DESCRIPTION The present invention relates to a methodology for selecting superagonists, antagonists and superantagonists of human interleukin-6 (hereinafter referred to also as h IL-6 or IL-6) based on three-dimensional modelling.

As is known, WO 92/21029 to Genentec Inc. teaches a method for determination of agonists or antagonists of growth hormones and ligands with a similar structural conformation. The potential agonists and antagonists are put into contact with a receptor for the hormone and this causes formation of a ternary complex consisting of a molecule of the potential agonist or antagonist and two molecules of such receptor for the hormone to be agonized or antagonized. Dimerization of receptors induced by a ligand molecule allows to conclude that the ligand has two different interaction sites (site 1 and site 2) , on which it is possible to operate using mutagenesis to generate agonists or antagonists.

It is known that the ligands in the group of cytokines similar to Interleukin 6 (IL-6), that is Oncostatin M (OS ) , Leukemia Inhibitory Factor (LIF) , Ciliary Neurotrophic Factor (CNTF), and Interleukin 11

(IL-11), induce the formation of a receptor complex or which the membrane molecule gp 130 is a part. In this receptor complex the specific receptor for each of these cytokines and the membrane molecule gp 130 are always present as common elements. It is thus possible to formulate the hypothesis that site 1 and site 2 bind to two different molecules in this class of hormones: site 1 to the specific receptor and site 2 to gp 130. Identification of the two sites is made possible, as will be seen more clearly from the following, by construction of a three-dimensional model of the receptor complex based on the functional similarity between seσuences cf

the human growth hormone (hGK) receptor and sequences of the receptors for the hormones m question. Isolation of variants that, with respect to the wild type hormone, have a greater affinity for the specific receptor (superagonists or superantagonists) is obtained by construction of filamentous phage libraries, for example M13, carrying the hormone, both in the wild type and mutant version.

According to the invention, the difference between the three-dimensional model, for example of IL-6, adopted here and the one adopted in WO92/21029 leads to identify different residues in helix A and C as constituents of site 2. In fact, according to present invention, for the construction of the IL-6 model not the growth hormone, but the structure of a different cytok ne was used as template.

Modelling of the human interleukin 6 molecule is performed as follows. It is known, from data available in scientific literature, that the a ino acidic sequence of human interleukin 6 shows similarities with that of the granulocyte colony stimulating factor (G-CSF) . The three-dimensional structure of bovine granulocyte colony stimulating factor (bG-CSF) , determined using X-ray crystallography, was used as template to develop a three- dimensional model of human IL-6 from residue 16 to 184. Firstly, the amino acidic sequence of human IL-6 was aligned with that of bG-CSF. On the basis of the αer ved alignment, the amino acidic residues in the bG-CSF three- dimensional structure were replaced by the corresponding residues of human IL-6 using molecular modelling program m a computerized interactive graphic unit. In the positions in which alignment involves either deletions or insertions (which suggests a different local structure m the interleukin 6 molecule) adjustments were made by applying the options provided by the molecular modelling program.

This three-dimensional model of interleukin 6, baseα on the bG-CSF structure, has enabled the identification of the two sites of interaction between human interleukin 6 and its two receptors: the low affinity receptor gp 80 (site 1) and the high affinity signal transducer receptor gp 130 (site 2) . The following procedure was used to identify the two sites. From sequence comparison it is known that all the members of the family of hematopoietic receptors are related to each other by the fact that they share a domain, known as the cytokme binding domain. This similarity of sequences also indicates a hign probability of structural similarity m corresponding parts of the various receptors, including the two interleukin 6 receptors, gp 80 and gp 130. The observation that the cytokines that bind to these receptors all have (or are predicted to have) a similar structure, that is a four helix oundle, strongly supports the notion that the interaction between these cytokines and their receptors, by means of the cytokme binding domain, must be very similar - albeit not identical - in biologically active complexes.

Considering that the three-dimensional structure of one of these receptor complexes (the complex made by growth hormone and the extra-cellular domain of the dimeπc receptor for the growth hormone, i.e. GHbp) nas been determined by means of X-ray crystallograpny, our oG-CSF built model of human interleukin 6 allows us to identify the potential sites of interaction between interleukin 6 and its two receptors gp 80 (site 1) and gp 130 (site 2) . This has been accomplisned, according to the present invention, by constructing a structural model of gp 80 and gp 130 based on the coorαinates furnished oy the X-ray christallographic structure of the growth hormone receptor, and by substituting m such complex the growth hormone with our bG-CSF built model of numan mterleukin-ό (see fig. 1) .

As is known, interleukin 6 is a polypeptide of 184 ammo acids which, as described, oelongs to the class of helical cytokines. Interleukin 6 is a multi-functional cytokme produced by various cell types. It acts as a differentiation and growth factor on cells of various lineages, such as for example cells in the immune system, hepatocytes, kidney cells, hematopoietic stem cells, kerat ocytes and neurones.

Production of superagonists of interleukin 6 would allow the use of therapeutic doses lower than those required with wild type interleukin 6 in the treatment of numerous serious diseases. In fact, interleukin 6 has important and promising applications in the treatment of breast cancer, leukemia, and infectious diseases or diseases connected with disorders of bone marrow progenitor cells.

In addition superagonists of IL-6 could be used in protocols for ex vivo expansion of hematopoietic progenitor cells both bone marrow transplantation and gene therapy.

On the other hand the production of antagonists or superantagonists of human interleukin 6 would allow inhibition of interleukin 6 in numerous diseases characterized by its excessive production, such as chronic autoimmune diseases, myeloma/plasmacytoma, post- menopausal osteoporosis and cancer cachexia.

The methodology for the selection of superagonists, antagonists or superantagonists of mterleukm-6, according to the present invention, comprises the following operations: comparing the ammo acid sequence of oovme granulocyte colony stimulating factor (bG-CSF) with the sequence of said hormone; and on the basis of the above comparison, formulating a three αimens onal oαel of saiα hormone, whicn allows the identification of residues that form the site of interaction witn the sυecific receotor 'Site 1)

and those that constitute the site of interaction with gp

130 (Site 2) respectively.

For selection of superagonists of interleukin 6, the methodology according to the ^' present invention further comprises the following additional operations: production of a series of phage libraries containing mutations of the following wild type residues of interleukin 6 (present in the form of fusion product with filamentous phage proteins) : helix A:

Ser 22, Glu 23, Asp 26, Arg 30, Leu 33, Ser 37, Arg 40,

Glu 42; loop AB:

Ser 52, Ser 53, Ala 56, Leu 57, Glu 59, Asn 60, Leu 62, Leu 64, Pro 65, Lys 66,

Met 67, Ala 68, Glu 69, Lys 70, Asp 71, Phe 74, Gin 75,

Ser 76; helix D:

His 164, Leu 165, Arg 168, Ser 169, Lys 171, Glu 172, Phe 173, Gin 175, Ser 176, Ser 177, Leu 178, Arg 179, Ala

180, Leu 181, Arg 182, Gin 183, Met 184.

- selection, from the mixed population of phages belonging to each individual phage library and expressing interleukin 6 mutants, of that or those with an affinity for the specific receptor greater than that of wild type interleukin; and

- identification of the best receptor binder amino acid sequence or sequences by sequencing of the DNA extracted from the selected phage particles. In this case, a series of phage libraries can be produced containing mutations of said wild type residues of interleukin 6 present as a fusion product with the M13 pill protein.

The methodology for selecting antagonists of interleukin 6 according to the present invention comprises - along with the operations indicated above for

- O -

molecular modelling of the human IL-6 protein and its receptor chains - the following operations:

- mutagenesis of the residues identified to form part of the site of interaction with gp 130 (Arg 30, Tyr 31, Gly 35, Ser 37, Ala 38, Ser 118, Lys 120, Val 121, Gin 124, Phe 125, Gin 127, Lys 128 and Lys 129), using conventional molecular biology techniques;

- evaluation of biological activity and affinity for the specific interleukin 6 receptor of the mutants produced as above, in order to identify variants of interleukin 6 whose affinity for the specific receptor is normal and that show reduction or loss of the biological activity; and

- evaluation of the above variants of interleukin 6 as antagonists for tne biological activity of wild type interleukin 6 on human cell lines.

In case of obtaining of superantagonists of interleukin 6 by combination of the variants of ammo acid sequences responsible for antagonist activity, identified as above, with amino acid mutations responsible for an increased affinity of the specific receptor for interleukin 6.

In the methodology for obtaining antagonists or superantagonists of interleukin 6, the mutagenesis of the residues identified as above can be performed using a molecular biology technique chosen from the group comprising Polymerase Cha Reaction, Primer Extension, Ol gonucleotide Directed Mutagenesis, and combinations tnereof. The present invention is not limited to the methodology for selection of superagonists, antagonists or superantagonists of interleukin 6. On the contrary, it extends to molecules ootamable by sa d methodology cf selection, i.e. to: superagonists of h IL-6, witn the exception of tne molecule called IL-6 IRA and carrying the following three substitutions Glnl7ΞIle/Serl76Arg/GInl83Ala; antagonists cf h IL-6,

witn the exception of three molecules with the following suostitutions :

Tyr31Asp/Gly35Phe/Serll8Arg/Vall21Asp (DFRD) Tyr31Asp/Gly35Phe/Serll8Phe/Vall21Asp (DFFD) Tyr31Asp/Gly35Phe/Serll8Leu/Vall21Asp (DFLD) ; and superantagonists of h IL-6, with the exception of the molecule called Santl and carrying the following seven substitutions : Tyr31Asp/Gly35Phe/Serll8Arg/Vall21Asp/ Glnl75Ile/Serl76Arg/Glnl83Ala. Up to this point a general description of the sub ect of the present invention has been given. With the aid of the following examples a detailed description of specific embodiments of the invention will now be given, with the purpose of giving a better understanding of the objects, characteristics, advantages and methods of application thereof.

Figure 1 shows a scheme illustrating the methodology applied to identify site 1 and site 2 in the case of human interleukin 6. Figure 2 shows the increase in potency of three superantagonists according to the invention, i.e. Sant 3, Sant 4 and Sant 5, over antagonist Tyr31Asp/Gly35Phe/ Serll8Arg/Vall21Asp (the one letter codes have been used the figure), with the increase of concentration. DEPOSITS

E.Coli K12 bacteria - transformed using the plasmid pHenΔhIL-6 containing, from the recognition site of the restriction enzyme Sail to that for the restriction enzyme Notl, a nucleotidic sequence cooing for the am o acid sequence of wild type human interleukin 6 - have been deposited on 10/6/1993 with The National Collection of Industrial and Marine Bacteria Ltd. (NCIMB) , Aberdeen, Scotland, UK, with access number NCIMB 40563. Example 1 Application of the methodology according to tne present invention for the selection of suoeraσonists os

- o - interleukin 6 by means cf mutagenesis cf ammo ac d residues tne AB loop

The strategy consists m construction of a hybrid gene containing all the region coding for hIL-6 (SEQ ID NO: 1) followed by the last 157 ammo acids of protein pill of the phage M13 and preceded by the sequence Pel B, which vectors the synthesized protein to the periplasmic space .

This construct allows the obtaining of phagemid particles displaying on their surface correctly folded and biologically active human interleukin 6.

A phage library was constructed containing mutations of residues Asp 71, Phe 74, Gin 75 and Ser 76 of interleukin 6, starting from the variant IL-6 IRA (substitutions Glnl75Ile/Serl76Arg/Glnl83Ala) described in WO95/00852 and having an affinity for the receptor approximately five times greater that that of wild type human interleukin 6, present in the form of fusion product with protein pill of filamentous phage M13. The library was constructed using the Primer Extension technique. The utagenic oligonucleotide is IL-6 DFQS, a 95 nucleotides oligo, whose sequence is SEQ ID NO: 2. Primer IL-6 DFQS introduces degenerations into codons coding for the ammo acids 71 (wild type Asp) , 74 (wild type Phe) , 75 (wild type Gin) and 76 (wild type Ser) . The oligonucleotide IL-6 AB primer, whose sequence is SEQ ID NO: 3, was used as primer for the Primer Extension reaction. The two oligonucleotides were annealed m vitro, and the annealed oligonucleotides were useo as substrate for a Primer Extension reaction. The double- stranαed DNA fragment thus ootamed was then digested and ligated into the plasmid pHenΔhIL-6 in order to replace the wild type sequence with the mutated ones. The ligation product was inserted m bacteria, yielding roughly three million independent transformants . The transformed bacteria were infected with the M13K07 helper

bacteriophage to generate the pnage liorary (a library of phasmids) .

The library underwent selection by incubation with magnetic beads coated with monoclonal antibody directed against shrIL-6R and in the presence of shrIL-6R and shrgpl30. The phasmid population eluted at pH 3.6 was then amplified in bacteria. After four cycles of selection-amplification, randomly selected phasmids were sequenced over the mutagemzed region, the corresponding mutant interleukin 6 proteins were produced in the periplasmic space of the appropriate bacterial strain and tested for interleukin 6 specific receptor binding. Table 1 shows that, using the methodology according to the present invention, it is possible to select variants of interleukin 6 having an additional increase in the affinity for the specific receptor, molecules with mutations both in helix D and in region A-B.

TABLE 1

Receotor binding properties variants of interleukin 6-

IRA cont,ammg additional mutations m the residues 71,

74, 75 and 76 of the region A-B

Position 71 74 75 76 Receptor binding ( ) wwiilldd ttyyppee A Asspp P Phhee G Giinn S Seerr 100%

IL-6IRA Asp Phe Gin Ser 450% phasmid D3- -3 Asp Tyr Phe He 2350°= pnasmid D4- -1 Asp Tyr Tyr Val 2750% phasmid D3- -7 Asp Phe Tyr He 2770% pphhaassmmiidd D D44-- -1199 A Asspp P Phhee T Tyyrr S Seerr 1800% pnasmid D4- -20 Asp Phe Tyr Lys 4200% phasmid D3- -16 Asp Phe Tyr Leu 1450% phasmid D4- -17 Asp Phe Phe He 2430%

Examole 2 ADDIIcation of the methodoloσv acccr ^■ dmσ to the Dresen t- mvention fo V c ^■ btai σ suDerantaσonists cf _nterJ L 'uKln 5

The four mutations Tyr31Asp/Gly3SPhe/Serll8Arg/ Vall21Asp (DFRD) confer antagonistic properties as described m WO95/00852. These four mutations were comomed with mutations capable of increasing the specific receptor binding capacity (described m example 1) , using the Polymerase Cham Reaction (PCR) molecular biology technique. More specifically: the super-binder mutations on helix D and region AB of the phasmid D 3-7 (described m example 1), to create the mutant protein Sant 3; the super-binder mutations on helix D and region AB of the phasmid D 3-3 (described in example 1), to create the mutant protein Sant 4; the super-binder mutations on helix D and region AB of the phasmid D 4-20 (described in example 1), to create tne mutant protein Sant 5.

The mutant proteins, containing nine (Sant 3 and Sant 5) or ten (Sant 4) amino acid substitutions, were tested both for their specific mterleukm-6 receptor binding, and for their ability to antagonize tne biological activity of mterleukm-6 on human hepatoma and myeloma cells. Table 2 and fig. 2 show the specific receptor binding properties of DFRD and of Sant 3, Sant 4 and Sant 5 along with the concentrations (expressed m nanogra s of mutant per milliliter of culture medium) of mutant necessary to inhibit 50% of interleukin 6 biological activity (hepatoma cells were stimulated with 4 nanograms of wild type interleukin 6 per milliliter of culture medium, while myeloma cells were stimulated with 0.1 nanograms of interleukin 6 per milliliter of culture medium, due to the higher sensitivity of tne latter cells to w ld type interleukin 6) .

TABLE 2

Inhibition of wild type mterleukin 6 bioloσical activity on both hum .an heoatoma and mve :loma cells as a function of the mutant antagonists' specific mterleukm -6 recector omdinσ capacity

Receptor 50% inhibition of interleukin 6 activity on:

Antagonist binding hepatoma cells myeloma cells

(% of wild type) Hep3B XG-1 DFRD 97% 164 ng/ml 190.0 ng/ml

Sant 3 2800% 2.4 ng/ml 1.85 ng/ml

Sant 4 2000% 2.7 ng/ml 3.90 ng/ml

Sant 5 4500% 2.3 ng/ml 2.45 ng/ml

As can be seen from the table, the introduction of the ammo acid substitutions described in example 1 has at once increased the specific receptor binding capacity of the parental mutant DFRD and decreased the amount of antagonist needed to inhibit 50% of wild type interleukin 6 biological activity on both cell lines tested, therefore generating very effective and strong interleukin 6 superantagonists.

SEQUENCE LISTING GENERAL INFORMATION (l) APPLICANT: ISTITUTO DI RICERCHE DI 3I0L0GIA MOLECOLARE P. ANGELETTI S.p.A. (11) TITLE OF INVENTION: A METHODOLOGY FOR SELECTING

SUPERAGONISTS, ANTAGONISTS AND SUPERANTAGONISTS OF HUMAN INTERLEUKIN-6 BASED ON RECEPTOR COMPLEX THREE DIMENSIONAL MODELLING

(ill)NUMBER OF SEQUENCES: 3 (iv) CORRESPONDENCE ADDRESS:

(A) ADDRESSEE: Societa Italiana Brevetti

(B) STREET: Piazza di Pietra, 39

(C) CITY: Rome

(D) COUNTRY: Italy (E) POSTAL CODE: 1-00186

(v) COMPUTER READABLE FORM:

(A) MEDIUM TYPE: Floppy disk 3.5" 1.44 MBYTES

(B) COMPUTER: IBM PC compatible

(C) OPERATING SYSTEM: PC DOS/MS DOS Rev 6.22 (D) SOFTWARE: Microsoft Word 6.0

(vm) ATTORNEY INFORMATION

(A) NAME: DI CERBO, Mario (Dr.)

(B) REGISTRATION NUMBER:

(C) REFERENCE: RM/X88471/PC-DC (ix) TELECOMMUNICATION INFORMATION

(A) TELEPHONE: 06/6785941

(B) TELEFAX: 06/6794692

(C) TELEX: 612287 ROPAT (1) INFORMATION FOR SEQ ID NO: 1: (l) SEQUENCE CHARACTERISTICS

(A) LENGTH: 555 base pairs

(B) TYPE: nucleic acid (C) STRANDEDNESS: single

(D) TOPOLOGY: linear (11) MOLECULE TYPE: DNA

(ill) HYPOTHETICAL: no (iv) ANTISENSE: no

(v) FRAGMENT TYPE: internal (vii) IMMEDIATE SOURCE:

(A) SYNTHESIS: production in bacteria (ix) FEATURE: (A) NAME: IL-6 cDNA

(C) IDENTIFICATION METHOD: polyacrylamide gel

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 1: CCA GTA CCC CCA GGA GAA GAT TCC AAA GAT GTA GCC GCC CCA CAC AGA 48 Pro Val Pro Pro Gly Glu Asp Ser Lys Asp Val Ala Ala Pro His Arg 1 5 10 15

CAG CCS CTC ACG AGC TCA GAA CGA ATT GAC AAA CAA ATT CGG TAC ATC 96 Gin Pro Leu Thr Ser Ser Glu Arg He Asp Lys Gin He Arg Tyr He 20 25 30 CTC GAC GGC ATC TCA GCC TTA AGA AAG GAG ACA TGT AAC AAG AGT AAC 144 Leu Asp Gly He Ser Ala Leu Arg Lys Glu Thr Cys Asn Lys Ser Asn

35 40 45

ATG TGT GAA AGC AGC AAA GAG GCA CTG GCA GAA AAC AAC CTG AAC CTT 192 Met Cys Glu Ser Ser Lys Glu Ala Leu Ala Glu Asn Asn Leu Asn Leu . 50 55 60

CCA AAG ATG GCT GAA AAA GAT GGA TGC TTC CAA TCT GGA TTC AAT GAG 240 Pro Lys Met Ala Glu Lys Asp Gly Cys Phe Gin Ser Gly Phe Asn Glu 65 70 75 80

GAG ACT TGC CTG GTG AAA ATC ATC ACT GGT CTT TTG GAG TTT GAG GTA 288 Glu Thr Cys Leu Val Lys He He Thr Gly Leu Leu Glu Phe Glu Val

85 90 95

TAC CTA GAG TAC CTC CAG AAC AGA TTT GAG AGT AGT GAG GAA CAA GCC 336 Tyr Leu Glu Tyr Leu Gin Asn Arg Phe Glu Ser Ser Glu Glu Gin Ala 100 105 110 AGA GCT GTC CAG ATG AGT ACA AAA GTC CTG ATC CAG TTC CTG CAG AAA 384 Arg Ala Val Gin Met Ser Thr Lys Val Leu He Gin Phe Leu Gin Lys

115 120 125

AAG GCA AAG AAT CTA GAT GCA ATA ACC ACC CCT GAC CCA ACC ACA AAT 432 Lys Ala Lys Asn Leu Asp Ala He Thr Thr Pro Asp Pro Thr Thr Asn 130 135 140

GCC AGC CTG CTG ACG AAG CTG CAG GCA CAG AAC CAG TGG CTG CAG GAC 48 Ala Ser Leu Leu Thr Lys Leu Gin Ala Gin Asn Gin Trp Leu Gin Asp 145 150 155 160

ATG ACA ACT CAT CTC ATT CTG AGA TCT TTT AAG GAG TTC CTG CAG TCC 52 Met Thr Thr His Leu He Leu Arg Ser Phe Lys Glu Phe Leu Gin Ser

165 170 175

AGC CTG AGG GCT CTT CGG CAA ATG TAG 5

Ser Leu Arg Ala Leu Arg Gin Met 180 (2) INFORMATION FOR SEQ ID NO: 2

(i) SEQUENCE CHARACTERISTICS

(A) LENGTH: 95 base pairs '(B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear

(ii) MOLECULE TYPE: synthetic DNA (iii) HYPOTHETICAL: no (iv) ANTISENSE: no (v) FRAGMENT TYPE: internal (vii) IMMEDIATE SOURCE:

(A) SYNTHESIS: oligonucleotide synthesizer

(ix) FEATURE:

(A) NAME: DFQS (C) IDENTIFICATION METHOD: poiyacrylamide gel

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 2: GTGAGAGCTC CAAAGAGGCA CTGGCAGAAA ACAACCTGAA CCTTCCAAAG ATGGCTGAA AANNSGGATG CNNSNNSNNS GGATTCAATG AGGAG (3) INFORMATION FOR SEQ ID NO: 3

(i) SEQUENCE CHARACTERISTICS

(A) LENGTH: 72 base pairs

(B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear

(ii) MOLECULE TYPE: synthetic DNA (iii) HYPOTHETICAL: no

(iv) ANTISENSE: yes

(v) FRAGMENT TYPE: internal

(vii) IMMEDIATE SOURCE:

(A) SYNTHESIS: oligonucleotide synthesizer

(ix) FEATURE:

(A) NAME: AB primer

(C) IDENTIFICATION METHOD: poiyacrylamide gel (xi) SEQUENCE DESCRIPTION: SEQ ID NO: 3:

GGCCTCTAGA TATACCTCAA ACTCCAAAAG ACCAGTGATG ATTTTCACCA GGCAAGTCTC 60 CTCATTGAAT CC 72