Eur. J. Biochem. 225, 635-640 (1994) 0 FEBS 1994

Covalent modification of the interleukin-5 receptor by isothiazolones leads to inhibition of the binding of interleukin-5 RenC DEVOS', Yves GUISEZ', Geert PLAETINCK', Sigrid CORNELIS', Jan TAVERNIER', Jos6 VAN DER HEYDEN', Louise H. FOLEY' and Julie E. SCHEFFLER'

Roche Research Gent, Gent, Belgium Hoffmann-La Roche Inc., New Jersey, USA

(Received July 21, 1994) - EJB 94 1096/3

Using a fusion protein of the human interleukin-5-receptor a chain (hILSRa) and the human IgG Cy3 chain (hILSRa-hy3), we have developed a solid-phase assay for high-flux screening of a collection of synthetic compounds. We report on the identification of isothiazolone derivatives as potent inhibitors of binding of interleukin-S (IL5) to the hILSRa, as measured in a solid-phase assay (soluble hlLSRa or hIL5Ra-hy3) or on COS-1 cells expressing the hILSRa on the cell membrane. The binding of hIL4 and human granulocyte macrophage colony-stimulating factor (hGM-CSF) to their respective receptors is not inhibited by the isothiazolones in similar assay systems. Scatchard analysis revealed that these compounds caused a decrease in affinity of the IL5Ra for lL5. The inhibition of binding ILS to its receptor by the isothiazolone derivatives is abrogated by free- sulfhydryl-containing compounds such as dithiothreitol, indicating that the isothiazolones react with the sulfhydryl group of free cysteine residues in the hIL5Ra. Mutation of Cys66 led to a receptor which still binds hILS, but which was insensitive to the inhibition by isothiazolones. Mutation of Cys249 and Cys296 to serine resulted in complete loss of IL-5-binding activity. The use of radio- labeled isothiazolone confirmed that Cys66, present in the first domain of the receptor, is the target for covalent modification leading to a decrease in affinity.

Interleukin-5 (IL5) is a cytokine which is responsible for the differentiation and growth of eosinophils and basophils from their progenitor cells (Sanderson, 1992). IL5 transduces its signal by binding to a specific receptor present on these cells. We and others (Tavernier et al., 1991 ; Kitamura et al., 1991) have demonstrated that the IL5 receptor is composed of two transmembrane polypeptide chains ; an IL5-binding a chain (ILSRa) and a non-115-binding p chain vc), which is shared with the IL3 and granulocyte macrophage colony- stimulating factor (GM-CSF) receptor. Both the ILSRa and the /Ic chain belong to the haematopoietic-growth-factor-re- ceptor family (Bazan, 1990). The binding of IL5 on the IL5Ra chain reveals its association with the fit chain, which results in an increase in binding affinity. We have also dem- onstrated (Tavernier et al., 1991, 1992) that a mRNA cod- ing for a soluble form of the human IL5Ra is generated in eosinophilic cells by alternative splicing, and that the corre- sponding recombinant soluble hILSRa has retained its ILS- binding capacity.

Some progress has been made in defining low-molecular- mass molecules which interfere with the activity of cytokines (Cooper and Masamune, 1992), however none of these com- pounds has been designed to act as a cytokine-binding antagonist or agonist. Using a hybrid molecule in which the soluble (s) hIL5Ra is fused to the heavy chain of human

Correspondence to R. Devos, Roche Research Gent, J. Pla- teaustraat 22, B-9000 Gent, Belgium

Abbreviations. IL, interleukin ; GM-CSF, granulocyte macro- phage colony-stimulating factor; shILSRa, soluble human ILS-re- ceptor a chain.

IgG3, we have developed a solid-phase assay for screening large numbers of chemical compounds in order to find an ILS-binding antagonist (Devos et al., 1993). Here we de- scribe the identification of a class of compounds, the isothia- zolones, that have the capacity to modify the hIL5-binding domain of the ILSRa, resulting in a dramatic loss of affinity for ILS. These compounds do not affect the binding of hIL4 or hGM-CSF to their respective receptors, in similar assay systems. We demonstrate that the isothiazolones react with a free sulfhydryl group in the first domain of the hIL5Ra molecule.

EXPERIMENTAL PROCEDURES

Reagents

Human recombinant IL5, soluble hILSRa, soluble hIL4R and hIL4 were produced in a baculovirus expression system and purified using published procedures (Devos et al., 1993 ; Le et al., 1988). Purified hGM-CSF (Escherichia coli) was provided by Biogen Inc. Radioiodination of the cytokines was with Iodogen (Pierce). All isothiazolone derivatives and the I4C-labeled isothiazolone were synthesized at Hoffmann- La Roche. ['"C]Iodoacetic acid (56 Cilmol) was purchased from Amersham.

Cytokine-binding assays

Binding of '251-labeled hILS or hIL4 was performed in a solid-phase assay using hIL5Ra-hy3 (hy3, human IgG Cy3

636

chain) or hIL4R-hy3 fusion proteins bound onto plastic wells of a microtiter plate (Dynatech Microlyte 2) with goat anti- (human IgG) antibody as described (Devos et al., 1993). In the presence of lOpg/ml shIL5Ra or shIL4R, the binding of hIL5 or hIL4, respectively, was inhibited by over 95%. Alternatively, the binding of hIL5 was measured using solu- ble hIL5Ra bound via the non-neutralizing monoclonal anti- body 6H1 (Devos et al., 1993). After washing with NaCVP, (0.14M NaC1, 2.7 mM KCI, 8 mM Na,HPO,, 1.5 mM KH,PO,), the bound radioactivity was measured in a micro- scintillation counter (Topcount, Packard Canberra Co.). For binding of hGM-CSF, COS-1 cells were transfected with plasmid pCDM8-hGM-CSFRa DNA as described (Tavernier et al., 1991). Scatchard plot analysis of binding hiL5 to hIL5Ra-hy3 and determination of the binding affinity was as described (Devos et al., 1993).

Modification of shIL5Ra with 14C-labeled isothiazolone

Purified shIL5Ra (5 pg) was allowed to react for 10 min at room temperature with 0.12 mM ['4C]isothiazolone H (40 Cdmol; Fig. 1) or ['4C]iodoacetic acid, and loaded onto a NAP-5 column (Pharmacia) equilibrated in NaCVPi. Frac- tions excluded in the void volume and containing radio- activity were concentrated using a Centricon 10 concentrator (Amicon). Next, the proteins were analysed by SDSPAGE, blotted onto a poly(viny1idene difluoride) membrane (Milli- pore), stained with Coomassie blue and exposed to X-ray film.

Mutagenesis of hIL5Ra

Mutagenesis of cysteine residues to serine in the soluble hIL5Ra was carried out by primer-mediated mutagenesis using the PCR method as described by Higuschi (1990). Primers on either side of the baculovirus transfer vector pVL1393 (Invitrogen corp.) containing the shIL5Ra cDNA sequence were employed. The final product was subcloned in pVL1393, and transfected into Sf9 cells using the Baculo- Gold transfection kit (Pharmingen). The presence of shIL5Ra mutants in the supernatant was detected by [35S]methionine (Amersham) labeling and immunoprecipitation with rabbit antiserum raised against shIL5Ra. For mutagenesis of the cysteine residues in the membrane-anchored form of hILSRa, a restriction-selection procedure was employed (Clontech) using the vector pSV-SPORT1 (Gibco-BRL) containing the hIL5Ra cDNA. The resulting mutant plasmids were trans- fected into COS-1 cells for analysis of the binding of hIL5. The presence of the mutations in all plasmids was established by DNA sequencing.

RESULTS

Identification of isothiazolones as inhibitors of hIL5 binding to hIL5Ra

Using a bivalent hIL5Ra-hy3 fusion receptor we have developed a solid-phase assay for screening compounds for inhibition of binding of hIL5 (Devos et al., 1993). Out of about 67 500 compounds, we identified one which exhibited significant inhibition. The IC,, of this compound, 2-(2-pyri-

ure R1 R7 I73

H H H H H H H CI CI H CI CI H Br CN CI

Fig. 1. Chemical structures of the isothiazolone derivatives used in this study.

100 I 0 hlL5

80 I O h'L4 n 70 x

60

50

40

30

v

c

.- c -

l o 0 * 0.0 1 0.1 1 10 100

lsothiazolone A (pg/ml) Fig. 2. Inhibition of the binding of hILS to hILSRa by isothiazo- lone A. 100 pl (4 nM) '251-hIL5 (0) or '251-hIL4 (0) in NaCW, containing 1% BSA was incubated at room temperature for 1 h in the presence of different concentrations of isothiazolone A, in a microtiter plate containing hIL5Ra-hy3 (0) or hIL4R-hy3 (0). After washing the plate, the bound radioactivity was measured. The per- centage inhibition was calculated relative to the radioactivity meas- ured in the absence of isothiazolone A (0% inhibition).

dyl)-4-isothiazolin-3-one (structure A, Fig. 1) and of related derivatives is approximately 4 pg/ml (20 pM). A plateau was reached at approximately 80% inhibition of binding of hIL5 at a concentration of isothiazolone of 10 pg/ml (Fig. 2). These compounds also inhibited the binding of hIL5 to the monovalent soluble hIL5Ra (Devos et al., 1993), and to the membrane-anchored hIL5Ra expressed on COS-1 cells (Tav- ernier et al., 1992). The binding of hIL4 to a hIL4R-hy3 fusion receptor in a similar assay to that used for hIL5 bind- ing, was not inhibited by the isothiazolones (Fig. 2). The binding of hGM-CSF on COS-1 cells transfected with the hGM-CSFRa was not influenced by the presence of 50 pg/ ml of isothiazolone structure A (data not shown).

637

1 .O

0.8

0)

? 0.6 u, \ U c 2 0.4 m

0.2

0.0

1 0:- lsothiazolone A

Qk, 0 o:+ lsothiazolone A '.

d, '\, Kd=0.22nM

O'.,, b Kd=8.60nM O'& I\

I I

0 50 100 150 200 250 300

Bound (pM1 Fig.3. Effect of isothiazolone A on the binding of hlILS on hIL5Ra as measured by Scatchard analysis. 100 p1 1Z51-hIL5 rang- ing from 17 pM to 35 nM was incubated (in triplicate) overnight at 4°C in NaClP, containing 1 % BSA in the absence (0) or presence (0) of 20 pg/ml isothiazolone A in a microtiter plate containing a suboptimal amount of hIL5Ra-hy3 (Devos et al., 1993). Scatchard analysis and determination of the Kd value was performed using the LIGAND program (Munson and Rodbard, 1980).

Affinity of the hIL5Ra for hIL5 in the presence of isothiazolone

We previously reported the affinity of hIL5Ra for hIL5 as measured in a solid-phase assay (Kd = 0.2 nM; Devos et al., 1993). When measured in the presence of isothiazolone A, this affinity was reduced dramatically (over 40-fold, Fig. 3). Prior incubation of the hIL5Ra-hy3 protein with the isothiazolone followed by washing still inhibited the subse- quent binding of hIL5 (data not shown). This indicated that these chemicals interact with the receptor rather than with hIL5. No competitive inhibition was observed in the presence of the isothiazolones, as the IC,, remained the same at con- centrations of hIL5 of 20, 4 and 0.2 nM. Addition of isothia- zolone to the hIL5Ra after incubation with hIL5 still inhib- ited the binding of hIL5, thus showing that the compound was able to dissociate the receptor-bound ligand (data not shown). These results suggest that the isothiazolones proba- bly interact covalently with the hIL5Ra molecule, causing a decrease in the affinity for hIL5.

Isothiazolones react with free sulfhydryl groups in hIL5Ra

The inhibition of binding of hIL5 on the hIL5Ra by the isothiazolones was completely abrogated by the presence of free sulfhydryl-containing compounds such as dithiothreitol (Fig. 4). This led us to conclude that these compounds are reacting with the sulfhydryl group of free cysteine residues in a polypeptide chain, resulting in the formation of a disulfide- bridged adduct. Such a reaction has been demonstrated by isolation and full characterization of the adduct between N- acetylcysteine with isothiazolone F (Scheffler, J. E., personal communication). The adduct between N-acetylcysteine and isothiazolone F was as potent for inhibition of binding of hIL5 on its receptor as the isothiazolone F itself (data not shown). Therefore, the disulfide-exchange reaction of the isothiazolonelN-acetylcysteine adduct with the sulfhydryl

40 t

f + lsothiazolone A -d

0 I I ' ' " ' . . ' I ' '"n"'' ' " ' ' ' " 1 ' -'dJ 0.1 1 10 100 1000

[Dithiothreitol] (pM) Fig. 4. Neutralization by dithiothreitol of the isothiazolone-A-in- duced inhibition of hIL5 binding to hILSRa. 100 pl (4 nM) 12*1- hIL5 containing different concentrations of dithiothreitol was incu- bated at room temperature for 1 h in the absence (0) or presence (0) of 12.5 pg/rnl isothiazolone A in a microtiter plate containing hIL5Ra-hy3. After washing, the bound radioactivity was measured.

group of free cysteine residues in the hIL5Ra molecule, is possibly as efficient as the reaction of the S in the isothiazo- lone ring. This has also been demonstrated by reaction of the adduct with N-acetylcysteine to afford N,N-diacetylcystine and the corresponding mercaptoacrylamide (Scheffler, J. E., personal communication).

Identification of the cysteine residue in the hIL5Ra modified by isothiazolone

The hIL5Ra molecule contains seven cysteine residues. In analogy with the other members of the cytokine-receptor family (Bazan, 1990) it is thought that the four cysteines in the second domain are involved in disulfide bridges (Fig. 5). To learn which of the three remaining cysteine residues is modified by the isothiazolone, we made use of 14C-labeled isothiazolone H, and of the fact that some preparations of shIL5Ra produced in a baculovirus-insect cell expression system contain molecules specifically cleaved by a protease. This cleavage results in a one-third (N-terminus containing Cys66) fragment, and a two-thirds fragment containing Cys249 and Cys296. Since both fragments are still held to- gether by a disulfide bridge, and can only be revealed by PAGE under reducing conditions (Fig. 6A), the position of the protease-cleavage site is located in one of the loops of the second domain of the receptor. The exact position of this cleavage site was determined by N-terminal amino acid se- quence analysis of the two-thirds fragment. The sequence obtained (LRSYQ) identified this position in the loop formed by the disulfide bridge connecting Cysll4 with Cys135 (Fig. 5). After reaction of a partially cleaved preparation of shIL5Ra with 14C-labeled isothiazolone H, the receptor was analysed by PAGE under non-reducing and reducing condi- tions. Fig. 6B shows that the label was found in both the intact receptor molecule, and in the smallest fragment con- taining Cys66 generated under reducing conditions. No label was found in the two-thirds fragment. This experiment also suggested that the disulfide bond of the adduct between the receptor and the isothiazolone is resistant to exchange by the sulfhydryl group of 2-mercaptoethanol.

638

PROTEASE

- D 0 M A I N

1 -

- n 0 M A I N

2 -

- D 0 M A I N

3 __

CELLMEMBRANE

Fig. 5. Topology diagram of the hILSRa. Three domains contain- ing approximately 100 amino acid residues each are shown. The marked arrows represent the presumed P-strands, and the lines repre- sent presumed loops connecting the P-strands and the domains as predicted by Bazan (1990). The first domain contains the cysteine residue (position 66) which is modified by isothiazolone compounds (arrow). The two disulfide bridges conserved in the cytokine recep- tor family and connecting the A and B strands and the D and E strands (dashed line) in the second domain are indicated. The loop connecting the A and B strands (arrow) i n this domain is also subject to attack by a protease derived from the baculovirus expression sys- tem. The cytokine-receptor consensus sequence WSXWS in the G strand of the third domain and the two cysteine residues (positions 249 and 296) are also indicated.

Mutational analysis of the cysteines in the hILSRa

To confirm that Cys66 is modified by the isothiazolones, we constructed hILSRu mutants in which either Cys66, Cys249, or Cys296 was mutated to serine. Soluble hIL5Ra mutants were constructed and expressed in baculovirus-in- fected insect cells: also mutants of the membrane-anchored

1 M 2 1 M 2 Fig.6. Modification of Cys66 in the soluble hIL5Ra by 14C-la- beled isothiazolone H. Soluble hIL5Ra was modified with ''C-la- beled isothiazolone H and analysed on a 10% polyacrylamide/SDS gel in the absence (lane 1) or presence (lane 2 ) of 2-mercapto- ethanol. The proteins were blotted on a poly(viny1idene difluoride) membrane and stained with Coomassie blue (A), and an autoradio- graph (B) was made. a, position of the intact soluble hIL5Ra; b, position of the C-terminal 2/3 fragment; c, position of the N-termi- nal 1/3 fragment. (M), ''C-labeled rainbow marker (Amersham).

25

&

$i 20 0

0 rn

c-

15 - V c 3 0

In

c

10

1 I

2 z 5

0

C66S

0.1 1 10 100

lsothiatolone A (pg/ml) Fig. 7. hIL5-binding capacity of cysteine to serine mutants of soluble hIL5Ra and the effect on inhibition by isothiazolone A. Wild-type soluble hIL5Ra (wt, 0 ) and cysteine to serine mutants (C66S, W ; C249S, A; and C296S, V) of the soluble hIL5Ra present in the supernatant of recombinant baculovims-infected Sf9 cells, were bound onto wells of a microtiter plate via a non-neutralizing monoclonal antibody 6H1 (Devos et al., 1993). The binding of lZ5I- hIL5 was measured in the presence of different concentrations of isothiazolone A.

receptor were synthesized and expressed on the surface of COS-1 cells. In both cases we observed that the C66S mutant of the hIL5Ra was still able to bind hIL5. The affinity of binding hIL5 of this mutant however was not affected by the isothiazolone (Fig. 7). Both C249S and C296S hIL5Ra mutants completely lost hIL5-binding capacity (Fig. 7). Structure prediction of the membrane-proximal domain of the hIL5Ra chain based on analogy with the growth-hormone receptor indicated that these two Cys residues are located in parallel P-strands (strand C and strand F, Fig. 5). Possibly, these two cysteine residues are involved in a third disulfide bridge, leaving Cys66 as the only free extracellular cysteine residue.

639

DISCUSSION

The extracellular part of the cytokine receptors contains domains which have been predicted to form a barrel structure consisting of seven P-strands and loops connecting them (Ba- zan, 1990). Most members of this family, like human growth- hormone receptor and IL6R contain two such domains form- ing a double-barrel structure. The ILSRa, GM-CSFRa and IL3Ra contain three such domains, while the /Ic chain con- sists of a duplicated double-barrel structure. Mutational analysis of the growth-hormone receptor (Bass et al., 1991) and the IL6R (Yawata et al., 1993) has demonstrated that the residues critical for the binding of the ligand are located on both domains in or near the hinge region between them. The N-terminal barrel in the hIL5Rc is also involved in the bind- ing of IL5, since deletion of this domain results in loss of IL5-binding capacity (data not shown). The modification of the free cysteine residue (Cys66) in this domain by isothiazo- lone yields a receptor with a 40-fold reduced IL5-binding affinity. Isothiazolones are compounds known to react with free sulfhydryl groups (Crow and Leonard, 1965; Crow and Gosney, 1970; Chan et al., 1970; Collier et al., 1990 a, b), and compounds like dithiothreitol neutralize the inhibitory effect of the isothiazolones. The adduct between N-acetylcys- teine and isothiazolone also inhibits the binding of IL5 to its receptor, indicating that a disulfide-exchange reaction can occur with Cys66 in the hIL5Ra. Other cysteine-modifying reagents such as iodoacetic acid, 5,5'-dithio-bis(2-nitroben- zoic acid) (Ellman's reagent), IV-ethylmaleimide, or 2,2'-di- thiodipyridine, did not influence the IL5-binding capacity of hIL5Ra (data not shown). This means that Cys66 is not ac- cessible to these chemicals, or that modification of Cys66 by these reagents does not interfere with the subsequent interac- tion of hIL5Ra with IL5. In contrast to the isothiazolones, I4C-labeled iodoacetic acid is not incorporated into non-re- duced, non-denatured hIL5Ra (data not shown). This indi- cates that the lack of inhibition of IL5 binding, observed with this sulfhydryl reagent is due to the inaccessability of free cysteine residues. Perhaps the bulky iodine atom in iodoace- tic acid does not allow access to the sulfhydryl of Cys66. Steric hindrance might also be the reason why isothiazolone derivatives containing a cyclohexyl group on the ring nitro- gen, are less inhibitory than the isothiazolones described in Fig. 1 (data not shown). Although the extracellular part of hIL4R (Idzerda et al., 1990) and of hGM-CSFRn (Gearing et al., 1989) contain multiple cysteine residues (seven and eleven cysteine residues, respectively, of which at least four are thought to be in disulfide bridges), the isothiazolones did not influence the binding of these cytokines to their recep- tors.

The second domain of hIL5Ru contains four cysteine res- idues conserved among the members of the cytokine family, which are predicted to form two disulfide bridges. Further- more, this domain contains a loop connecting the A and B strands, which in the case of the hIL5Ra is much longer (10 residues) compared to the other cytokine receptors. We have demonstrated that this loop in the hIL5Ra can be specifically cleaved (between Arg125 and Lys126) by a protease. Finally, the third, membrane-proximal, barre! structure contains the motif WSXWS, which is also conserved in the cytokine-re- ceptor family. The integrity of this motif is essential for li- gand binding (Miyazaki et al., 1991) and underscores the important role of the membrane-proximal domain in the proper folding of the cytokine receptors. Two cysteines are located in this domain of the hILSRa, which perhaps form a

disulfide bridge connecting the loop between the G and the F strand with strand C. Substitution by serine of either of these cysteine residues resulted in a receptor which lacks IL5-binding capacity. Probably the substitution of one of these cysteine residues may disturb the overall conformation, perhaps by preventing the formation of an interconnecting disulfide bond.

The present study shows that a class of compounds, the isothiazolones, cause a dramatic fall in affinity of the hIL5Ra for hIL5. Isothiazolones are known to react with nucleophiles including thiols, and it is shown that they react with a free cysteine in the extracellular part of the hIL5Ra. These com- pounds are widely used as industrial biocides, and their anti- microbial activity is thought to be due to their interaction with intracellular accessible thiols such as gluthathione (Col- lier et al., 1990 a, b). More recently, these compounds were also found to be inhibitors for the alginate gene activiation in Pseudomonas aeruginosa (Roychoudhury et al., 1993). The isothiazolones may be useful as thiol-specific reagents in other proteins and may provide a general tool to identify cysteine residues involved in protein-protein interactions.

We thank Annick Verhee, Tania Tuypens and Ina FachC for their technical help during part of this work; Johan Bostoen for prepara- tion of the figures; Prof. Joel Vandekerckhove for protein sequenc- ing; Prof. Walter Fiers, Dr Michael Steinmetz and Dr Ann Welton for helpful discussions.

REFERENCES Bass, S. H., Mulkerrin, M. G. & Wells, J. A. (1991) A systematic

mutational analysis of hormone-binding determinants in the hu- man growth-hormone receptor, Proc. Natl Acad. Sci USA 88,

Bazan, J. F. (1990) Structural design and molecular evolution of a cytokine receptor superfamily, Proc. Nut1 Acud. Sci USA 87,

Chan, A. W. D., Crow, W. D. & Cosney, 1. (1970) Isothiazole Chem- istry-VIII; Base-catalyzed dimerization of N-alkyl-3-isothiazo- lones, a synthesis of 2,4-bismethylene-1,3-dithietanes, Tetrahe- dron 26, 1493-1502.

Collier, P. J., Ramsey, A,, Waigh, R. D., Douglas, K. T., Austin, P. & Gilbert, P. (1990a) Chemical reactivity of some isothiazolone biocides, J. Appl. Bucteriol. 69, 578-584.

Collier, P. J., Ramsey, A. J., Austin, P. & Gilbert, P. (1990b) Growth inhibitory and biocidal activity of some isothiazolone biocides, J. Appl. Bucteriol 69, 569-517.

Cooper, K. & Masamune, H. (1992) Cytokine modulation as a me- dicinal chemistry target, Annu. Rep. Med. Chem. 27, 209-218.

Crow, W. D. & Leonard, N. J. (1965) 3-isothiazolone-cis-3- thiocyanoacrylamide equilibria, J. Org. Chem. 30, 2660- 2665.

Crow, W. D. & Gosney, I. (1970) Isothiazole Chemistry-1X; Selec- tivity in carbanion attack on N-ethyl-3-isothiazolone, Tetruhe- dron 26, 1463-1473.

Devos, R., Guisez, Y., Cornelis, S., Verhee, A,, Van der Heyden, J., Manneberg, M., Lahm, H.-W., Fiers, W., Tavernier, J. & Plae- tinck, G. (1993) Recombinant soluble human interleukin-5 (hIL- 5 ) receptor molecules : cross-linking and stoichiometry of bind- ing to IL-5, J Biol. Chem 268, 6581 -6578.

Gearing, D. P., King, J. A,, Gough, N. M. & Nicola, N. A. (1989) Expression cloning of a receptor for human granulocyte-macro- phage colony-stimulating factor, EMBO J. 8, 3667 -3676.

Higuschi, R. (1990) Recombinant PCR, in PCR Protocols: a guide to methods and applications (Innis, M. A., Gelfand, D. H., Snin- sky, J. J. & White, T. J., eds) pp. 177-183, Academic Press, San Diego.

Idzerda, R. L., March, C. J., Mosley, B., Lyman, S. D., Vanden Bos, T., Gimpel, S. D., Din, W. S., Grabstein, K. H., Widmer, M. B., Park, S. L., Cosman, D. & Beckmann, M. P. (1990) Human

4498 -4502.

6934-6938.

639

DISCUSSION

The extracellular part of the cytokine receptors contains domains which have been predicted to form a barrel structure consisting of seven P-strands and loops connecting them (Ba- zan, 1990). Most members of this family, like human growth- hormone receptor and IL6R contain two such domains form- ing a double-barrel structure. The ILSRa, GM-CSFRa and IL3Ra contain three such domains, while the /Ic chain con- sists of a duplicated double-barrel structure. Mutational analysis of the growth-hormone receptor (Bass et al., 1991) and the IL6R (Yawata et al., 1993) has demonstrated that the residues critical for the binding of the ligand are located on both domains in or near the hinge region between them. The N-terminal barrel in the hIL5Rc is also involved in the bind- ing of IL5, since deletion of this domain results in loss of IL5-binding capacity (data not shown). The modification of the free cysteine residue (Cys66) in this domain by isothiazo- lone yields a receptor with a 40-fold reduced IL5-binding affinity. Isothiazolones are compounds known to react with free sulfhydryl groups (Crow and Leonard, 1965; Crow and Gosney, 1970; Chan et al., 1970; Collier et al., 1990 a, b), and compounds like dithiothreitol neutralize the inhibitory effect of the isothiazolones. The adduct between N-acetylcys- teine and isothiazolone also inhibits the binding of IL5 to its receptor, indicating that a disulfide-exchange reaction can occur with Cys66 in the hIL5Ra. Other cysteine-modifying reagents such as iodoacetic acid, 5,5'-dithio-bis(2-nitroben- zoic acid) (Ellman's reagent), IV-ethylmaleimide, or 2,2'-di- thiodipyridine, did not influence the IL5-binding capacity of hIL5Ra (data not shown). This means that Cys66 is not ac- cessible to these chemicals, or that modification of Cys66 by these reagents does not interfere with the subsequent interac- tion of hIL5Ra with IL5. In contrast to the isothiazolones, I4C-labeled iodoacetic acid is not incorporated into non-re- duced, non-denatured hIL5Ra (data not shown). This indi- cates that the lack of inhibition of IL5 binding, observed with this sulfhydryl reagent is due to the inaccessability of free cysteine residues. Perhaps the bulky iodine atom in iodoace- tic acid does not allow access to the sulfhydryl of Cys66. Steric hindrance might also be the reason why isothiazolone derivatives containing a cyclohexyl group on the ring nitro- gen, are less inhibitory than the isothiazolones described in Fig. 1 (data not shown). Although the extracellular part of hIL4R (Idzerda et al., 1990) and of hGM-CSFRn (Gearing et al., 1989) contain multiple cysteine residues (seven and eleven cysteine residues, respectively, of which at least four are thought to be in disulfide bridges), the isothiazolones did not influence the binding of these cytokines to their recep- tors.

The second domain of hIL5Ru contains four cysteine res- idues conserved among the members of the cytokine family, which are predicted to form two disulfide bridges. Further- more, this domain contains a loop connecting the A and B strands, which in the case of the hIL5Ra is much longer (10 residues) compared to the other cytokine receptors. We have demonstrated that this loop in the hIL5Ra can be specifically cleaved (between Arg125 and Lys126) by a protease. Finally, the third, membrane-proximal, barre! structure contains the motif WSXWS, which is also conserved in the cytokine-re- ceptor family. The integrity of this motif is essential for li- gand binding (Miyazaki et al., 1991) and underscores the important role of the membrane-proximal domain in the proper folding of the cytokine receptors. Two cysteines are located in this domain of the hILSRa, which perhaps form a

disulfide bridge connecting the loop between the G and the F strand with strand C. Substitution by serine of either of these cysteine residues resulted in a receptor which lacks IL5-binding capacity. Probably the substitution of one of these cysteine residues may disturb the overall conformation, perhaps by preventing the formation of an interconnecting disulfide bond.

The present study shows that a class of compounds, the isothiazolones, cause a dramatic fall in affinity of the hIL5Ra for hIL5. Isothiazolones are known to react with nucleophiles including thiols, and it is shown that they react with a free cysteine in the extracellular part of the hIL5Ra. These com- pounds are widely used as industrial biocides, and their anti- microbial activity is thought to be due to their interaction with intracellular accessible thiols such as gluthathione (Col- lier et al., 1990 a, b). More recently, these compounds were also found to be inhibitors for the alginate gene activiation in Pseudomonas aeruginosa (Roychoudhury et al., 1993). The isothiazolones may be useful as thiol-specific reagents in other proteins and may provide a general tool to identify cysteine residues involved in protein-protein interactions.

We thank Annick Verhee, Tania Tuypens and Ina FachC for their technical help during part of this work; Johan Bostoen for prepara- tion of the figures; Prof. Joel Vandekerckhove for protein sequenc- ing; Prof. Walter Fiers, Dr Michael Steinmetz and Dr Ann Welton for helpful discussions.

REFERENCES Bass, S. H., Mulkerrin, M. G. & Wells, J. A. (1991) A systematic

mutational analysis of hormone-binding determinants in the hu- man growth-hormone receptor, Proc. Natl Acad. Sci USA 88,

Bazan, J. F. (1990) Structural design and molecular evolution of a cytokine receptor superfamily, Proc. Nut1 Acud. Sci USA 87,

Chan, A. W. D., Crow, W. D. & Cosney, 1. (1970) Isothiazole Chem- istry-VIII; Base-catalyzed dimerization of N-alkyl-3-isothiazo- lones, a synthesis of 2,4-bismethylene-1,3-dithietanes, Tetrahe- dron 26, 1493-1502.

Collier, P. J., Ramsey, A,, Waigh, R. D., Douglas, K. T., Austin, P. & Gilbert, P. (1990a) Chemical reactivity of some isothiazolone biocides, J. Appl. Bucteriol. 69, 578-584.

Collier, P. J., Ramsey, A. J., Austin, P. & Gilbert, P. (1990b) Growth inhibitory and biocidal activity of some isothiazolone biocides, J. Appl. Bucteriol 69, 569-517.

Cooper, K. & Masamune, H. (1992) Cytokine modulation as a me- dicinal chemistry target, Annu. Rep. Med. Chem. 27, 209-218.

Crow, W. D. & Leonard, N. J. (1965) 3-isothiazolone-cis-3- thiocyanoacrylamide equilibria, J. Org. Chem. 30, 2660- 2665.

Crow, W. D. & Gosney, I. (1970) Isothiazole Chemistry-1X; Selec- tivity in carbanion attack on N-ethyl-3-isothiazolone, Tetruhe- dron 26, 1463-1473.

Devos, R., Guisez, Y., Cornelis, S., Verhee, A,, Van der Heyden, J., Manneberg, M., Lahm, H.-W., Fiers, W., Tavernier, J. & Plae- tinck, G. (1993) Recombinant soluble human interleukin-5 (hIL- 5 ) receptor molecules : cross-linking and stoichiometry of bind- ing to IL-5, J Biol. Chem 268, 6581 -6578.

Gearing, D. P., King, J. A,, Gough, N. M. & Nicola, N. A. (1989) Expression cloning of a receptor for human granulocyte-macro- phage colony-stimulating factor, EMBO J. 8, 3667 -3676.

Higuschi, R. (1990) Recombinant PCR, in PCR Protocols: a guide to methods and applications (Innis, M. A., Gelfand, D. H., Snin- sky, J. J. & White, T. J., eds) pp. 177-183, Academic Press, San Diego.

Idzerda, R. L., March, C. J., Mosley, B., Lyman, S. D., Vanden Bos, T., Gimpel, S. D., Din, W. S., Grabstein, K. H., Widmer, M. B., Park, S. L., Cosman, D. & Beckmann, M. P. (1990) Human

4498 -4502.

6934-6938.

640

interleukin 4 receptor confers biological responsiveness and de- fines a novel receptor superfamily, J. Exp. Med. 171, 861 - 873.

Kitamura, T., Sato, N., Arai, K.-I. & Miyajima, A. (1991) Expres- sion cloning of the human IL-3 receptor cDNA reveals a shared p subunit for the human IL-3 and GM-CSF receptors, Cell 66,

Le, H. V., Ramanathan, L., Labdon, J. E., Mays-Ichinco, C. A., Syto, R., Arai, N., Hoy, P., Takebe, Y., Nagabhushan, T. L. & Trotta, P. P. (1988) Isolation and characterization of multiple variants of recombinant human interleukin 4 expressed in mammalian cells, J. Biol. Chem. 263, 10817-10823.

Miyazaki, T., Maruyama, M., Yamada, G., Hatakeyama, M. & Tani- gushi, T. (1991) The integrity of the conserved ‘WS motif’ com- mon to the IL-2 and other cytokine receptors is essential for ligand binding and signal transduction, EMBO J. 10, 3191- 3197.

Munson, P. J. & Rodbard, D. (1980) Ligand: a versatile computer- ized approach for the characterization of ligand binding systems, Anal. Biochem. 107, 220-239.

Roychoudhury, S., Zielinski, N. A,, Ninfa, A. J., Allen, N. E., Jung- heim, L. N., Nicas, T. I. & Chakrabarty, A. M. (1993) Inhibitors

1165 - 11 74.

of two-component signal transduction systems : Inhibition of al- ginate gene activation in Pseudomonas aeruginosa, Proc. Natl Acad. Sci USA 90, 965-969.

Sanderson, C. J. (1992) Interleukin-5, eosinophils, and disease,

Tavernier, J., Devos, R., Comelis, S., Tuypens, T., Van der Heyden, J., Fiers, W. & Plaetinck, G. (1991) A human high affinity in- terleukin-5 receptor (ILSR) is composed of an ILS-specific a chain and a p chain shared with the receptor for GM-CSF, Cell 66, 1175-1184.

Tavernier, J., Tuypens, T., Plaetinck, G., Verhee, A., Fiers, W. & Devos, R. (1992) Molecular basis of the membrane-anchored and two soluble isoforrns of the human interleukin 5 receptor a sub- unit, Proc. Natl Acad. Sci USA 89, 7041 -7045.

Yawata, H., Yasukawa, K., Natsuka, S., Murakami, M., Yamasaki, K., Hibi, M., Taga, T. & Kishimoto, T. (1993) Structure-function analysis of human IL-6 receptor: dissociation of amino acid resi- dues required for IL-6-binding and for IL-6 signal transduction through gp130, EMBO J. 12, 1705-1712.

Blood 79, 3101 -3109.