BIAcore 3000 instrument (BIAcore AB, Uppsala, Sweden). The immobilization of biotinylated HS onto a streptavidin-coated CM4 sensor chip was performed according to an established protocol (Ricard-Blum et al., 2004). The actual binding interactions were recorded at 258C in PBS, pH 7.4, containing 0.01% (v/v) P20 surfactant (BIAcore AB). For complete regeneration, 2.5 min injections of different protein concentrations at a flow rate of 60 ml/min were followed by 5 min dissociation periods in buffer and a pulse of 1 M NaCl. The maximum response signals of protein binding to the HS surface, corresponding to the plateaus of the respective sensorgrams, were used for Scatchard plot analysis and for the calculation of equilibrium dissociation constants (Kd values).

Chaotrope-induced unfolding experiments Protein solutions (300 nM) were prepared with stock solutions of guanidine in PBS. After a 2 h equilibration period, the fluorescence emission spectra were recorded on a Perkin-Elmer LS50B fluorometer as described previously (Goger et al., 2002; Rek et al., 2002). The wavelength at the emission maximum was plotted against the concentration of guanidine and analyzed using the Boltzmann equation contained in the Origin software package (Microcal Inc.). The temperature was maintained constant during all fluorescence experiments by coupling to an external water bath.

Protein cross-linking After equilibration of 10 mM wtRANTES and RANTES mutant solutions in PBS for 10 min, a 20-fold molar excess of ethylene glycol-bis (succinimidylsuccinate) was added. After 30 min, the cross-linking reaction was stopped with 1 mM Tris, pH 7.4. The cross-linked oligomerization and aggregation products were separated by gel electrophoresis on NuPageTM 4 – 12% Bis – Tris gels (Invitrogen) run in MES buffer according to manufacturer’s instructions and visualized by silver staining.

Chemotaxis assay RANTES mutant-directed cell migration was investigated using a 48-well Boyden chamber system (Neuroprobe) equipped with 5 mm PVP-coated polycarbonate membranes. wtRANTES and RANTES mutant dilutions ranging from 0.1 to 100 nM in RPMI 1640 containing 20 mM HEPES, pH 7.3, and 1 mg/ml BSA were placed in the lower wells of the chamber in triplicates, including wells with buffer alone. Fifty microliters of THP-1 cell suspension (European collection of cell cultures) in the same medium at 2 106 cells/ml was placed in the upper wells. After a 2 h incubation period at 378C and 5% CO2, the upper surface of the filter was washed in Hank’s balanced salt solution. The migrated cells were fixed in methanol and stained with Hemacolorw (Merck). The migrated cells of five 400 magnifications per well were counted and the mean of three independently

conducted experiments was background-corrected plotted against the chemokine concentration.

and

Animals Lewis rats were purchased from Janvier (Le-Genest-St-Isle, France) or bred in our own colony. They had unlimited access to rat chow and water. Animals were used for experiments at the age of 6 to 8 weeks. All animal experiments were approved by the Review Board of the Government of Oberbayern. Treatment of animals conformed to the ARVO Statement on the Use of Animals in Ophthalmic and Vision Research.

Active immunization with uveitogenic peptide Uveitogenic peptide PDSAg was derived from the sequence of bovine retinal S-Antigen (aa 342–355, FLGELTSSEVATEV) and purchased from Biotrend (Cologne, Germany). Lewis rats were immunized into both hind legs with a total volume of 200 ml emulsion containing 15 mg PDSAg in complete Freund’s adjuvant, fortified with Mycobacterium tuberculosis strain H37RA (BD, Heidelberg, Germany) to a final concentration of 2.5 mg/ml. For treatment, groups of four rats (n ¼ 4 rats; n ¼ 8 eyes) received daily interperitoneal injections of 100 mg/rat H23K mutant dissolved in 0.5 ml PBS (or PBS only as control) from day 1 after active immunization until day 19. The time course of disease was determined by daily examination of animals with an ophthalmoscope. Uveitis was graded clinically as described (de Smet et al., 1993), and the average clinical score of all eyes is shown per group and day. Clinical examination and scoring considered only inflammation and infiltration of cells into the anterior chamber of the eye, whereas involvement of the posterior part of the eye (vitreous, retina) has to be determined finally by histology. Uveitis was graded clinically as follows: score 0, no signs of inflammation; 0.5, dilated iris vessels; 1, peripupillar infiltrates; 2, pupil covered with fibrin; 3, hypopyon and 4, hypopyon with hemorrhage. The average clinical score of all eyes was calculated per group and day (course of disease). Results

Fold conservation of RANTES mutants Recombinant expression of RANTES in E. coli leads to the retention of the initiating methionine residue which renders RANTES a potent inhibitor of monocyte migration, the so-called Met-RANTES (Proudfoot et al., 1996). Unlike proinflammatory RANTES (designated as wtRANTES), all RANTES mutant proteins expressed in E. coli, like in this study, carried the N-terminal methionine extension and thus exhibited chemotactic inhibitory activity (see what follows). The three N-terminal deletion mutants RANTES (3-68) V49K/E66S, RANTES (9-68) V49K/E66S and RANTES (9-68) V49R/E66S failed to express under the earlier described conditions (N-terminal truncation mutants that also act as receptor antagonists have so far only been synthesized). Far-UV CD measurements and SELCON analysis were used to check for mutation-induced changes in the overall secondary structure of the proteins. Except for RANTES A22K and RANTES 15LSLA18 V49R/E66S, which both displayed a signal intensity too low to be analyzed with SELCON, according to the raw CD spectra 369

B.Brandner et al.

and further SELCON analysis, all other proteins were properly folded. RANTES A22K was probably too highly aggregated to be analyzed with CD. Compared with wtRANTES, the a-helical content was slightly decreased in all mutants, whereas a small increase in the b-sheet content was observed for all mutants (data not shown). For investigating the overall 3-D fold of the mutants, steady-state fluorescence spectra were recorded. wtRANTES yielded an emission maximum of 340.3 nm upon excitation at 280 nm, which is typical for a protein containing a partially solvent-exposed tryptophan residue such as Trp57 on top of the C-terminal a-helix. All RANTES mutants exhibited no or only a marginal shift in their emission maxima to the red, indicating a compact 3-D structure of these proteins (Table I). As an exception, RANTES 15LSLA18 V49R/E66S exhibited a large emission wavelength shift to 346.1 nm referring to partial denaturation of this mutant.

Effect of mutations on RANTES stability The wavelength at maximum intrinsic fluorescence emission proved to be a very sensitive parameter for chaotrope-induced unfolding studies as expressed by the large red shift upon guanidine hydrochloride (Gdn.HCl) addition. The emission wavelength shift upon Gdn.HCl-induced denaturation together with the fitted parameters x0 (unfolding transition) and dx (cooperativity) is displayed in Table I. The wt protein was highly resistant to chemical denaturation with an unfolding transition (x0) at 6.6 M guanidine. The mutants RANTES H23K, RANTES A22K, RANTES A22K/N46R, RANTES N46K and RANTES N46R exhibited only slightly lower midpoints than wtRANTES, ranging from 5.8 to 4.5 M Gdn.HCl, indicating a high protein stability and thus, a compact overall fold. The two mutants RANTES 15LSLA18 V49R/E66S and RANTES T43K, however, unfolded at much lower guanidine concentrations with unfolding transitions at 2.4 and 1.9 M guanidine, respectively.

Chemical cross-link oligomerization studies Mutant and wtRANTES proteins were equilibrated in PBS, and their oligomerization/aggregation state was determined using the cross-linking agent ethylene glycol-bis succinimidyl succinate. Whether the proteins formed irregular aggregates or regular oligomers at high concentrations could not be determined by this approach. However, the general applicability of our cross-link approach for determining the chemokine oligomerization/aggregation state was demonstrated by reproducing the multimerization state of RANTES mutants with known oligomerization behavior as well as of further chemokines (data not shown). For all cross-linking experiments, a protein concentration of 10 mM was chosen because wtRANTES multimerizes into distinct oligomers and non-separable aggregates at this concentration. The cross-linked proteins showed a clearly distinct oligomerization/aggregation pattern (Fig. 1) which did not correlate with the number of basic amino acid residues introduced into the chemokine. The mutations A22K and N46K present in four mutants enhanced aggregation compared with wtRANTES, whereas all other mutations reduced the multimerization tendency. In RANTES H23K and RANTES T43K, the monomer and dimer prevailed over the trimer and tetramer, whereas the mutation E66S present in RANTES V49R/E66S and RANTES 15LSLA18 V49R/E66S rendered the proteins mainly monomeric and dimeric. The single mutations N46R and Q48K had the most potent disaggregating effect on RANTES—the mutants were nearly exclusively present as monomers.

Binding of RANTES A22K and H23K to HS One aim of this protein engineering study was to improve the affinity of RANTES towards its natural GAG ligand with the

Table I. Fluorescence parameters of chaotrope-induced unfolding of wtRANTES and RANTES mutants

lmax

wtRANTES RANTES A22K RANTES H23K RANTES T43K RANTES N46K RANTES N46R RANTES Q48K RANTES A22K/N46R RANTES N46K/Q48K RANTES V49R/E66S RANTES 15LSLA18 V49R/E66S

Fit parameters

Native protein (nm)

Unfolded protein (nm)

x0 [Gdn.HCl] dx [Gdn.HCl]

340.3 340.5 341.0 342.2 343.9 342.5 343.7 340.4 342.9 343.8 346.1

364.4 357.3 358.0 357.7 356.5 359.1 359.0 356.5 357.6 361.5 360.4

6.6 + 0.3 5.7 + 0.1 5.8 + 0.1 1.9 + 0.8 4.6 + 0.1 4.4 + 0.1 3.5 + 0.4 4.8 + 0.1 3.7 + 0.1 3.7 + 0.3 2.4 + 0.5

0.7 + 0.1 0.5 + 0.1 0.2 + 0.1 0.7 + 0.1 0.2 + 0.1 0.7 + 0.1 1.0 + 0.4 0.3 + 0.1 0.4 + 0.1 1.1 + 0.4 0.7 + 0.5

Chaotrope-induced unfolding curves were fitted using the Boltzmann equation. lmax, emission wavelength maximum upon excitation at 280 nm; x0, unfolding transition at concentration of Gdn.HCl; dx, slope of the unfolding curve.

370

Fig. 1. Oligomerization and aggregation of wtRANTES and RANTES mutants. Protein solutions (10 mM) were cross-linked with ethylene glycol-bis succinimidyl succinate, separated by SDS–PAGE and silver-stained. This SDS–PAGE is representative of three independent cross-linking experiments. 1, RANTES A22K; 2, RANTES H23K; 3, RANTES T43K; 4, RANTES N46K; 5, RANTES N46R; 6, RANTES V49R/E66S; 7, RANTES 15LSLA18 V49R/E66S; 8, RANTES Q48K; 9, RANTES A22K/N46R; 10 RANTES N46K/Q48K; 11, wtRANTES; 12, PageRulerTM protein ladder (Fermentas); 13, wtRANTES not cross-linked.


purpose to generate specific protein-based GAG antagonists (Potzinger et al., 2006). We have used isothermal fluorescence titration experiments to investigate the affinity of all RANTES mutants produced in this study; the results of which will be published elsewhere (Rek et al., manuscript in press). In the following experiments, we have focused on the two most stable single-site RANTES mutants with different oligomerization behavior and with the highest improvement in GAG affinity, which were also to be tested in an in vivo model for their anti-inflammatory activity (see what follows). In isothermal fluorescence titration studies, both RANTES mutants were found to bind to soluble GAG oligosaccharides with a Kd value of 200 nM which corresponds to an affinity increase more than double compared with wtRANTES (Kd ¼ 450 nM) (Rek et al., manuscript in press). Since we were also interested in the binding kinetics of these RANTES mutants with respect to HS, the natural co-receptor of chemokines, we have used surface plasmon resonance (SPR) to determine Kd values as well as off rates (koff ) of the RANTES A22K and H23K mutants. Interestingly, recording interactions with chip-immobilized HS under otherwise identical physiological buffer conditions (e.g. PBS), we found that wtRANTES displayed a bellshaped Scatchard plot in SPR concentration-dependent experiments (Fig. 2), rendering Kd analysis impossible. The RANTES A22K and H23K mutants, on the other hand, exhibited Kd values of 317.5 and 44.5 nM, respectively. Analyzing individual SPR sensorgrams with respect to koff (Fig. 3) gave 5.9 1024 s21 for Met-RANTES, 3.8 1023 s21 for H23K and 2.04 1022 s21 for A22K. For wtRANTES, a koff value of 1.02 1023 s21 was obtained. Unmodified but N-methionylated RANTES apparently remains much longer on HS than any other RANTES mutant tested in this study. At buffer conditions in which RANTES exists primarily in its monomeric form, similar to those used for RANTES structure determination ( pH ,4), no binding could be detected for any RANTES protein investigated in this study (data not shown).

Fig. 2. SPR analyses of Met-RANTES and RANTES A22K and H23K binding to HS. Scatchard plots and binding constants (Kd values) obtained from concentration-dependent SPR experiments.

Fig. 3. SPR sensorgrams used for the analysis of koff. Representative normalized SPR sensorgrams of 400 nM wtRANTES, Met-RANTES, H23K and A22K mutants. For the calculation of koff values, a whole set of chemokine concentrations was analyzed using the BIAavaluation software, and the respective koff values were finally averaged.

Influence of mutations on receptor activation For the monitoring of receptor activation in vitro, a chemotaxis assay with the pro-monocytic THP-1 cell line was chosen, which carries all three GPCR for wtRANTES, CCR1, CCR3 and CCR5. As expected, wtRANTES attracted THP-1 cells most potently at a concentration of 10 nM. Both RANTES mutants, A22K and H23K, exhibited completely impaired chemotaxis on monocytes compared with wtRANTES, which exhibited the bell-shaped curve typical for chemokines (Fig. 4). The reason for the mutants being chemotactically inactive was apparently the N-terminal methionine, which was also found to impair Ca2þ mobilization of these proteins (data not shown).

Testing of RANTES mutants in a model of autoimmune uveitis Clinical disease was scored daily by visual inspection of the eyes of rats treated with RANTES mutants or PBS as

Fig. 4. Dose-dependent chemotaxis experiments of THP-1 cells using wtRANTES and A22K and H23K mutants in a conventional modified Boyden chamber set-up. A typical dose-dependent chemotactic response was observed for wtRANTES, whereas both mutants proved to be chemotactically inactive.

371

B.Brandner et al.

Fig. 5. Time course of experimental autoimmune uveitis. Rats (n ¼ 4) induced with 15 mg PDSAg in complete Freund’s adjuvant, fortified with Mycobacterium tuberculosis strain H37RA, were treated from days 1–19 either with PBS or with 100 mg RANTES H23K per day. The average clinical score of all eyes is displayed per group and day.

controls. As can be seen from Fig. 5, treatment with H23K resulted in earlier recovery from ocular inflammation compared with the control group, which was followed by a delayed onset of uveitis in some previously unaffected eyes. This could be explained by the prevalence of HSPGs on immigrating leukocytes which act as a non-specific ‘neutralizing’ sink for the H23K mutant, thereby disrupting the therapeutic effect of the protein at increasing numbers of infiltrated leukocytes. Interestingly, A22K did not show any effect in this animal model (data not shown). Discussion Different routes of interfering with chemokine-driven, pathological inflammatory responses have been taken in the past. In this protein engineering study, we have investigated the combined effect of increased GAG-binding affinity and N-methionylation on the biophysical as well as on in vivo properties of RANTES. It is known that the retention of a single methionine residue at the N-terminus of RANTES leads to a functional receptor antagonist (Proudfoot et al., 1996). This modification by itself, although effective in several animal models, was, however, not sufficient to take Met-RANTES into clinical trials. Knocking out GAG binding of chemokines (in combination with N-methionylation) was pioneered by the Proudfoot group, and its therapeutic effect based on saturating leukocytes with non-sticky chemokines is still under intensive consideration (Wells et al., 2006). We have chosen to knock-in higher GAG-binding affinity, thereby creating a chemokinebased GAG antagonist which, in combination with N-methionylation, was expected to show anti-inflammatory activity in vivo. For this purpose, a thorough biophysical analysis of a number of RANTES mutants was performed. For engineering increased affinity towards HSPG co-receptors into RANTES, the amino acids Thr43, Asn46, Gln48 and Val49 next to the RANTES GAG-binding epitope 44 RKNR47 were considered for site-directed mutagenesis. An expansion of this linear epitope was proposed by engineering the N-proximal loop containing Ala22 and His23 (Fig. 6). GAG-binding protein domains are usually characterized by 372

Fig. 6. Molecular structure of the RANTES monomer (derived from the NMR structure, PDB entry 1HRJ) indicating the known GAG-binding motif 44 RKNR47 (green) as well as the sites A22 and H23 (red) where mutations were introduced in this study.

either a linear sequence or spatial vicinity of basic residues which interact with negatively charged sulfate groups of GAGs. These electrostatic interactions determine the affinity between GAGs and proteins. Therefore, on the basis of the X-ray structure of RANTES, further basic amino acids were introduced into the GAG-binding domain on the chemokine (this region of the protein is highlighted in Fig. 6). For this purpose, the flexible lysine residue was used more frequently than the rather rigid arginine residue in order to allow a better induced fit upon GAG ligand binding. In isothermal fluorescence titration experiments, all RANTES mutants, except for N46K/Q48K, gave higher affinities towards a GAG oligosaccharide (Rek et al., manuscript in press). The two single-site mutants which gave the highest improvement in GAG affinity, A22K and H23K, were chosen for intensive equilibrium and kinetic interaction studies with HS by SPR. As shown in Fig. 6, these modifications created, in addition to the well-known linear 44RKNR47 GAG-binding epitope, an extended 3-D GAG-binding epitope which was found to be responsible for higher GAG-binding affinity. The GPCR antagonistic effect of N-methionylation was not affected by the A22K and H23K mutations as determined by a chemotaxis assay (Fig. 4). Interestingly, engineering a lysine residue into position 22 further enhanced the ability of RANTES to form oligomers/ aggregates, whereas a lysine at position 23 significantly reduced the tendency to form aggregates (Fig. 1). With respect to GAG binding, A22K gave an almost 10-fold lower affinity compared with H23K as determined by SPR. Apparently, the RANTES – RANTES interaction site is very sensitively interfaced with the GAG interaction site. With respect to residual time on GAGs, both mutants were found in SPR measurements to dissociate from HS much quicker than both Met-RANTES and wtRANTES (Fig. 3). GAGs were shown to induce wtRANTES oligomers (Hoogewerf et al., 1997; Johnson et al., 2004a). GAG-induced aggregation might therefore be responsible for the extended residual time of Met-RANTES and wtRANTES on HS compared with the A22K and H23K mutants. Although these RANTES variants exhibited higher GAG-binding affinity compared


with wtRANTES owing to the artificial 3-D GAG-binding epitope created in the A22K and H23K mutants, it is hypothesized that a deficiency in GAG-induced oligomerization has been introduced by these mutations. This could also explain the only minor therapeutic effect of the mutants in the autoimmune uveits model: the A22K mutant, which remained significantly shorter on HS than the H23K mutant (Fig. 3), did not show any anti-inflammatory effect in the rat model of autoimmune uveitis. The transient therapeutic effect of the H23K mutant in this model is thus explained by its higher affinity towards GAGs compared with A22K and wtRANTES. In addition, the H23K mutant has partially retained the potential to form GAG-induced oligomers, which is reflected in the improved residual time on GAGs compared with A22K. Future protein engineering experiments are therefore needed to improve on the residual time of therapeutic RANTES molecules on GAGs. Acknowledgements We are grateful to Simi Ali for providing the pRSET A/RANTES plasmid and we are indebted to John Gallagher for providing size-fractionated heparin.

Proudfoot,A.E., Fritchley,S., Borlat,F., Shaw,J.P., Vilbois,F., Zwahlen,C., Trkola,A., Marchant,D., Clapham,P.R. and Wells,T.N. (2001) J. Biol. Chem., 276, 10620–10626. Proudfoot,A.E., Handel,T.M., Johnson,Z., Lau,E.K., LiWang,P., Clark-Lewis,I., Borlat,F., Wells,T.N. and Kosco-Vilbois,M.H. (2003) Proc. Natl Acad. Sci. USA, 100, 1885– 1890. Rek,A., Geretti,E., Goger,B. and Kungl,A.J. (2002) Biophys. Biochem., 2, 319–340. Ribeiro,S. and Horuk,R. (2005) Pharmacol. Ther., 107, 44–58. Ricard-Blum,S., Feraud,O., Lortat-Jacob,H., Rencurosi,A., Fukai,N., Dkhissi,F., Vittet,D., Imberty,A., Olsen,B.R. and van der Rest,M. (2004) J. Biol. Chem., 279, 2927– 2936. Rot,A. (1992) Immunol. Today, 13, 291– 294. Simmons,G., Clapham,P.R., Picard,L., Offord,R.E., Rosenkilde,M.M., Schwartz,T.W., Buser,R., Wells,T.N. and Proudfoot,A.E. (1997) Science, 276, 276– 279. Skelton,N.J., Aspiras,F., Ogez,J. and Schall,T.J. (1995) Biochemistry, 34, 5329– 5342. Slimani,H., Charnaux,N., Mbemba,E., Saffar,L., Vassy,R., Vita,C. and Gattegno,L. (2003) Biochim. Biophys. Acta, 1617, 80–88. Sreerama,N. and Woody,R.W. (1993) Anal. Biochem., 209, 32–44. Tumova,S., Woods,A. and Couchman,J.R. (2000) J. Biochem. Cell. Biol., 32, 269–288. Wells,T.N.C., Power,C.A., Shaw,J.P. and Proudfoot,A.E.I. (2006) Trends Pharmacol. Sci., 27, 41– 47. Witt,D.P. and Lander,A.D. (1994) Curr. Biol., 4, 394–400. Received March 30, 2009; accepted March 31, 2009 Edited by Thomas Kiefhaber

Funding This work was supported by the Austrian Science Fund P17872 B09, by the Austrian Ministry of Education, Science and Arts (GEN-AU programme, APP-II) and by the Deutsche Forschungsgemeinschaft SFB 571. References Ali,S., Palmer,A.C., Banerjee,B., Fritchley,S.J. and Kirby,J.A. (2000) J. Biol. Chem., 275, 11721–11727. Appay,V., Brown,A., Cribbes,S., Randle,E. and Czaplewski,L.G. (1999) J. Biol. Chem., 274, 27505–27512. Blanpain,C., Doranz,B.J., Bondue,A., Govaerts,C., De Leener,A., Vassart,G., Doms,R.W., Proudfoot,A. and Parmentier,M. (2003) J. Biol. Chem., 278, 5179–5187. Butcher,E.C. (1991) Cell, 67, 1033–1036. Czaplewski,L.G., et al. (1999) J. Biol. Chem., 274, 16077–16084. de Smet,M.D., Bitar,G., Roberge,F.G., Gery,I. and Nussenblatt,R.B. (1993) J. Autoimmun., 6, 587–599. Gao,W., Topham,P.S., King,J.A., Smiley,S.T., Csizmadia,V., Lu,B., Gerard,C.J. and Hancock,W.W. (2000) J. Clin. Invest., 105, 35– 44. Gerard,C. and Rollins,B.J. (2001) Nat. Immunol., 2, 108– 115. Goger,B., Halden,Y., Rek,A., Mosl,R., Pye,D., Gallagher,J. and Kungl,A.J. (2002) Biochemistry, 41, 1640–1646. Gong,J.H., Uguccioni,M., Dewald,B., Baggiolini,M. and Clark-Lewis,I. (1996) J. Biol. Chem., 271, 10521–10527. Hoogewerf,A.J., Kuschert,G.S.V., Proudfoot,A.E.I., Borlat,F., Clark-Lewis,I., Power,C.A. and Wells,T.N.C. (1997) Biochemistry, 36, 13570–13578. Johnson,Z., et al. (2004a) J. Immunol., 173, 5776– 5785. Johnson,Z., et al. (2004b) Biochem. Soc. Trans., 32, 366– 377. Middleton,J., Neil,S., Wintle,J., Clark-Lewis,I., Moore,H., Lam,C., Auer,M., Hub,E. and Rot,A. (1997) Cell, 91, 385– 395. Pakianathan,D.R., Kuta,E.G., Artis,D.R., Skelton,N.J. and Hebert,C.A. (1997) Biochemistry, 36, 9642–9648. Park,P.W., Reizes,O. and Bernfield,M. (2000) J. Biol. Chem., 275, 29923– 29926. Potzinger,H., Geretti,E., Brandner,B., Wabitsch,V., Piccinini,A.M., Rek,A. and Kungl,A.J. (2006) Biochem. Soc. Trans., 34, 435– 437. Power,C.A. and Proudfoot,A.E. (2001) Curr. Opin. Pharmacol., 1, 417– 424. Proudfoot,A.E. and Borlat,F. (2000) Methods Mol. Biol., 138, 75–87. Proudfoot,A.E., Power,C.A., Hoogewerf,A.J., Montjovent,M.O., Borlat,F., Offord,R.E. and Wells,T.N. (1996) J. Biol. Chem., 271, 2599– 2603. Proudfoot,A.E., Buser,R., Borlat,F., Alouani,S., Soler,D., Offord,R.E., Schroder,J.M., Power,C.A. and Wells,T.N. (1999) J. Biol. Chem., 274, 32478– 32485.

373